Datasets:

TomTBT

/

pmc_open_access_section

like 1

PMC3016259

21162716

Introduction Bladder cancer is the fourth most commonly diagnosed malignancy in men and the ninth most commonly diagnosed malignancy in women, (NCI annual report 2009). Urinary bladder neoplasm can be grouped into three different categories: Superficial, invasive and metastatic. At presentation, 75% of the tumors are superficial, 20% are invasive and up to 5% have de novo metastasis. The wall of the bladder is lined with cells called transitional cells. More than 90% of urothelial cancers in the bladder are transitional cell carcinomas (TCC). Other important histologic types include squamous cell carcinoma and adenocarcinoma [ 1 ]. At presentation, tumors are usually limited to the bladder mucosa (Ta) or submucosa (T1). These tumors can be removed by transurethral resection (TUR), but tend to recur in 50-70% of the patients. Measures to decrease this high recurrence rate include intravesical chemotherapy and immunotherapy (BCG - Bacillus Calmet-Guerin). These treatments decrease the recurrence rate, but are associated with side effects and frequent failures [ 1 ]. The target population of this study is patients with superficial bladder cancer refractory to conventional therapies. Conventional therapies have focused on mass cell killing without specific targeting and often cause damaging and severe side effects to normal tissues. The development of targeted therapeutic strategies based on human cancer gene therapy is an attractive approach. Based on early studies of our group and others, the transcriptional regulatory sequences of the H19 and IGF2 genes emerged as candidates for cancer targeted therapy. H19 and IGF2 (the human P3 and P4 promoters) are onco-fetal genes and are oncogenes [ 2 - 4 ], expressed in the fetus and in a broad spectrum of tumors, but rarely in normal adult tissues [ 5 - 7 ]. H19 is a paternally-imprinted, oncofetal gene that encodes a RNA (with no protein product) acting as a "riboregulator" [ 8 ], which is expressed at substantial levels in embryonic tissues, in different human tumor types, and marginally or not expressed in the corresponding tissues of the adult [ 6 , 9 ]. The 67-aa IGF2 is a member of the insulin like growth factor family that is involved in cell proliferation and differentiation [ 10 ]. The human IGF2 gene contains 9 exons (E1-9) and 8 introns [ 10 , 11 ], and is transcribed from 4 different promoters (P1-P4) producing 4 different transcripts [ 11 - 13 ]. All four transcripts share a common coding region and a common 3.9 kb 3-UTR, but variable 5-UTRs [ 11 ]. IGF2 is an imprinted gene that is almost exclusively expressed from the paternal allele [ 14 - 16 ]. The P3 and P4 promoters are the major IGF2 promoters during embryogenesis and tumor development, while P1 is exclusively active in adult liver tissue and P2 activity is rarely detected in adult human tissue [ 10 ]. Increased expression of IGF2 as a result of the loss of its imprinting is frequently seen in a variety of human tumors [ 16 - 18 ]. In addition, abnormal signal transduction and/or promoter activation was reported as a major mechanism for the IGF2 overexpression in a variety of tumors including bladder carcinoma, hepatocellular carcinoma, breast cancer, ovarian cancer and prostate cancer [ 19 - 22 ]. The human H19 gene lies within 200 kb downstream of the paternally expressed IGF2 gene at 11p.15.5. These two genes are frequently coordinately regulated, both in terms of their common expression pattern and are reciprocal imprinting. Enhancers located downstream of H19 stimulate transcription of both genes [ 23 ]. We have shown that IGF2 or H19 are significantly expressed in 50-84% of human bladder carcinomas, respectively [ 7 , 24 ]. Our group has previously reported the construction of single promoter vectors expressing diphtheria toxin A-chain gene, under the control of IGF2-P4 or H19 regulatory sequences (IGF2-P4-DTA and H19-DTA). We showed that these constructs were able to selectively kill tumor cell lines and inhibit tumor growth in vitro and in vivo in accordance to the transcriptional activity of the above-mentioned regulatory sequences [ 7 , 25 ]. Moreover, our group used this therapeutic approach (using H19-DTA) in a successful treatment of a patient suffering from bladder cancer for a period of over 6 years [ 25 ], a phase I/IIa clinical trial using this therapeutic approach has been successfully completed [ 26 ] and the FDA has approved the initiation of following phase IIb clinical trial. However, there are TCC cells that do not express H19 and as a result, there are patients that could not match this treatment. Thus for the first time, in the present study, a double promoter DTA-expressing vector was created, carrying on a single construct two separate genes expressing the diphtheria toxin A-fragment (DTA), from two different regulatory sequences, H19 and IGF2-P4 ('H19-DTA-P4-DTA' vector). This novel approach, create a new family of plasmids regulated by two regulatory sequences, which in their natural genome position are both proximately located and are reciprocally imprinted. This is a new biology concept, which mimics the unique biology reciprocity relations phenomenon of IGF2 and H19. This vector was then used to transfect and to eradicate tumor cells in culture or to inhibit tumor growth ( in vivo ), in heterotopic and orthotopic bladder tumor models. The activity of the double promoter vector was tested and compared to the activity of the single promoter vectors. The results showed enhanced activity of the double promoter vector, H19-DTA-P4-DTA, relative to the single promoter expression vectors carrying either DTA sequence alone.

Materials and methods Cell culture The human bladder carcinoma cell line T24P was obtained from the American Type Culture Collection (ATCC; Rockville, MD). The human bladder carcinoma cell line HT-1376 was kindly provided by Prof W. Schulz, Heinrich-Heine University of Dusseldorf, Germany. Cells were grown to confluency in a humidified incubator with 5% CO2 in polystyrene culture flasks and were maintained in DMEM-F12 (1:1) medium containing 10% fetal calf serum. RNA Isolation, cDNA Synthesis and PCR RNA was extracted from cell lines or frozen tissue blocks, using the RNA STAT-60TM Total RNA/mRNA isolation reagent, according to the manufacture's instructions. The RNA was treated by RNAse-free DNAse I to eliminate any contaminating DNA. Total cDNA was synthesized from 2 μg total RNA in 20 μl reaction volume with 10 ng/μl of the oligo-(dT)15 primer and 10 units/μl M-MLV Reverse Transcriptase according to the manufacturer instructions. 2 μl of cDNA samples were taken for the amplification of the different transcripts using the different primers. The amplification conditions were 95°C for 2 min, followed by 30 cycles of 94°C for 30 sec, 59°C for 45 sec and 72°C for 60 sec, and finally 72°C for 5 min. The PCR reactions were carried out in 25 μl volumes in the presence of 6 ng/μl of each of the forward and the reverse primers using 0.05 units/μl of Taq polymerase according to the kit instructions (Takara). The forward (5'-CCGGCCTTCCTGAACA) and reverse (5'-TTCCGATGGTGTCTTTGATGT) primers designed for the detection of H19 RNA are spanning exons 2-3 and from exon 5 respectively, in order to validate that the PCR product is of the H19 RNA transcript and not from the endogenous H19 gene. The primers designed for the detection of IGF2-P4 RNA were designed to bind at exon 6 (5'-TCCTCCTCCTCCTGCCCCAGCG), for the P4 transcript in the forward direction and the reverse primer (5'- CAGCAATGCAGCACGAGGCGAAGCC) was designed to bind the 3' end of exon 7 and the 5' end of exon 8 without the introns in between. The integrity of the cDNA was assayed by PCR analysis of the ubiquitous, cell cycle independent, histone variant, H3.3 [ 7 ]. The PCR products were separated by electrophoresis on 2% gel agarose, and detected by ethidium bromide dye. Quantitative Real time PCR (qRT-PCR) Human TCC samples were obtained from patients undergoing TUR or radical cystectomy at Hadassah Hospital (Hadassah Hebrew University Medical Center, Jerusalem, Israel), following permission of the local IRB. Samples were analyzed using Mx3000p qRT-PCR detection system and its appropriate software Mx3000p qRT-PCR Software version 3.20 (Stratagene, La. Jolla, CA). Samples contained 10 μl of absolute blue qRT-PCR master mix (ABgene, Epsom, UK), 2 μl of samples, 500 nM of primers and 100 nM of TaqMan MGB probes (Applied Biosystems, Foster City, CA, USA) [ 27 ]. Amplification was done by an initial step of enzyme activation at 95°C, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. The amount of FAM fluorescence released from each tube was measured as a function of the PCR cycle number. To estimate the sensitivity of the real-time PCR procedure, three separate plasmid DNA controls were used with 10 fold serial dilutions of known quantities. For H19 analysis, starting from 0.2 ng (9 × 10 7 copies) up to 0.2 × 10 -7 ng (≤ 9 copies of plasmid DNA) were used. For IGF2-P4 analysis, starting from 0.2 ng (3 × 10 7 copies) up to 0.2 × 10 -7 ng (≤ 3 copies of plasmid DNA) were used. Simultaneous amplifications of standard dilution series were then performed. The number of target copies was determined using the standard curve created in the same run. The qRT-PCR assays were accepted when a positive signal was detected in all positive control dilutions and no signal was detected in the negative sample controls. The threshold for high expression level was set as >10,000 DNA copies number (per 0.2 μg c-DNA). These experiments were performed in triplicates. DIG-labeled Probe Synthesis A PCR strategy was used to generate template DNA for synthesis of labeled RNA probes. Forward primers for the human H19 and IGF2-P4 genes were designed. Each primer contain Sp6 promoter sequence in its 5'-end. Accordingly, a reverse primer was also designed with T7 promoter sequence incorporated in its 5'-end. The PCR products obtained for the H19 and IGF2-P4 transcript were purified from the gel by the DNA and Gel Band Purification Kit (Amersham), and used as templates for the PCR-based incorporation of T7 and Sp6 RNA polymerase promoter. The PCR conditions used to generate the T7/Sp6 templates were the same as described earlier for the synthesis of H19 and IGF2 specific transcripts. The PCR products (containing T7 and Sp6 promoters) were purified from the gel, sequenced and found to be identical to the relevant published sequences in the gene bank. 100 to 200 ng from the purified products were used as templates for the T7 and Sp6 polymerase (2 units/μl), according to the manufacturer instructions in the presence of 2 units/μl RNase inhibitor. T7 and Sp6 promoters were respectively used to drive the synthesis of the antisense and the control sense Digoxigenin-labeled UTP probes. The resulting probes were treated by 2 units of RNase free DNase I, pelleted and resuspended in appropriate volume of DEPC-treated double distilled water. The sizes of the synthesized probes were analyzed by running on 4% denaturing agarose minigel, and their labeling efficiency was determined by dot blot analysis. In situ hybridization (ISH) The non radioactive ISH washing and treatments were as described in [ 7 ]. Each section was rehydrated by 30 μl of the hybridization solution containing about 30 ng of DIG labeled RNA probe at 52°C. The ISH was performed on successive slides of TCC tissue for H19 and IGF2-P4 transcripts. The intensity of hybridization signal was indicated as (0) for no staining, (+1) for weak, (+2) for moderate and (+3) for strong signals. The distribution of the hybridization signal was referred to as up to one third of the cells, + (1), one to two thirds, ++ (2), and more than two thirds, +++ (3). Therefore the total scoring (intensity + quantity) for each sample varied from 0 (no expression) to 6 (very high expression). Low expression was set as total scoring of 0 < X < 3 and high expression was set as total scoring of 3 ≤ X ≤ 6. Plasmid construction The H19-Luc plasmid which contains the luciferase gene under the control of the human H19 promoter region from nucleotide -818 to + 14 was prepared as described [ 28 ]. The H19-Luc plasmid was digested with XbaI and NcoI, and the insert of the luciferase gene (luc) was replaced by the Diphtheria toxin A chain (DTA) coding region to yield the H19-DTA construct. The DTA gene was prepared from the pIBI30-DT-A plasmid (kindly donated by Dr. Ian Maxwell, University of Colorado, USA). The human IGF2-P4 promoter from the Hup4 vector (described in [ 11 ]) (a kind gift from Prof. P.E. Holthuizen, University of Utrecht, The Netherlands) were constructed by GENEART into the pGL3 basic vector (Luc-1) (Promega, Madison, MI), which lacks any eukaryotic promoter and enhancer sequences and carries the Kanamycine resistance gene (insert 812 bp), using BstEII and Hind III restriction sites, resulting in the expression vector P4-Luc. The DTA containing vector P4-DTA was designed by replacing the luciferase gene in P4-Luc with the DTA gene between the XbaI and NcoI restriction sites. Each of the cloned promoters and the DTA gene were sequenced and compared to the published sequences of the gene bank. We constructed double promoter expression plasmids, carrying on a single construct two separate genes expressing the diphtheria toxin, from two different regulatory sequences, as follows: H19 + IGF2-P4 promoters (hereinafter "H19-DTA-P4-DTA"; depicted in Figure 1 ). A double promoter control constructs was created, using the same strategy, expressing the luciferase reporter gene ('H19-Luc-P4-Luc'). The double promoter expression plasmids were cloned by GENEARTTM, (Germany) Transfection Cationic polymer (jetPEI) transient transfection The in vitro jetPEITMtransfection reagent compact the DNA into positively charged particles capable of interacting with anionic proteoglycans at the cell surface and entering cells by endocytosis. The transfection procedure was done as recommended by the manufacturer (Polyplus-transfection, France). A total of 0.1 × 10 6 cells/well were grown overnight in a twelve-well Nunc multidish (75 mm). For each well, 2 μg DNA and 4 μl of the jetPEI (N/P = 5) were diluted separately with 50 μl of 150 mM NaCl each, and vortex-mixed gently. The jetPEI solution was added at once to the DNA solution, the mixture was vortex-mixed for 10 seconds and the mixture was incubated for 15 minuets at room temperature. The 100 μl jetPEI/DNA mixture was then applied drop-wise onto the serum containing medium of each well. The transfection experiment was stopped after 48 hours. Luciferase activity The cells were harvested and the luciferase activity was determined using the luciferase Assay System kit (Promega). The light output was measured using a Lumac Biocounter apparatus. The total protein content of the lysates was determined by the Bio-Rad protein assay reagent and the results were normalized to the total protein and expressed as Light units/μg protein. LucSV40 (Luc-4) was used as a positive control for the efficiency of transfection as it contains the SV40 promoter and enhancer, while Luc-1 that lacks any regulatory sequences was used as a negative control to determine the basal nonspecific luciferase expression, which was found to be negligible in all of the cell lines. All experiments were done in triplicates and the results expressed as mean and standard error. In vitro targeted therapy The cells were cotransfected with 2 μg of the LucSV40 control vector and with the indicated amounts of the DTA expressing vector (H19-DTA, P4-DTA or the DTA double promoter expressing vector H19-DTA-P4-DTA). The same cells were additionally transfected with 2 μg LucSV40 alone in the same experiment. The H19-DTA, P4-DTA and H19-DTA-P4-DTA cytotoxic activity was determined by calculating the % of decrease in the cotransfected LucSV40 activity compared to that of LucSV40 transfected alone in the same cell type. The total protein content of the lysates was determined by the Bio-Rad protein assay reagent and the results were normalized to the total protein and expressed as Light units/μg protein. Therefore the reduction in luciferase activity, reflect the inhibition of protein synthesis activity by the DTA. In vivo targeted therapy animal models All surgical procedures and the care given to the animals were approved by the local committee for animal welfare. Animals were kept in the Hebrew University's animal facility with water and food ad librum (all experimental research on animals follow internationally recognized guidelines). The histopathological examinations of the different tumors were performed in consultation with a trained pathologist. Heterotopic nude mice model Confluent T24P and HT-1376 human bladder carcinoma cells were trypsinized to a single cell suspension and resuspended in PBS. 2 × 10 6 T24P cells or HT-1376 cells (in 150 μl volume) were subcutaneously injected into the back of female CD1 nude mice, 6-8 weeks old. 10 days after cell inoculation the developing tumors were measured in two dimensions and randomized to different treatments. Animals were separated to different groups of the same size (n = 6). The ability to inhibit tumor growth by the single promoter DTA expression vectors (P4-DTA, H19-DTA) and by the double promoter DTA expression vector (H19-DTA-P4-DTA) was tested. Intratumoral injections of 25 μg of either DTA expressing constructs (treatment groups) or Luc expressing constructs (control groups) were given 10, 12 and 14 days after cells inoculation. In vivo Jet-PEI a 22 kDa linear form of polyethylenimine (PEI) was used as a transfection enhancer reagent. PEI/DNA complexes with a ratio of PEI nitrogen to DNA phosphate of 6 were prepared in a solution of 5% w/v glucose according to the manufacture's instructions. Tumor dimensions were measured, and the tumor volume was calculated according to the formula width 2 × length × 0.5. The animals were sacrificed 3 days after the last treatment, the tumors were excised and their ex-vivo weight and volume were measured. Samples of the tumors were fixed in 4% buffered formaldehyde and processed for histological examination for evidence of necrosis and persistent tumor. Computerized measurements of tumor surface area and of the necrotic surface area were made using the Image Pro Plus software (Media cybernetics, Silver Springs, USA). Orthotopic bladder cancer model Female CD1 nude mice, 6-8 weeks old were used to develop orthotopic superficial bladder tumors. Mice were anesthetized with intra-peritoneal injection of ketamine (85 mg/kg) and xylazine (3 mg/kg). The bladder was catheterized with a 24 gauge catheter, than drained and its mucosa was mildly disrupted with 0.1 ml HCl 0.1N for 15-sec. (The bladder is rather resistant to implantation of cells, and therefore it is necessary to create abrasions in the bladder mucosa of the anesthetized rodent either by acid, in order to increase "tumor take" [ 29 ]). The acid was immediately neutralized with 0.1 ml NaOH 0.1N, and the bladder was washed three times with 0.1 ml PBS. Subsequently, a 0.1 ml suspension of PBS containing 10 × 10 6 T24P human bladder carcinoma cells was instilled into the bladder. The urethra was ligated with 6/0 silk suture to assure that cells were retained in the bladder. After 2 hours the sutures were removed and the bladders were evacuated by spontaneous voiding. Four healthy mice were left without T24P cells instillation. Seven days after cell instillation, the animals were anesthetized and the bladders were catheterized the same way. The bladders were washed three times with 0.1 ml of PBS. Animals were separated to different groups of the same size (n = 6). Mice of the DTA treatment groups received 20 μg of the toxin vector H19-DTA-P4-DTA. The control group received 20 μg of the reporter vector H19-Luc-P4-Luc. A group of 4 mice were kept with no treatment. The same treatments were repeated after 3 days. The in vivo -jetPEITM was used as a transfection enhancer agent. For preparation of the solution, 2.4 μl of the jetPEI (N/P ratio = 6) in 50 μl glucose 5% (w/v) were mixed with 20 μg of treatment plasmids respectively, in 50 μl of 5% glucose solution. The resulting mixture was vortex-mixed and left for 10-15 minutes at room temperature and subsequently instilled into the mice bladder transurethrally using the catheter as described above. The animals were sacrificed 4 days after the last plasmid instillation, their bladders were removed and the serosal surface and the adjacent sex glands were dissected carefully. Samples of the tumors were fixed in 4% buffered formaldehyde and processed for histological examination for evidence of necrosis and persistent tumor. Computerized measurements of tumor surface area and of the necrotic surface area were made using Image Pro Plus software (Media cybernetics, Silver Springs, USA). Other samples were frozen by liquid nitrogen and stored at -80°C to be analyzed by RT-PCR for evidence of IGF2, H19, luciferase and DTA mRNA expression.

Results Expression of IGF2-P4 and H19 transcripts in human bladder carcinoma tissues determined by ISH or by RT-PCR The human IGF2-P4 and H19 regulatory sequences are highly active in a variety of human cancers. In this study we present an approach for targeted therapy of bladder carcinoma by driving the DTA expression under the control of IGF2-P4 and H19 regulatory sequences. To evaluate the possible use of IGF2-P4 and H19 regulatory sequences for targeted therapy of bladder cancer, we determined the expression of IGF2-P4 and H19 transcripts by RT-PCR, qRT-PCR and ISH. Human TCC samples were obtained from patients undergoing TUR or radical cystectomy at Hadassah Hospital, following permission of the local IRB. The samples were first tested for H19 and IGF2-P4 overall expression by RT-PCR or by ISH (Table 1 ). 38 out of 39 TCC samples examined by RT-PCR showed positive IGF2-P4 transcripts expression and 37 out of 39 TCC samples showed positive H19 expression. Accordingly, 24 out of the 28 TCC samples examined by ISH showed positive IGF2 expression from IGF2-P4 (Figure 2A ), and 27 out of the 28 TCC samples showed positive H19 expression (Figure 2B ) (Table 1 ). Taken together the PCR and ISH results show that 62 out of 67 (92.5%) and 64 out of 67 (95.5%) positively expressed varying levels of IGF2-P4 and H19, respectively. Comparison of the expression levels of IGF2-P4 and H19 transcripts in human TCC samples detected by ISH and by qRT-PCR qRT-PCR and ISH techniques were applied to determine and quantity the level of H19 and IGF2-P4 in human TCC samples. Human TCC samples (n = 29) were examined by qRT-PCR and the expression level of H19 and IGF2-P4 specific transcripts was determined for each sample by the total number of DNA copies (per 0.2 μg c-DNA). Table 2 demonstrates that high levels of IGF2-P4 and H19 transcripts were found in 83% (24/29) and in 90% (26/29) of the tumor samples, respectively. However the total combined expression of both IGF2-P4 and H19 transcripts, were detected at high expression levels in 100% (29/29) of the tumor samples. Additional human TCC samples (n = 28) were examined by ISH and the expression levels of IGF2-P4 and H19 transcripts were determined by the intensity of the hybridization signal and by the quantity of the stained cells. Table 3 shows that out of 28 TCC samples, high expression levels of H19 and IGF2-P4 were found in 75% (21/28) and 50% (14/28) of the TCC samples, respectively. However when the overall combined expression analysis of the intensity and quantity of both transcripts H19 + IGF2-P4 was determined, then 100% (28/28) of the samples showed positive expression and 26 out of 28 TCC samples (96%) showed high expression levels. Expressing DTA from two different regulatory sequences, using a 'double promoter strategy' As described, high levels of H19 and IGF2-P4 transcripts were detected in TCC samples. Furthermore, enhanced expression was clearly exhibited for a combined expression of both transcripts (H19 + IGF2-P4). Therefore, we decided to further investigate the combination use of H19 and IGF2-P4 regulatory sequences for driving toxin gene expression. A double promoter expression vector was created, carrying on a single construct two separate genes expressing the diphtheria toxin A (DTA), from two different regulatory sequences, H19 and IGF2-P4 promoters ("H19-DTA-P4-DTA"; depicted in Figure 1 ). In vitro DTA expression by a single construct containing DTA genes separately expressed from H19 and IGF2-P4 regulatory sequences The activity of the double promoter construct H19-DTA-P4-DTA was first tested in vitro by determining its ability to lyse two different human bladder carcinoma cell lines, relative to the single promoter constructs. Anti-tumor therapeutic activity was determined by measuring the inhibition of luciferase activity following co-transfection with LucSV40. T24P and HT-1376 TCC cells were co-transfected with the indicated vectors (H19-DTA, P4-DTA, or H19-DTA-P4-DTA) in a dose-response manner at the indicated concentrations (Figure 3 ) and with 2 μg of LucSV40. Luciferase activity as an indicator of survival of the transfected cells was determined and compared to that of cells transfected with LucSV40 alone. H19-DTA or P4-DTA was able to drive the expression of the DTA gene and thus reduce luciferase activity in a dose-response manner. H19-DTA-P4-DTA, however, exhibited far enhanced efficiency in lysing the cancer cell lines, relative to each of the single promoter constructs, in T24P cells (Figure 3A-B ) and in HT-1376 cells (Figure 3C-D ). The double promoter expressing vector H19-DTA-P4-DTA was able to reduce the LucSV40 activity to more than 70% at concentrations as low as (0.005 μg/well) in T24P (Figure 3B ) and HT-1376 (Figure 3D ) cells, respectively. Less significant inhibition was obtained by H19-DTA or P4-DTA at the same concentrations (0.005 μg/well) in T24P (Figure 3B ) and HT-1376 (Figure 3D ) cells. In vivo tumor growth inhibition by the double promoter vector in bladder cancer animal models We used the double promoter construct, H19-DTA-P4-DTA assessing its tumor growth inhibition activity, by DTA expression in vivo using heterotopic and orthotopic animal models for bladder cancer. Expression of IGF2-P4 and H19 transcripts in heterotopic subcutaneous tumors In order to develop a model for heterotopic bladder tumors, T24P or HT-1376 human bladder cancer cells were subcutaneously injected into the dorsa of 6-7 weeks old CD-1 (nude) female mice. Tumors were developed 10 days after cell injection, dissected and total RNA was extracted from the tumors. The expression of IGF2-P4 and H19 RNA was determined by RT-PCR analysis. High expression of IGF2-P4 and H19 RNA was found in the heterotopic tumors induced by T24P cells (Figure 4A lanes 1-2) or by HT-1376 cells (Figure 4B lanes 1-2). Moreover there was no H19 and IGF2 expression in normal control mice (lane 3). Interestingly, the expression of H19 and IGF2-P4 RNA in the heterotopic tumors was higher compared to the in vitro expression of T24P cells (lane A4) or HT-1376 cells (lane B4) used for inoculation. Tumor growth inhibition by the double promoter vector in heterotopic bladder carcinoma model The tumor growth inhibition activity of H19-DTA-P4-DTA was tested in heterotopic bladder tumors, induced by T24P cells. T24P cells were subcutaneously injected into the back of 6-7 weeks old CD-1 female mice in order to develop a model for heterotopic bladder cancer. 10 days after subcutaneous cell inoculation, the mice developed measurable heterotopic tumors for testing. The therapeutic potency of the vectors was tested by direct intratumoral injection of 25 μg of the DTA expression vectors (H19-DTA, P4-DTA, or H19-DTA-P4-DTA), or of the control vectors (H19-Luc, P4-Luc, or H19-Luc-P4-Luc) into each heterotopic bladder tumor. Tumors sizes were determined and the in vivo fold increase of the tumor size was calculated prior to each treatment and before sacrifice. Three injections of H19-DTA or P4-DTA (Figure 5 ) at two-day intervals were able to inhibit tumor development by 49% (P = 0.001) and 55.5% (P = 0.005), respectively compared to H19-Luc and P4-Luc treatments. However, three injections of the double promoter plasmid H19-DTA-P4-DTA at two-day intervals inhibited tumor development by 70% (P < 0.001) compared to H19-Luc-P4-Luc treatment (Figure 5 ). The double promoter construct thus exhibited enhanced ability to inhibit tumor development in vivo , compared to each of the single-promoter constructs (H19-DTA, or P4-DTA). To confirm the difference between the H19-DTA-P4-DTA and H19-Luc-P4-Luc groups, tumors were excised and their ex vivo volume and weight were determined as well. Mice treated with H19-DTA-P4-DTA exhibited a 61% (P < 0.001) reduction of the ex-vivo tumor volume (Figure 6A ) and a 54% (P = 0.002) reduction of the ex-vivo tumor weight (Figure 6B ) compared to H19-Luc-P4-Luc treated mice. The consistency of the results, in measurements of the ex-vivo tumors as well, eliminates any unrelated difference of the measurements (such as subcutaneous inflammation swelling due the necrosis reaction, etc.). In vivo tumor growth inhibition of orthotopic bladder tumors by the double promoter vector Transurethral implantation of human bladder cancer cells into the mouse bladder (orthotopic model) provides a more relevant tool for the investigation of the biology and therapy of bladder cancer than subcutaneous implantation of bladder cancer cells (heterotopic model). Therefore, a mouse model was developed by intravesical instillation of T24P human bladder carcinoma cells onto the wall of the mouse bladder in vivo . This model was then used for testing the tumor growth inhibition activity of the double promoter H19-DTA-P4-DTA vector. Treatment of the orthotopic tumors Considerably large tumors were obtained 14 days after the T24P cells inoculation. As shown in Figure 7A high expression of both H19 and IGF2-P4 was determined by RT-PCR, in orthotopic bladder tumors, sacrificed 14 days after cells inoculation. By this time the tumors already started to invade the lamina propria as well as the superficial and deep muscle (Figure 7B ). These tumors would not therefore be suitable to start the treatment by the DTA therapeutic constructs because it does not resemble the stage at which most of the cases in human (more than 75%) consult the physician. Therefore, the treatment was started 7 days after cells inoculation, which was enough to develop smaller and less invasive orthotopic tumors than after 14 days. The treatment group (n = 6) was intravesically treated with 20 μg of H19-DTA-P4-DTA and the control group (n = 6) received 20 μg of H19-Luc-P4-Luc. Three days later the same treatments were repeated. Additional four control healthy mice were intravesically treated with HCL/NaOH at the beginning of the experiment with no additional following treatments. The animals were sacrificed at the end of the experiment (4 days after the second treatment), their bladders were processed for assessment of tumor sizes and for PCR and histology analyses (see Materials and Methods). As can be seen in Figure 8 , two treatments of 20 μg of H19-DTA-P4-DTA in three day intervals were able to inhibit tumor growth significantly as reflected by measuring the size of the tumors and by bladders weight. Tumor area of each bladder was macroscopically determined, using the ImagePro Plus software for measurement and analysis of the tumor area. The average size of the H19-DTA-P4-DTA treated tumors at the end of the experiment was 86% smaller (Figure 8A ) than that of the H19-Luc-P4-Luc treated ones (6.37 ± 2.1 mm 2 and 44.6 ±8.5 mm 2 respectively) (P < 0.001). As shown in Figure 9B , the group treated with the reporter vector showed usually more than one large lesion, with different grades of invasion. In contrast, only small tumors were detected in the H19-DTA-P4-DTA treated bladders (Figure 9E ). Inhibition of tumor growth was also reflected in bladders weight (Figure 8B ). The mean bladder weight of H19-DTA-P4-DTA treated mice was 40 ± 9 mg compared to 120 ± 20 mg in the control group. The mean bladder weight of the healthy mice was 30 ± 3 mg (P < 0.001). Expression of DTA and Luc RNA in mouse orthotopic treated bladder tumors At the end of the experiment, bladders were excised and total RNA was extracted from each tumor. RNA samples from the treated tumors were analyzed by RT-PCR for DTA and for luciferase mRNA expression. Figure 10A (lanes 1-2) shows high luciferase expression after treatment with the H19-Luc-P4-Luc reporter vector. The PCR revealed DTA mRNA expression in H19-DTA-P4-DTA treated tumors (lanes 3-4) but not in the luciferase treated tumors (lanes 1-2). This indicates that the tumors were efficiently transfected by the H19-DTA-P4-DTA vectors and that the human H19 and IGF2-P4 promoters were activated and DTA was produced. Necrosis in H19-DTA-P4-DTA treated bladder, as a result of the diphtheria toxin activity, is shown in figure 10B In vitro enhanced activity of the double promoter H19-DTA-P4-DTA construct compared to combination of the single promoter constructs The presence of an enhanced anti-cancer activity of the double promoter construct H19-DTA-P4-DTA was tested in the human bladder cancer cell lines T24P and HT-1376. T24P and HT-1376 cells were co-transfected with 2 μg of LucSV40 and either (a) the concentrations indicated (Figure 11 ) of single-promoter constructs H19-DTA + P4-DTA in combination, or (b) the same amount of H19-DTA-P4-DTA as for one of the single-promoter constructs. The total amount of DNA co-transfected in samples receiving both single promoter constructs was therefore twice than the cells transfected with H19-DTA-P4-DTA. Luciferase activity was determined and compared to that of cells transfected with LucSV40 alone. The double-promoter construct H19-DTA-P4-DTA exhibited enhanced efficiency in lysing the cancer cell lines, relative to the combined activity of both single promoter constructs (H19-DTA + P4-DTA), in T24P cells (Figure 11A-B ). Very similar results were obtained in HT-1376 cells (Figure 11C-D ). In vivo additive activity of the double promoter construct compared to combination of the single promoter constructs The presence of an additive tumor growth inhibition activity of the double promoter construct H19-DTA-P4-DTA was tested in vivo in a nude mice heterotopic bladder cancer model (described hereinabove). The therapeutic potency of the vector was tested by 3 intratumoral injections, at two-day intervals, of 25 μg of H19-DTA-P4-DTA or of the control vector (H19-Luc-P4-Luc), into each heterotopic bladder tumor. Tumor size was determined and in vivo fold increase of the tumor size was calculated at the end of each treatment. To test whether the in vivo tumor growth inhibition activity of H19-DTA-P4-DTA was augmented-than-additive, an additional group of T24P tumor-containing mice was treated with three injections of 25 μg each of single-promoter constructs H19-DTA + P4-DTA in combination. The total amount of DNA co-transfected administered was therefore twice (50 μg) than the H19-DTA-P4-DTA group. As can be seen in Figure 12 , tumor development in mice receiving both H19-DTA and P4-DTA plasmids was inhibited by 63.4% (P = 0.001) compared to combined H19-Luc + P4-Luc treated mice. However, an enhanced effect was observed in mice treated with the double-promoter construct H19-DTA-P4-DTA, wherein tumor development was inhibited by nearly 70% (P = 0.005) compared to mice treated with the control plasmid H19-Luc-P4-Luc. Figure 12 summarizes all T24P heterotopic bladder cancer model results. H19-DTA-P4-DTA clearly exhibits enhanced activity compared to each of the single promoter plasmids alone and also superior to their combined activity. As can be seen in Figure 12 , mice intratumorally treated with higher dose as 50 μg of the double-promoter construct H19-DTA-P4-DTA (same total amount of the combined single promoter plasmids), showed enhanced inhibition of more than 80%. Thus, the H19-DTA-P4-DTA vector exhibits augmented-than-additive in vivo tumor growth inhibition activity, compared to the combined activity of both single-promoter constructs (H19-DTA and P4-DTA).

Discussion The present work shows the successful use of a double promoter expressing vector, carrying on a single construct two separate DNA sequences expressing the diphtheria toxin A-fragment (DTA), from two different regulatory sequences, selected from the cancer-specific promoters H19 and IGF2-P4. This construct was used to transfect and to eradicate tumor cells in culture ( in vitro ) or tumors developed in animal models ( in vivo ) of bladder carcinoma. Cancer is a multigene and multi-factorial disease. The last decade has seen the emergence of numerous multigene expression profiles that aim to outdo traditional predictive and prognostic factors (reviewed by [ 30 ]). However, targeted therapies such as Herceptin and Avastin, are targeting one specific protein. Further personalized and targeted therapies should be considered, targeting more than one target (protein or a gene). Accordingly, several chemotherapies nowadays are administrated as cocktails, in combination with radiotherapy (reviewed by [ 31 ]) and in combination with targeted agents (reviewed by [ 32 ]). Therefore, we applied an innovative approach using on a single construct more than one specific marker gene which are differentially expressed in tumor cells, for targeted cancer therapy. IGF2 and H19 are reciprocally imprinted and are highly expressed in a broad spectrum of tumors, but rarely in normal adult tissues [ 7 , 33 ]. Using a single promoter (e.g. an H19 promoter or an IGF2-P4 promoter) alone for expression of a cytotoxic gene presents several unresolved problems. For one, not every tumor cell of a given type of cancer is positive for expression via the H19 promoter or the IGF2-P4 promoter sequences. Thus, such therapy could fail in a sizable proportion of patients, even without accounting for tumor mutagenesis. Determination of responsiveness to such constructs would involve the costly and difficult step of genotyping individual tumors. Tumors are known to exhibit significant genomic instability and heterogeneity. Thus, even individuals with an H19-expressing tumor, for example, may contain some cancer cells that have downregulated or abrogated H19 expression via mutation. Therefore, expressing the cytotoxic gene from a single promoter in such patients may result in temporary and partial tumor regression that will rapidly be reversed when the cells containing these mutations survive and rapidly multiply. Therefore the use of double promoter expressing vectors is highly novel. Tumor cells can express high levels of H19 and IGF2, or only one of those genes. That way, majority of the tumor cells could efficiently express the diphtheria toxin. This novel approach, create a new family of plasmids regulated by two regulatory sequences, which in their natural genome position are both proximately located and are reciprocally imprinted. This is a novel biology concept, which mimics the unique biology reciprocity relations phenomenon of IGF2 and H19. Once introduced into target tissue, the plasmid vectors have several advantages over viral vectors (reviewed by [ 34 , 35 ]): (1) the plasmids have no potential to be infectious; (2) they possess levels of expression per cell that are equivalent to some viral vectors that persist as extra-chromosomal elements; (3) the lack of immunogenicity, thus allowing for repeated treatments; (4) plasmids transfect mainly dividing cells, with contrast to most viral vectors that, except for retroviral vectors, which transfect both dividing and non dividing cells; and finally (5) the long-term stability, safety and the lack of need special treatments or storage requirement of the plasmid vectors. In this study, the therapeutic potential of the vectors was tested in TCC of the bladder. The bladder has long been thought to be an ideal target for DNA based therapy because it is easily accessible by catheter and is largely a self-contained "bag-like" organ. While the protective glycosaminoglycan (GAG) that is present in the normal bladder mucosa interferes with the plasmid transfection [ 36 ], it is not present in the bladder tumor, allowing efficient transfection of principally the tumor urothelium. In the same way an orthotopic model can be designed, which the bladder can then be easily approached by catheter. In cancer gene therapy, direct DNA injection is currently a reliable, reproducible, and simple technique for intratumoral gene transfer [ 37 ]. We transferred the plasmids into cell lines and into the target tissue of the animal models, as complex with the linear cationic polyethylenimine (jetPEI) as a transfection reagent. This method was chosen based on previous studies of our group showing relatively high levels of transfection efficiency, in vitro , in vivo and lately in TCC patients as part of a phase I/IIa bladder cancer clinical trial [ 25 , 27 ]. JetPEI condenses the DNA into positively charged particles capable of interacting with anionic proteoglycans at the cell surface and entering by endocytosis [ 38 ]. Subunit A of the diphtheria toxin (DTA), a highly potent poison, was chosen as an effector molecule. When only the cDNA coding for the A-fragment is expressed, the released DT-A toxin from the lysed cells will not be able to enter neighboring cells in the absence of the DT-B fragment [ 39 ]. This approach not only will insure high killing activity but will be of great advantage against any unintended toxicity to non-target normal cells. Moreover, introduction of DTA DNA sequence under the control of regulatory sequences of genes differentially expressed in tumors but not in adjacent non-tumor cells will selectively favor the specificity of the treatment. Over plurality of cancer specific promoters, H19 and IGF2-P4 regulatory sequences were selected for targeting cancer cells. The H19 and IGF2-P4 regulatory sequences are expected to be good candidates for specifically inducing the expression of DTA in target tumor cells but not in cells of normal tissue. They are known to be differentially over-activated in various tumor types and to show no or minimum activity in the surrounding normal tissue [ 40 , 41 ]. This is in addition to the known autocrine/paracrine mode of IGF2 mitogen action in the development of a wide range of human malignancies. Accordingly, destruction of the H19 and IGF2 expressing tumor cells not only will eliminate part of the tumor but will also diminish the supply of mitogenic IGF2 to neighboring tumor and non-tumor cells and may lead to arrest of tumor growth and prevent following metastases process [ 42 , 43 ]. Based on previous results of our group demonstrating efficient treatment of TCC using either H19-DTA or IGF2-P4-DTA vector [ 25 ], it appeared that TCC tumors could be efficiently treated by each of these vectors. Based on this assumption we hypothesized that by using double promoter expression vector, which the expression of DTA is controlled by more than one regulatory sequence, a higher therapeutic potential is expected, if the tumor shows high specific expression from more than one of the above mentioned regulatory sequences (H19 or IGF2-P4). In order to determine the applicability of this assumption, the first stage was to explore the expression level of each of the mentioned regulatory sequences and then compare it to the combined expression level (from the two regulatory sequences). First, the overall expression of H19 and IGF2-P4 was analyzed by ISH and RT-PCR in 67 human TCC samples. Taken together the PCR and ISH analyses results show (Table 1 ) that 62 out of 67 (92.5%) and 64 out of 67 (95.5%) positively expressed varying levels of IGF2-P4 and of H19, respectively. Next, the quantitative expression was further analyzed by ISH and by qRT-PCR. Out of 29 TCC samples detected by qRT-PCR, (Table 2 ), high levels of IGF2-P4 and H19 transcripts were found in 83% (24/29) and in 90% (26/29) of the tumor samples, respectively. Moreover, the total combined expression of both IGF2-P4 and H19 transcripts was detected at high expression levels in 100% (29/29) of the tumor samples. Out of 28 TCC samples detected by ISH (Table 3 ), high levels of IGF2-P4 and H19 transcripts were found in 50% (14/28) and 75% (21/28) of the TCC samples respectively. When the overall combined expression analysis of the intensity and quantity of both transcripts H19 + IGF2-P4 was determined, then 100% (28/28) of the samples showed positive expression and 26 out of 28 TCC samples (96%) showed high expression. Thus, both ISH and qRT-PCR detections confirmed that by analyzing the combined expression from two promoters, 100% of the samples show positive expression and nearly 100% show high expression. These results clearly support the rationale of our hypothesis, which DTA could be extensively expressed from more than one specific regulatory sequence. Therefore, we further investigated the combination use of H19 and IGF2 regulatory sequences for driving toxin gene expression in therapeutic vectors for bladder cancer treatment. The double promoter construct H19-DTA-P4-DTA exhibited far superior efficiency in vitro (Figure 3 ), in lysing human bladder carcinoma cell lines, relative to each of the single promoter constructs carrying either DTA DNA sequence alone (H19-DTA or P4-DTA). Therefore we further evaluated the therapeutic potential of the double promoter toxin vector in heterotopic and orthotopic mouse models. 1. Heterotopic bladder cancer model was used to evaluate tumor growth inhibition of the double promoter vectors compared to that of the single promoter vectors. The advantages of this model are its rapidity, reproducibility, accessibility and visibility of tumors. When using immuno-deficient animal like the nude type mice, human cell lines can be employed and better simulation of human tumor is obtained. H19-DTA-P4-DTA exhibited superior ability to inhibit heterotopic tumor development by 70% (P < 0.001) compared to H19-DTA or P4-DTA activity (Figure 5 ). Additional Ex-vivo measurements of tumors weight and volume, re-confirmed the difference between the H19-DTA-P4-DTA and control groups. The consistency of the results, by measuring of the ex-vivo tumors as well (Figure 6 ), eliminates any unrelated difference of the measurements (such as subcutaneous inflammation swelling due to necrosis reaction, etc.). 2. The disadvantage of the heterotopic model is the weak correlation in histology and clinical course between this model and the clinical disease. Therefore by inducing orthotopic TCC tumors in mice bladders, tumors resemble human bladder tumors by their histology, by the clinical course of TCC (local tumor growth, invasion, and metastatic activity), and by the ability to treat bladder tumors intravesically, the same way human bladders are clinically treated. Therefore we evaluated the feasibility of intravesical therapy of H19-DTA-P4-DTA, in nude mice orthotopic bladder cancer model. The average size of the H19-DTA-P4-DTA treated tumors was 86% smaller than that of the H19-Luc-P4-Luc treated ones (P < 0.001) (Figure 8A ) and there was also significant difference in mean bladders weight (P < 0.001) (Figure 8B ). Only small tumors were detected in the H19-DTA-P4-DTA treated bladders (Figure 9 ), compared to large lesions and with different grades of invasion in the group treated with the reporter vector. However the tumors were not completely destructed and it should be stressed that in patients with bladder cancer, the tumors are first surgically completely resected and the purpose of the following intravesical treatment is therefore to treat any possible remaining tumor cells and to prevent tumor recurrence. The inhibition of tumor progression resulted exclusively from the toxic effect of the diphtheria toxin. This was confirmed by RT-PCR determining mRNA expression of DTA only in heterotopic tumors treated with DTA expressing vector (Figure 10A ), and by performance of cellular necrosis in H19-DTA-P4-DTA treated tumors compared to the H19-Luc-P4-Luc treated and non-treated ones (Figure 10B ). In addition, all of the tested orthotopic tumor samples showed high expression of H19 and IGF2-P4 transcripts (Figure 7 ), thus it strongly prove the assumption that the orthotopic tumor cells activate the H19 and IGF2-P4 promoters and therefore drive the expression of DTA within the cells and in consequence triggering their necrosis. Finally, we dealt with the question whether transfection of both single promoter vectors (expressing the diphtheria toxin) in combination, may exhibit better efficacy than transfection of the double promoter construct. The use of double promoter vectors was previously described [ 44 ] as a convenient tool for evaluation of the activity of a gene of interest by monitoring a reporter gene activity simultaneously expressed on the same construct. However an additive activity of the double promoter vector versus combination of two single promoter vectors was never demonstrated. Therefore the presence of an additive anti-cancer effect of the double promoter constructs H19-DTA-P4-DTA was tested in vitro , in human TCC cells and in vivo , in heterotopic bladder cancer mice. In vitro enhanced activity of the double promoter vector H19-DTA-P4-DTA (Figure 11 ) was exhibited in T24P bladder cancer cells. A superior activity of the double promoter vector in lysing the cancer cell lines was exhibited, relative to the combined activity of both single promoter constructs (H19-DTA + IGF2-P4-DTA), in a dose response manner. It should be stressed that the total amount of DNA co-transfected in cells receiving both single promoter constructs was therefore twice than the cells transfected with the double promoter constructs. Thus, H19-driven and IGF2-P4-driven DTA-encoding sequences presented on a single expression vector (H19-DTA-P4-DTA), exhibited enhanced protein synthesis inhibition activity, relative to expression vectors carrying either DTA sequence alone when tested against bladder cancer cells. Augmented-than-additive activity of the double promoter vectors H19-DTA-P4-DTA (Figure 12 ) was further exhibited in vivo , in heterotopic tumors induced by T24P bladder cancer cell lines. Heterotopic tumors treated with combination of total amount of 50 μg of both single promoter H19-DTA and -P4-DTA constructs, were inhibited by 63% (P = 0.001) compared to combined H19-Luc + P4-Luc, control treated mice (Figure 12 ). However, an enhanced effect was observed in mice treated with only 25 μg of the double-promoter construct H19-DTA-P4-DTA, wherein tumor development was inhibited by 70% (P = 0.005) compared to the mice treated with the control plasmid H19-Luc-P4-Luc. Tumors treated with higher dose as 50 μg of the double-promoter construct H19-DTA-P4-DTA (same total amount of the combined single promoter plasmids), showed enhanced inhibition of at least 80% (Figure 12 ). Thus, the H19-DTA-P4-DTA vector exhibited augmented-than-additive in vivo anti-cancer activity, compared to the combined activity of both single-promoter constructs (H19-DTA and P4-DTA.

Conclusions In this study double promoter expression vector were used, expressing DTA from two different regulatory sequences, H19 and IGF2-P4. Several reasons support this strategy. First, IGF2-P4 and H19 are reciprocally imprinted and are exclusively expressed at high levels in cancer cells and not in normal cells. We demonstrated that combined expression from the two separate regulatory sequences, showed complementary expression profile, in which nearly 100% of tumor samples expressed high levels from at least one of the regulatory sequences. By that the DTA could be better expressed in larger number of cancer cells and therefore enhance the tumor inhibition activity. Second, H19 and IGF2 play major role in tumor development. By selective killing of cancer cells, which express H19 and IGF2, the treated tumor cells as well as the neighboring tumor cells (as IGF2 mediate its effect in autocrine/paracrine manner) are at least partly deprived of their IGF2 supply. By that the targeted destruction of cancer cells expressing IGF2 or H19, companied by enhanced bystander effect, may lead to arrest of tumor growth and prevent following metastases process. Overall, the double promoter vector, H19-DTA-P4-DTA, exhibited augmented-than-additive anti-cancer activity relative to single promoter expression vectors carrying either DTA sequence alone, when tested against bladder tumor cells. As H19 and IGF2-P4 are expressed at very high levels in a broad spectrum of different cancers, therefore we propose a double promoter expression approach for targeted cancer therapy. According to this approach patients will be treated with specific double promoter expression toxin vector which are under the control of the IGF2-P4 and H19 regulatory sequences, differentially expressed in those cancers. Moreover, our proposed treatment may be applied in combination with other cancer therapy methods, such as chemotherapy and radiology. This approach should be tested in appropriate animal models.

Background The human IGF2-P4 and H19 promoters are highly active in a variety of human cancers (including bladder cancer), while existing at a nearly undetectable level in the surrounding normal tissue. Single promoter vectors expressing diphtheria toxin A-fragment (DTA) under the control regulation of IGF2-P4 or H19 regulatory sequences (IGF2-P4-DTA and H19-DTA) were previously successfully used in cell lines, animal models and recently in human patients with superficial cell carcinoma of the bladder (treated with H19-DTA). However this targeted medicine approach could be limited, as not all cancer patients express high levels of H19. Hence, a double promoter DTA-expressing vector was created, carrying on a single construct two separate genes expressing the diphtheria toxin A-fragment (DTA), from two different regulatory sequences, selected from the cancer-specific promoters H19 and IGF2-P4. Methods H19 and IGF2-P4 gene expression was tested in samples of Transitional Cell Carcinoma (TCC) of the bladder by in-situ hybridization (ISH) and by quantitative Real-Time PCR (qRT-PCR). The therapeutic potential of the double promoter toxin vector H19-DTA-IGF2-P4-DTA was tested in TCC cell lines and in heterotopic and orthotopic animal models of bladder cancer. Results Nearly 100% of TCC patients highly expressed IGF2-P4 and H19, as determined by ISH and by qRT-PCR. The double promoter vector exhibited superior tumor growth inhibition activity compared to the single promoter expression vectors, in cell lines and in heterotopic and orthotopic bladder tumors. Conclusions Our findings show that bladder tumors may be successfully treated by intravesical instillation of the double promoter vector H19-DTA-P4-DTA. Overall, the double promoter vector exhibited enhanced anti-cancer activity relative to single promoter expression vectors carrying either gene alone.

List of abbreviations ATCC : American type culture collection; BCG : Bacillus Calmet-Guerin; DTA/DT-A : Diphtheria toxin A chain; H19-Luc-P4-Luc : Reporter vector expressing each luciferase under the control of a different promoter: H19 or IGF2-P4; H19-DTA-P4- DTA : Therapeutic (double promoter) vector expressing each DTA under the control of a different promoter: H19 or IGF2-P4; IGF2 - Insulin like growth factor 2; ISH : In situ hybridization; Luc : Luciferase; P4 : Human IGF2 P4 promoter; H19-Luc : Reporter vector expressing the luciferase under the control of human H19 promoter; P4-Luc : Reporter vector expressing the luciferase under the control of human IGF2 P4 promoter; H19-DTA : Therapeutic (single promoter) vector expressing the DTA under the control of H19 promoter; P4-DTA : Therapeutic (single promoter) vector expressing the DTA under the control of IGF2 P4 promoter; PCR : Polymerase chain reaction; PEI : Polyethylenimine; TCC : Transitional cell carcinoma; qRT-PCR : quantitative real-time polymerase chain reaction Competing interests The authors declare that they have no competing interests. Authors' contributions DA - conducted the study and conceived of the study, participated in design, coordination, data interpretation, performed the statistical analysis, and drafted the manuscript. AH - conceived of the study, participated in design, interpretation of data and critically revised the manuscript. All authors read and approved the final manuscript.

Acknowledgements We thank Professor Ofer Gofrit from the Department of Urology, Hadassah Hebrew University Medical Center, Jerusalem, Israel for providing TCC samples from patients.

CC BY

no

2022-01-12 15:21:44

J Transl Med. 2010 Dec 16; 8:134

oa_package/01/13/PMC3016259.tar.gz

PMC3016260

21167064

Background Non-small-cell lung cancer (NSCLC) is one of the most frequent human malignancies, constituting about 80% of all lung tumors. NSCLC can be divided into genetic subsets on the basis of the activating mutations that they harbor; each of these subsets may correspond to patient cohorts that are likely to benefit from treatment with specific inhibitors[ 1 ]. Activating mutations in the epidermal growth factor receptor (EGFR), affecting hotspots within exons that code for the tyrosine kinase domain, can be found in 10-40% of NSCLC patients, mostly in adenocarcinomas, with the higher frequency observed in Asian patients[ 1 , 2 ]. About 50% of mutated patients harbor in-frame deletions in exon 19, (around codons 746 to 750) and 35-45% show the substitution of leucine 858 by an arginine in the exon 21. The remaining mutants are insertions in exon 20 (5%) and uncommon substitutions spanning exons from 18 to 21, such as L861Q[ 3 , 4 ]. These specific mutations are related to a higher sensitivity to the tyrosine kinase inhibitors (TKIs) erlotinib and gefitinib[ 4 - 7 ], whereas the EGFR T790 M mutation in exon 20 is observed in 50% of cases with acquired resistance to erlotinib and gefitinib[ 8 ] and has also been detected in 38% of patients with de novo drug resistance[ 9 ]. Molecular biology techniques, such as SARMS or direct automatic sequencing, are currently used to detect EGFR mutations in formalin-fixed, paraffin-embedded tissues (FFPET). In our experience, in-frame deletions in exon 19 are detected by fragment analysis of fluorescently labeled PCR products, and L858R mutations in exon 21 by TaqMan assay. Mutations are then confirmed by direct sequencing[ 10 , 11 ]. However, the routine use of these methods in clinical laboratories is still often limited by financial and technical constraints. Moreover, their sensitivity depends on the quality and the quantity of tumoral cells in FFPET. In a previous study, we developed a highly sensitive molecular method for detecting EGFR mutations in NSCLC samples containing as few as eight tumor cells[ 10 ]. The development of antibodies that specifically detect mutant EGFR protein by IHC would be an easy pre-screening test to complement the molecular assays currently used for the assessment of EGFR mutations in NSCLC. Yu et al[ 12 ] have developed mutation-specific rabbit monoclonal antibodies (mAb) against EGFR with the E746_A750 deletion in exon 19 or the L858R point mutation in exon 21 for IHC application (Cell Signaling Technology Inc., Danvers, MA, USA). In the present study, these two rabbit mAbs were used to assess EGFR mutations in five NSCLC cell lines and in tumor biopsies from 78 stage IV NSCLC patients. The results were then compared with those obtained by other molecular analyses[ 10 , 11 ].

Methods Sources of cell lines and culture The PC-9 lung tumor cell line was kindly provided by Roche (Basel, Switzerland); the A549 and H460 cell lines were purchased from the American Type Culture Collection. Tissue culture materials were obtained from Biological Industries (Kibbutz Beit Haemek, Israel) and Invitrogen (Paisley, Scotland, UK). H1650 and H1975 were kindly provided by Dr. Herbert Haack and Dr. Katherine Crosby (Cell Signaling Technology, Inc.). We received five slides of the H1975 cell line and five of the H1650 cell line with 4-μm sections for IHC analysis from the Cell Signaling Technology laboratory. Study population and tumor pathology Twenty-six stage IV NSCLC patients had been seen at the USP Dexeus University Institute, and 52 had been previously screened for EGFR mutations and treated with erlotinib as part of the Spanish Lung Adenocarcinoma Data Base (SLADB)[ 11 ]. All of these 52 patients were known to have EGFR mutations, while the remaining 26 patients had not been previously screened. All patients provided written informed consent. Approval was obtained from the institutional review board and the ethics committee at each hospital. Table 1 shows patient characteristics. Four-μm sections of the FFPET specimens were stained with H/E and histologically examined. All samples were classified according to the 2004 WHO classification[ 13 ]: 5 undifferentiated large cell carcinomas and 3 small cell neuroendocrine carcinomas, 1 squamous cell carcinoma and 69 adenocarcinomas, of which 55 showed a single pattern and 14 presented mixed aspects. We further evaluated the adenocarcinoma subtype as follow: 36 adenocarcinomas with a glandular pattern, 20 with a solid aspect, 6 with a partial papillary differentiation, 1 with micropapillary aspects and 6 with a partial bronchioloalveolar pattern (Table 1 ). DNA extraction and mutation analyses Tumor cells (8 to 150) were captured by laser microdissection (Carl Zeiss MicroImaging GmbH, München, Germany) into 10 μL of PCR buffer (Ecogen, Barcelona, Spain) plus proteinase K and incubated 4 hours to overnight at 60°C. Proteinase was inactivated at 95°C for 10 min, and the cell extract submitted to PCR. DNA from the cell line PC-9 was used as a mutated control for exon 19, and wt control for exons 20 and 21. DNA from the H1975 cell line was used as a wt control for exon 19, and mutated control for exons 21/20. EGFR gene mutations in exons 19 and 21 were analyzed by our sensitive methodology as previously described[ 10 ]. Exons 19 and 21 of the EGFR gene were amplified by a nested PCR. Sequencing was performed using forward and reverse nested primers with the ABI Prism 3100 DNA Analyzer (Applied Biosystems, Foster City, CA, USA). In addition to sequencing, EGFR deletions in exon 19 were determined by length analysis of fluorescently labeled PCR products. The collected data were evaluated with the GeneScan Analysis Software (Applera, Norwalk, CT, USA). Finally, EGFR mutation (L858R) in exon 21 was also determined by TaqMan ® Assay (Applied Biosystems). The L861Q mutation was detected by direct sequencing. Immunohistochemical analysis The following antibodies were used for the IHC analysis (Cell Signaling Technology, Inc.): EGF Receptor (D38B1), EGFR E746-A750 deletion specific (6B6) and EGFR L858R mutant-specific (43B2). The FFPET samples were cut serially at 4 μm and the sections were introduced in the stainer and automatically deparaffinized (Leica Microsystems BondMAX Automated Immunostainer, Wetzlar, Germany). The reactives were added automatically, treating the samples with EDTA buffer (pH 9.0) (Bond Epitope Retrieval Solution 2, Leica Microsystems) as antigen retriever and processed for 30 min. The slides were incubated with the antibodies against EGF receptor and EGFR mutations at a dilution of 1:100 for 60 minutes. After the sections were treated with the streptavidin-biotin-peroxidase complex method (Bond Polymer Refine Detection, Leica Microsystems) with diaminobenzidine (DAB) as a chromogen and counterstained with hematoxylin. IHC expression of mAbs against EGFR was evaluated using the following scoring, as previously described[ 14 ]: 0 = negative or faint staining in <10% of tumor cells; 1 = weak staining in >10% of cancer cells; 2 = moderate staining; 3 = strong staining. A score of 0 was considered negative, a score of 1 was considered weakly positive, and a score of 2 or 3 was considered highly positive (Additional File 1 , Figure S1). Statistical analyses The absolute and relative frequencies of qualitative variables were calculated in percentages. The sensitivity and specificity of the EGFR test by IHC was determined in comparison with PCR-based results. All analyses were performed using SPSS v 16.0 software (SPSS Inc., Chicago, IL).

Results EGFR mutation analysis in NSCLC patients We screened EGFR mutations in 78 FFPET samples from NSCLC patients by a methodology described elsewhere[ 10 ], which involves fragment analysis (exon 19), Taqman assay (exon 21) and sequencing. Twenty-six samples were analyzed in the Pangaea Biotech Oncology Laboratory and 52 from a previous study[ 11 ] were analyzed in the Catalan Institute of Oncology, Hospital Germans Trias i Pujol. Twenty-two samples (28%) were wt EGFR, 29 (37%) had a deletion in exon 19, and 27 (35%) had mutations in exon 21. Of the 29 patients with the exon 19 deletion, 17 (59%) had 15-bp deletions (16 with del E746-A750 [ELREA] and 1 with del E746-A750 [ELREA] + T751I), and 12 (41%) had rare deletions of 9-bp, 12-bp, 18-bp, 21-bp or 24-bp. Of the 27 patients with exon 21 mutations, 25 (93%) had the L858R mutation and 2 (7%) had the L861Q mutation (Additional File 1 , Table S1). IHC analysis of mutation-specific mAbs against EGFR in human NSCLC cell lines We analyzed by IHC five human NSCLC cell lines with known EGFR gene status. In the two cell lines with wt EGFR (H460 and A549), we found positive (score 3) expression of EGFR (D38B1) protein (100%) and negative (score 0) expression of EGFR E746-A750 deletion specific (6B6) and EGFR L858R mutant-specific (43B2). In the two cell lines with exon 19 deletion (15 bp) (H1650 and PC9), expression of EGFR (D38B1) protein and EGFR E746-A750 deletion specific (6B6) was positive (score 3) (100%) and expression of EGFR L858R mutant-specific (43B2) was negative (score 0). The cell line H1975 with exon 21 mutation (L858R) showed positivity (score 3) for EGFR (D38B1) and EGFR L858R mutant-specific (43B2) (100%) and negativity (score 0) for EGFR E746-A750 deletion specific (6B6). (Table 2 , Figure 1 ). IHC analysis of mutation-specific mAbs against EGFR in NSCLC patients In 22 tumor tissues with wt EGFR, we found high expression of EGFR (D38B1) protein (score 2 or 3) in 8 cases (36%) and weak positivity (score 1) in 4 cases (18%). All the cases were negative for EGFR E746-A750 deletion specific (6B6) and EGFR L858R mutant-specific (43B2) proteins. In the 29 patients with exon 19 deletions, high expression of EGFR E746-A750 deletion-specific (6B6) protein (score 2 or 3) was observed in 17/17 cases (100%) with 15-bp deletion (16 with ELREA and 1 with ELREA + T751I). Of the 12 cases showing uncommon deletions in exon 19, nine (75%) samples were completely negative (score 0) and 3 (25%) were weakly positive (score 1). All cases were negative (score 0) for EGFR L858R mutant-specific (43B2) protein. In the 27 patients with exon 21 mutations, IHC with the mAb against the L858R mutation showed high positivity for the protein (score 2 or 3) in 25/27 (93%) and in 100% of the 25 samples with the L858R substitution; however, it failed to identify the L816Q mutation (0/2 cases). In addition, all 27 samples were negative for the EGFR E746-A750 deletion-specific (6B6) protein (Tables 2 and 3 , Figures 2 and 3 ).

Discussion EGFR is a member of the ErbB family of receptor tyrosine kinases, which also includes HER2/neu, HER3, and HER4[ 15 ]. Activating mutations in the tyrosine kinase domain, involving mainly exons 19 and 21, play an important role in lung oncogenesis and tumor progression and are related to the clinical efficacy of EGFR TKIs such as gefitinib or erlotinib[ 5 , 9 , 11 ]. Analysis of these mutations has become an important tool for targeted therapy in lung cancer 3 , and in recent years many efforts have been made to find a more specific and sensitive methodology to detect them[ 10 , 16 - 18 ]. Nevertheless, these techniques are relatively expensive for routine use in clinical laboratories, and depend on the quality of the samples. IHC is a standardized assay of simple methodology and high sensitivity and specificity, and the development of specific antibodies against EGFR mutation proteins might be useful for the diagnosis and treatment of lung cancer. In 2009 Yu et al [ 12 ] first generated two mAbs against E746-A750del and L858R point mutation from New Zealand rabbits and evaluated them by Western blotting, immunofluorescence and IHC. They tested these antibodies in a series of cell lines and in tumor tissues from patients with primary NSCLC, with known and unknown EGFR mutations, comparing the IHC results with DNA sequencing. They found that IHC with these mutation-specific antibodies for EGFR mutations showed a sensitivity of 92% and a specificity of 99%. Recently, five studies[ 14 , 19 - 22 ] examined the presence of EGFR mutations in NSCLC by IHC using the same two rabbit mAbs and reported sensitivity ranging from 36% to 100% and specificity ranging from 94% to 99% (Table 4 ). Kato et al [ 20 ] analyzed 70 gefitinib-treated NSCLC patients. Although a high sensitivity and specificity for these mAbs were described, IHC staining was not significantly correlated with overall survival. A very exhaustive analysis of the role of EGFR in NSCLC was recently reported by Ilie et al [ 19 ]. They assessed EGFR status in a tissue microarray (TMA) of 61 lung adenocarcinomas by IHC, fluorescent in situ hybridization (FISH) and direct sequencing and compared their results with those of conventional methods performed on whole-tissue sections. The authors reported a specificity of 92% for the mAb against the E746-750 deletion. Kawahara et al [ 21 ] reported a sensitivity of 83% for the L858R mutation antibody and 79% for the exon 19 deletion antibody. Brevet et al [ 14 ] reported a sensitivity of 84.6% and a specificity of 98.9% for E746-A750 and a sensitivity of 95.2% and a specificity of 98.8% for L858R. Kitamura et al [ 22 ] reported a sensitivity of 36% and a specificity of 97% for L858R and a sensitivity of 40% and a specificity of 99% for E746-A750. In the present study, we found a sensitivity of 100% and a specificity of 100% for the L858R exon 21 mutation antibody and a sensitivity of 63% and a specificity of 100% for the 15-bp deletion antibody. Table 4 summarizes the clinicopathological characteristics of patients and the findings of seven studies examining EGFR mutations by IHC, including the present study. Although the most common EGFR mutations are the 15-bp ELREA deletion in exon 19 and the L858R substitution in exon 21[ 2 , 3 ], other less frequent deletions have been identified[ 4 , 6 , 8 , 23 ]. Using DNA sequencing, Yu et al [ 12 ] detected only two cases with uncommon deletions in exon 19; E746-T751 del stained positive and L747-A750 negative for IHC. In the present study, we had 12 samples with uncommon deletions in exon 19 (9-bp, 12-bp, 18-bp, 21-bp and 24-bp) and 2 samples with the uncommon exon 21 L816Q mutation. In these samples, IHC for both mutation-specific antibodies was not able to detect the alteration.

Conclusions IHC with the mutation-specific rabbit mAbs against EGFR is a simple and standardized assay which could prove useful as a first, quick screening of NSCLC patients. However, although these antibodies seem to be quite reliable for the detection of patients carrying the most common EGFR mutations[ 12 ], they were not able to detect other EGFR gene mutations, such as 9-bp, 12-bp, 18-bp, 21-bp or 24-bp deletions or the L861Q substitution[ 14 ]. In consequence, if the antibodies are to be used in clinical practice, molecular biology techniques will be needed to further analyze the IHC-negative patients. However, the generation of a refined panel of antibodies able to detect both the frequent and the uncommon EGFR exon 19 deletions and exon 21 mutations as well as the resistance mutation T790 M in exon 20 could lead to the universal application of IHC for detecting EGFR mutations in NSCLC patients, as part of the routine IHC work-up of lung adenocarcinomas.

Background Immunohistochemistry (IHC) with mutation-specific antibodies may be an ancillary method of detecting EGFR mutations in lung cancer patients. Methods EGFR mutation status was analyzed by DNA assays, and compared with IHC results in five non-small-cell lung cancer (NSCLC) cell lines and tumor samples from 78 stage IV NSCLC patients. Results IHC correctly identified del 19 in the H1650 and PC9 cell lines, L858R in H1975, and wild-type EGFR in H460 and A549, as well as wild-type EGFR in tumor samples from 22 patients. IHC with the mAb against EGFR with del 19 was highly positive for the protein in all 17 patients with a 15-bp (ELREA) deletion in exon 19, whereas in patients with other deletions, IHC was weakly positive in 3 cases and negative in 9 cases. IHC with the mAb against the L858R mutation showed high positivity for the protein in 25/27 (93%) patients with exon 21 EGFR mutations (all with L858R) but did not identify the L861Q mutation in the remaining two patients. Conclusions IHC with mutation-specific mAbs against EGFR is a promising method for detecting EGFR mutations in NSCLC patients. However these mAbs should be validated with additional studies to clarify their possible role in routine clinical practice for screening EGFR mutations in NSCLC patients.

Competing interests The authors declare that they have no competing interests. Authors' contributions SS, MT, RR participated in the design of the study and its writing. MAM, CQ, IDA, CM, JBA, SB carried out the molecular genetic studies. JLGL, UJ, DI, TM, SV, CC, RGC, BM have made substantial contributions to acquisition of data. JJS, MT, RR, SS have made substantial contributions to analysis and interpretation of data. SS, SRC carried out the immunoassays. JJS performed the statistical analysis. All authors read and approved the final manuscript. Supplementary Material

Acknowledgements The authors thank Herbert Haack and Katherine Crosby (Cell Signaling Technology, Inc., Danvers, MA, USA) for providing the monoclonal antibodies used in the study and Ignacio Wistuba (Departments of Thoracic/Head and Neck Medical Oncology and Pathology, The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA) for comments on an earlier version of the manuscript.

CC BY

no

2022-01-12 15:21:44

J Transl Med. 2010 Dec 18; 8:135

oa_package/be/8d/PMC3016260.tar.gz

PMC3016261

21129187

Background In ancient times, medicinal plants have been used all over the world as unique sources of medicines and may constitute the most common human use of biodiversity [ 1 , 2 ]. According to the World Health Organization, 80% of people in developing countries still depend on local medicinal plants to fulfill their primary health needs [ 3 ]. Besides that, there is a global consensus on the benefits of phytopharmacy and at present medicinal plants occupy a key position in plant research and medicine. These facts associated with the progressive loss of traditional knowledge, due to rural exodus, and with the threats to which Plant Genetic Resources (PGR) are exposed, make the efforts to study and preserve PGR relevant in every respect. In this context, several conservation studies have been performed [ 4 - 6 ]. Like most African countries, Mozambique is an important repository of biological diversity. This diversity is used by ca. 90% of the country's population to fulfill its housing, food, energy and health needs. According to [ 7 ], in Mozambique approximately 15% of the total PGR (ca. 5,500 plant species) is used by rural communities for medical purposes and plays a key role in basic health care. Despite a long history of medicinal plants use in Mozambique, research on this subject is still incipient [ 8 - 10 ] and poorly disseminated, focusing mainly on medicinal plant markets and trade issues from Maputo province [ 7 ]. The work presented in this article reports on the utilization of medicinal plants in the Canhane village, district of Massingir, Province of Gaza. The last survey in the region dates from 1960-70 [ 11 , 12 ]. Canhane village is located 32° 09' 30" E and 24° 4' 30" S (Figure 1 ). With an extension of 7,200 ha, the village has a flat landscape with slopes ranging from 0 to 2% and altitudes from 95 m N to 200 m S [ 13 ]. The climate is semi-arid with two seasons: (i) dry season (April/May to October/November), with temperatures varying from 14.5°C to 28.5°C and a maximum annual precipitation of 67.9 mm; and (ii) hot and rainy season (October/November to April/May), with temperatures ranging from 19.9°C to 32.8°C and a maximum annual precipitation of 370 mm [ 14 ]. The humidity index may vary between -50 and -70, the negative values indicating the dryness of the region [ 15 ]. The soils are essentially sandy with a low to moderate percentage of organic matter (0-3%) and thus poor for agriculture. The village has 1357 inhabitants (51% women, 49% men) the great majority belonging to the Valoyi ("Witch doctor") family from the Changana ethnic group [ 16 , 17 ]. The community has poor access to water resources, health services (the closest health center is located in the Massingir village, seven Km away from Canhane), trading and communications, an obsolete energy system and an unsuccessful school system. Due to the lack of a local health center, traditional medicine plays an important role in basic health care. The main activity is agriculture, followed by livestock and fisheries. Handicraft is a tertiary activity. The major habitat types of Canhane are woodlands, savannah and grasslands [ 18 , 19 ]. Currently, the vegetation communities are at different levels of degradation mainly due to human practices (e.g. production of firewood, charcoal and grazing). The over-exploitation of resources and the limiting environmental conditions seem to be associated with the decay of the resilient capacity of the ecosystems as evidenced by the occurrence of great devastated areas [ 17 ]. With this study, we intended to contribute to the conservation and valorization of the local floristic and cultural heritage. It should be noted that the study area is of particular importance, since it is located in the heart of the Limpopo National Park, which together with Kruger National Park (South Africa) and Gonarezhou National Park (Zimbabwe) constitute the Great Limpopo Transfrontier Park and Conservation Area (GLTP). The study reports on 53 medicinal plant species and their traditional applications.

Methods Ethnobotanical data collection The work was initiated with a meeting between the researchers, the community leader and the Commission for Social Management from Canhane Village, in order to: i) explain the aim and importance of the work and its integration on the Community-based Development Program; ii) get cooperation and permission to use the cultural heritage; iii) collect information for structuring the interviews; iv) give orientations for the selection of informants by age and gender; and v) plan the field activities. Eleven informants (six men and five women) were selected as the best traditional knowledge holders. The selection criteria were based on the size of the study site, the vegetation homogeneity and on the indications provided by the community. Due to reasons related to beliefs and myths, it was not possible to get the information directly from Witch doctors. However, it should be highlighted that most of the Canhane inhabitants belong to the Valoyi ("Witch doctor") family. Using standard methods [ 20 , 21 ], the data was collected through intensive structured interviews and complemented with semi-structured interviews in local language (i.e. Changana). These included: common and local name of the plant, applications, parts of the plant used, methods of preparation and administration routes. Translation to Portuguese was validated by linguistic specialists. Taxonomic identification The medicinal plants reported by the informants were collected during three field surveys (in October of 2007 and in March and November of 2008). The team was accompanied by two local guides with a deep knowledge of local flora. Species identification was done during the field visits and by comparing voucher specimens with specimens deposited at the Herbarium of the Faculty of Sciences, Universidade Eduardo Mondlane (LMU, Maputo, Mozambique). The scientific names were confirmed through specialized bibliography [ 22 - 25 ] as well as the African Plant Database [ 26 ], Tropicos database [ 27 ] and the International Plant Names Index [ 28 ]. Additional information was gathered from the study of numerous herbarium specimens, mainly from the Tropical Research Institute Herbarium (LISC, Lisbon, Portugal).

Results and Discussion Medicinal Plants' Survey A total of 53 plant species distributed over 47 genera and 31 families were reported by the 11 informants (Table 1 ). All the reported species grew naturally in the area, reflecting the social importance of the local floristic resources. Most of the identified plants were shrubs or trees (15 spp. or 28.3%), herbs and trees (11 spp. for each category or 20.8%), and shrubs (nine species or 17.0%). The best represented families were Fabaceae (six species), Euphorbiaceae (four species) and Tiliaceae (three species). Altogether the 53 species were used to treat 50 different human health problems (Table 1 ), the great majority of which (75.5%) having more than one medical application. The most cited species were Euclea racemosa (ca. 82%), Colophospermum mopane , Cucumis sp. and Elephantorrhiza elephantina (ca. 73% each species), Cassia abbreviata and Cissus quadrangularis (ca. 64% each species), Aloe marlothii, Maerua edulis, Secamone parvifolia and Terminalia sericea (ca. 55% each species) and Boscia albitrunca, Gossypium herbaceum and Gymnosporia heterophylla (ca. 46% each species) (data not shown). The number of medicinal plants and their potential applications reflect the rich ethnomedicinal knowledge in the Canhane community. Similar potentialities were found in other African countries like Cameroon [ 29 ] and Ethiopia [ 30 - 32 ] as well as in non-African countries [ 33 - 35 ]. Certainly, there is a lot more knowledge to exploit on the topic in Mozambique. More than half of the reported species (54.7%) were used for stomach and intestine related disturbances (Table 2 ). Of these, almost 38% were used to treat diarrhea and dysentery, a major concern in the region. In fact, in Mozambique diarrhea has for a long time been associated with a complex array of illnesses. Amongst them, dysentery and cholera usually have a high mortality rate if not treated promptly [ 10 ]. The use of traditional medicinal plants seems to play a major role in controlling diarrhea-associated diseases. Around 23% of the surveyd species were used as analgesic, anti-inflammatory or anti-pyretic and for wound treatment, 15% for dentistry and 11% for gynecology-related problems. Approximately 9% of the reported species were used to treat ear diseases and hemorrhoids, 8% for burns, cough, debility and malnutrition, epilepsy, eye diseases and malaria, and 6% for heart problems. Only one species, Ximenia americana (ca. 2%) was used against HIV-AIDS. Thus, looking at the three major national health concerns, namely diarrhea and dysentery, malaria and HIV-AIDS a considerable number of potentialities are available for the first group (11 species), while moderate (four species) and low (one species) alternatives can be exploited for malaria and HIV-AIDS. In fact, several pharmacological studies of these three groups of human ailments are available for most of the species reported in the present survey [ 36 - 42 ]. With the exception of six species ( Blepharis diversispina, Grewia flavescens, Guibourtia conjugata, Hermannia micropetala, Loeseneriella crenata, Zanthoxylum humile ), all species under study have been reported as medicinal plants in other African countries [ 11 , 43 - 46 ]. While the use of G. flavescens and Z. humile by traditional healers has been reported in India and Mozambique, respectively [ 7 , 47 ], as far as our literature review goes, four species (i.e. B. diversispina, G. conjugata, H. micropetala , and L. crenata ) were reported here for the first time. Of these, only two genera have been associated with ethonomedicine: the genus Blepharis [ 48 ] and the genus Loeseneriella ( L. obtusifolia ). Thus, these species constitute new potential sources of natural medicines. From the 53 species, nine were reported previously by [ 10 ] and 3 by [ 7 ] in studies conducted in the province of Maputo. Besides that, several other species belonging to 11 genera ( Aloe, Asparagus, Boscia, Cissus, Crinum, Cucumis, Ficus, Grewia, Maerua, Secamone, Strychnos ) were also reported as medicinal species [ 7 ]. The potential medicinal plant markets from the southern provinces of Maputo and Gaza seem to be different. This may reflect the rich ethonomedicinal potential which exists in the entire country. A comparative analysis with local specific ethnobotanical literature [ 11 , 12 , 46 ] and complementary information gathered from the LISC Herbarium plant collections, identified 25 different plant species used for medicinal purposes (Table 3 ) of which only two, Combretum imberbe and Lannea schweinfurthii , are common to those reported in this study. Regarding their applications, similarities were found for C. imberbe (stomach disorders) and L. schweinfurthii (diarrhea and stomach disorders). According to the available data, C. imberbe was also used to treat schistosomiasis and L. schweinfurthii to treat tuberculosis, while in our survey they were additionally indicated for the treatment of toothache ( C. imberbe ), anemia and malaria ( L. schweinfurthii ). Because the older surveys did not specifically target medicinal plants, we believe that our data are more accurate in what concerns the applications of these two species. This fact may also explain why the great majority of the species reported 40 years ago (23 out of 25 or 92%) does not overlap with those identified in this survey. However, the possibility of loss of genetic resources and/or traditional knowledge should also be considered. The great majority of the identified species (46 spp. or 86.8%) were also used for other purposes than medicine (Table 4 ; Figure 2 ). The major groups of applications were associated with beliefs and myths (26 spp. or ca. 49%) or used as food (24 spp. or ca. 45%). Wood production, handicraft and veterinary were the third major class of application, with 10 (ca. 19%), 9 (ca. 17%) and 8 (ca. 15%) species, respectively. This reinforces the socio-economic importance of the reported species, placing them in a privileged position for conservational aspects and income-generating purposes. Plant parts used, methods of preparation and administration routes Several plant parts were used (Table 1 ), the most frequent being roots (38.8%), followed by leaves (17.5%), stems (13.6%), fruits (8.8%), bark (5.8%), sap (5.8%), combinations of plant organs (3.9%), branches (2.9%) and seeds (2.9%). Regarding the methods of preparation (Figure 3 ), in many cases (38%) a combination of methods was used. The most common method was decoction (25%), followed by direct consumption (10%), infusion (6%), crushing (5%), grinding (5%), maceration (4%), scraping (2%), heating (2%), burning (1%), cutting (1%) and juice (1%). Fifty nine percent of the medicines were administered orally, 31% topically and only 10% through vaccine, bath, enema, eyewash and necklace (ca. 2% for each mode) (Figure 4 ). In general, the results seem to follow the pattern of medicinal plant uses in Africa [ 26 , 28 , 49 ] except that in Canhane, instead of leaves, roots occupy the top position which is concordant with the results from [ 7 ]. Consistent with the findings of [ 28 , 49 ] in Kenya and Ethiopia respectively, is the lack of standardized dosage and quality control. Conservational aspects In general, the community was conscientious and motivated regarding conservational issues and had adopted sound measures for the rational use of medicinal plants. Conservation in farms or home gardens was performed for the most commonly used plants, namely Aloe marlothii, A. zebrina, B. albitrunca, C. mopane, C. zeyheri, E. racemosa, Ficus sycomorus, Flueggea virosa, Grewia hexamita , G. monticola , H. micropetala, Sclerocarya birrea and T. sericea . Additionally, the intensity and frequency of exploitation was controlled and there were local rules to protect native plant species, particularly Adansonia digitata , B. discolor , Cissus cornifolia , C. mopane , E. elephantina , F. sycomorus , F. virosa , G. monticola , G. conjugata , Manilkara mochisia , S. birrea , and Strychnos madagascariensis . Other conservation measures included community guards in protected places to control fires and logging, mostly due to South African migrants. On the other hand, trading was controlled and confined to the village.

Results and Discussion Medicinal Plants' Survey A total of 53 plant species distributed over 47 genera and 31 families were reported by the 11 informants (Table 1 ). All the reported species grew naturally in the area, reflecting the social importance of the local floristic resources. Most of the identified plants were shrubs or trees (15 spp. or 28.3%), herbs and trees (11 spp. for each category or 20.8%), and shrubs (nine species or 17.0%). The best represented families were Fabaceae (six species), Euphorbiaceae (four species) and Tiliaceae (three species). Altogether the 53 species were used to treat 50 different human health problems (Table 1 ), the great majority of which (75.5%) having more than one medical application. The most cited species were Euclea racemosa (ca. 82%), Colophospermum mopane , Cucumis sp. and Elephantorrhiza elephantina (ca. 73% each species), Cassia abbreviata and Cissus quadrangularis (ca. 64% each species), Aloe marlothii, Maerua edulis, Secamone parvifolia and Terminalia sericea (ca. 55% each species) and Boscia albitrunca, Gossypium herbaceum and Gymnosporia heterophylla (ca. 46% each species) (data not shown). The number of medicinal plants and their potential applications reflect the rich ethnomedicinal knowledge in the Canhane community. Similar potentialities were found in other African countries like Cameroon [ 29 ] and Ethiopia [ 30 - 32 ] as well as in non-African countries [ 33 - 35 ]. Certainly, there is a lot more knowledge to exploit on the topic in Mozambique. More than half of the reported species (54.7%) were used for stomach and intestine related disturbances (Table 2 ). Of these, almost 38% were used to treat diarrhea and dysentery, a major concern in the region. In fact, in Mozambique diarrhea has for a long time been associated with a complex array of illnesses. Amongst them, dysentery and cholera usually have a high mortality rate if not treated promptly [ 10 ]. The use of traditional medicinal plants seems to play a major role in controlling diarrhea-associated diseases. Around 23% of the surveyd species were used as analgesic, anti-inflammatory or anti-pyretic and for wound treatment, 15% for dentistry and 11% for gynecology-related problems. Approximately 9% of the reported species were used to treat ear diseases and hemorrhoids, 8% for burns, cough, debility and malnutrition, epilepsy, eye diseases and malaria, and 6% for heart problems. Only one species, Ximenia americana (ca. 2%) was used against HIV-AIDS. Thus, looking at the three major national health concerns, namely diarrhea and dysentery, malaria and HIV-AIDS a considerable number of potentialities are available for the first group (11 species), while moderate (four species) and low (one species) alternatives can be exploited for malaria and HIV-AIDS. In fact, several pharmacological studies of these three groups of human ailments are available for most of the species reported in the present survey [ 36 - 42 ]. With the exception of six species ( Blepharis diversispina, Grewia flavescens, Guibourtia conjugata, Hermannia micropetala, Loeseneriella crenata, Zanthoxylum humile ), all species under study have been reported as medicinal plants in other African countries [ 11 , 43 - 46 ]. While the use of G. flavescens and Z. humile by traditional healers has been reported in India and Mozambique, respectively [ 7 , 47 ], as far as our literature review goes, four species (i.e. B. diversispina, G. conjugata, H. micropetala , and L. crenata ) were reported here for the first time. Of these, only two genera have been associated with ethonomedicine: the genus Blepharis [ 48 ] and the genus Loeseneriella ( L. obtusifolia ). Thus, these species constitute new potential sources of natural medicines. From the 53 species, nine were reported previously by [ 10 ] and 3 by [ 7 ] in studies conducted in the province of Maputo. Besides that, several other species belonging to 11 genera ( Aloe, Asparagus, Boscia, Cissus, Crinum, Cucumis, Ficus, Grewia, Maerua, Secamone, Strychnos ) were also reported as medicinal species [ 7 ]. The potential medicinal plant markets from the southern provinces of Maputo and Gaza seem to be different. This may reflect the rich ethonomedicinal potential which exists in the entire country. A comparative analysis with local specific ethnobotanical literature [ 11 , 12 , 46 ] and complementary information gathered from the LISC Herbarium plant collections, identified 25 different plant species used for medicinal purposes (Table 3 ) of which only two, Combretum imberbe and Lannea schweinfurthii , are common to those reported in this study. Regarding their applications, similarities were found for C. imberbe (stomach disorders) and L. schweinfurthii (diarrhea and stomach disorders). According to the available data, C. imberbe was also used to treat schistosomiasis and L. schweinfurthii to treat tuberculosis, while in our survey they were additionally indicated for the treatment of toothache ( C. imberbe ), anemia and malaria ( L. schweinfurthii ). Because the older surveys did not specifically target medicinal plants, we believe that our data are more accurate in what concerns the applications of these two species. This fact may also explain why the great majority of the species reported 40 years ago (23 out of 25 or 92%) does not overlap with those identified in this survey. However, the possibility of loss of genetic resources and/or traditional knowledge should also be considered. The great majority of the identified species (46 spp. or 86.8%) were also used for other purposes than medicine (Table 4 ; Figure 2 ). The major groups of applications were associated with beliefs and myths (26 spp. or ca. 49%) or used as food (24 spp. or ca. 45%). Wood production, handicraft and veterinary were the third major class of application, with 10 (ca. 19%), 9 (ca. 17%) and 8 (ca. 15%) species, respectively. This reinforces the socio-economic importance of the reported species, placing them in a privileged position for conservational aspects and income-generating purposes. Plant parts used, methods of preparation and administration routes Several plant parts were used (Table 1 ), the most frequent being roots (38.8%), followed by leaves (17.5%), stems (13.6%), fruits (8.8%), bark (5.8%), sap (5.8%), combinations of plant organs (3.9%), branches (2.9%) and seeds (2.9%). Regarding the methods of preparation (Figure 3 ), in many cases (38%) a combination of methods was used. The most common method was decoction (25%), followed by direct consumption (10%), infusion (6%), crushing (5%), grinding (5%), maceration (4%), scraping (2%), heating (2%), burning (1%), cutting (1%) and juice (1%). Fifty nine percent of the medicines were administered orally, 31% topically and only 10% through vaccine, bath, enema, eyewash and necklace (ca. 2% for each mode) (Figure 4 ). In general, the results seem to follow the pattern of medicinal plant uses in Africa [ 26 , 28 , 49 ] except that in Canhane, instead of leaves, roots occupy the top position which is concordant with the results from [ 7 ]. Consistent with the findings of [ 28 , 49 ] in Kenya and Ethiopia respectively, is the lack of standardized dosage and quality control. Conservational aspects In general, the community was conscientious and motivated regarding conservational issues and had adopted sound measures for the rational use of medicinal plants. Conservation in farms or home gardens was performed for the most commonly used plants, namely Aloe marlothii, A. zebrina, B. albitrunca, C. mopane, C. zeyheri, E. racemosa, Ficus sycomorus, Flueggea virosa, Grewia hexamita , G. monticola , H. micropetala, Sclerocarya birrea and T. sericea . Additionally, the intensity and frequency of exploitation was controlled and there were local rules to protect native plant species, particularly Adansonia digitata , B. discolor , Cissus cornifolia , C. mopane , E. elephantina , F. sycomorus , F. virosa , G. monticola , G. conjugata , Manilkara mochisia , S. birrea , and Strychnos madagascariensis . Other conservation measures included community guards in protected places to control fires and logging, mostly due to South African migrants. On the other hand, trading was controlled and confined to the village.

Conclusions This study shows the social importance of the floristic richness in the Canhane village, particularly regarding the significance of medicinal plants in primary healthcare. This is reflected in the great diversity of plants used for medical purposes as well as in the wide range of their applications and associated procedures. The data compiled in this study are a contribution to the documentation of PGR at the national and regional level and can serve as a basis to develop larger and interdisciplinary studies.

Background Medicinal plants are used by 80% of people from developing countries to fulfill their primary health needs, occupying a key position on plant research and medicine. Taking into account that, besides their pharmaceutical importance, these plants contribute greatly to ecosystems' stability, a continuous documentation and preservation of traditional knowledge is a priority. The objective of this study was to organize a database of medicinal plants including their applications and associated procedures in Canhane village, district of Massingir, province of Gaza, Mozambique. Methods In order to gather information about indigenous medicinal plants and to maximize the collection of local knowledge, eleven informants were selected taking into account the dimension of the site and the fact that the vegetation presents a great homogeneity. The data were collected through intensive structured and semi-structured interviews performed during field research. Taxonomical identification of plant species was based on field observations and herbarium collections. Results A total of 53 plant species have been reported, which were used to treat 50 different human health problems. More than half of the species were used for stomach and intestine related disturbances (including major diseases such as diarrhea and dysentery). Additionally, four species with therapeutic applications were reported for the first time, whose potential can further be exploited. The great majority of the identified species was also associated with beliefs and myths and/or used as food. In general, the community was conscientious and motivated about conservational issues and has adopted measures for the rational use of medicinal plants. Conclusions The ethnomedicinal use of plant species was documented in the Canhane village. The local community had a rich ethnobotanical knowledge and adopted sound management conservation practices. The data compiled in this study show the social importance of the surveyed plants being a contribution to the documentation of PGR at the national and regional level.

List of abbreviations GLTP: Great Limpopo Transfrontier Park and Conservation Area; PGR: Plant Genetic Resources. Competing interests The authors declare that they have no competing interests. Authors' contributions The design, planning, field survey and taxonomic analysis was coordinated and conducted by TF. AR and TF performed the data processing and analysis. The taxonomic revision was done by MMR and JT. Data from other geographical regions and from 1960-70 was retrieved by MMR, JT and TF. Literature retrieval was done by AR and MMR. AR wrote the manuscript, which was revised by MMR and TF. All authors read and approved the manuscript.

Acknowledgements The authors would like to express their gratitude to the Canhane community, particularly the 11 informants for their unreserved efforts in transmitting traditional local knowledge, the Canhane Community Lodge for coordinating the work with the community and for the logistics, the direction of Paulo Samuel Kankhomba Primary school for selecting the students and Non-Governamental Organization LUPA. Acknowledgments are also due to botanical collectors, António Zacarias and Ernesto Macamo, and translators, Arminda Mfumo, dr. David Langa and dr. Orlando Bahule. Thanks to Dr. José Manuel Mota Cardoso (Veterinary Hospital, Eduardo Mondlane University, Mozambique) for medical terminology and language revision, Dr. Katharina Pawlowsi (Stockholm University) for the language revision, Dr. Cristina Duarte (Tropical Research Institute, Portugal) for the taxonomic revision, and Ezequiel Correia for preparing the distribution map of the study area. This work was supported by a grant from Fundo Aberto - Universidade Eduardo Mondlane and the Swedish International Development Agency (Research Funding).

CC BY

no

2022-01-12 15:21:44

J Ethnobiol Ethnomed. 2010 Dec 3; 6:33

oa_package/db/28/PMC3016261.tar.gz

PMC3016262

21134261

Background Clopidogrel is a widely prescribed thienopyridine for the prevention of atherothrombotic complications following acute coronary syndromes (ACS) or percutaneous coronary interventions (PCIs) [ 1 ]. Clopidogrel is a prodrug that has no intrinsic antiplatelet activity without activation by hepatic metabolism through the cytochrome P450 (CYP) system [ 2 ]. Multiple CYP enzymes have been implicated in this process, but recently the CYP2C19 enzyme has assumed predominance as it is involved in both sequential oxidative steps [ 3 ]. The possibility of drug interactions limiting clopidogrel's efficacy was raised several years following in vitro statin and clopidogrel studies, but without definitive clinical confirmation of increased adverse outcomes [ 4 - 6 ]. Recently, mechanistic in vitro studies have suggested that proton pump inhibitors (PPIs) may diminish clopidogrel's clinical efficacy via CYP2C19 competitive inhibition [ 7 - 10 ]. Consistent with these in vitro observations, several clinical studies have shown higher cardiovascular events in clopidogrel patients exposed to PPIs compared to those not exposed [ 11 , 12 ], leading the Food and Drug Administration (FDA) [ 13 ] and the European Medicines Agency [ 14 ] to issue public alerts recommending the avoidance of prescribing PPIs in patients who also take clopidogrel. However, as the studies have largely been nonexperimental, the possibility of a spurious association due to bias needs to be attentively considered. This is a clinically important question as many cardiac patients are also at high risk of gastrointestinal (GI) bleeding (due to age, smoking and concomitantly prescribed drugs), and PPIs may substantially mitigate this risk [ 15 ]. Previous reviews of this interaction question have appeared [ 16 - 20 ], but they have not (1) been systematic, (2) provided a critical analysis of methodological issues or (3) integrated these safety concerns with prior knowledge of clopidogrel's time frame of action. The present review addresses these issues by performing a systematic and critical analysis of all clinical studies of this putative interaction.

Methods We reviewed the MEDLINE and EMBASE electronic databases from January 1, 2005, to October 7, 2010, without any language restriction, combining search terms for clopidogrel ("clopidogrel" OR "Plavix" OR "thienopyridine"), PPIs ("PPI" OR " omeprazole" OR "lansoprazole" OR "pantoprazole" OR "esomeprazole" OR "rabeprazole") with those for cardiovascular outcomes ("mortality" OR "cardiovascular disease" OR "heart disease" OR "CAD" OR "MI" OR "UA" OR "coronary angiography" OR "coronary restenosis" OR "PCI" OR "stroke") and drug interaction ("interaction" OR " inhibition" OR " CYP2C19"). References of relevant identified articles were hand-searched for additional studies. Abstracts from medical organization conferences (American Heart Association, American College of Cardiology, European Society of Cardiology and Transcatheter Cardiovascular Therapeutics) were manually searched from 2005. A priori it was decided that only articles reporting clinical outcomes would be retained. Mechanistic in vitro studies measuring platelet aggregation were therefore excluded. Two investigators independently reviewed articles for inclusion and study quality. Both reviewers examined the methodology component independently of the study results, but given the publicity surrounding the individual studies it was impossible to ensure that the reviewers were totally blinded to the outcomes. It was planned to resolve disagreements by consensus, but there was perfect agreement between the investigators on their initial assessments. This systematic review was performed according to the PRISMA guidelines (see online Additional file 1 ). Overall study quality was defined as high for well-performed randomized clinical trials (RCTs) as they exhibit the best internal validity. Observational studies are considered of lower quality as they have less internal validity due to numerous potential biases and consequently could attain a maximum moderate quality score. By following published guidelines to improve the reporting of observational studies [ 21 ] and specifically to mitigate against potential biases [ 22 ], we qualitatively evaluated all the observational studies for their propensity for the most common and important biases (selection, confounding and misclassification), thereby allowing their final classification into moderate- or low-quality levels. Overall observational study quality was judged to be moderate if the propensity for bias was low and specific methods to minimize bias were employed (explicit consideration of the time dependency of drug exposure and appropriate clinical outcome assessment to minimize misclassification, multiple methods to assess confounding and multivariate sensitivity analyses to test the independence and robustness of their results). Alternatively, studies were judged to be at moderate risk for bias and hence a low quality score. The quality score of unpublished studies was systematically reduced by one grade owing to the absence of formal peer review. Isolated abstracts were considered to be of low quality as there are insufficient details to fully evaluate these reports. Finally, whenever possible, an external quality measure that compared survival curves from the clopidogrel efficacy randomized trials to the observational hazard ratios was performed to check for consistency in the etiologically relevant time windows. Since study heterogeneity existed with respect to research methods, quality, study populations and health care systems, it was decided a priori that quantitative data pooling was inappropriate.

Results Our literature search found 54 publications, and 18 of these studies [ 11 , 12 , 23 - 38 ] fulfilled our inclusion criteria (see Figure 1 ). One study was experimental with random PPI allocation [ 25 ], while the others were nonexperimental. Study details are presented in Table 1 . Two studies [ 12 , 36 ] with an otherwise low propensity for bias considered a primary endpoint of only nonfatal myocardial infarction (MI), ignoring fatal myocardial infarction, which has been a recurrent component of the composite measure of clopidogrel efficacy. Although an increased risk with the clopidogrel-PPI combination for nonfatal myocardial infarction was observed in both studies, mortality was actually lower in the PPI-exposed group in one study [ 12 ] and was not reported in the other [ 36 ], such that the more logical and standard combined endpoint of cardiovascular death and nonfatal MI could not be reliably evaluated. The lack of justification for their primary outcome and lack of sensitivity analyses regarding potential different adverse outcomes explains the lower assigned quality ratings for these two studies. Among the full published studies judged to be methodologically weak, several potential biases (selection, residual confounding, immortal time and interpretative) were noted. Ten of 13 studies judged to be of low quality reported a statistically significant association for a harmful clopidogrel-PPI interaction, while none of the five moderate-quality studies found an association ( P = 0.007). Each individual study evaluated a combined exposure to any PPI for the assessment of their primary endpoint, with the exception of the experimental study [ 25 ], which was restricted to omeprazole. One study [ 12 ] did report a difference between pantoprazole and nonpantoprazole PPIs in their association with adverse clinical outcomes, but the corrected proper statistical analysis actually revealed no difference [ 39 ]. No other study demonstrated a clinically significant difference in clinical outcomes among the different PPIs.

Discussion Our systematic review identified 18 clinical studies of a clopidogrel-PPI interaction, with the majority (10) reporting a statistically significant result. Multiple sources of heterogeneity (different populations, outcomes assessed, drug exposure methods and study quality) existed between the studies, preventing a formal quantitative analysis of all studies. However, a stratified analysis based on study quality demonstrated an inverse correlation between study quality and a positive outcome. Moreover, none of the 18 studies found a clinically meaningful difference between the various PPIs. Therefore, high-quality evidence supporting a clinically significant clopidogrel-PPI interaction is presently lacking. Consequently, while good prescribing patterns are to be endorsed for all medications, recent edicts to avoid PPIs in cardiac patients with a clinical indication for their use seem poorly justified. While our criteria for judging study quality are standardized and were applied a priori and independently of the study results, their validity would be enhanced by some empirical assessment. Certainly, these quality criteria have face and content validity. In judging the presence of construct validity, it may be helpful to review what is known about clopidogrel efficacy from high-quality clinical trials. The Clopidogrel in Unstable Angina to Prevent Recurrent Events Trial (or CURE) was the seminal randomized trial [ 1 ] showing the benefits of clopidogrel in acute coronary syndromes, namely, a 1.9% absolute reduction in the primary endpoint of cardiovascular death, nonfatal MI or stroke with clopidogrel which was evident by 3 months with no or very little additional advantage at later time points (see Figure 2 ). In a PCI trial [ 40 ], clopidogrel again was shown to exhibit its beneficial results principally in the first 30-90 days. A lack of long-term benefit in the non-acute setting was confirmed in the randomized CHARISMA trial [ 41 ], which showed no advantages in stable cardiac or high-risk patients at any time over a 28-month follow-up period. Contrast these results with those from the observational studies of the putative drug interaction attributing 9% [ 11 ] and 18% [ 28 ] absolute increases in adverse outcomes to the inhibition of the protective effects of clopidogrel. Another study [ 29 ] attributed a greater than 15% absolute difference in all-cause mortality to this putative association. This represents a many-fold larger effect than the observed clopidogrel benefits in randomized trials. Moreover, in these studies, no significant difference was seen between the clopidogrel-PPI group and the clopidogrel-alone group within the therapeutically active 30-to 90-day window (see Figure 2 ). The observed adverse effects attributed to the drug interaction were therefore occurring at a time when clopidogrel has not been shown to be therapeutically active. It therefore seems likely that the observed differences between PPI users and nonusers are unrelated to clopidogrel inhibition, but rather are related to residual confounding from unmeasured prognostic variables as measured baseline prognostic factors were markedly different between the groups. Supporting this hypothesis, a post hoc analysis of the CREDO study [ 26 ] showed a 2% absolute increase in adverse cardiac events even among placebo patients who were exposed to PPIs. Unfortunately, the other observational studies have not presented survival functions, so a similar evaluation of the biological plausibility and consistency of their results cannot be made. It bears repeating that the only high-quality randomized trial permitting an unconfounded assessment of a causal association [ 25 ] did not show any evidence of harm. Moreover, it may be argued that post hoc analyses of clinical trial data offer better internal validity than other observational studies because of a greater homogeneity in the patient population, increased and higher-quality baseline clinical data and better outcome validation, thereby potentially mitigating, but certainly not eliminating, possible biases [ 22 ]. Our evaluation of the present studies empirically confirmed this association, with the post hoc clinical trial [ 30 ] receiving a higher score by our independent criteria providing further external support of our quality scale. Again, this higher-quality study did not show evidence for the putative interaction. Our systematic review provides a reproducible, prespecified and comprehensive review of the totality of the evidence in the literature of a potential interaction between clopidogrel and PPIs, thereby facilitating informed decision making. However, our systematic review may be criticized for our inability to ensure total blinding of methodological assessment from outcomes. As with any systematic review, we are also limited by the quality of the original publications. While we found a paucity of high-quality clinical evidence to support a clopidogrel-PPI interaction either collectively or individually, the power to find such an association is limited and does not preclude the possibility that later high-quality studies may appear to confirm this putative drug association. The issue of adverse outcomes associated with concomitant use of clopidogrel and PPIs has been assessed in previous reviews [ 16 - 20 ], as well as by American and European regulatory agencies [ 13 , 14 ], which concluded that the interaction is clinically significant. However, previous reviews have been dominated by studies with surrogate laboratory endpoints, were not systematic in identifying the clinical evidence, did not consider the putative interaction effect as a function of our previous knowledge of clopidogrel efficacy and did not provide a critical analysis of the methodological issues of each study. Lower-quality observational studies tend to overestimate effect size because of bias [ 42 ]. In the case of regulatory agencies, the threshold to report putative clinical adverse effects is naturally low in an effort to protect public health. However, with the recent spate of controversies regarding the cardiovascular safety of approved drugs such as rofecoxib and rosiglitazone, there is perhaps a strong desire for these agencies to avoid appearing inactive and rather to be perceived as proactive. Nevertheless, it is disconcerting that the potential positive benefits of avoiding gastrointestinal bleeds in high-risk patients appear to have been minimized if not ignored and that regulatory decisions have been made not on a systematic assessment of the clinical evidence, but rather largely on surrogate laboratory data and isolated clinical studies of questionable validity. One regulatory agency [ 14 ] has targeted specific PPIs as being culpable without a single valid clinical study to support this claim. Correctly or not, the impression is that the safety of the regulatory agencies from criticism has become more important than the critical assessment of public health safety. Beyond our specific conclusion that the evidence for a clinically meaningful clopidogrel-PPI interaction is poor, our critical analysis of this potential drug interaction leads to several general observations regarding pharmacoepidemiological research and subsequent policy decisions. First, selection bias or confounding by indication remains a potential threat to all observational research of both intended and unintended effects. Second, residual confounding is likely to be persistent when multivariate analysis is limited to data routinely collected from purely administrative data sets lacking the necessary clinical granularity. Post hoc analyses of randomized trials are similarly subject to the same potential biases, but the magnitude may be less owing to both a more homogeneous source population (from the strict inclusion and exclusion criteria, which favor the recruitment of populations with only one major health issue), the availability of more extensive baseline clinical data (decreasing residual confounding) and more systematic outcome assessment through adjudication (decreasing misclassification or information bias), thereby allowing a more rigorous assessment of the independent contribution of any putative association. Third, studies of surrogate endpoints (for example, platelet inhibition) may provide interesting avenues for future outcomes research, but these studies should not be used for clinical and regulatory decision making. The perils of surrogate markers have been repeatedly demonstrated in cardiovascular medicine, with specious examples ranging from ventricular ectopy suppression to lowering of glycosylated hemoglobin. Fourth, when possible, cohort studies should present survival functions, which may help in determining the reasonableness of conclusions by ensuring that events occur within the etiologically relevant time windows as demonstrated by RCTs. Finally, the time-varying nature of drug exposure must be meticulously tracked and sensitivity analyses must be performed to assess the robustness of conclusions to residual confounding or measurement error. Readers should reasonably expect that the peer review process provides some guidance and discussion about potential biases in observational research. Too often there is a tendency to explain discordant study results not by considering biases, but by assuming that the results are due to different populations, interventions or outcomes [ 43 , 44 ]. In our opinion, the probability of such effect modification is often small compared to the probability of bias. While some biases are subtle, others become more obvious with attentive reading and consideration of the totality of the evidence. The need to improve the reliability and value of medical research has been well recognized by many groups [ 43 - 45 ], and several guidelines have been published [ 21 , 22 ], including those that pertain to observational research, but more specific guidelines for designing and interpreting pharmacoepidemiological studies may be helpful.

Conclusions This systematic review of the putative drug interaction between clopidogrel and PPIs identified 18 clinical studies. Rigorous and extensive evaluation of the study methodologies demonstrated that the majority were of low quality, and only these studies reported a positive association between clopidogrel-PPI exposure and increased cardiovascular events. Supporting the notion that bias may be responsible for a spurious association, the increase in cardiovascular events attributed to this interaction appears to be occurring within a therapeutic window when clopidogrel has not been shown to be etiologically active. Conversely, all studies assessed as high quality did not demonstrate this association. No study was able to isolate any individual PPI as being more prone to adverse cardiac outcomes. Careful attention to the methodological details of published studies is therefore mandatory to avoid reaching questionable scientific conclusions. In light of this systematic review, and given the potential benefit of PPIs for patients at high risk of gastrointestinal bleeding, regulatory agencies may wish to reevaluate their recent edicts regarding this putative interaction.

Background Recently, several publications have investigated a possible drug interaction between clopidogrel and proton pump inhibitors (PPIs), and regulatory agencies have issued warnings despite discordant study results. In an attempt to clarify the situation, we performed a systematic review with a critical analysis of study methodologies to determine whether varying study quality (that is, bias) could explain the discordant results. Methods A systematic review of all studies reporting clinical outcomes was performed using an electronic literature search of the MEDLINE and EMBASE databases, abstracts from the major cardiology conferences and a hand-search of bibliographies from identified articles. Each study was evaluated for its risk of bias according to a prespecified quality measure scale. Results A total of 18 studies were identified. Ten of 13 studies judged to be of low scientific quality reported a statistically positive interaction between clopidogrel and the general class of PPIs, and each concluded this was likely a clinically meaningful effect. None of the five studies judged to be of moderate or high quality reported a statistically significant association. Multiple sources of heterogeneity (that is, different populations, outcomes assessed, drug exposure methods and study quality) prevented a formal quantitative analysis of all studies. An increased risk of bias was observed in the positive studies, resulting in an inverse correlation between study quality and a reported statistically positive interaction (10/13 versus 0/5; P = p = 0.007). There was also no clinical evidence for a positive interaction according to specific PPIs. Conclusion The observed association between clopidogrel and PPIs is found uniquely in studies judged to be of low quality and with an increased risk of bias. High-quality evidence supporting a clinically significant clopidogrel/PPI interaction is presently lacking.

Competing interests The authors declare that they have no competing interests. Authors' contributions JMB conceived the project. JMB and JPAL developed the protocol jointly and independently abstracted the data. JMB did the statistical analysis. JMB and JPAL jointly wrote all drafts and the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1741-7015/8/81/prepub Supplementary Material

Acknowledgements This work did not benefit from any external funding. Dr Brophy is a funded research scientist of Le Fonds de la Recherche en Santé du Québec.

CC BY

no

2022-01-12 15:21:44

BMC Med. 2010 Dec 6; 8:81

oa_package/13/82/PMC3016262.tar.gz

PMC3016263

21176133

Background Obesity is the epidemic of our time, with sharply and steadily rising rates [ 1 , 2 ]. The major adverse consequences of obesity including type 2 diabetes, atherosclerosis and essential hypertension, when added together, account for a large number of disease related deaths [ 3 , 4 ]. If the obesity-related cancer cases are added to this number, obesity-related mortality by far exceeds that of other common diseases [ 5 ]. The latter indicates the urgent need to develop novel efficient therapeutic modalities for this condition. The common denominator in the pathogenesis of the co-morbidities of obesity is the presence of an active, low-grade inflammatory process [ 6 ]. Despite evidence linking obesity to alterations in inflammatory response, little is known about the specific effects of obesity on the immune system. Recently, there has been a greater appreciation of the role of epigenetics, meiotically and mitotically heritable changes in gene expression that are not coded in the DNA sequence itself, in the immune and inflammatory responses [ 7 - 9 ]. Therefore, we hypothesize that DNA methylation changes play a role in obesity induced immune dysfunction. The goal of this study was to characterize DNA methylation profile in peripheral blood leukocytes in obese versus lean subjects using a genome wide approach. Identification of methylation changes in specific genes will provide important targets for further study into the mechanisms of obesity's effect on the immune system and the potential to develop new therapies to treat multiple obesity comorbidities.

Methods Subjects The genome wide methylation analysis was conducted in seven obese and seven age-matched lean controls. These 14 subjects were identified from the participants (n = 534) in the Lifestyle, Adiposity, and Cardiovascular Health in Youth (LACHY) study using the following inclusion criteria: (1) African American (AA) ancestry; (2) male; (3) having leukocyte DNA available; (4) obese cases having a body mass index (BMI) ≥ 99 th percentile for age and sex and lean controls having BMI ≤ 10 th percentile for age and sex. The LACHY study consisted of roughly equal numbers of AA and European American (EA) adolescents aged 14 to 18 years of both sexes recruited from high schools in the Augusta, Georgia area [ 10 ]. The replication cohort included 46 obese (BMI ≥ 30 kg/m 2 or BMI ≥ 95 th percentile for age and sex if age ≤ 18) and 46 lean (BMI ≤ 22 kg/m 2 or BMI ≤ 40 th percentile for age and sex if age ≤ 18) AA males selected from three cohorts, the Blood Pressure (BP) stress study (n = 603) [ 11 ], the Georgia Cardiovascular twin study (n = 1,183) [ 12 ] and the Prevention of Hypertension in African American Teens (PHAT) study (n = 262) [ 13 ]. Both the BP Stress study and the twin study are on-going longitudinal studies which have followed the subjects more than 10 years. Both studies included roughly equal numbers of AAs and EAs or males and females. The BP stress study was established in 1989 with subjects aged to 7 to 16 years at baseline and the twin study was established in 1996 with subjects aged 7 to 25 years at baseline [ 11 , 12 ]. The PHAT study was a cross-sectional study and consisted of AA males and females aged 14 to 20 years [ 13 ]. Subjects in all the three studies were also recruited from Augusta, GA area. The obese and lean subjects were identified based on the following criteria: (1) having leukocyte DNA available; (2) AA males; (3) only one twin subject from a twin pair was selected if both twins met the criteria; (4) if multiple visits (with multiple leukocyte DNA) were available for a subject, this subject had to be obese or lean on all the visits to be included in the replication sample and the leukocyte DNA collected at the visit when the subject were the most obese or most lean was used. For all four cohorts self identification by self-reports of each subject or by a parent if the subject was under 18 years of age was used to classify ethnicity according to previously described criteria [ 14 ]. Subjects in all the four studies were overtly healthy, free of any acute or chronic illness on the basis of parental reports and were not on anti-hypertensive, lipid lowering, anti-diabetic and anti-inflammatory medications [ 10 - 13 ]. The Institutional Review Board at the Medical College of Georgia approved the studies. Informed consent was obtained from all subjects and by parents if subjects were less than 18 years of age. Measurements For all the four cohorts, height and weight were measured by standard methods using a wall-mounted stadiometer and a scale, respectively. BMI was calculated as weight/height 2 . Systolic BP (SBP) and diastolic BP (DBP) were measured with Dinamap monitors, using an appropriately sized BP cuff placed on the subject's right arm. BP measurements were taken at 11, 13, and 15 minutes, during a 15-minute supine relaxation period. The average of the last two readings was used to represent SBP and DBP values [ 10 - 13 ]. Fasting peripheral blood samples in the LACHY cohort and non-fasting peripheral blood samples in the other three cohorts were collected. The buffy coat and plasma samples were separated and stored at -80°C. DNA was extracted from the buffy coat. In the LACHY cohort, fasting glucose levels were measured using Ektachem DT II system (Johnson and Johnson Clinical Diagnostics, Rochester, NY, USA) and fasting insulin was assayed in duplicate by specific radioimmunoassay (Linco Research, Inc., St Charles, MO, USA) [ 10 ]. QUICKI (quantitative insulin-sensitivity check index) was calculated to index insulin sensitivity using the following formula: 1/[log(fasting insulin, μU/ml) + log(fasting glucose, mg/dl)]. Out of the screening sample which included the seven obese cases and seven lean controls selected from the LACHY cohort, fasting glucose and insulin levels were not available for one case and one control. Genome wide methylation chip The HumanMethylation27 BeadChip from Illumina (Illumina, San Diego, CA, USA) was used. This chip can quantitatively measure 27,000 CpG sites, covering more than 14,000 well-annotated genes at single-CpG resolution. Each chip can accommodate 12 samples. After bisulfite treatment, 200 ng of the converted DNA was whole genome amplified (WGA) and enzymatically fragmented. The bisulfite-converted WGA-DNA samples were purified and applied to the BeadChips. Image processing and intensity data extraction were performed according to Illumina's instruction http://www.illumina.com/products/infinium_humanmethylation27_beadchip_kits.ilmn . Each methylation data point is represented by fluorescent signals from the methylated and unmethylated alleles. DNA methylation beta values are continuous variables between 0 (completely unmethylated) and 1 (completely methylated), representing the ratio of the intensity of the methylated bead type to the combined locus intensity. Initial array processing and quality control were performed with BeadStudio software. The microarray data discussed in this paper have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE25301 http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE25301 . Pyrosequencing The methylation levels of the selected six CpG sites from the six genes in the replication cohort were determined by pyrosequencing technology, a rapid and robust method for quantitative methylation analysis. After bisulfite treatment, 10 ng of the converted DNA was used in a PCR reaction to amplify the target region. One of the PCR primers was biotin labeled. Single-stranded biotinylated PCR products were prepared for sequencing by use of the Pyrosequencing Vacuum Prep Tool according to the manufacturer's instructions. The PCR products (each 10 μl) were sequenced by Pyrosequencing PSQ96 HS System (Pyrosequencing-Qiagen) following the manufacturer's instructions. The methylation status of each locus was analyzed individually as a T/C SNP using QCpG software (Biotage, Kungsgatan, Sweden). PCR primers and sequencing primers for these six genes are listed in Additional file 1 . All the samples were assayed in the same run and in a random sequence. Statistical analysis For the genome wide methylation analysis, the Limma package [ 15 ] was used to analyze each CpG site for differential methylation between obese and lean subjects. CpG sites on the × and Y chromosomes were excluded from the analysis. Each CpG site was assigned a raw P -value based on a moderated t statistic. To correct for multiple testing, the set of raw P -values were converted to false discovery rates (FDR) according to Benjamini and Hochberg [ 16 ]. For the six CpG sites in the replication cohort, a Student's t -test was used to investigate whether their methylation levels differ between the obese and the lean group. Linear regression was further used to adjust the potential effect of age on methylation levels. We combined the genome wide and the replication steps for these six CpG sites using the weighted z score-based meta-analysis approach implemented in the package METAL [ 17 ]. Prior to analysis, methylation levels of the CpG sites in the replication cohort were log-transformed or square root-transformed to obtain a better approximation of the normal distribution. Preliminary analyses, t-tests and regression analyses were done using STATA 8 (StataCorp, College Station, TX, USA). Gene Ontology analysis was conducted with the FatiGO tool [ 18 ]. FatiGO takes two lists of genes and converts them into two lists of GO terms. Then a Fisher's exact test for 2 × 2 contingency tables is used to check for significant over-representation of GO terms in one of the sets with respect to the other one. Multiple testing correction (indexed by adjusted P- values) to account for the multiple hypothesis tested (one for each GO term) is applied to reduce false positives. Since at least two CpG sites were included for the majority of genes in this genome wide chip, we selected the CpG sites with the lowest p value to represent this gene.

Results Table 1 displays the general characteristics of the screening sample. At this age range, obese subjects already showed significantly higher insulin resistance. Figure 1 is a volcano plot showing the raw P -values for all CpG sites versus mean methylation difference between the obese group and the lean group. We did not observe any CpG sites survive multiple testing correction with the most significant CpG site with a FDR of 0.14 and a raw P- value of 5 × 10 -6 . Table 2 lists the top 10 most significant CpG sites and one additional CpG site which showed the largest methylation difference between cases and controls (-27.1% for CREB3L3 ). Since at least two CpG sites were included for the majority of the genes in this genome wide methylation chip, the P -value for the second CpG site in this gene was also listed in Table 2 . Out of the 11 CpG sites listed in Table 2 (under the column "more significant CpG site"), we selected six CpG sites for validation in the replication cohort using the following inclusion criteria: (1) the most significant differentially methylated CpG site, which is the CpG site locating at -156 of the UBASH3A (ubiquitin-associated and SH3 domain-containing protein A) gene; or (2) the CpG sites in those genes of which the second CpG site also showed a raw P < 0.05. These include the CpG sites from the TRIM3 (Tripartite motif-containing 3), CTSZ (Cathepsin Z preproprotein), HIPK3 (Homeodomain interacting protein kinase 3), CDH5 (Cadherin 5 type 2 preproprotein) and CREB3L3 (cAMP responsive element binding protein 3-like 3) genes. The validation was conducted on the more significant CpG site (with Illumina ID 13517, 17029, 23241, 25599, 8829 and 23739) in each gene by pyrosequencing. The general characteristics of the replication cohort are displayed in Table 3 . Since the DNA from one obese subject failed on four pyrosequencing assays, the data from this subject was excluded from the analysis. Table 4 lists the results in the replication cohort. The findings from the genome wide methylation chip were validated for the CpG sites in the UBASH3A, TRIM3, HIPK3 and CDH5 genes. The methylation status of the CpG site in the UBASH3A gene was 3.3% higher ( P = 0.0019) and the methylation status of the CpG site in the TRIM3, HIPK3 and CDH5 genes was 1.21% ( P = 0.0004), 7.31% ( P = 0.0135) and 3.09% ( P = 0.045) lower in the obese versus the lean group. These significant findings persisted after adjustment of age for the association between obesity and methylation status of the CpG site in the UBASH3A gene ( P = 0.008) and TRIM3 gene ( P = 0.001), but changed to borderline significance for HIPK3 gene ( P = 0.05), and disappeared for the CpG site in the CDH5 gene ( P = 0.095). The results of the meta-analysis for these six CpG sites on the data from the genome wide step and the replication step are shown in Additional file 2 Although the pyrosequencing assay was designed to target one specific CpG site for each gene, some of the assays covered several surrounding CpG sites. The results of these CpG sites are also listed in Table 4 . The correlation within samples among the multiple CpG sites measured within a gene is listed in Additional file 3 . Gene Ontology analysis was performed to test whether some common functional trends in molecular functions and biological processes were associated with the genes exhibiting differences between obese cases and lean controls in the genome wide chip. We included those genes with a raw P ≤ 0.01 to the first list (n = 298) and included all the other genes in the second list. As expected from a pilot study in 14 subjects, we did not observe any GO categories survive multiple testing. Table 5 lists the GO categories with raw P -value less than 0.05. Interestingly, we observed enriched functional processes that are potentially relevant for inflammatory response with immune response (GO:0006955), cell activation (GO:0001775), cytokine production (GO:0001816), response to biotic stimulus (GO:0009607) and antigen binding (GO:0003823) among the top GO categories. The results support that obesity might lead to methylation changes in genes involved in inflammatory pathways.

Discussion In this study we aimed to identify obesity related methylation changes in peripheral blood leukocytes using a genome wide approach in youth and young adults who are free of obesity comorbidities. The primary findings of this study are increased methylation levels at one CpG site in the UBASH3A gene and decreased methylation level at one CpG site in TRIM3 gene in obese subjects compared with lean controls. The protein encoded by the UBASH3A gene is the "ubiquitin-associated and SH3 domain-containing protein A", which was previously also known as T-cell ubiquitin ligand (TULA) and suppressor of T-cell signaling 2 (Sts-2). It is expressed predominantly in T-cells, where it has a suppressing effect on T-cell signaling and activation. It acts in part by inhibiting c-CBL-mediated downregulation of protein tyrosine kinases (PTKs) that are activated upon T-cell receptor stimulation [ 19 ]. UBASH3A is also capable of promoting T-cell apoptosis [ 20 ]. UBASH3A and UBASH3B (another protein in this family) gene double knock-out mice were hyperresponsive to T-cell receptor stimulation with increased cytokine secretion, although mice lacking UBASH3A were normal in all respects including T-cell function [ 21 ]. However, whether the expression of the UBASH3A gene is regulated by DNA methylation is unknown. No CpG island exists in this gene and the CpG site showing higher methylation levels in obese subjects from this study locates at the promoter region with a distance of 156 bp to the transcription start site. There is a possibility that methylation of this CpG site or other CpG sites with methylation levels correlated with this CpG site inhibits the interactions between DNA sequence and nuclear proteins, resulting in decreased gene expression. In-silico analysis of the region of this CpG site using TFSEARCH software [ 22 ] did not find this CpG site located at any known transcription factor binding sites. However, methylation of this CpG site may suppress gene transcription by recruiting methylcytosine-binding proteins that in turn associate with large protein complexes containing corepressors and histone deacetylases. The binding of these complexes to DNA may lead to a change in the chromatin structure from an active to an inactive form [ 23 ]. This speculation needs to be confirmed. TRIM3 is one member of the TRIM protein family. These proteins share a conserved tripartite architecture and have a variety of cellular functions including cell proliferation, differentiation, oncogenesis, apoptosis, immune signaling and have been implicated in autoimmune diseases [ 24 ]. In peripheral blood leukocytes, TRIM3 is mainly expressed in macrophages and can be up-regulated by viral infection [ 25 , 26 ]. Similar to UBASH3A , the role of methylation in TRIM3 gene expression has not been reported yet. The TRIM3 gene has 1 CpG island located in the promoter and exon 1 (nucleotides -446 to 576), one CpG island spanning intron 2 and exon 3 (nucleotides 7759 to 8334), and one CpG island in exon 13 and the downstream region (nucleotides 24883 to 25610). The CpG site showing lower methylation levels in obese subjects in both the scanning cohort and the replication cohort locates in the promoter CpG island (348 bp upstream to the transcription start site). The methylation status of the other CpG site which showed significant differential methylation levels between obese cases and lean controls in the genome wide methylation analysis also locates in this CpG island (331 bp upstream to the transcription start site). Because the pyrosequencing assay also measured the methylation levels of the -331 CpG site (Illumina ID: 19207), we were able to show that also for this site obese subjects had lower methylation levels in comparisons with the lean subjects (age adjusted P = 0.022 as shown in Table 4 ). Similar to the CpG site in the UBASH3A gene, these two CpG sites are not located at any known transcription factor binding sites. Future studies will be needed to test whether the density of the methylation or the methylation of specific CpG sites in this CpG island has effect on TRIM3 gene expression. The observed DNA methylation differences between obese cases and controls were relatively small. They were 10.1% and 2.54% in the genome wide step and 3.3% and 1.2% in the replication step for the UBASH3A gene and the TRIM3 gene, respectively. This modest level of differences raises an important question: what is the biological significance of changes in methylation on this degree? In this study, we used the DNA from leukocytes, which represent different cell populations with distinct epigenetic profiles. It is plausible that only the change in the epigenetic profile of one specific cell type is related to obesity. In this regard, the actual epigenetic differences might be more substantial than reported here but only present in this specific blood leukocyte cell type. Future studies on epigenetic profiling of various types of cell populations in the leukocytes are warranted to gain a greater understanding of the epigenetic dysregulation in obesity. Furthermore, this kind of study will be able to identify the cells specifically reflecting obesity-associated methylation changes, which is of great interest in itself. Transcriptional profiling studies will be very valuable in understanding whether the DNA methylation status differences are associated with differences in gene expression. Unfortunately, cellular RNA is not available for the samples used in the current study, which were selected from several existing cohorts. There are several strengths of this study. First, we selected obese cases and controls with extreme phenotypes, which maximizes the power to make discoveries. Second, we focused on youth and young adults with the distinct advantage that the results are not confounded by obesity comorbidities or use of medication, both of which are very common in adult subjects with obesity. Third, a hypothesis-free genome wide approach was used. This approach supersedes the limitations imposed by candidate gene methylation studies and allows searching the whole genome in an unbiased manner. Interpretation of these data is also limited by several constraints. First, in this study we hypothesize that obesity will lead to methylation changes in the DNA of leukocytes, which further lead to obesity related co-morbidities. However, our study design cannot determine whether the identified methylation changes are the cause or the consequence of obesity. Future studies on subjects who changed their body size status will be needed to clarify the causal directions. On the other hand, epigenetic regulation is tissue specific. In this regard, the target tissue to identify epigenetic variations responsible for obesity should be the hypothalamus of the brain rather than peripheral leukocytes. Second, in the current study, the Infinium HumanMethylation27 Beadchip was selected because of its quantitative measure at each CpG site. However, the limited coverage of this genome wide chip will restrict the findings to certain CpG sites within certain genes. Future studies should use chips with more complete coverage of the genome such as the recently released 450K Infinium Methylation BeadChip from Illumina. Third, the current study is a small pilot study with the genome wide step conducted only in seven obese and seven lean subjects. Future studies with much larger sample size are warranted to discover a more complete profile of obesity related methylation changes. Fourth, although all the obese cases in the genome-wide step were free of clinical diseases such as CVD and diabetes, some of them already showed evidence of insulin resistance. It is possible that insulin resistance may be the factor triggering methylation changes and not the obesity per se . Recent literature supports the postulate that glucose restriction alters gene expression through epigenetic mechanisms [ 27 ]. We conducted further analysis within the obese cases and did not observe that any of the CpGs showed methylation difference between cases with impaired fasting glucose (n = 3) versus those with normal glucose levels (n = 3), with the most significant CpG site having an adjusted P- value of 0.85. However, the limited sample size prevents us from drawing any conclusion based on this analysis. Unfortunately, fasting glucose was not measured in the replication cohort. Future studies with large sample sizes and more detailed phenotypes will be needed to clarify this question.

Conclusion In this study, we identified several reproducible changes in DNA methylation of peripheral blood leukocytes between obese cases and lean controls. This study provides evidence that obesity is associated with methylation changes in blood leukocyte DNA. Further studies are warranted to determine the causal direction of this relationship as well as whether such methylation changes can lead to immune dysfunction. Such studies will have the ability to identify new insight into disease etiology and provide new targets for prevention of obesity related diseases such as cardiovascular diseases and type 2 diabetes.

Background Despite evidence linking obesity to impaired immune function, little is known about the specific mechanisms. Because of emerging evidence that immune responses are epigenetically regulated, we hypothesized that DNA methylation changes are involved in obesity induced immune dysfunction and aimed to identify these changes. Method We conducted a genome wide methylation analysis on seven obese cases and seven lean controls aged 14 to 18 years from extreme ends of the obesity distribution and performed further validation of six CpG sites from six genes in 46 obese cases and 46 lean controls aged 14 to 30 years. Results In comparison with the lean controls, we observed one CpG site in the UBASH3A gene showing higher methylation levels and one CpG site in the TRIM3 gene showing lower methylation levels in the obese cases in both the genome wide step ( P = 5 × 10 -6 and P = 2 × 10 -5 for the UBASH3A and the TRIM3 gene respectively) and the validation step ( P = 0.008 and P = 0.001 for the UBASH3A and the TRIM3 gene respectively). Conclusions Our results provide evidence that obesity is associated with methylation changes in blood leukocyte DNA. Further studies are warranted to determine the causal direction of this relationship as well as whether such methylation changes can lead to immune dysfunction. See commentary: http://www.biomedcentral.com/1741-7015/8/88/abstract

Abbreviations AA: African American; BMI: Body Mass Index; BP: Blood Pressure; CDH5: Cadherin 5 type 2 preproprotein; CREB3L3: cAMP responsive element binding protein 3-like 3; CTSZ: Cathepsin Z preproprotein; DBP: Diastolic Blood Pressure; EA: European American; FDR: False Discovery Rate; HIPK3: Homeodomain interacting protein kinase 3; LACHY: Lifestyle, Adiposity, and Cardiovascular Health in Youth; PHAT: Prevention of Hypertension in African American Teens; QUICKI: Quantitative Insulin-Sensitivity Check Index; SBP: Systolic Blood Pressure; TRIM3: Tripartite motif-containing 3; UBASH3A: Ubiquitin-associated and SH3 domain-containing protein A; WGA: Whole Genome Amplified. Competing interests The authors declare that they have no competing interests. Authors' contributions XW had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. XW and HS contributed to the study concept and design. XW, SS, HS, HS and DM analyzed and interpreted the data. HZ, YD, BG, FT and GH acquired the data. XW drafted the manuscript. XW, HS, GH and BLM critically revised the manuscript for important intellectual content. XW and SS did the statistical analysis. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1741-7015/8/87/prepub Supplementary Material

Acknowledgements The participants in this study were recruited by several NIH funded projects including HL69999, HL56622, HL077230 and HL64157. XW is also funded by the American Heart Association (0730156N) and the NIH (HL086530).

CC BY

no

2022-01-12 15:21:44

BMC Med. 2010 Dec 21; 8:87

oa_package/53/91/PMC3016263.tar.gz

PMC3016264

21143851

Conclusions CVF has great potential as a source for biomarkers for pathologies of the female genital tract. Many details about the complex physiology of the female genital tract, even under normal circumstances, are yet unknown and only few proteomics studies have been executed on this topic [ 39 , 46 , 47 ]. Nevertheless, CVF may yield a wealth of information about the "normal" functioning of the female genital tract and proteomic analysis may improve our understanding of micro-environmental changes due to the influence of multifaceted conditions such as menstrual cycle, age or pregnancy. Despite the progress made in the discovery phase (table 1 ), not many of the potential biomarkers find their way towards a clinical application. Only the AmniSure ® ROM test for the diagnosis of PPROM, based upon the detection of placental α-microglobulin-1, has been further validated in clinical trials and is FDA approved. The problem in the validation of biomarkers and the subsequent development of biomarker based diagnostic tests, often lies in the transition from discovery to validation phase. Frequently, too little effort has been made in setting up statistical relevant experiments and important aspects such as sample size, inter- and intraindividual and technical variabilities are often underestimated or not well defined. This inevitably leads to results which lack statistical power and to potential biomarkers which generally do not survive further validation [ 153 , 178 , 188 , 189 ]. In order to obtain statistically significant and practically useable highly specific and sensitive CVF biomarkers, several steps must be undertaken. 1) It is essential to analyze and characterize the CVF proteome precisely to be able to define the physiological environment in which experiments are conducted [ 18 ]. 2) Also, the inter- and intraindividual variability must be determined, which is critical in order to determine the number of samples needed to obtain statistically relevant results and to distinguish real biomarkers from natural occurring biophysiological variations. So far, although quantitative and comparative experiments were performed on CVF, not much attention has been given to the contribution of biological variation [ 16 , 24 ]. A good understanding of the effect of factors affecting the proteome composition (e.g. age, use of contraceptives, menstrual cycle, etc.) is therefore essential and should be taken into account during sample collection. Additionally, the choice of control groups in the experimental setup can have a drastic effect on the eventual results. For example, some markers (e.g. fibronectin, C-reactive protein, prolactin) for preterm birth are often found in preterm premature rupture of membranes (table 1 ). Nevertheless, markers for preterm birth which were found to be highly specific, may be in fact low specific if PPROM patients were included in the control group. Up to now, most studies using CVF as clinical samples, used antibody dependent techniques such as ELISA and Western blotting, although these are only capable to provide very limited amounts of information so that important data are inevitably missed. Nevertheless, many of these studies improved our understanding of different pathologies and conditions. However, the use of proteomic technology has the advantage to analyze many different proteins at the same time and can yield large amounts of information. We strongly believe that quantitative and comparative proteomic analysis of the CVF proteome will yield new complementary insights and information about the biophysiological nature of the female genital tract which may lead to further progression in our understanding of gynecological pathologies. All together, this may result in a better management of these diseases and thus, hopefully, will reduce associated morbidity and mortality for women, mothers and children.

Cervicovaginal fluid has an important function in the homeostasis and immunity of the lower female genital tract. Analysis of the cervicovaginal fluid proteome may therefore yield important information about the pathogenesis of numerous gynecological pathologies. Additionally, cervicovaginal fluid has great potential as a source of biomarkers for these conditions. This review provides a detailed discussion about the human cervicovaginal proteome and the proteomics studies performed to characterize this biological fluid. Furthermore, infection-correlated pathological conditions of the female genital tract are discussed for which cervicovaginal fluid has been used in order to identify potential biomarkers. Recent years, numerous studies have analyzed cervicovaginal fluid samples utilizing antibody-based technologies, such as ELISA or Western blotting, to identify biomarkers for preterm birth, premature preterm rupture of membranes, bacterial vaginosis and cervical cancer. The present article will discuss the importance of proteomic technologies as alternative techniques to gain additional meaningful information about these conditions. In addition, the review focuses on recent proteomic studies on cervicovaginal fluid samples for the identification of potential biomarkers. We conclude that the use of proteomic technology for analysis of human cervicovaginal fluid samples is promising and may lead to the discovery of new biomarkers which can improve disease prevention and therapy development.

Female genital tract physiology The female genital tract is characterized by a unique immunologic micro-environment. It is essential for human reproduction that the immune system of the female genital tract is modulated correctly as it needs to tolerate the presence of sperm and fertilized oocytes without starting an immune reaction. Nevertheless, a strong immune system in the genital tract is very important to protect the interior organs and the fetus or embryo against the large pathogenic stress [ 1 ]. This obligation of a dual immune system, one that tolerates and one that reacts efficiently, is a biological challenge. Failure to successfully accomplish this challenge can lead to specific gynecopathological conditions which eventually may result in severe complications. The lower female genital tract (vagina and ectocervix) is lined with a protective non-keratinized stratified squamous epithelium, whereas the higher genital tract (uterus, oviducts and ovaria) is lined with a columnar epithelium connected with tight junctions. The mucosa creates a physical barrier for invading microorganisms due to the production of a glycocalyx (i.e. a hydrophilic glycoprotein layer) which allows for hydration of the mucosa which hinders the adherence of bacteria on epithelial cells. Nevertheless, when bacteria are able to reach the mucosa and adhere to it, they will be removed by cell shedding [ 1 - 8 ]. The mucosa of the lower female genital tract is covered by numerous commensal bacteria which form an important protective factor. The normal vaginal bacterial flora consists predominantly of Lactobacillus spp. (e.g. Lactobacillus jenensii, Lactobacillus crispatus , Lactobacillus iners ) but (facultative) anaerobic species such as Gardnerella vaginalis are also present although in much lower concentrations. Glycogen from exfoliated cells is first metabolized to glucose by epithelial cells which is then used by the vaginal flora to produce lactic acid, thereby maintaining a low vaginal pH (3.5-4.7). Also, some Lactobacillus spp. (e.g. Lactobacillus crispatus, Lactobacillus rhamnosus and Lactobacillus acidophilus ) produce H 2 O 2 at concentrations lethal for exterior microorganisms. Additionally, commensals compete with nonresident bacteria for available nutrients and some bacteria produce broad-spectrum antimicrobial peptides (i.e. bacteriocins). Together, these factors exert a selective antimicrobial activity which inhibits the growth of nonresident bacteria without affecting the survival of the commensal flora [ 1 , 2 , 7 , 9 - 11 ]. The presence of adaptive immunity mechanisms in the female genital tract is known, although detailed information is scarce. Langherhans (dendritic) cells, which reside in the genital mucosa, and epithelial cells are able to present antigens to T lymphocytes and thus can elicit an adaptive immune response [ 3 , 12 ]. Also, plasma cells are localized nearby submucosal glands and secrete, after activation, immunoglobulins (primarily IgG and secretory IgA) into the cervicovaginal fluid (CVF; see below). These immunoglobulins can recruit and activate other immune cells, hinder bacterial adherence and opsonize pathogens. Additionally, mucosal-associated lymphoreticular tissue (MALT), which consists mainly of T-lymphocytes and monocytes/macrophages, is located in the lamina propria of the cervix [ 2 ]. Another vital element of the immune system of the female genital tract is CVF. CVF is made of (i) vulvar secretions from sebaceous, sweat, Bartholins and Skene glands, (ii) plasma transudate through the vaginal wall, (iii) exfoliated cells, (iv) bacterial products, (v) cervical mucus (vi) endometrial and oviductal fluids and (vii) secretions from vaginal immune cells. The latter three are influenced by sex steroid hormones e.g. during the menstrual cycle and pregnancy [ 13 , 14 ]. CVF consists mainly of water and contains many different factors such as cholesterol, lipids, mucin, carbohydrates, amino acids, proteins and inorganic ions. It covers the lower female genital tract and hydrates the mucosa, creating a physical barrier for microbial invasion. Many studies have used CVF as samples for the analysis of proteins and found correlations between different expression profiles and specific pathologies. It is therefore not surprising that during the last 5 years several comprehensive proteomics studies have been performed on CVF in order to catalogue the CVF proteome and to look for potential biomarkers. Cervicovaginal fluid as potential biomarker source Over the last 10 years, the potential of using body fluid for the identification of diagnostic or prognostic biomarkers has been recognized [ 15 ]. This is also the case for CVF, where the low cost and ease of sample collection, circumvention of the risk associated with biopsies and the possibility of observing high numbers of patients using multiple samples are clear advantages of this body fluid [ 16 - 18 ]. Four different methods are frequently applied for the collection of CVF: 1) cervicovaginal washings (i.e. the cervicovagina is rinsed with washing buffer after which the fluid is collected), 2) cervicovaginal swabs (i.e. swabs or brushes are applied to the mucosal wall and rotated to collect the CVF), 3) cervicovaginal wicks (i.e. wicks such as tampons, strips or sponges are inserted in the female genital tract which absorb cervicovaginal secretions) and 4) diaphragm-like devices (i.e. cups placed over the cervix which collect cervical fluid). Few studies have been performed on the reliability and validity of the CVF collection methods. Snowhite et al . [ 19 ] compared the wicking method with cervicovaginal washings and found that cervicovaginal washings are the superior sample collection method for cytokine assays in terms of reproducibility. Another research group demonstrated that cervicovaginal washings are better samples as compared to swabs for assessment of HIV1 RNA in cervicovaginal secretions [ 20 ]. Lastly, collection of samples using cups has the advantage that ready-to-use samples are obtained. Because the secretions are not absorbed (cfr. swabs and wicks), no sample recovery steps need to be undertaken and protein loss is minimized. In addition, the cervical secretions are not diluted in a washing buffer during sample collection (cfr. cervicovaginal washings). However, because the cups are placed over the cervix, this sample collection method has the disadvantage that only cervical secretions and no vaginal secretions are collected. These examples indicate that the sample collection method may influence results. Therefore, well defined and standardized sampling methods may help avoiding misinterpretation of results and decrease the technical variation of the experimental setup. So far, plasma/serum is the most analyzed body fluid compared to other body fluids such as cerebrospinal fluid, urine, saliva, bronchoalveolar lavage fluid, nipple aspirate fluid, tear fluid and amniotic fluid [ 18 ]. A great advantage of plasma/serum is that these samples contain a plethora of potential biomarkers since this body fluid is in contact with nearly all organ systems. Nevertheless, some important disadvantages are inherent to the use of plasma/serum. Due to the presence of plasma/serum in the entire body, biomarkers often lack discriminatory power for a particular organ system and thus are not specific enough. Also, since the volume of plasma/serum is substantial, potential biomarkers are strongly diluted, resulting in lowered sensitivity. Moreover, it is well known that serum and plasma protein concentrations have a large dynamic range which often implies extensive sample preparation and separation steps. All together, these disadvantages result in a large inter- and intravariability of plasma/serum samples [ 16 ]. Other body fluids such as cerebrospinal fluid, urine or CVF are less subjected to these restraints. For instance, it can be expected that biomarkers found in CVF, being a "proximal" fluid, show more specificity and sensitivity for gynecopathological conditions as compared to blood [ 21 ]. This is due to 1) the relatively small volume/organ ratio of CVF, thereby increasing the possibility that potential biomarkers are less diluted, thus increasing their sensitivity and 2) CVF is a body fluid that is specific for the female genital tract and hence the CVF proteome will be related to conditions of these organs [ 16 , 18 , 21 , 22 ]. CVF is therefore an interesting body fluid to use for the development of biomarkers for gynecological pathologies. One must keep in mind however that many factors may significantly influence the protein composition of the CVF. For example, secretions of the genital tract are notably influenced by varying levels and ratios of estrogens and progesterone during the menstrual cycle [ 23 ] resulting in variable amounts of protein yield (for instance, we measured between 0.58 mg and 13.37 mg in cervicovaginal washings). Therefore it is practically impossible to quantify the normal amount of total protein content in healthy CVF. Although one can expect some changes in the amount of protein content for pathological conditions, these are often unnoticeable due to the large normal fluctuations. Because of this variation, it is often difficult to distinguish real biomarkers from natural occurring biophysiological variations [ 16 , 24 ]. For example, analysis of microbicide induced cytokine deregulation is often hindered by the menstrual cycle [ 4 , 25 - 27 ], pathological conditions, sexual intercourse [ 28 ], age [ 29 ], and the use of contraceptives [ 30 - 33 ] or hormonal substitution therapies. If one disregards these effects, they will inevitably lead to biased or confounded results. The use of experimental designs of high-quality and well determined and clinically validated criteria are required to detect real product induced cytokine changes from background levels [ 34 ]. The use of body fluid as biomarker source implies the application of proteomic methodologies, because body fluids do not contain a corresponding genome or transcriptome and therefore gene expression cannot be analyzed [ 17 ]. Most studies on the identification of potential biomarkers are restricted to the use of antibody dependent techniques (e.g. ELISA or Western blotting). Thus far, only a limited number of large proteomic studies have been performed on CVF and the majority of these studies focused on the characterization of the CVF proteome. Only during the last few years quantitative methods were used to detect differences in protein profiles between particular conditions [ 14 , 22 , 35 - 45 ]. Proteomics techniques for the characterization of the human cervicovaginal proteome Generally, a proteomic biomarker study can be divided in several steps which are critical to obtain clinically and statistically significant results. (i) The characterization and analysis of the proteome, which also includes analysis of variation, is a first essential stage followed by (ii) differential analysis of the proteome, including statistical analysis to identify potential biomarkers (discovery phase). In a next phase (iii) the potential markers need to be validated on other independent cohorts (validation phase) after which (iv) the implementation possibilities of the validated markers in a clinical environment can be investigated [ 16 , 18 ]. Only the last 5 years, studies started to investigate the human CVF proteome to characterize the protein composition and to detect protein expression differences between specific patient conditions (table 1 ) using comprehensive antibody-independent proteomics techniques. Venkataraman et al . [ 38 ] published one of the first human CVF proteomics papers and demonstrated that human CVF may play a critical role in the host defense against HIV. The authors showed that CVF contained intrinsic anti-HIV activity from which the greater part was found in the cationic protein/peptide fraction. This fraction was analyzed using two-dimensional (native acid urea-PAGE followed by Tricine SDS-page) gel electrophoresis after which spots were identified using MALDI-TOF/TOF tandem mass spectrometry. The authors identified 18 polypeptides including known antimicrobial peptides/proteins such as calgranulin A/B, lysozyme, human neutrophil peptide (HNP) 1, cathepsin G and several histones [ 2 ]. However, none of these cationic polypeptides alone had potential anti-HIV activity. In addition, cationic fraction-depleted CVF did not contain anti-HIV activity but this activity was completely restored after adding the isolated cationic fraction. This indicates that the anti-HIV activity is almost entirely dependent on the CVF cationic fraction and probably is the result of a synergistic combination of different molecules. Also, it can be expected that, apart from the 18 proteins identified in this study, the cationic fraction from CVF is constituted of other proteins and peptides (such as elafin, see later) which could also play a role in the anti-HIV activity from CVF. Di Quinzio et al . [ 40 ] used full human CVF instead of a particular fraction for their proteomics analysis. Cervicovaginal swabs from five pregnant women were collected and separated using 2D (isoelectric focusing (IEF) followed by SDS-PAGE) gel electrophoresis. Subsequently, protein spots common to all five gels were tryptically digested and identified using MALDI-TOF mass spectrometry. Alternatively, the samples were first separated using capillary reversed phase (RP)-HPLC followed by electrospray ion trap tandem mass spectrometry. The authors identified 15 different proteins common to the five samples. These proteins function in different areas such as blood transport, structural integrity, fatty acid metabolism, calcium binding, inflammation, proteinase inhibition and oxidative stress defense [ 40 ]. The group further developed their proteomics platform to allow the analysis of differences in the CVF proteome caused by normal labour and pregnancy. In a first study, the authors compared the 12-29kDa CVF proteome fraction from pregnant women at 26-30 days before labour onset, 1-3 days before onset and during natural term labour. They identified seven different proteins (cystatin-A, interleukin-1 receptor antagonist, glutathione S-transferase P, peroxiredoxin-2, thioredoxin, copper-zinc superoxide dismutase, and epidermal fatty-acid binding protein) [ 39 ]. In a follow-up study, interleukin-1 receptor antagonist (IL-1RA), was validated and shown to decrease in concentration during the course of the pregnancy. Also, the concentrations of IL-1RA were 6-fold lower in case of preterm premature rupture of membranes (see below), which indicates that IL-1RA functions in the remodeling of the fetal membranes [ 46 ]. Furthermore, the group used the same techniques to analyze the 25-45 kDa human CVF proteome fraction and compared expression profiles between pregnant women at 14-17 days, 7-10 days and 0-3 days before natural labour and during labour. The authors found temporal changes in the expression of five proteins (serpin B3, serpin B1, annexin A3, collagen α2 type IV and albumin) which correlate with natural occurring term labour [ 47 ]. Recently, the same research group showed that 1) expression of α-enolase was significantly increased during spontaneous labour at term [ 48 ] and 2) parturition results in increased oxidative stress. This last phenomenon was reflected in the CVF since the total antioxidant capacity of CVF decreased drastically during labour, largely due to a decline in copper zinc superoxide dismutase and thioredoxin-4 CVF concentrations. The combination of total antioxidant capacity of CVF and copper zinc superoxide dismutase concentration showed high specificity and sensitivity for the prediction of labour within three days [ 49 ]. These studies are good examples of how proteomics may yield important information about complex biological phenomena such as pregnancy. Although the studies from Venkataraman et al . [ 38 ] and Di Quinzio et al . [ 40 ] were among the first to use proteomics techniques to mine the CVF proteome, their analyses were restricted to analysis of CVF proteome subfractions. In contrast, Dasari et al . [ 42 ] performed the first large comprehensive analysis of the CVF proteome. The authors used 2D (strong cation exchange (SCX) followed by RP)-HPLC or SDS-PAGE followed by RP-HPLC as sample separation methods. CVF proteins or peptides were identified using ESI-Q/TOF tandem mass spectrometry combined with several search engines. The study primarily aimed to identify CVF proteins and 150 different proteins were detected of which the majority functioned either as immune and defense-related or as metabolic proteins. The authors also compared serum and amniotic fluid protein sets with the proteins identified in the study. 77 proteins were unique to CVF, whereas 56 proteins were also found in serum and 17 in amniotic fluid. Additionally, they found that there is a significant degree of complementarity between the experimental setups as well as between the different search engines (Sequest, X!Tandem and OpenSea) used for the identification of CVF proteins. For example, only 38% of the spectra were identified by all three search engines, indicating that implementation of different search algorithms can result in a significant increase in number of identifications. Finally, the authors emphasized the need for biological and technical replicates in order to identify as many proteins as possible from the CVF proteome [ 42 ]. The same research group published the first comparative quantitative proteomics study on CVF. Pereira et al . [ 43 ] used 2D (IEF followed by SDS-PAGE) differential gel electrophoresis (2D-DIGE) to analyze differences in CVF protein expression levels between women with preterm-labour, spontaneous preterm birth patients and controls. Alternatively, they performed also 2D (SCX followed by RP) HPLC followed by ESI-Q/TOF to identify present proteins and to quantify protein expression semiquantitatively using spectral counting. During these experiments, the authors identified 205 proteins. Of those, 28 (2D-LC-MS/MS analysis) and 17 (2D-DIGE) proteins showed statistically significant changes between preterm labour patients, patients with spontaneous preterm birth and controls which can be considered as potential biomarkers [ 43 ]. This study is further discussed in more detail below. Tang et al . [ 45 ] also performed a characterization of the CVF proteome using 2D (IEF followed by SDS-PAGE) gel electrophoresis and MALDI-TOF/TOF analysis for the identification of CVF proteins. 59 different proteins were thus identified. The authors showed the presence of polymorphonuclear cells (e.g. neutrophils and eosinophils) in CVF and demonstrated that these immune cells could be activated and secreted proteins in the lower genital tract. Additionally, they showed that CVF contains a large fraction (47%) of proteins which are also found in plasma, which could be indicative for the permeability status of genital tract mucosa. Since the permeability is often increased in case of cervicovaginal infection, these plasma proteins could eventually be used together with secreted proteins from activated neutrophils and eosinophils (e.g. lactoferrin, myeloperoxidase, human neutrophil lipocalin and eosinophilic cationic protein) as markers for cervicovaginal inflammation [ 45 ]. The proteomics study which resulted in the largest set of protein identifications was performed by Shaw et al . [ 44 ]. The authors were able to identify 685 proteins from human CVF by means of 2D (SCX followed by RP)-HPLC and SDS-PAGE followed by RP-HPLC. Tandem mass spectrometric analysis was executed using an ESI-linear ion trap after which proteins were identified and functionally classified. Due to the complementary nature of the techniques and the incorporation of several technical replicates the authors were able to identify a large number of CVF proteins. Also, a less stringent identification methodology was used in this study, which may also account for this high number. Additionally, the authors focused on the kallikrein family (i.e. family of 15 secreted serine proteases with tryptic or chymotryptic activity) and confirmed the presence of different kallikreins (kallikrein 6, 7, 10, 11, 12 and 13) in CVF by ELISA [ 44 ]. Further research on this topic by the same authors revealed that kallikreins may play a physiologically active role in the desquamation of vaginal epithelial cells and activation of cervicovaginal antimicrobial proteins. The authors also hypothesized that cervicovaginal kallikrein expression may result in pathological conditions (e.g. PPROM) [ 50 , 51 ]. Klein et al . [ 14 ] performed proteome analyses on CVF samples from pregnant women who had symptoms of pending preterm birth. The authors identified 40, mainly high abundant, proteins using a shotgun proteomics approach using nanoscale RP-HPLC coupled to an ESI-ion trap mass spectrometer and showed that this method yields reproducible and reliable results. Additionally, they analyzed CVF using 2D-PAGE and demonstrated that the 2D-PAGE profiles are very similar for different samples, but that the spot intensities showed a certain degree of intersample variation. Our group [ 22 ] also used proteomics technology to characterize the CVF proteome. During a colposcopy, the cervicovagina is rinsed with a 5% acetic acid solution in order to make HPV infected lesions visible. Normally, this washing fluid is discarded but was used by our group to perform CVF proteomics studies. Using C 4 -RP-HPLC fractionation on the protein level followed by C 18 -RP-HPLC on the peptide level, and MALDI-TOF/TOF mass spectrometry, we were able to identify 339 proteins. During these experiments, we also determined the technical variability of the platform. Three technical replicates were analyzed and 68% of the different proteins identified, were detected in all three replicates. Analysis of the chromatographic separations showed a high degree of reproducibility (coefficient of variation was 0.56%). This led to the conclusion that the largest contribution to the technical variation is made by the mass spectrometric analysis. Also, using an in-house build relational database, we compared the largest proteomics studies on CVF and determined a set of proteins which are identified in the vast majority of the studies, independent of patient physiology or used analytical methods. This overlapping protein set was the first step towards the delineation of the CVF core proteome (see below) [ 22 ]. Finally, two studies have been performed on cervical mucus [ 36 , 37 ]. Because cervical mucus is part of the CVF (see above), these studies can be seen as an analysis of a subfraction of the CVF proteome. Andersch-Björkamn et al . [ 36 ] used 1D-PAGE and SDS-agarose composite gel electrophoresis to separate proteins and subsequently performed RP-nano-HPLC peptide separations. Peptides were identified using ESI-FTICR tandem mass spectrometry. The authors identified 178 proteins from cervical mucus. Furthermore, they analyzed protein and mucin composition and the mucin O -glycosylation of the cervical mucus at different time points during the menstrual cycle. The results showed that cervical mucus was similar before and after ovulation, but differed at the time of ovulation. This observation was also seen in alterations of the O -glycosylation of the mucus. The physiological significance of these changes is not yet understood [ 36 ]. Panicker et al . [ 37 ] recently performed a study on the human cervical mucus proteome. The authors used 2D-PAGE and 1D-PAGE followed by nanocapillary RP-HPLC. Tandem mass spectrometry was performed using an ESI-Q-TOF to characterize the cervical mucus proteome. Immunodepletion of albumin and immunoglobulins did not result in a significant improvement. Although the authors noticed an overlap between the different experimental setups, each platform yielded a certain amount of unique proteins, indicating the importance of using different techniques in order to maximize proteome coverage. The authors also characterized the phosphorylation and glycosylation of cervical mucus proteins and found 14 proteins which were posttranslationally modified [ 37 ]. The studies described above primarily aimed to catalogue the CVF proteome. Therefore, in what follows, some upcoming facts of the CVF proteome will be reviewed in more detail. Beside these studies, four proteomic studies on CVF have been performed which focused on the detection of differences in protein expression profiles due to HIV resistance and pending preterm birth [ 35 , 41 , 52 , 53 ]. These quantitative studies are discussed further below. The cervicovaginal proteome: numbers and facts 852 different proteins were identified in diverse antibody-independent proteomics studies on human CVF and cervical mucus (listed in additional file 1 ) [ 14 , 22 , 36 - 38 , 40 , 42 - 45 ]. Also, during our latest experiments we were able to identify an additional 256 CVF proteins (unpublished data), making it a total of 1108 different CVF proteins which are identified thus far (listed in additional file 2 ). 252 and 257 proteins were also found in respectively plasma [ 54 , 55 ] and amniotic fluid [ 56 - 58 ] (shown in additional file 2 ). 733 proteins from the CVF proteome were not identified in plasma or amniotic fluid. Additionally, 223 proteins were identified in cervical mucus [ 36 , 37 ]. Of those, 51 proteins were uniquely identified in cervical mucus (additional file 2 ) while the greater part (172 proteins) was also identified in studies on CVF [ 14 , 22 , 38 , 40 , 42 - 45 ]. This was to be expected because cervical mucus is a subfraction of CVF. This suggests that analysis of subfractions of the CVF proteome may still yield new identifications [ 22 ]. Functional classification (Figure 1A ) of the CVF proteome shows a great diversity of biological roles of which "protein metabolism and modification" and "immunity and defense" are the largest categories (respectively 17% and 10%). Classification based upon cellular localization (Figure 1B ) shows that most proteins are present in the cytoplasm or in the extracellular region (respectively 21% and 20%) Figure 2 gives a schematic overview of the proteome structure of the CVF. Although exceptions and deviations from this generalized scheme are most likely, it may help in explaining the difficulties encountered during CVF biomarker discovery. Roughly, a proteome consists of a plethora of proteins which are present in either higher concentrations (hence easier to identify) or lower concentrations (usually difficult to identify). The difference between the highest and the lowest protein concentration in a proteome is described as the dynamic range, which is specific for a particular sample and important to take into account when performing comprehensive proteomics studies. Indeed, the higher the dynamic range of the sample, the more difficult it is to describe its proteome as complete as possible. Implementation of different separation dimensions, optimization of the techniques and sample preparation may increase the sensitivity and resolving power of the proteomics platform, thus allowing the detection of more proteins. Because CVF is an extracellular fluid with an important role in the innate immunity, it can be expected that physiological relevant proteins are extracellular, immunological molecules. We call these "characteristic" proteins since, based on their function and localization, they represent the fundamental characteristics of CVF. Both high and low abundant CVF proteins may fulfill a physiological function in the cervicovagina. For example, we found that the most abundant protein in CVF is S100A9 [ 22 ] which forms with S100A8 the calprotectin heterodimer. Calprotectin is a potent cervicovaginal antimicrobial peptide and inhibits bacterial growth by sequestration of zinc. Its importance is also reflected in the numerous physiological and pathological processes (e.g. cervical cancer, pregnancy and labour) for which correlations have been found with differences in calprotectin expression levels [ 59 ]. Examples of low abundant physiological relevant proteins are β-defensins or histones. In response to infectious stimuli, histones can create neutrophil extracellular traps (NET) (i.e. protrusions made of a DNA backbone in which histones, proteases and antimicrobial peptides/proteins are embedded) which are efficient in trapping and eliminating pathogens [ 60 - 62 ]. However, a substantial fraction of high and low abundant proteins may also originate from aberrant processes. For instance, sampling of CVF may cause disruption of the epithelial cell layer (e.g. when using brushes or tampons) and the continuous shedding of the epithelial cell layer may also bring in dead cells. These processes introduce many intracellular proteins (e.g. metabolic or cytoskeletal proteins) in the CVF which have no significant biological function in this body fluid. We therefore call these proteins "non-characteristic" since their function and localization do not match the CVF related characteristics. In general, it can be assumed that due to the extracellular nature of the CVF and its importance in the innate immunity, the characteristic proteins will contain larger amounts of extracellular and immunological fractions as compared to the non-characteristic proteins. Another part of the CVF is made of non-characteristic low abundant proteins of which the concentration shows large intra- and interindividual variations. These concern often intracellular proteins which take part in metabolic processes (e.g. glycolysis). Although the higher abundant characteristic proteins may contain very interesting biomarkers, potential biomarkers usually reside in the lower abundant part of the proteome. Therefore, one of the main challenges is the isolation of the biomarker from the numerous surrounding background proteins. How can these two be distinguished? Since background proteins are usually sporadically and randomly detected, they show no correlation with a particular physiological or pathological condition. In contrast, there must be a clear and significant association between alterations in the concentration of a biomarker and a specific condition or pathology. Taking all of the above into account, the CVF proteome can roughly be divided into two subfractions: the core and the variable proteome. The core proteome consists of proteins which are identified in most proteomics studies and therefore mainly represents high abundant (characteristic and non-characteristic) proteins, although characteristic low abundant proteins are also present. Hence, we expect the fraction of characteristic proteins to be high in the core proteome. We recently described a set of 136 proteins which are frequently identified, independent of patient physiology or used analytical methods. This subproteome has an increased extracellular immunological protein fraction [ 22 ], indicating that it is more relevant for CVF (as compared to the complete set of overlapping and non-overlapping proteins). It can therefore be considered as a first version of the core proteome. In contrast, the variable proteome contains proteins which are infrequently identified and often represent non-characteristic low abundant proteins. Therefore, the extracellular and immunological fraction is decreased in this part. This subdivision may explain some dissimilarities seen between the different proteomics studies on CVF. The fractions of immunological and extracellular proteins are lower in the 'larger' studies of Shaw et al . [ 44 ] and Zegels et al . [ 22 ] (respectively 11.72% and 12.57%) as compared to the 'smaller' studies from Pereira et al . [ 43 ] and Dasari et al . [ 42 ] (respectively 18.95% and 19.61%). The same holds for the extracellular fraction [ 22 ]. In a recent experiment, we identified 706 proteins from which 267 proteins were new identifications (unpublished results). Only 5.33% of these proteins fell into the "immunity and defense" functional category, which is much lower than was found in any other study (see above). A general trend is noticeable in proteomic analysis of CVF towards the identification of increasing quantities of different proteins using comprehensive techniques. This tendency was facilitated by the drastic technological evolution which proteomics has undergone the last 5 to 10 years. The development of sensitive, high resolution mass spectrometers with fast scan rates allows the coverage of dynamic ranges up to six orders of magnitude. The dynamic range of the experimental setup was further enhanced by improving HPLC separation prior to mass spectrometric analysis (e.g. increased sensitivity and reproducibility and use of columns with higher peak capacity). Moreover, progress has been made in multidimensional separations in order to further increase the peak capacity. All together, the technological developments in proteomics resulted in analytical platforms which can now cover larger parts of the proteome [ 63 ]. This is also reflected in the studies on the CVF proteome where increasing numbers of proteins were identified with the contemporary techniques (Figure 3 ). It is the common opinion that by digging deeper into the CVF proteome, more potential biomarkers could be isolated and multi-protein expression profiles can thus be analyzed. With this reasoning however, one must take into account some inherent disadvantages of large proteomic studies. In Figure 3 , the total number of proteins identified and the numbers of immunological or extracellular proteins (which are characteristic for the CVF proteome) in the different studies are plotted. From this, it is clear that while the total number of identified proteins increases exponentially, the immunological and extracellular fraction only increases linearly. This graph indicates therefore that the majority of newly obtained identifications have no known function or their function in the CVF is questionable or not biologically relevant. We therefore hypothesize that although large number of proteins are identified in the recent sensitive large proteomics studies, the vast majority of newly identified proteins are most likely part of the variable proteome which consist mainly of non-characteristic proteins. Thus by digging deeper into the CVF proteome, the challenge of identifying specific biomarkers among this large amount of non-characteristic proteins increases significantly. Optimization of sample collection methods to minimize cell lysis may be a step forward to reduce the amount of non-characteristic proteins. On the other hand, epithelial cell shedding is an essential process in the removal of pathogens and many other factors (e.g. hormonal regulation, non-oral contraceptives) will always cause cell disruption and lysis, prior to sample collection. Cervicovaginal fluid biomarkers with potential clinical applications The lower genital tract is exposed to a large microbial pressure and forms an important entry gate for a wide variety of pathogens such as Candida spp., Trichomonas vaginalis , human immunodeficiency virus (HIV) and human papillomavirus (HPV). In contrast, the upper female genital tract is believed to be sterile under normal circumstances. Nevertheless, pathogens such as Neisseria gonorrhoeae and Chlamydia trachomatis can occasionally ascend from the lower to the higher genital tract and can cause pathogenic conditions such as pelvic inflammatory disease (PID). The clinical impact of these infections may not be underestimated since they can lead to serious conditions such as preterm birth, increased susceptibility to sexually transmitted infections (STI), infertility and cancer [ 1 , 2 ]. Therefore, early diagnosis is essential for the successful treatment of the underlying pathology and to prevent irreversible complications. For this reason, there is a high demand for new diagnostic tools, such as biomarkers, which allow for better and earlier treatment of gynecological pathologies [ 17 ]. In the following paragraphs some important infection-correlated pathologies are discussed for which potential biomarkers were discovered using human CVF samples. Table 2 gives an overview of these potential biomarkers. Human papillomavirus infection and cervical cancer HPV is one of the most common sexually transmitted infections. It is estimated that 70-80% of the sexually active female population will acquire HPV before the age of 50 [ 64 ] and 80% of the HPV-infected patients will spontaneously clear the virus within 2 years [ 65 , 66 ]. HPV is associated with high morbidity and mortality worldwide, since it can cause precancerous lesions and cervical cancer. The latter is the second most observed cancer in women and is caused for nearly 100% by HPV-infection [ 67 ]. There are more than 150 HPV types and almost 40 of them infect the anogenital region. HPV can be divided in low and high risk HPV. The most common low-risk types are type 6 and 11. The most common high-risk types are 16, 18, 45, 31 and 33). Cervical cancer is mainly caused by an infection with a high risk type [ 68 ]. Infection of squamous or glandular epithelial cells with HPV causes morphological alterations which lead to the formation of squamous or glandular intra-epithelial lesions. The severity of this precancerous condition depends on the extent, the site as well as on the depth of the infected epithelial layer [ 67 , 69 ]. Because of the high prevalence, the severe consequences and the lack of a good treatment of advanced cervical carcinoma, the emphasis of managing this condition lies on prevention and diagnosis in a stadium as early as possible. The recently introduced prophylactic HPV vaccination will prevent about 98% of infections with HPV 16 and 18 but will also in part, due to so called cross protection, prevent HPV 31 (89%), HPV 33 (82%), HPV 45 (100%) and any high risk HPV type 16/18/31/33/35/39/45/51/52/56/58/59/66/68 (70%). As such, the HPV vaccination will prevent about 70 - 80% of all cervical cancers [ 70 , 71 ]. Recently, clinical trials started with a nine-valent HPV vaccine which could potentially prevent 90% of all cervical cancers but information on this topic is still scarce. Therefore, it is still needed to screen women between the ages of 25 and 65 on a frequent basis by means of a Pap test (liquid based cytology) preferable with a type specific HPV testing. Screening using the classical Pap test has led to a considerable reduction of the mortality, yet there are still some notable shortcomings. The current screening is (1) labour intensive, (2) difficult to automatize, (3) vulnerable to inter- and intra-observer variability and (4) is not very sensitive [ 65 , 72 , 73 ]. Moreover research has demonstrated that despite intensive screening campaigns, the coverage in most Western countries stays low (e.g. 59% in Belgium) [ 74 ]. Testing for the presence of HPV DNA or RNA will increase the sensitivity, but it cannot predict progression to intra-epithelial lesions and eventually cervical cancer. Taking all the above into account, this implicates that alternative screening methods (e.g. based upon protein biomarkers present in CVF) with both a higher sensitivity and specificity are needed. The identification of a highly sensitive and specific protein biomarker from CVF would mean a substantial improvement both on medical and social-economical field because an accurate biomarker for the detection of cervical neoplasia will result undoubtedly in a more precise and efficient screening which correlates with a further reduction of precancerous morbidity and cervical cancer morbidity and mortality [ 75 ]. Additionally, proteomic analysis of CVF could lead to biomarkers for the evaluation of patients after treatment [ 76 ] or to prognostic biomarkers which could predict the evolution from precancerous low-grade squamous intra-epithelial lesions (LSIL) to either high-grade squamous intra-epithelial lesions (HSIL) or clearance of the lesions. These prognostic biomarkers would benefit the screening method as the number of follow-up patients could be reduced significantly (up to 80%). Also, identification of proteins involved in the clearance of the virus could be of major importance for the future development of new antiviral therapies. However, only few studies are performed on human CVF for the identification of such biomarkers as most large proteomic studies were restricted to cervical cancer tissue [ 77 - 82 ]. Studies which used CVF as samples for their analysis did not use antibody-independent proteomics techniques, but mainly ELISA or Western blotting, and were therefore restricted to the analysis of a small number of proteins taking part in the immune response against HPV. In table 1 , several proteins are listed which are associated with HPV infection, cervical cancer and cancer prognosis and progression. Nevertheless, these proteins are primarily correlated with the prognosis of the disease but do not allow for a sensitive and/or specific prediction of healthy women which are at high risk. It is our opinion that such potential biomarkers have the highest chance to be discovered in a proteome wide approach, using proteomics techniques on a large number of samples. HIV-resistance More than 25 million people have already died because of HIV since the discovery of the virus in 1981, making it one of the most disastrous epidemics in human history. In 2008, 33.4 million people were living with HIV and 2.7 million people got infected by the virus [ 83 , 84 ], mainly due to heterosexual HIV transmission [ 84 - 86 ]. One of the largest problems remains the absence of an effective therapy. Also, despite the enormous efforts made, the development of a prophylactic AIDS-vaccine will likely not be available in the recent years to come [ 87 ]. This leads to an increased importance for microbicides (i.e. chemical entities that can prevent or reduce transmission infections) which can be applied locally (e.g. vaginally or rectally) as a potential alternative approach in HIV prevention [ 9 , 88 , 89 ]. However, the efficacy trials often gave rise to unsatisfactory results like enhanced HIV-transmission, toxicity and immune activation [ 41 , 90 ]. To overcome these problems, new powerful microbicides need to be identified which preferentially occur naturally and act under these in vivo circumstances against sexually transmitted HIV infections [ 41 ]. Analysis of samples obtained from exposed seronegative individuals (ESNs) may yield such new, therapeutically relevant microbicides. ESNs are individuals (<5% of the population) from high-risk cohorts that remain IgG-seronegative despite frequent HIV-exposure [ 91 ] and therefore show a certain degree of in vivo HIV-resistance. ESNs are found between commercial sex workers, hemophiliacs receiving HIV-contaminated blood, healthcare workers, children from HIV-infected mothers, intravenous drug users or they are the seronegative partner in a discordant couple [ 92 - 94 ]. Some biological factors are already associated with the ESN-status. Those include innate host factors like mutations of chemokine receptors (e.g. CCR5-Δ32), the upregulation of chemokines such as RANTES, MIP-1α and -1β due to genetic polymorphisms, specific human leukocyte antigen haplotypes (e.g. HLA-B*27 and HLA-B*57), natural killer cell activity regulated by killer Ig like receptor/HLA interaction, the presence of autoantibodies and/or alloantibodies and more efficient immunological effects after toll like receptor stimulation. Also, particular properties of the acquired immune system of the host show correlation with the ESN-status such as cytotoxic and helper T lymphocyte responses against HIV epitopes, humoral immune responses with the production of neutralizing anti-HIV antibodies and the presence of soluble inhibitory factors [ 92 - 99 ]. Generally, none of these associations alone can be held fully responsible for the observed HIV-resistance, and many unknown factors are yet to be discovered [ 41 ]. Current studies speculate that ESNs inhibit HIV-infection in a very early phase of viral transmission [ 100 , 101 ]. HIV-resistance may for example result from mechanisms present at the viral entry gate, which often is the mucosa of the female genital tract. Among these mechanism, proteins or peptides present in CVF of ESNs may play an important role [ 41 ]. Several studies on female ESNs have used CVF as choice of sample for analysis. It has been described that cervicovaginal lavages from ESNs sometimes contain gp41 and p24 HIV antigens, suggesting that there was an HIV exposure, but no seroconversion could be detected [ 102 ]. Anti-HIV IgA and IgG antibodies were detected in CVF obtained from heterosexual ESN women [ 103 - 113 ]. Additionally, it has been shown that levels of the HIV suppressive β-chemokine RANTES levels were significantly higher in CVF from ESNs [ 100 , 114 ]. Moreover, CVF contains proteins/peptides with intrinsic anti-HIV activity such as defensins, lactoferrin, lysozyme, cathelicidin and SLPI [ 22 , 92 , 115 , 116 ]. The proteomic study from Venkataraman et al . on CVF showed that the cationic fraction of CVF has natural anti-HIV activity. The authors speculated that this activity is the result from a complex synergism between different proteins in CVF [ 38 ]. It is possible that differences in protein expression patterns of those anti-HIV proteins/peptides may result in the observed HIV-resistance of some ESN cohorts. However, conventional techniques such as ELISA or Western blotting are incapable to analyze complex differential patterns from multiple proteins. Therefore, proteomics studies may aid in the elucidation of these factors. Indeed, recently, Burgener et al . [ 35 ] studied HIV-resistance in ESN persons and used 2D-DIGE as a quantitative proteomics technique to compare protein expression patterns from ESNs with those from control groups. The authors identified 16 differentially expressed proteins with different biological functionalities such as protease inhibitors (e.g. serpin B3/B4/B13/B1 and cystatin A) and proteins with an immunological role (e.g. protein S100A7 and complement component 3). Iqbal et al . [ 41 ] used surface enhanced laser desorption ionization (SELDI)-TOF mass spectrometry for the analysis of the CVF proteome of a large population of ESNs and control groups. The authors demonstrated that elafin/trappin-2 is significantly upregulated in ESNs [ 41 ]. Elafin/trappin-2 is an antiprotease from the whey acidic protein (WAP) family, of which SLPI is also a member, that can be found in CVF (see above). It functions as an anti-inflammatory protein by counteracting proteases secreted by neutrophils. Due to the cationic nature of elafin/trappin-2 it can destabilize bacterial cell walls or viral envelopes, thus exerting a certain degree of antimicrobial activity [ 117 ]. At about the same time, elafin/trappin-2 was identified as a new anti-HIV factor which directly interacted with the virus for its inhibition, thus making it an interesting protein for further investigation [ 118 ]. Hence, both studies confirm each other, thereby increasing the possibility that elafin/trappin-2 effectively plays a significant role in HIV-resistance. Surprisingly, this protein was not among the differentially expressed proteins in the study of Burgener et al . [ 35 ] where a different technique was used (2D-PAGE vs. SELDI). These results indicate once again that different proteomics techniques complement each other and are necessary to uncover further other CVF proteins correlating with in vivo HIV-resistance. Moreover the use of samples obtained from dissimilar ethnological cohorts may yield different results as it becomes more evident that several different mechanisms may lead to HIV-resistance. Bacterial vaginosis Bacterial vaginosis (BV) is a frequently observed condition by which the normal vaginal flora is disrupted. The prevalence of this pathology in Europe is 5-30%. Other geographical regions, especially those with a lower socio-economic status like the sub-Saharan part of Africa, show higher prevalences (over 50%) [ 119 - 121 ]. BV is a non-inflammatory condition, although it is often classified as a vaginitis, characterized by the loss of hydrogen peroxide forming Lactobacillus spp. (e.g. L. crispatus, L. acidophilus, L. rhamnosus) and by an increased growth of anaerobic species (e.g. Atopobium vaginae, Mobiluncus spp., Prevotella spp.). Also, whereas a healthy vaginal flora contains only few different and predominantly Lactobacillus species, the vaginal flora correlated with BV is polymicrobial in nature [ 119 ]. The question remains whether the loss of the Lactobacillus spp. is due to the introduction of a yet unknown factor which results in the passive growth of the anaerobic species or, vice versa, whether there is a massive influx of anaerobic species which eradicate the lactobacilli [ 122 , 123 ]. Although BV is not classified as a sexually transmitted infection, the condition has several features indicating that it may be caused by an exogenous source [ 124 ]. Nevertheless, the etiology of BV is not yet well understood. BV can lead to severe health problems and early diagnosis can prevent serious complications. A significant correlation exists between the presence of BV and preterm birth and it has been shown that treatment of BV may reduce the risk on preterm birth [ 119 , 120 , 124 , 125 ]. BV has also been linked with PID, miscarriage [ 124 ] and infertility [ 126 ] increased susceptibility for HIV and HPV [ 124 , 127 - 132 ], herpes simplex virus type 2 [ 133 - 135 ] and other lower genital tract pathogens [ 120 ]. The mechanism by which BV leads to those pathological conditions is not well known. It has been suggested that increased susceptibility to HIV and genital tract infections may be attributed to the production of succinate by anaerobes, which can inhibit neutrophilic function [ 136 ]. Also, some anaerobes produce sialydases which effect immune cells (e.g. inhibition of phagocytes) [ 137 ] and have mucinase activity [ 138 ]. Additionally, Ureaplasma urealyticum can produce proteases which can cleave IgA [ 139 ]. It has also been shown that CVF from some seropositive and seronegative women with BV has the capability to enhance HIV replication in vitro . This activity has been attributed to a certain soluble and protease sensitive HIV-inducing factor (HIF) which has not been identified yet. The general opinion hypothesizes that HIF is a mixture of products derived from BV associated bacteria on the one hand and host factors such as certain cytokines (e.g. myeloid-related protein 8) on the other hand [ 140 , 141 ]. No large proteomics studies have been performed on CVF from women with BV, but several ELISA experiments were executed. The results show that BV CVF samples are deficient in antimicrobial activity due to lower expression levels of several antimicrobial peptides such as SLPI, defensins, lysozyme and lactoferrin [ 142 - 147 ]. Furthermore, many studies also detected differences in expression levels of some proinflammatory cytokines correlated with BV (see table 1 ). These studies suggest that BV leads to the disruption of the normal innate and adaptive immune response and induces an inflammatory environment in the lower genital tract which can lead to increased susceptibility for HIV and other genital tract infections [ 120 , 128 , 148 ]. However, since only antibody dependent studies were performed, analysis was limited to a few well characterized proteins. Nevertheless, little is known about expression differences of multiple proteins, which may include new or unexpected ones. Differential analysis of the CVF proteome using proteomics approaches, may yield more information about the perturbation of the female genital tract immunity due to BV and further clarify the etiology of the disease. Preterm Birth As mentioned before, conditions such as bacterial vaginosis and other infections of the female genital tract are the main cause of preterm birth. Other risk factors include multiple pregnancy or uterine contractions of cervical incompetence. Nevertheless, a significant part of the observed preterm births are idiopathic, indicating that the etiology of preterm birth is still not well understood [ 149 - 151 ]. Preterm birth, defined by the WHO as birth occurring before 37 completed weeks of gestation, is a major cause of neonatal morbidity and mortality [ 152 , 153 ]. Recent reports estimate the incidence of preterm birth 9.6% worldwide, with the highest rates in Africa and North America (respectively 11.9% and 10.6%) and the lowest in Europe (6.2%). The impact of preterm birth on the health of the neonate can be very severe on short-term (e.g. respiratory stress syndrome, brain injury) as well as on long-term (e.g. chronic lung disease, sensory deficits, learning disabilities) thus affecting the child's development and the physical and psychological health of the infant's environment [ 153 - 155 ]. Preterm birth is difficult to treat, therapy is often inefficient because the underlying cause is seldom determined and the diagnosis is frequently made too late [ 154 , 156 - 159 ]. Due to the clinical importance of preterm birth, prevention strategies need more attention in the management of this condition [ 157 , 160 ]. Currently this is achieved by assessing the risk of the occurrence of preterm birth using several parameters or markers. History of preceding preterm birth is the most important risk factor and may relate to cervical anatomical problems. In addition, presence of any of the causal factors mentioned before drastically increases the chance on preterm birth [ 149 , 161 ]. CVF fetal fibronectin (FFN) concentrations are also correlated with the risk for preterm labour/birth, but quantification of CVF FFN alone has not enough discriminatory power [ 162 - 168 ]. Combination of this test with cervical ultrasound measurements however showed better correlations [ 166 , 169 - 172 ]. Several other markers have been correlated with the risk on preterm birth/labour, all of which are summed in table 1 . Nevertheless, low specificity and precision are inherent to these primary predictors. Therefore, new markers need to be identified which allow for more precise and specific risk assessment [ 149 , 157 ]. Besides, these markers may give additional information about the etiology and/or pathophysiology of this condition which can be used in the development of alternative therapies [ 43 ]. As mentioned before, Pereira et al . [ 43 ] identified potential biomarkers for preterm birth/labour in a proteomics study. These included fibronectin, several S100 proteins (calgranulin A, B and C), acute-phase reaction proteins (annexin A3, α-1-antitrypsin and α-1-acid glycoprotein) and proteins involved in cellular organization and motility (e.g. profilin-1, thymosin β-4, rho GDI 2). Recently, the same research group looked for biomarkers for intra-amniotic infection, which is an important risk factor for preterm birth and may lead to severe neonatal pathologies. 170 CVF samples were classified in three different groups: intra-amniotic infection and preterm birth; no intra-amniotic but preterm birth; no intra-amniotic infection and preterm labour. The samples underwent 2D-LC (SCX/RP) combined with ESI-QTOF mass spectrometry. Quantification and comparison of protein expression levels between the three groups was achieved by spectral counting. The authors found 26 potential biomarkers for intra-amniotic infection (statistical significant expression differences), which included acute phase reactants (e.g. α1-antitrypsin), immune modulators (e.g. lysozyme and α-defensin), amniotic fluid proteins (e.g. haptoglobin) and extracellular matrix-signaling proteins (e.g fibronectin). 13 of those were confirmed as potential biomarkers for intra-amniotic infection in immunoassays performed on the 170 samples. Using four of these proteins (α1-acid-glycoprotein, insulin-like growth factor binding protein-1, calgranulin C and cystatin A) in a logistic regression model, a fairly accurate differentiation between patients with intra-amniotic infection and those without intra-amniotic infection could be obtained. Although further clinical validation of these potential biomarkers is required, these results may lead to the development of new diagnostic assays for intra-amniotic infection. Also, the authors showed that the etiology of preterm birth caused by intra-amniotic infection is different than preterm birth without infection [ 52 ]. Shah et al . [ 53 ] used the stable isotope labeling by amino acids in cell culture (SILAC) technique to label proteins from an endocervical and vaginal cell-line (respectively End1 and Vk2). After the labeling process, a labeled cellular secretome (i.e. a "standard secretome") from vaginal and endocervical cells could be purified. CVF proteins can then be relatively quantified in test samples in a precise and accurate way by means of this "standard secretome" and multiple reaction monitoring (MRM). This technique is called stable isotope labeling proteome (SILAP). Normally, MRM requires the addition of labeled synthetic proteotypic peptides, which are often very expensive, but the use of SILAP has the advantage that the labeled "standard secretome" can be used instead. Nevertheless, since the concentration of the peptides in the "standard secretome" is not accurately known, it is not possible to absolutely quantify proteins in the test samples. A set of 15 candidate CVF biomarkers were thus relatively quantified and expression profiles were compared between healthy controls and preterm birth cases. This resulted in three proteins which were significantly upregulated (desmoplakin isoform 1, stratifin and thrombospondin), but the possible role of these proteins in the pathogenesis of preterm birth is not yet known [ 53 ]. Preterm premature rupture of membranes Many studies have been performed on preterm premature rupture of membranes (PPROM) which is an important risk factor of preterm birth (see above). PPROM is defined as "rupture of the fetal membranes prior to onset of labour in a patient who has a gestational age of less than 37 weeks" and has an incidence of 3-5% of all pregnancies [ 173 - 178 ]. A number of CVF proteins and peptides are associated with this condition and are listed in table 1 . Also other non-protein markers are associated with PPROM such as lactate [ 173 , 179 , 180 ], urea [ 181 ] and creatinine [ 181 - 183 ]. However, no differential proteomics studies have been performed focusing on this condition. Nevertheless, it can be expected that, as is the case with the preterm birth in general, the use of proteomic technology may result in new insights in the etiology of PPROM and the isolation of clinical applicable biomarkers. Amnisure ® International LLC (Cambridge, MA) has developed the FDA approved AmniSure ® ROM test for the diagnosis of PPROM. This simple immunoassay identifies placental α-microglobulin-1 which is abundant in amniotic fluid (2000-25000 ng/ml), but almost undetectable in cervicovaginal secretions (5-25 ng/ml). Only in the case of PPROM, higher concentrations are detectable in CVF. Analysis of clinical application possibilities of the test has shown a sensitivity of 99% and a specificity of 88-100% [ 184 , 185 ]. This test is a good example of the potential of CVF samples as biomarker sources and their applicability in clinical diagnosis. Although the test is still under investigation, the clinical use of such non-invasive test may eventually replace traditional diagnostic methods [ 177 ]. Classical approaches versus new techniques The vast majority of potential CVF biomarkers listed in table 1 was identified and validated using antibody-dependent classical approaches such as Western blotting or ELISA. Only the last few years, the use of proteomics as a new alternative technique for biomarker discovery has gained more interest due to the evolution of techniques such as chromatography and mass spectrometry from low reproducible and highly specialized methods to more user friendly, robust systems. The use of comprehensive proteome characterization and analysis has the great advantage that a huge amount of proteins (at least those which fall in the dynamic range of the applied techniques) are under consideration as potential biomarkers. Other techniques such as ELISA require the use of antibodies, which may restrict investigation to those biomarkers for which antibodies are available and purchased. Additionally, antibody-based techniques are sometimes incapable of discriminating between different subtle protein forms, whereas mass spectrometry can solve this problem. For example, it has been published that two α-defensins (HNP 1 and 2) are very sensitive and specific biomarkers for intra-amniotic infection. However, using ELISA, it is not possible to discriminate between HNP1 and 2 due to high sequence similarity. In contrast, due to the mass difference between HNP1 and 2, mass spectrometry can differentiate between both forms [ 186 , 187 ]. Nevertheless, this does not imply that proteomics is superior to antibody dependent methods in biomarker discovery experiments and makes them completely obsolete. Indeed, antibody-based validation is often required to confirm proteomics results. Also, as can be seen in table 1 , a large part of the biomarkers identified in non-proteomics studies, are not identified in any of the CVF proteome characterization studies mentioned above. Therefore, techniques such as Western blotting or ELISA are rather complementary then inferior to proteomics. Abbreviations 2D-DIGE: 2D differential gel electrophoresis; BV: bacterial vaginosis; CVF: cervicovaginal fluid; ESI: electrospray ionization; ESN: exposed seronegative individuals; FFN: fetal fibronectin; FTICR: fourier transform ion cyclotron resonance; HIF: HIV-inducing factor; HIV: human immunodeficiency virus; HLA: human leukocyte antigen; HNP: human neutrophil peptide; HPV: human papillomavirus; HSIL: high-grade squamous intra-epithelial lesions; IEF: isoelectric focusing; IL-1RA: interleukin-1 receptor antagonist; LC: liquid chromatography; LSIL: low-grade squamous intra-epithelial lesions; MALDI: matrix assisted laser desorption ionization; MIP: macrophage inhibitory protein; MRM: multiple reaction monitoring; MS: mass spectrometry; ND: not determined; NET: neutrophil extracellular traps; PAGE: polyacrylamide gel electrophoresis; PID: pelvic inflammatory disease; PPROM: preterm premature rupture of membranes; Q: quadrupole; RANTES: regulated on activation, normal T-cell expressed and secreted; RP: reversed phase; SCX: strong cation exchange; SELDI: surface enhanced laser desorption ionization; SILAC: stable isotope labeling by amino acids in cell culture; SILAP: stable isotope labeling proteome; SLPI: secretory leukocyte protease inhibitor; STI: sexually transmitted infection; TOF: time-of-flight; WAP: whey acidic protein Competing interests The authors declare that they have no competing interests. Authors' contributions GZ was responsible for the research and the writing of the manuscript. GAAVR and WAAT wrote a significant part about proteomics and HPV infection, provided advice and reviewed the manuscript. XWMVO was responsible for the supervision and thorough review of the manuscript. All authors read and approved the final manuscript. Supplementary Material

Acknowledgements GZ is supported by a doctoral fellowship of the Research Foundation-Flanders (FWO). GAAVR is funded by a doctoral fellowship of the Agency for Innovation by Science and Technology in Flanders (IWT).

CC BY

no

2022-01-12 15:21:44

Proteome Sci. 2010 Dec 8; 8:63

oa_package/97/25/PMC3016264.tar.gz

PMC3016265

21122155

Introduction Measures of mortality are basic indicators for population health assessment, and are ideally derived using data from death registration systems. However, such data are not available for Indonesia,[ 1 ] the world's fourth largest population[ 2 ]. Indonesia has experienced steady socio-economic development over the past five decades, and has undergone demographic and epidemiologic transition resulting in important changes to population size, structure and health profile[ 3 ]. Timely and accurate information on mortality and causes of death is necessary to understand the direction and pace of these changes. However, civil registration in Indonesia has not yet realised its utility as a reliable data source for mortality measurement[ 4 ]. Under these circumstances, intermittent household surveys have been the principal source for mortality assessment in Indonesia[ 5 ]. (see Additional File 1 ) Available evidence on adult mortality and causes of death from surveys in Indonesia is limited on account of incomplete reporting of deaths[ 6 ]. Hence, indirect demographic methods and model life tables have been used to estimate mortality in Indonesia[ 6 , 7 ]. Also, as part of the Global Burden of Disease Study, the World Health Organization estimated mortality by age, sex and cause in Indonesia in 2001[ 8 ]. Total mortality was estimated using the WHO Model Life Table System,[ 9 ] and cause of death patterns for Indonesia were assumed to approximate a combination of patterns experienced in Singapore, Thailand, India and the Philippines[ 10 ]. Such estimates are not useful for regular monitoring of mortality levels and trends in Indonesia. The Indonesian Ministry of Health has identified key measures of childhood and maternal mortality, and life expectancy at birth as indicators to chart health development in its 'Healthy Indonesia 2010' Plan[ 11 ]. These indicators are required at national, provincial and district level, as well as for urban centres. There is also a need to measure mortality from tuberculosis and HIV/AIDS for the United Nations Millennium Development Goals (MDGs)[ 12 ]. Also, In 2003-2004, an assessment of district health system performance in Indonesia identified a critical lack of information on mortality[ 13 ]. Therefore, the compilation of such data became a priority for the health sector. The Indonesian Mortality Registration System Strengthening Project (IMRSSP) was launched in 2006, towards developing a routine data source. The National Institute of Health Research and Development (NIHRD) coordinated IMRSSP activities in Indonesia, with technical assistance from the School of Population Health, University of Queensland. The broad goals of IMRSSP are to improve reporting of deaths and causes of death from health facilities and community based health centres, based on international standards. The initial objectives were to develop, test and implement cause of death reporting mechanisms in two pilot sites in Central Java. This paper reports the methods and processes developed and implemented in the IMRSSP, as well as an analysis of data collected in the two sites during 2006-2007.

Methods The urban municipality of Surakarta and the predominantly rural district of Pekalongan were chosen as the pilot sites for the project. The actual field implementation was preceded by a preparatory phase that involved several steps. At first, an inter-sectoral working group was constituted, which included representatives from NIHRD, the administrative and health sectors at central, provincial and district levels, and academic representatives. The working group first reviewed existing death registration procedures, and subsequently proposed a streamlined mechanism for death notification and registration at local level.(see Additional File 2 ) These procedures also incorporated international norms for reporting the cause of death, and compilation of mortality statistics[ 14 ]. The operational plans were then disseminated to the two pilot project areas for orientation of field staff. Extensive training programs were conducted for field interviewers, medical certifiers, and coders. This capacity building included in-class training sessions supplemented by instructional manuals and operational guidelines, [ 15 - 18 ] as well as on-site supervision and quality control throughout the implementation of IMRSSP. Following a six month preparatory phase, data collection under IMRSSP was launched in January 2006 in the two sites. Baseline population data by age and sex for each area were obtained from the local offices of the Badan Pusat Statistik (Statistics Indonesia). For deaths in hospitals, attending physicians issued a death notification form to relatives, and also completed standard medical certificates of cause of death. For deaths outside hospitals, local health centre personnel first issued a death notification form to relatives, and later conducted household interviews to ascertain more details on the cause, using an Indonesian adaptation of draft versions of international standard verbal autopsy (VA) questionnaires[ 19 ]. Completed questionnaires were reviewed by trained health centre physicians who applied specific guidelines to attribute causes of death from VA [ 17 , 19 ]. All medical certificates of cause of death from hospitals and VA based death certificates from health centres are submitted to the district health office. Over here, trained coders apply rules to select and code the underlying cause for each hospital or home death, and these records are entered in the IMRSSP electronic database for subsequent analysis. Completeness of data capture by IMRSSP was assessed in a sample of villages and urban wards in each site, through triangulation of data on deaths from different sources. In addition to the health sector data (i.e. IMRSSP), information on deaths during 2007 was obtained from two additional sources; the civil registration system; and deaths identified during an independent household survey conducted in December 2007. Data were matched across the three sources, using variables such as the name(s), age at death, gender, address and date of death. Allowances were made for matching the month rather than the exact date of death, and allowing a five year margin on the reported age at death from different sources, if other variables were matched (see Additional File 3 for details). Completeness of IMRSSP data was assessed as the proportion of IMRSSP deaths out of the total list of unique deaths derived from triangulation. Life tables for males and females were developed for each study population site from the observed age-specific death rates. In addition, life tables were also computed after adjusting observed death rates using the measures of completeness described above. (details in Additional File 3 ) Underlying causes of reported deaths were aggregated by age and sex according to the ICD Mortality Tabulation List 1 [ 20 ] for primary tabulations of leading causes of death. Also, observed cause-specific mortality rates were standardised by age using the WHO international population standard, [ 21 ] and compared with similar rates reported for Indonesia in the Global Burden of Disease 2004 project[ 22 ].

Discussion From a policy perspective, the initial findings presented for the two sites clearly indicate that in addition to the Millennium Development Goals, non-communicable diseases are also health priorities for Central Java. The adjusted under-five mortality rates (34-40 per 1000 live births) are at the upper end of the range reported by the Indonesian Demography and Health Survey program (32; 95% CI 21, 42) for Central Java for the period 1997-2007 [ 23 ]. For adults, the cause-specific mortality patterns highlight the magnitude of mortality from stroke and ischaemic heart disease in both urban and rural populations, and their contrast to the GBD estimates for Indonesia. The observed proportionate mortality from stroke in both sites is higher than reported (15%) by a recent national sample survey conducted in 2007, which also collected data on causes of death using VA methods[ 24 ]. These findings should lead to further epidemiological investigation of potential nutritional or behavioural determinants of cardiovascular disease in Indonesia, as conducted elsewhere[ 25 ]. Similarly, appropriate primary and secondary prevention strategies are required for diabetes, at least in urban areas. Also, cancer mortality needs more attention, in view of what is known on smoking trends in Indonesia,[ 26 ] and the current absence of population cancer registries in Indonesia[ 27 ]. The data also indicate the persistent burden from infectious diseases in the rural population. These findings could also be useful for future regional and global burden of disease estimates, and for comparative risk factor assessment. While these findings represent the first local evidence on the estimated levels of mortality and causes of death within Indonesia, there are some important data limitations, including the completeness of IMRSSP data, and the validity of information on causes of death. Incomplete data from both sites (particularly Surakarta) is a matter of concern (see Additional file 3 ), despite the reported crude death rates of 5.8 - 6.8 per 1000 for the two sites being much higher than the crude death rate of 3.3 per 1000 reported from a national survey in 2007[ 28 ]. The completeness of data could be improved by strengthening the registration guidelines to require both de facto and de jure registration. There is also the need to sensitise all stakeholders about the need for reporting deaths at the extremes of age. The most important issue would however, be the need to strengthen intersectoral collaboration, given that over 90% of deaths are likely to be recorded either by the administrative or the health sector. (See Additional file 3 ) Such collaboration can be achieved by the health system being proactive in periodically obtaining lists of deaths recorded by the local administration. Also, occasional coordination sessions could be conducted to orient registration and health sector personnel about their roles in IMRSSP. Based on the findings from this research, the Government of Indonesia has recently published a joint regulation from the Ministries of Home Affairs and Health, authorising collaborative efforts between offices from each ministry at all levels, to strengthen mortality and cause of death registration in Indonesia[ 29 ]. Another potential limitation is the validity of reported causes of death. About 37% of deaths in Surakarta and 9% in Pekalongan are medically certified in hospitals, and it would be reasonable to assume accurate reporting of the cause for these deaths. However, research elsewhere has identified problems even in medically certified deaths, although the observed misclassification patterns were compensatory[ 30 , 31 ]. On the other hand, the challenges in attributing causes of death using VA methods are well known[ 32 ]. Hence, continued implementation of IMRSSP should also include specific efforts to measure the reliability and where possible, validity of reported causes, both from VA and medical certification. Also, forensic and/or police data systems should be explored to identify deaths from injuries that are missed by the current IMRSSP data collection processes, leading to regular data sharing between these systems. Similarly, information should also be shared between registration and other health programs that have their own case and mortality notification systems (e.g. maternal and child health; tuberculosis control). The results from such activities would improve data completeness, and enable appropriate interpretation and therefore utility of cause-specific mortality data from the project. This project has developed a model to strengthen death registration systems as a source of empirical data to measure total and cause-specific mortality in Indonesia. From an operational perspective, the health sector was targeted for strengthening mortality notification. This approach was adopted since deaths either occur in health facilities, or are brought to the notice of community health personnel in the course of their functions, or, as per the recent legislation, documentary evidence of death is required from health personnel. Also, the processes for collection, compilation and analyses of data on age, sex and cause-specific mortality require health personnel with specific training, which was provided in this project. Nevertheless, these functions are best served when integrated with the legal and administrative processes for vital registration. Hence, in recognition of the systematic approach adopted in the IMRSSP, and the urgent need for local evidence on mortality across Indonesia, the project has been expanded to sites located on four other major islands of Indonesia; Sumatra, Kalimantan, Sulawesi, and Papua in 2007-2008; and to sites in Bali and Nusa Tenggara Timur (NTT) in 2009. (see Figure 3 ).

Conclusions Implementing cause of death reporting in conjunction with death registration appears feasible in Indonesia Improved inter-sectoral collaboration between registration and health departments, along with a role for local offices of Statistics Indonesia (BPS) in the compilation, processing and submission of vital statistics would increase the availability of mortality data at the district and provincial levels. Continued implementation of IMRSSP over the next 5 to 10 years, supported by appropriate operational research activities to measure and improve data quality is necessary to develop a routine source of mortality data for health policy and monitoring in Indonesia.

Background Mortality statistics from death registration systems are essential for health policy and development. Indonesia has recently mandated compulsory death registration across the entire country in December 2006. This article describes the methods and results from activities to ascertain causes of registered deaths in two pilot registration areas in Central Java during 2006-2007. The methods involved several steps, starting with adaptation of international standards for reporting causes of registered deaths for implementation in two sites, Surakarta (urban) and Pekalongan (rural). Causes for hospital deaths were certified by attending physicians. Verbal autopsies were used for home deaths. Underlying causes were coded using ICD-10. Completeness of registration was assessed in a sample of villages and urban wards by triangulating data from the health sector, the civil registration system, and an independent household survey. Finally, summary mortality indicators and cause of death rankings were developed for each site. Findings A total of 10,038 deaths were registered in the two sites during 2006-2007; yielding annual crude death rates of 5.9 to 6.8 per 1000. Data completeness was higher in rural areas (72.5%) as compared to urban areas (52%). Adjusted life expectancies at birth were higher for both males and females in the urban population as compared to the rural population. Stroke, ischaemic heart disease and chronic respiratory disease are prominent causes in both populations. Other important causes are diabetes and cancer in urban areas; and tuberculosis and diarrhoeal diseases in rural areas. Conclusions Non-communicable diseases cause a significant proportion of premature mortality in Central Java. Implementing cause of death reporting in conjunction with death registration appears feasible in Indonesia. Better collaboration between health and registration sectors is required to improve data quality. These are the first local mortality measures for health policy and monitoring in Indonesia. Strong demand for data from different stakeholders can stimulate further strengthening of mortality registration systems.

Findings The contrasting socio-economic profiles and access to health care between urban Surakarta and rural Pekalongan serve as background information to study mortality differentials between the two sites. For indicators such as urbanization, literacy rates, and access to water supply and sanitation, Pekalongan approximates the national average for Indonesia. (see Table 1 ). Analysis of the population age structure indicated that Surakarta has a relatively older population than Pekalongan. Overall, data collection covered a total of 123 registration units and was accomplished through the services of 36 health facilities; which indicates the administrative and management support required for IMRSSP implementation. The assessment of completeness of IMRSSP data in a sample of units demonstrated a marked variability in recording of information of deaths by the three different data sources. Figure 1 shows the distribution of deaths from the three sources used to compile the complete list of deaths from the sample units. The figure indicates that registration data from the desa/kelurahan (See Additional File 2 for definitions) offices has the highest proportion of completeness (> 85%), and also that combining data from registration and IMRSSP would result in over 90% completeness. The summary mortality indicators in Table 2 indicate a three year differential between adjusted male and female life expectancy at birth in each site. Although there are only marginal differences in the risks of dying before age 5 years between the two sites, adult mortality rates are considerably higher in Pekalongan. This differential results in lower life expectancy at birth for both males and females in rural Pekalongan as compared to urban Surakarta. Table 3 compares the proportionate mortality distributions by cause between the two sites. While stroke is the leading cause, the presence of ischaemic heart disease and chronic obstructive lung disease among the leading causes is probably a reflection of lifestyle factors and environmental exposures that are common to both areas. However, the presence of tuberculosis and diarrhoeal diseases among leading causes of death in Pekalongan highlights the importance of infectious diseases in rural areas of central Java. Comparisons between observed age-standardised death rates per 100,000 population by cause (for males and females combined) from each field site with estimated rates for Indonesia reported in by the World Health Organization's Global Burden of Disease (GBD) study in 2004 [ 22 ] are presented in Figure 2 . In addition to the major differences between the observed death rates and the WHO modelled estimates, particularly for stroke and ischaemic heart disease; the observed rates highlight the important sub-national differentials in cause-specific mortality from tuberculosis, diabetes, and chronic obstructive lung disease. Competing interests The authors declare that they have no competing interests. Authors' contributions CR and SS conceptualised the design of IMRSSP, contributed to data analysis and interpretation. CR drafted the initial version of the manuscript. CR, SS, SD, S, YW, LP and SK participated in the data collection, analysis and interpretation. TA and JI facilitated data management and quality control, and participated in data analysis and interpretation. ADL contributed to data interpretation. All co-authors contributed to revisions and preparation of the final version of the manuscript. Supplementary Material

Acknowledgements The project was conducted within the routine program of activities of various government departments in Indonesia. External support for field activities was provided through a grant from the Australian Agency for International Development (ROU 13652), which included support to the University of Queensland. Funds were also available through a grant from the World Health Organization (Grant No. INO GPE 002/RB/05/EC1.P1) which supported activities during the preparatory phase.

CC BY

no

2022-01-12 15:21:44

BMC Res Notes. 2010 Dec 2; 3:325

oa_package/0a/62/PMC3016265.tar.gz

PMC3016266

21138553

Background Tobacco use is one of the leading causes of preventable mortality in industrialized nations [ 1 ]. In 2002, World Health Organization [ 2 ] reported that about one in eight deaths in Japan are due to smoking (about 100,000 deaths a year), and Japan had some of the most lenient anti-tobacco laws among developed nations. Since then, considerable efforts have been made to promote smoking cessation activities. The national project, 'Health Japan 21' includes tobacco control, and the prevention of passive smoking was included in the Health Promotion Law implemented in 2003 [ 3 ]. Furthermore, in 2006, smoking cessation treatment by physicians was approved under the national health care insurance in Japan via the recognition of 'nicotine dependence' as a disease [ 4 ]. However, in 2008, it was shown that 36.8% of Japanese adult men and 9.1% of women still smoke regularly [ 5 ]. Given this situations, it is important for health professionals in Japan to further promote smoking cessation program. Smoking is accepted as a well-established risk factor for many oral diseases including oral cancer and periodontal disease [ 6 ]. Oral health professionals are in a unique position to advise smokers to quit by providing effective counseling on the various aspects of tobacco-induced diseases. Dentists and dental hygienists have regular contact with smokers and a great potential for helping their patients to quit smoking; yet, this potential is often underutilized [ 7 - 14 ]. It has also been suggested that Japanese health professionals may lack sufficient awareness of their position as role models for dental patients, in comparison to western developed countries [ 15 , 16 ]. Concurrently with the inclusion of smoking cessation intervention by physicians into the national health care insurance plan in Japan, a smoking cessation program was initiated at Suidobashi Hospital, Tokyo Dental College, in 2006. The program is a new multi-disciplinary approach to smoking cessation, which includes interventions by the medical and dental professionals. While the program has been a significant addition to our patient care [ 17 ], it was our general feeling that more effort was needed to implement smoking cessation care to patients receiving dental treatments. As part of an effort to further promote smoking cessation activities, this study assessed smoking-related perceptions, attitudes and practices of dental school hospital-based health professionals.

Methods A cross-sectional survey of health professionals was conducted at Suidobashi Hospital, Tokyo Dental College, Tokyo, Japan in July 2010. Among the practicing professionals, those who were providing direct care to patients were asked to participate in the study. Questionnaire A novel questionnaire was developed in order to assess smoking-related perceptions, attitudes and practices of health professionals. Specifically, the basic structure of the questionnaire items was generated in reference to the published studies [ 13 , 18 ] and to the Clinical Practice Guideline by U.S. Department of Health and Human Services [ 19 ]. Where required, one dentist, fluent in both Japanese and English performed the forward translation. A survey study in Japan [ 18 ] was also used as a guide for translation, as some of the wording in the questionnaire was similar to the material being translated. The base questionnaire was piloted with a small group of health professionals (two dentists, one dental hygienist, one nurse and one physician) in order to ensure the clarity and comprehensiveness of the questionnaire. The final and refined version (See additional file 1 : Items in the survey questionnaire) was composed of the following 11 items. In Items 1 to 4, participant demographics were assessed. In Item 5, smoking status and history were requested. Items 6 and 7 inquired about participants' experiences with smoking cessation care training and their willingness to receive it. In Item 8, the provision of smoking assessment or cessation care to patients was assessed, with four subscale items on a 10-point scale, asking the participants how often they performed the intervention specified from never (0%) to always (100%). Where appropriate, the answers were dichotomised; the cut-off point was a median score of 50%. In Item 9, we assessed participants' perceptions regarding smoking or smoking cessation care, with seven subscale items on a 5-point scale, asking if they strongly disagree (1) or agree (5) with a given statement. Where appropriate, the answers were dichotomized ('strongly disagree', 'somewhat disagree', 'not sure either way' versus 'somewhat agree' and 'strongly agree'). In Item 10, the participants' potential barriers were assessed. In the final Item 11, the participants were asked to express their views on smoking or smoking cessation care. Ethical considerations Approval for the study protocol and the survey implementation was obtained from the Suidobashi Hospital's review board. A formal ethics review was waived by the institutional review board, because the research involved submitting a questionnaire to adults who complied of their own free will and, any information that would identify participants was avoided. Written informed consent was obtained from all participants. Statistical analysis All the data obtained was examined and the responses were coded. The data was then descriptively analyzed and an appropriate test was applied, using a statistical package, InStat 3.1 (GraphPad, La Jolla, CA). Chi-squared test and one-way analysis of variance (ANOVA) with Tukey-Kramer multiple comparisons test were used. A p value of less than 0.05 was considered as statistically significant.

Results Characteristics of study participants In total, 129 questionnaires were distributed, of which 105 were returned. Of these 93 (72%) were fully completed responses from practicing professionals. Demographic data were listed in Table 1 . The participants comprised predominantly oral health professionals; 58% were dentists and 24% were dental hygienists. 16% were registered nurses. Only small number of physicians (n = 2) participated, mainly because of the dental school-based hospital setting. Therefore, data for nurses and physicians were pooled to represent medical professionals for the subsequent analysis. Overall, nearly two-thirds (62%) of participants were women. About 65% of dentists were men, but all other professionals comprised women. About two-thirds of participants were aged 20-39 years. There was wide variety in the number of years after receiving professional qualifications. All medical professionals had at least 5 years of practicing experience, while 41% of dental professionals had fewer than 4 years of experience. Altogether, 18% of participants reported being current smokers (17% of men and 19% of women), 15% reported being ex-smokers and 67% reported never smoking. Higher smoking rates were reported by dentists compared with other professionals (p = 0.0199, Chi-squared test for independence). None of the physicians or nurses who participated in the study was current smoker. Past training in smoking cessation care The majority of participants reported that they have had never received training in smoking cessation care. Only 8% had previous training, and 71% indicated that they would like to receive it. Smoking assessment or cessation care We next assessed the provision of smoking assessment or cessation care to patients. Among the four components of a brief intervention, 'asking smoking status' was the most frequently performed activity, followed by 'advise to quit'. (Table 2 ). Dentists and medical professionals asked about smoking status significantly more than dental hygienists (p < 0.05, p < 0.01, respectively; one-way ANOVA with Tukey-Kramer multiple comparisons test). Medical professionals were more likely to advise their patients to quit smoking than dental hygienists (p < 0.05). 'Assess willingness to make a quit attempt' and 'assist in quit attempt' were the interventions implemented for less than one-quarter of their patients. There were no statistically significant differences in any of those interventions among the health professionals. There were no significant differences in provision of smoking assessment or cessation care between smokers and non-smokers, as assessed by Chi-squared test. Perception of smoking and smoking cessation The participants' perceptions regarding smoking or smoking cessation care were assessed (Table 3 ). There was a statistically significant difference among overall scores (p < 0.0001). 'It is important to ask smoking status' and 'oral health professionals should participate more in smoking cessation care' were the subscale items with relatively high scores. There were no significant differences in these perceptions between professional types. When the relationship with smoking status was assessed, a statistically significant difference was noted in responses to 'oral health professionals should participate more in smoking cessation care' (p = 0.0404, Chi-squared test). Non-smokers were more likely to agree with this notion than smokers. Perceived barriers to smoking cessation care 'Lack of knowledge and training' (61%) was identified as a central barrier to smoking cessation care, followed by 'few patients willing to quit' (45%) (Figure 1 ). No statistically significant associations were noted between the demographic aspects including years in practice and the barriers that limit the health care professionals from providing smoking cessation care. While many of the individual comments in Item 11 were related to the perceived barriers expressed in the previous item, they also reflected a general positive attitude toward provision of smoking cessation care.

Discussion This study aimed to identify smoking-related perceptions and practices of health care professionals at a dental-school based hospital in Japan. Despite a general positive perception towards smoking cessation, the provision as well as the extent of smoking cessation care by the study participants was found to be relatively low. The findings of the present study support the notion that potential of oral health professionals in smoking cessation is underutilized [ 7 - 14 ]. Oral health professionals have a role in encouraging their patients to quit smoking and helping them, via referral pathways. Also, dental staff should be trained and given the resources to deliver the advice themselves [ 20 ]. By the mid 1990's, a quarter of tobacco users in the United States had been advised to quit by their dentist [ 21 ]. However, even in northern Europe and North America, a substantial proportion of dental professionals either do not provide smoking cessation interventions or provide them on a very limited basis [ 22 , 23 ]. Little information is available on smoking cessation practices of oral health professionals in Asia. In a survey study of 152 dentists practicing in Malaysia [ 24 ], only 17.9% of them were actually involved in smoking cessation counselling. Consistent with these reports, the dentists and dental hygienists in the present study indicated that smoking cessation interventions such as 'assess willingness to make a quit attempt' and 'assist in quit attempt' were implemented for less than one-quarter of their patients who smoke. The efficacies of smoking-cessation intervention in dental or medical-dental settings were shown in Japan [ 4 , 17 ], suggesting the feasibility of integrating dentists and dental hygienists in a medical smoking cessation intervention. Encouragingly, in a recent survey study of 435 certified periodontists of the Japanese Society of Periodontology, a relatively high percentage (54.7%) of the participants indicated that they provide some form of smoking cessation care to their patients [ 25 ]. Initiatives are needed to further promote smoking cessation care in our practice setting. Smoking behaviour among health professionals can be a significant obstacle in promoting smoking cessation initiatives [ 13 , 22 ]. A survey conducted among members of the Japan Medical Association in 2008 showed that the prevalence of cigarette smoking among 3,486 Japanese physicians was 15.0% for men and 4.6% for women [ 26 ]. A recent survey study conducted among 1000 dentists in Japan revealed that the prevalence of regular smokers was 25% for men and 3% for women [ 27 ]. In the present study, prevalence of smokers among women (19%) was relatively higher than that reported for the general population [ 5 ] or for Japanese dentists in private practice [ 27 ]. However, this needs to be interpreted with caution, since nearly two-thirds of the participants were women. The overall prevalence of smokers was higher than that (14.7%) reported for the certified periodontists [ 25 ]. Tobacco-using health professionals, including dentists, were reported to be less proactive than their non-using counterparts [ 15 , 28 - 30 ]. This trend was supported by our finding that non-smokers were more likely to feel that oral health professionals should participate more in smoking cessation care, when compared to smokers. These results suggested that further effort in smoking cessation among the health professionals in our hospital is needed. We found that interventions by the dental hygienists in smoking cessation care were somewhat limited when compared to other health professionals (Table 2 ). This may be due to the fact that the dental hygienists working in our hospital have not been formally integrated into the smoking cessation treatment program. Given the importance of the role of oral health personnel in smoking prevention and control, an active participation of dental hygienists in the smoking cessation treatment program is necessary. In the present study, lack of training and knowledge was considered to be the major barrier that potentially limited the involvement of the health professionals in smoking cessation activities. Also, more than one third of the participants perceived that few patients would be willing to quit. Removing such barriers can be difficult [ 31 ]. A Cochrane Review [ 32 ] on the training of health professionals in smoking cessation concluded that health professionals who were trained were better at delivering smoking cessation interventions. In the present study, no statistically significant associations were noted between the barriers and the demographic aspects including years in practice. Given that only 8% of the participants had prior training in smoking cessation care, it is necessary to offer opportunities for adequate training. Rosseel et al. [ 13 ] reported that an atmosphere supportive of advice among colleagues is important in the implementation of strategies which support smoking cessation. Support from our own practice members and organization may increase the efficacy of the health professionals in smoking cessation care. This study has several inherent limitations. The data were collected via a self-reported questionnaire. Like all questionnaires, the possibility of both intentional and unintentional mis-reporting threatens the validity and reliability of the findings. Relatively small number of participants allowed limited statistical analysis of data. However, the present study adds to the existing literature by providing the salient information regarding the perceptions of and potential barriers to smoking cessation care among multiple health professions in Japan.

Conclusions We identified a need for further promotion of smoking cessation activities by the hospital-based health professionals. Hospital policies and faculty education providers need to provide staff with appropriate training and create an atmosphere supportive of smoking cessation activities.

Background Smoking is currently accepted as a well-established risk factor for many oral diseases such as oral cancer and periodontal disease. Provision of smoking cessation care to patients with oral problems is a responsibility of health care professionals, particularly dentists and dental hygienists. This study examined the smoking-related perceptions and practices of dental school hospital-based health professionals in Japan. Findings A cross-sectional study design was used. The sample was formed from dentists, dental hygienists, physicians and nurses of a dental school hospital in Tokyo, Japan (n = 93, 72%). Participants were asked to complete an 11-item questionnaire assessing demographic variables and smoking history, provision of smoking cessation advice or care, attitudes about smoking cessation, and perceived barrier(s) to smoking cessation care. Eighteen percent of participants reported being current smokers and 15% reported being ex-smokers, with higher smoking rates reported by dentists compared with other health professionals (p = 0.0199). While recognizing the importance of asking patients about their smoking status, actual provision of smoking cessation advice or care by participants was relatively insufficient. Interventions such as 'assess willingness to make a quit attempt' and 'assist in quit attempt' were implemented for less than one-quarter of their patients who smoke. Non-smokers were more likely to acknowledge the need for increased provision in smoking cessation care by oral health professionals. 'Lack of knowledge and training' was identified as a central barrier to smoking cessation care, followed by 'few patients willing to quit'. Conclusions A need for further promotion of smoking cessation activities by the health professionals was identified. The findings also suggest that dentists and dental hygienists, while perceiving a role in smoking care, do require training in the provision of smoking cessation care to hospital patients. In order to overcome the potential barriers, it is necessary to provide staff with appropriate training and create an atmosphere supportive of smoking cessation activities.

Consent Written informed consent was obtained from the participants for publication of this manuscript and accompanying materials. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Competing interests The authors declare that they have no competing interests. Authors' contributions AS designed the study, constructed the survey questionnaire, performed data analyses, and drafted the manuscript. AM, FU and MK contributed to data collection and analysis. MN interpreted the study and edited the manuscript. TM and TI oversaw procedures. All authors have read and approved the manuscript. Supplementary Material

CC BY

no

2022-01-12 15:21:44

BMC Res Notes. 2010 Dec 7; 3:329

oa_package/2e/a1/PMC3016266.tar.gz

PMC3016267

21138554

Materials and methods Strains Trichoderma reesei QM9414 (ATCC26921) and the following descendents lacking parts of light sensing pathways were used through this study: Δenv , Δblr1 and Δblr2 , missing the open reading frames of the respective genes [ 15 ]. Strains were maintained on malt agar plates (30 g L -1 malt extract, 1 g L -1 peptone, 20 g L -1 agar) in darkness at 30°C. Preculture preparation and shake flask cultures Conidia from two weeks old plates were harvested with sterile distilled water. An adequate volume of this suspension to reach 10 8 conidia L -1 final concentration was transferred to 750 ml Erlenmeyer flasks containing 200 ml of Mandels-Andreotti basal medium [ 39 ] prepared in 0.1 M citrate-phosphate buffer (pH 5.0) and supplemented with peptone to a final concentration of 1 g L -1 . As carbon source, 10 g L -1 Solka Floc 200 (International Fiber Corporation, North Tonawanda, NY, USA) was added. After 3 days of cultivation in constant darkness at 30°C on a rotary shaker (250 rpm) flasks were used to inoculate fermentation medium at 10% (v/v). Fermentation Every strain was analyzed in a 30 L double-walled stainless steel laboratory fermenter (Biostat C-DCU 3, B Braun Biotech, Germany) equipped with pO2, pH, temperature, pressure and foam sensors in 20 L working volume. Agitation was provided by a top-drive agitator with 3 Rushton impellers each with 6 blades. Medium components (10 g L -1 Solka Floc and 5 g L -1 wheat distiller's grain, 0.83 g L -1 KH 2 PO 4 and 0.83 g L -1 (NH 4 ) 2 SO 4 ) were suspended in tap water sterilized in the mounted device and subsequently ten flasks of the precultures were combined aseptically into the sterile inoculating equipment resulting in 2 L of inoculum, which was transferred into the fermentor. pH was continuously controlled and adjusted to 5.8 by addition of 10% (v/v) NH 4 OH or 10% (v/v) H 3 PO 4 . pO 2 level was kept on 30% by cascade control of air flow rate as first priority, varying between 5 and 12 L min -1 and velocity of agitation as secondary, shifting between 250 and 600 rpm. Beyond 30% of dissolved oxygen content (DO), air flow rate and agitation worked on the minimal values for both variables. Temperature was set to 28°C. Operation parameters (pH, volume of added base and acid, pO 2 , stirring velocity and air flow rate) were recorded by MFCS/win 2.0 (B Braun Biotech) software. To control foaming, ionic antifoam emulsion (Sigma-Aldrich Antifoam A) was added automatically. Fermentations were carried out in duplicate. Samples were withdrawn regularly, centrifuged (3400 g, 5 minutes) and supernatants were analyzed for enzyme activities and extracellular protein content. Biomass measurements were carried out daily from fermentation broth (see below). Western blot analysis Proteins from 0.5 ml fermentation supernatant were precipitated by adding 1 ml of 96% ethanol. Western blotting was performed according to standard protocols [ 40 ]. Briefly, after centrifugation the pellet was dissolved in 0.2 ml SDS buffer; 10 μl representing equal amounts of culture filtrate were loaded into a 7.5% SDS-PAGE gel and run at 15 mA. Transfer to nitro-cellulose membranes (Hybond-C Extra, Amersham Biosiences, UK) was done by semidry electroblotting and antibodies against the major cellulase CBH1 as well as horseradish peroxidase-conjugated anti-mouse IgG (Promega, Madison, US) were used for analysis of expression of cellulases. Analytical assays For the Filter Paper Activity (FPA) assay 0.5 ml suitably diluted supernatant to liberate approximately 1 mg glucose equivalent was mixed with 1.0 ml of 0.05 M Na-acetate buffer (pH 4.8) and a 6 × 1 cm strip of Whatman grade 1 filter paper was added. After incubation for 1 hour at 50°C, 3 ml of dinitrosalicylic acid reagent [ 41 ] was added, kept at 100°C for 5 minutes, diluted with 16 ml of distilled water and reducing sugar content was measured at 550 nm. The filter paper unit (FPU) was defined as the amount of glucose released given in μmol min -1 . To measure xylanase activity, 0.1 ml properly diluted supernatant to liberate approximately 0.2 μg xylose was added to the mixture of 0.4 ml 0.05 M citrate buffer (pH 5.3) and 0.5 ml of 1% (w/v) birchwood xylan (Sigma Aldrich) solution prepared the same buffer. After incubation at 50°C for 10 minutes the reaction was terminated by adding 1.5 ml of dinitrosalicylic acid reagent and then the mixture was kept at 100°C for 5 minutes. Absorbance was measured at 550 nm. A xylose calibration curve was used to calculate the activity which was defined as the amount of xylose released given in μmol min -1 . β-glucosidase activity was assayed according to the procedure of [ 42 ] using 4-nitrophenyl-β-D-glucopyranoside (Sigma-Aldrich) as substrate. For endo-glucanase activity measurement tenfold diluted supernatant was added to Azo-CM-cellulose solution (S-ACMC; Megazyme International Ltd., Ireland). Procedure was carried out according to the manufacturer's instructions. Extracellular protein content was determined by Coomassie Blue G250 reagent [ 43 ] using Bovine Serum Albumin as standard. All measurements were carried out at least in duplicate. For biomass determination 100 ml whole fermentation broth was filtered onto a dry and preweighed filter cloth and washed with distilled water. Filtration was carried out in triplicate. After drying at 105°C for 6 hours the filter cloth containing filter cake was weighed again. For further analysis 0.5 g homogenous sample from each filter cake was subjected to a two step sulfuric acid hydrolysis to release N-glucosamine from fungal cell wall: the sample was incubated with 60% (w/v) sulfuric acid at room temperature for 24 hours with subsequent dilution to 1 N and hydrolysis at elevated temperature (121°C for 1 hour) [ 44 ]. The cooled mixture was then neutralized by addition of 1 N solution of NaOH and total volume was measured. A portion was centrifuged at 3400 g for 5 minutes and supernatants were analyzed for N-glucosamine by the method of Blix [ 45 ] as described by Bussari et al . [ 46 ]. N-glucosamine content was calculated according to a calibration curve prepared with reagent grade N-glucosamine.HCl (Sigma-Aldrich). Cell mass is expressed as mg N-glucosamine/ml fermentation broth. Statistical analyses were performed with STATISTICA 8.0 software (StatSoft, Inc., Tulsa, OK, USA). The significance level for t-test for independent samples was set to a p-value of 0.05.

Discussion Transcriptional responses given to light reflect the changes in the ecological niche that surrounds fungi. T . reesei as a saprophyte mainly resides in the inside of decaying plant material, that is, in darkness. Presence of light means the open surface of the habitat where the successful dissemination of conidia is expected or existence of possible mating partner is assumed, but also where (energy consuming) measures for protection from harmful UV-light or desiccation have to be taken. The different physiological requirements of growth on the surface or within its substrate are reflected by the metabolic differences between these conditions [ 10 , 33 ], which can be exploited for strain improvement. One of our most surprising results in this study was the finding that in contrast to transcription of cbh1/cel7a , which decreases in deletion mutants of blr1 and blr2 [ 15 ], cellulase activity in the culture medium of the respective strains was nevertheless increased or at least remained at wild-type levels in case of Δblr1 . Interestingly, comparable effects have been observed for N. crassa (M. Schmoll, manuscript in preparation). Hence, this study will be the basis for further research to reveal the molecular basis for this phenomenon. In fact, in accordance with the results presented here, light has previously been reported to influence protein secretion in N. crassa [ 34 ], however a role of the photoreceptors in this process was not known so far. These results also decreased the applicability of our initial hypothesis that several cellulase genes (other than cbh1/cel7a, which is positively regulated by BLR1 and BLR2 [ 15 ]) might be directly regulated by binding of BLR1 and/or BLR2 to their promotors as transcription factors. Nevertheless, at present such a direct regulation cannot be fully ruled out. The finding of a discrepancy between transcription of the major cellulases and detected cellulase activities in the culture medium is particularly interesting because it suggests that cellulolytic enzymes may not be exclusively regulated on the transcriptional level [ 35 ] and references therein), but also posttranscriptionally - presumably in response to light. Although the signal transduction pathway triggering cellulase gene expression is far from being well-established, first insights are already available. Besides the light response pathway, signals related to sulphur metabolism as well as heterotrimeric G-protein signaling and the cAMP-pathway are involved in regulation of cellulase gene transcription. Addition of the organic sulphur source methionine decreases transcription of cbh1/cel7a below detection limits only in light [ 36 ]. Deletion of the G-protein alpha subunit GNA1 leads to strongly increased transcription of cbh1/cel7a in darkness [ 24 ] and constitutive activation of GNA3 causes considerably increased transcription of this gene in light [ 23 ]. Hence the relevance of the signals transmitted by GNA1 and GNA3 as well as the sulphur signal must be dependent on the light status, which is perceived and transmitted by the photoreceptors BLR1 and BLR2 [ 15 ]. The results presented here could be interpreted in a way that the extent of cellulase transcription is set in response to environmental signals (as transmitted for example by G-proteins via cAMP and phosphorylation to transcription factors), but that the distribution of resources for the energy consuming process of translation and secretion is at least in part governed by the photoreceptors BLR1 and BLR2 and possibly by ENVOY. The fact that these proteins and their orthologues regulate multiple targets [ 16 , 32 , 33 ] supports this hypothesis. Therefore it will be interesting to learn, whether or not the efficiency of strains engineered for high transcriptional activity of cellulase promotors - such as for example by deletion of gna1 - but also strains resulting from random mutagensis, can still be improved by deletion of blr2 , which might act negatively on secretion. Additionally, elucidation of the mechanism responsible for the impact of components of the light signaling pathways on the secretion machinery warrants further investigations. It is interesting that different effects have been observed for deletion of blr1 , blr2 and env1 . On the one hand this finding indicates that BLR1 and BLR2 do not act as a complex under our experimental conditions (mainly darkness) in their regulatory function targeting cellulase gene expression, but have individual functions. On the other hand it also suggests that although induction of env1 transcription is abolished upon deletion of blr1 or blr2 [ 15 ], the consequences of deletion of env1 are not similar to those of deletion of these photoreceptors. Hence BLR1 and BLR2 do not exert their function via ENV1, which confirms the hypothesis proposed earlier [ 15 ]. The increased efficiency of the enzyme mixture secreted by Δenv1 further renders a closer investigation of the postulated coregulation of cellulases [ 35 ] and references therein] interesting, especially with respect to different light conditions and the regulators involved in transmission of this signal. The lower efficiency of Δblr1 in terms of protein secretion could be due to a counteracting effect of BLR1 compared to BLR2 in secretion. However, in N. crassa it has been shown that in darkness the homologue of BLR2, WC-2, is present in excess over WC-1 [ 37 ]. Consequently, the effect seen for Δblr1 is more likely to be due to an increased availability of BLR2, whose binding partner has been removed resulting in additional BLR2 proteins, now free to exert their negative function on secretion. Effects of light on metabolic processes have been shown in numerous fungi and for many target mechanisms [ 10 , 12 ]. Therefore the function of the main components of the light perception machinery in regulation of plant cell wall degrading enzymes is not without precedent. A connection between carbon sensing and the function of BLR-1 and BLR-2 has been suggested for T. atroviride [ 31 ] and different roles in these metabolic functions have been detected: While both proteins are required to adjust the intensity of the response to a certain carbon source, BLR-1 is responsible for carbon source selectivity [ 38 ]. For ENV1 a regulatory function in cellulase transcription had been shown in T. reesei [ 20 ]. Although the most obvious functions of BLR1, BLR2 and ENV1 have been observed and studied in light, it also has been shown clearly that these proteins and their homologues in other fungi additionally have functions in darkness [ 15 , 32 , 33 , 38 ]. Hence the investigation of the light response machinery as done in this study, but also of its downstream targets, with respect to industrial fermentations opens up a new strategy for strain improvement aimed at more efficient biofuel production.

Conclusions With this study we demonstrated that components of light signaling pathways have an important role also in darkness related to carbon sensing, and highlight that factors which have been yet not associated with industrial cellulase expression still can have impact on it. From a practical point of view these modified strains can be used to enhance productivity, thus lower price of enzyme production which is essential for second generation biofuel breakthrough. Further studies to elucidate the complex signaling and regulatory network through which cellulase transcription and consequently expression can be triggered are already in progress. Our study thus provides new insights into this machinery, which can be exploited to enhance biotechnological fermentation at different stages of regulation in fungi.

Background In nature, light is one of the most important environmental cues that fungi perceive and interpret. It is known not only to influence growth and conidiation, but also cellulase gene expression. We therefore studied the relevance of the main components of the light perception machinery of Trichoderma reesei ( Hypocrea jecorina ), ENV1, BLR1 and BLR2, for production of plant cell wall degrading enzymes in fermentations aimed at efficient biosynthesis of enzyme mixtures for biofuel production. Findings Our results indicate that despite cultivation in mostly dark conditions, all three components show an influence on cellulase expression. While we found the performance of the enzyme mixture secreted by a deletion mutant in env1 to be enhanced, the higher cellulolytic activity observed for Δblr2 is mainly due to an increased secretion capacity of this strain. Δblr1 showed enhanced biomass accumulation, but due to its obviously lower secretion capacity still was the least efficient strain in this study. Conclusions We conclude that with respect to regulation of plant cell wall degrading enzymes, the blue light regulator proteins are unlikely to act as a complex. Their regulatory influence on cellulase biosynthesis involves an alteration of protein secretion, which may be due to adjustment of transcription or posttranscriptional regulation of upstream factors. In contrast, the regulatory function of ENV1 seems to involve adjustment of enzyme proportions to environmental conditions.

Findings The ascomycete Trichoderma reesei (anamorph of Hypocrea jecorina ) is one of the most prolific cellulase producing microorganisms, its efficient enzyme mixture being used in several processes of textile, food and pulp and paper industries [ 1 - 3 ]. Moreover a new market potential is arising with the commercialization of cellulosic ethanol plants: however, a main bottleneck for the economic success of the production of the second generation biofuels is the price of cellulolytic enzymes [ 4 ]. Strain improvement in T . reesei for plant cell wall degrading enzyme production can become more efficient with the use of the genome sequence [ 5 , 6 ]. Interestingly, analysis of the genome of T. reesei revealed an unexpectedly low number of genes encoding cellulolytic enzymes - despite the high efficiency of the cellulase mixture produced by this fungus. Besides improving the produced enzymes themselves or the efficiency of the promotors by which their expression is controlled, one strategy to elucidate the underlying mechanisms responsible for this high efficiency of T. reesei can be the investigation and exploitation of signal transduction processes [ 7 , 8 ] during growth on cellulosic substrates. Signaling mechanisms greatly contribute to successful adaptation and survival by receiving and interpreting numerous biotic and abiotic factors one of which is light. In contrast to plants, which utilize light as energy, for fungi light is merely a source of information. Blue light affects or initiates a number of physiological processes in fungi in general and also in Trichoderma , e.g. growth, conidiation and numerous metabolic pathways [ 9 , 10 ]. Many effects of light are common within the fungal kingdom and also the pathways of light sensing and its elements often share significant homology [ 11 ]. The photobiology of Trichoderma spp. has been investigated in considerable depth for decades [ 12 ]. Orthologues of the well studied Neurospora crassa photoreceptor genes wc-1 and wc-2 [ 13 ] genes were described in Trichoderma atroviride [ 14 ] and subsequently also in T. reesei [ 15 ]. The T. reesei blue light regulators (BLR1 and BLR2) have similar structural domains (PAS/LOV) and light independent regulatory roles were also reported for these proteins [ 15 ]. In T. atroviride also BLR independent light sensing routes have been proposed [ 16 ]. ENVOY, a PAS/LOV domain protein in T. reesei , which shares similarity with the N. crassa photoreceptor VIVID [ 17 - 19 ] is crucial in light tolerance and modulates cellulase transcription in a light dependent manner [ 20 ]. Recently, also an influence of the two photoreceptors BLR1 and BLR2 on cellulase gene transcription has been shown [ 15 ] suggesting that these regulators act positively on this process. In Trichoderma , also cAMP levels are responsive to light [ 21 ] and cAMP is involved in regulation of cellulase levels [ 22 ], which indicates an action via phosphorylation of transcription factors by cAMP dependent protein kinase A. Moreover, two G-protein alpha subunits (GNA1 and GNA3, which impacts cAMP levels) have been shown to exert a considerable light dependent influence on transcription of the major cellulase gene cbh1/cel7a [ 23 , 24 ]. However, since these high levels of transcription did not result in an equally high production capacity of the respective mutant strains (M. Schmoll, unpublished results), further (presumably light-dependent) regulatory checkpoints at the level of translation and/or secretion can be expected. Based on these findings we assumed that the major components of the light response pathway (BLR1, BLR2 and ENV1) could be crucial regulators or checkpoints in (light dependent) production of extracellular enzymes. Therefore we aimed to investigate the relevance of the light signaling machinery, which obviously plays an important role in cellulase regulation, for industrial fermentations. Analysis of strains defective in light sensing showed that the blue light regulatory proteins BLR1, BLR2 and ENV1 are indeed involved in regulation of expression and secretion of plant cell wall degrading enzymes, even in the predominantly dark conditions of a biotechnological steel fermentor. Consequently, our study revealed new targets for improvements of cellulase production by modification of the light signaling machinery. Presence of promotor motifs associated with light response in genes encoding plant cell wall degrading enzymes Because of the reported influence of light as well as of ENVOY, BLR1 and BLR2 on regulation of cellulase expression [ 15 , 20 ] a direct regulation of cellulolytic genes by the transcription factors BLR1 and BLR2 or and indirect impact of ENVOY can be assumed. In order to get a first guideline, whether the genes encoding plant cell wall degrading enzymes could be subject to direct regulation by the two photoreceptors or other factors associated to light response, we screened the promotors of genes encoding proteins presumably involved in plant cell wall degradation for the presence of light responsive promotor motifs. The region within 1000 bp upstream of the ATG codon was searched for specific motifs (Table 1 ). Gene sequences were obtained from the Trichoderma reesei , v2.0 genome database http://genome.jgi-psf.org/Trire2/Trire2.home.html . A complex consisting of N. crassa WC-1 and WC-2, the homologues of T. reesei BLR1 and BLR2, was reported to bind light-response elements (LREs) with the consensus sequence of GATNC--CGATN, where N can be any nucleotide but the same in both repeats [ 25 ]. An LRE motif was only accepted if its length did not exceed 50 bp. As both BLR1 and BLR2 have the characteristic to function as zinc-finger GATA factors, we also screened for the HGATAR (H = C, T, A) consensus sequence [ 26 ]. However, binding of these transcription factors to such GATA-sequences was not reported so far. The EUM1 motif which had been identified in the env1 as well as the N. crassa vvd promotors and thereafter detected in the cellobiohydrolase promotors cbh1 ( cel7a ) and cbh2 ( cel6a ) [ 20 ] was searched because of the light-dependent complex formation detected on EUM1 in both the env1 and gna3 promotor [ 23 , 27 ]. The proteins constituting the EUM1 binding complex are currently unknown. The analyzed genes were associated with possible activities (Table 1 ), however, in some cases the lack of knowledge on cellular localization of beta-glucosidases implicates uncertainty as to their involvement in extracellular substrate degradation. As presented in Table 1 the EUM1 sequence was found in the promoters of genes for all types of cellulolytic activities. These findings may be interpreted to correlate with the effect of light on cellulase transcription as detected earlier. The highest number of EUM1 motifs is present in the promoter of cel74a , which encodes a xyloglucanase. LRE motifs were only found in the promoters of three genes, two of which encode beta-glucosidases ( cel1b , cel3c ), albeit the localization of the encoded proteins is unknown. The third LRE was found in the promoter of a xylanase gene ( xyn3 ). GATA binding sites were found in most promoter regions. Since GATA factors are wide spread regulation elements and due to the high number of motifs found it is hard to predict any relation to light modulated cellulase gene expression, although binding of the photoreceptor complex cannot be excluded. Interestingly, the promoter of cel5b encoding an endo-glucanase was the only one where none of these consensus sequences associated with light response was found. Consequently, while in some cases LRE motifs would render a direct regulation by the photoreceptors BLR1 and BLR2 supposable and several EUM1 motifs suggest binding of (as yet uncharacterized) light regulatory factors, other mechanisms regulating the production of cellulolytic enzymes are likely to contribute to the modulated output in light and darkness. Cultivation of mutants in env1 , blr1 and blr2 for investigation of cellulase gene expression In order to obtain preliminary information on whether the transcriptional data on the influence of env1 , blr1 and blr2 on cellulase regulation [ 15 , 20 ] would correspond to the cellulolytic efficiency of the enzyme mixture secreted by the respective deletion strains, we first performed shake flask cultures. We found that all three strains showed a significantly higher specific cellulase activity on Mandels-Andreotti medium with 1% (w/v) microcrystalline cellulose than the wild-type strain up to 120 hours of cultivation (data not shown). Considering the transcription data for cbh1 in these strains [ 15 ] this higher efficiency was surprising. Therefore we chose to use all three strains for analysis of their efficiency in a laboratory scale fermenter and again tested them in shake flask cultures for cellulase expression in the medium to be used for fermentation, which confirmed the results above. In fermenter cultivations, similar patterns of pH adjustment and oxygen supply were observed for each strain. After approximately 5 hours of lag phase, pO 2 and pH began to decrease indicating growth of the fungus. The fermentation parameters recorded suggested that the modified strains have no increased oxygen and inorganic nitrogen demand. Fermentations were aborted after 73 hours of run when the oxygen level had reached 50 - 60% and acid addition was detected. Enhanced enzyme production by strains with defects in the light response pathway The degradative potential of the enzyme mixtures secreted by wild-type and mutant strains was first characterized using dyed Azo-CM-cellulose (Megazyme), which revealed the effectiveness of the whole cellulase enzyme mixture, especially of endo-1,4-β-D-glucanases. (Figure 1A ). Peaks of endo-glucanase activity in the culture medium were reached between 48 and 54 hours and after that slightly declined. Highest activity was measured in case of strain Δblr2 followed by Δenv1 , which are 59.0% (p-value 3·10 -6 ) and 44.2% (p-value 8.5·10 -5 ) higher than the wild-type QM9414, respectively. Between Δblr1 and wild type no significant difference was found (p-value 0.33). Analysis of filter paper activity (FPA), which determines the amount of glucose liberated from cellulose, was performed in order to obtain indications as to the extent of exo-cellulase/cellobiohydrolase activity. Maximal activity values were reached after 48 hours in case of each strain, thereafter activity values slightly decreased which is presumably due to protease activity in the medium (Figure 1B ). Highest activities were reached by Δenv1 and Δblr2 , (1.45 and 1.37 FPU mL -1 ), respectively, which are 21.9% (p-value 4.4·10 -5 ) and 15.3% (p-value 2.5·10 -4 ) higher than those of the parental strain QM9414 (1.218 FPU mL -1 ). Δblr1 produced 31.9% (p-value 3·10 -6 ) less FPU ml -1 than that observed with QM9414. Hence the increased degradative potential of Δenv1 and Δblr2 is only in part due to enhanced expression of the cellobiohydrolases, but seems to be mainly caused by an improved overall efficiency of the endoglucanase mixture secreted. Nevertheless, it must be considered that FPA also involves the action of endoglucanases and β-glucosidase and can thus not be seen as exclusive determination of cellobiohydrolase activity. Since the preferred substrate for yeast needed for the final step of conversion of plant material into ethanol is glucose, we also determined β-glucosidase activity (on pNPG). Initially, curves had the same slope for each strain, but Δblr2 reached an 88.4% (p-value 1.2·10 -5 ) higher peak value than the wild-type. Activities of Δblr1 and Δenv1 were practically equal (p-value 0.29), approximately 10% higher than QM9414 (p-values 3.3·10 -4 and 8.5·10 -4 , respectively; Figure 1C ). This result is in accordance with the findings described above, although it should be kept in mind that also cell wall bound β-glucosidase [ 28 , 29 ] may contribute to endoglucanase activity and FPA in vivo . Finally, we also assessed the potential of the light response mutant strains to degrade the hemicellulosic part of a given plant material. Highest xylanase activities were observed between 48 and 52 hours of fermentation. Surprisingly, despite its high efficiency on cellulose, Δblr2 (101.2 ± 7.5 IU/ml) showed approximately the same xylanase activity as QM9414 (103.8 ± 1.8 IU/ml; p-value 0.52) which may be because of the different regulation pattern of hemicellulolytic enzymes. A comparable effect has already been observed in case of T . atroviride mutants created by random mutagenesis that possessed enhanced FPA production properties but were deficient in xylanase secretion [ 30 ]. Peaks of Δenv1 (123.2 ± 2.0 IU/ml) and Δblr1 (116.7 ± 5.9 IU/ml) were 18.7% (p-value 2.6·10 -4 ) and 12.4% (p-value 6·10 -3 ) higher than that of QM9414. The positive effect of env1 gene deletion was consistent with the different activities suggesting that one downstream pathway of the regulatory output of ENVOY are the pathways involved in plant cell wall degradation. The fermentation profiles of the strains show that the function of the regulatory mechanisms causing enhanced cellulase production in the mutant strains becomes most significant after 30 hours of fermentation. While for QM9414 and Δblr1 enzyme activities stagnate shortly thereafter (possibly because a critical level of activity for sustaining certain nutrient levels is reached), enzyme production/secretion as reflected by still increasing activities continues in Δblr2 and Δenv1 . Specific performance of enzyme mixtures In order to enable a correlation of the performance of the secreted enzyme mixture (i. e. U/mg of secreted protein) with growth and total protein secretion capacity of the individual strains, we determined biomass and protein content of the culture medium during fermentation (Table 2 ). Biomass production of strain Δblr1 was considerably (37.5% (p-value 0.05) after 24 h and 41.4% (p-value 0.015) after 48 h) higher than that of the wild-type QM9414 throughout the fermentation. In contrast, neither Δblr2 nor Δenv1 showed a significant difference to the wild type (p-values > 0.1). Biomass specific filter paper activity was the highest for Δenv1 , followed by Δblr2 . Both BLR1 and BLR2 have been reported to regulate growth of Trichoderma atroviride and T. reesei on solid media [ 15 , 31 ], however similar effects of deletion of either of these genes have been observed. Nevertheless biomass formation has not been studied on cellulose containing liquid media in T . reesei . Our results show that biomass formation of Δenv1 is only marginally reduced (p-value 0.11) compared to that of the wild type, which is consistent with the finding that growth is not affected by the deletion in darkness [ 32 ]. To illustrate the performance of the cellulase mixture (U/mg of total secreted protein), protein specific endoglucanase activity and FPA was also calculated. The enhanced FPA of Δblr2 was in line with the highest protein concentration measured among the strains (Table 2 ). Considering the biomass formation of this strain, this result indicates that the high cellulolytic activity found in the culture medium of this strain is mainly due to an enhanced protein secretion capacity. An alternative interpretation would be that the higher levels of extracellular proteins observed for this strain are due to autolysis. However, the differences in secretion in terms of proteins secreted per biomass are already detectable for all strains after 24 hours of cultivation, when the fungi are clearly in their production phase and are actively growing (Table 2 ), which renders this hypothesis unlikely. Despite the high endoglucanase activity values of Δenv1 , only slightly increased protein concentration was observed. Therefore, in contrast to Δblr2 , the enzyme mixture secreted by Δenv1 seems to be more efficient than that of the other strains investigated in this study. Since in several promotors of plant cell wall degrading enzymes of T. reesei promotor motifs known to be involved in light response in other fungi have been detected (Table 1 ), light dependent adjustment of the proportions of the respective enzymes could be one way to achive this enhanced efficiency. For the photoreceptor mutants even decreased endoglucanase and FPA performance (EG activity per mg of secreted protein) was obtained. Interestingly, while the biomass-specific FPA for Δblr1 is clearly lower than for all other strains, including wild-type, the protein specific FPA is comparable to Δblr2 . Consequently, the rather low efficiency of this strain despite its enhanced growth may be due to a decreased protein secretion efficiency caused by deletion of blr1 . Analysis of proteins secreted by mutant strains In order to analyze whether the increased efficiency is due to an altered expression of the major cellobiohydrolase of T. reesei CBH1/Cel7a, Western blotting was performed (Figure 2 ). In case of Δ blr2 clearly higher abundance of CBH1 was observed, indicating that the increased cellulolytic efficiency of this strain is at least in part due to this enzyme. For Δ env1 the effect was less clear, suggesting that the improvement of the cellulase mixture found for this strain is likely to be caused by altered regulation of expression of the pool of enzymes contributing to cellulose degradation. Considering the data above, these results correspond well with the enhanced protein secretion capacity of Δblr2 and the obviously improved enzyme proportions for Δenv1 . Competing interests The authors declare that they have no competing interests. Authors' contributions MGK performed fermentations (together with ZM and GN) and sequence analysis and participated in drafting the manuscript. AS performed Western blot analysis. KR supervised the fermentations. MS conceived of the study, participated in its design and coordination and wrote the final version of the manuscript. All authors read and approved the manuscript.

Acknowledgements We want to thank Alfredo Herrera-Estrella for the gift of strains Δblr1 and Δblr2 and we gratefully acknowledge Christian P. Kubicek for critically reading the manuscript. This work was supported by the Austrian Research Fund (FWF), grant P21072 to MS. MGK was supported by the COST action BIOBIO (FP602).

CC BY

no

2022-01-12 15:21:44

BMC Res Notes. 2010 Dec 7; 3:330

oa_package/ea/e4/PMC3016267.tar.gz

PMC3016268

21172017

Background Chronic lymphocytic leukemia (CLL) is the most common adult leukemia in the western population. Although genetic factors are considered to contribute to CLL etiology, at present genomic aberrations identified in CLL are limited compared with those identified in other types of leukemia, which raises the question of the degree of genetic influence on CLL. We performed a high-resolution genome scanning study to address this issue. Findings Using the restriction paired-end-based Ditag Genome Scanning technique, we analyzed three primary CLL samples at a kilobase resolution, and further validated the results in eight primary CLL samples including the two used for ditag collection. From 51,632 paired-end tags commonly detected in the three CLL samples representing 5% of the HindIII restriction fragments in the genomes, we identified 230 paired-end tags that were present in all three CLL genomes but not in multiple normal human genome reference sequences. Mapping the full-length sequences of the fragments detected by these unmapped tags in seven additional CLL samples confirmed that these are the genomic aberrations caused by small insertions and deletions, and base changes spreading across coding and non-coding regions. Conclusions Our study identified hundreds of loci with insertion, deletion, base change, and restriction site polymorphism present in both coding and non-coding regions in CLL genomes, indicating the wide presence of small genomic aberrations in chronic lymphocytic leukemia. Our study supports the use of a whole genome sequencing approach for comprehensively decoding the CLL genome for better understanding of the genetic defects in CLL.

Findings CLL (Chronic lymphocytic leukemia) is an incurable disease mainly affecting the B cell lineage in the western population, with a median age of diagnosis of 72 year old [ 1 ]. Determining the cause of CLL is crucial for understanding the acquisition and for clinical diagnosis, treatment and prognosis of CLL. Genetic factors have been linked to the etiology of CLL. Cytogenetic analyses identified chromosomal abnormalities including del11q23 affecting the ATM gene, tri12, del 13q14, and del17p13 affecting TP53 gene [ 2 ]. In addition, CGH studies found gains and losses in Xp11.2-p21 and Xq21-qter [ 3 ]. Molecular studies identified three genes: IgVH , CD38 and ZAP-70 that correlate with CLL prognosis [ 4 - 6 ]. A CLL-specific microRNA signature was also identified, suggesting that microRNA deletion could be involved in CLL [ 7 ]. SNP array studies identified 2q21.2, 6p22.1 and 18q21.1 abnormalities that follow a Mendelian inheritance pattern [ 8 ]. Whole genome association studies also identified multiple loci at 2q37.3, 8q24.21, 15q21.3 and 16q24.1 that appear to be associated with genetic susceptibility to CLL [ 9 ]. Although evidence supports the involvement of genetic factors in CLL, the frequency of genomic aberrations identified in CLL is relatively lower than those observed in the leukemias affecting other types of hematopoietic lineages [ 10 ]. This information suggests that the CLL genome is relatively intact with fewer aberrations than other types of leukemia. Alternatively, more genomic aberrations may exist in CLL but these could mainly be small lesions in the CLL genome that are difficult to detect using conventional technologies due to their limited resolution. With the rapid progress of genome sequencing technologies, enthusiasm is increasing for pursuing comprehensive detection of genomic aberrations in cancer by sequencing cancer genomes. In the case of CLL, a critical issue is to know the degree of genomic aberrations in order to justify the use of whole genome sequencing approach to analyze CLL genome. We reasoned if we can scan certain CLL genomes at sufficient high resolution and at reasonable genome coverage, we should gain first-hand information to estimate the degree of genomic aberrations in CLL. We recently developed the DGS (Ditag Genome Scanning) technique that uses next-generation DNA sequencing technologies to collect paired-end sequences from restriction DNA fragments across a genome [ 11 ]. Using this technique, we analyzed CLL genomes. Nine samples of peripheral blood from untreated CLL patients diagnosed in Northwestern University Lurie Cancer Center and University of Chicago Medical Center were used in this study, of which three were used for paired-end tag collection, and eight including two used in paired-end tag collection were used for full-length sequencing analysis (Additional file 1 : Supplemental Table S1). Informed consent was made by the patients, and the use of clinical CLL samples was approved by the institutional review board of University of Chicago and Northwestern University following institutional guidelines. The detailed experimental process followed the published protocol [ 11 ] and outlined in Figure 1 . Briefly, mononuclear cells were isolated from each CLL peripheral blood or bone marrow sample by using NycoPrepTM A solution (Axis-Shield). Human genomic DNA was extracted from mononuclear cells by using QIAamp DNA Blood Kit (QIAGEN) following the manufacturer's protocol. To generate the DGS library, genomic DNA was fractionated by HindIII restriction digestion. The restriction fragments were dephosphorylated by CIP and cloned into pDGS- HindIII vector that contains two MmeI sites next to the HindIII cloning site. The genomic library was digested by MmeI to release two tags from the cloned DNA fragments. The tag-vector-tag fragments were then gel-purified, and re-ligated to form a ditag library. Ditags were released from the vectors by HindIII digestion, gel-purified, and concatemerized by using T4 DNA ligase (Promega). The concatemers at 200 to 500 bps were agarose-gel-purified and used for ditag sequencing by using a 454 GS20 sequencer (454 Life Sciences). Ditags were extracted from the resulting sequences based on the HindIII sites. Same ditags were combined to generate a unique ditag with the corresponding copy numbers. To generate the reference ditag database, virtual HindIII restriction fragments were generated from known human genomic sequences. Two 16-bp virtual tags were extracted from the 5' and the 3' ends of each virtual fragment, and connected to form a reference ditag representing the virtual DNA fragment. The following sequences were used to extract the reference ditags: 1. Human genome reference sequences (hg18): http://hgdownload.cse.ucsc.edu/goldenPath/hg18/bigZips/ 2. Human dbSNP 126: ftp://ftp.ncbi.nih.gov/snp/organisms/human_9606 3. Chimpanzee genome reference sequences (PanTro2): http://hgdownload.cse.ucsc.edu/goldenPath/panTro2/bigZips/ 4. Human GM15510 fosmid paired-end sequences: http://www.ncbi.nlm.nih.gov/Traces/trace.cgi?&cmd=retrieve&val=CENTER_PROJECT%20%3D%20%22G248%22&size=0&retrieve=Submit 5. Celera human genome sequences: http://www.ncbi.nlm.nih.gov/genomeprj/1431 6. Venter genome sequences: ftp://ftp.ncbi.nih.gov/pub/TraceDB/Personal_Genomics/Venter/ 7. Watson genome sequences: ftp://ftp.ncbi.nih.gov/pub/TraceDB/Personal_Genomics/Watson/ 8. Reference ditags were also extracted from HindIII fragments of E.coli K12 genome sequences to eliminate the ditags from E. coli DNA contaminated during library construction process. Initial ditag mapping was performed with perfect match between experimental ditags and hg18 reference ditags. For the unmapped experimental ditags, a single-base mismatch in each single tag of the ditag was allowed to compensate for possible sequencing error or SNP. To identify the unmapped ditags related with homopolymer generated by 454 sequencing chemistry, the unmapped ditags with more than two homo-bases were stretched, e.g. AAA -> AAAA, or shortened, e.g. AAA -> AA, and mapped to reference ditags again. For the ditags remaining unmapped, they were mapped to the reference ditags of other sequence sources in the ditag reference database. The ditags remaining unmapped after these processes were defined as the unmapped ditags. Unmapped ditag sequences were used to design sense primers and antisense (reverse/complementary) primers, with four extra bases CAGC added to the 5' end of sense primer and CGCC to the 5' end of antisense primer. Genomic DNA digested by HindIII was used as the templates for PCR amplification. PCR was performed with 35 cycles at 95°C 30 sec, 57°C 60 sec, and 72°C 3 min, followed by extension at 72°C for 10 min. The amplified products in each reaction were cloned into pGEM-T vector (Promega), transformed into E. coli TOP10 (Invitrogen), and plated in a single well of the 48-well Qtrays (Genetix). Four clones from each transformation were amplified by colony-PCR using M13F and M13R primers, and sequenced by Big-Dye Terminator v3.1 Cycle Sequencing Kit (ABI) using M13F primer. For the sequences that did not reach the full-length, second sequencing reactions were performed using M13R primer. To determine the genomic aberrations, each full-length sequence was mapped to hg18 using BLAT at a minimum of 90% identity as the cut-off. The paired-end ditags were collected from three CLL samples. Genomic DNA from each sample was fractionated by HindIII digestion, which provides 3,561-bp resolution on average across the genome based on hg18 sequences [ 11 ]. Unique paired-end ditags of 272,193, 320,283, and 307,547 was collected from each CLL sample, covering 32%, 34% and 38% HindIII fragments in each CLL genome respectively. Comparing the three ditag sets shows that between 87,968 and 108,579 ditags are present between two CLL samples, and 51,632 ditags are commonly present in all three CLL samples (Table 1A ). The ditags present only in individual CLL sample could be the ditags representing individual genomic differences, the ditags potentially originating from experimental artifacts, or ditags detected in one but not in others due to unsaturated ditag collection in each CLL under the sequencing scale. The 51,632 ditags detected in all three CLL samples cover 5% of genomic DNA fragments commonly detected in the three CLL genomes. In order to provide high confidence for further downstream studies, we focused on the 51,632 common ditags for further mapping analysis. We compared the 51,632 common ditags with multiple known human genome sequences, including the human genome reference sequence hg18, human SNP, human GM15510 genome sequences, chimpanzee genome sequences that are highly homologous to the humans, Watson genome sequences, and Venter genome sequences. Of the 51,632 ditags used for the mapping, 98.3% (50,799) map to hg18 that represent normal genomic fragments in the CLL genomes, 0.4% (230) are unmapped ditags that represent potential genomic aberrations commonly present in all three CLL genomes, and the remaining ditags map to other genomes that represent normal genome variations (Table 1B ). To determine the types of genomic aberrations for the unmapped ditags, we generated full-length sequence for the restriction DNA fragment detected by the unmapped ditags by using the "ditag-PCR" method, in which the ditag sequences were used as PCR sense and antisense primers to amplify the original DNA fragment that derived the unmapped ditag. We performed 192 reactions in eight CLL samples including two used in ditag collection and six additional CLL samples. Under the conditions that a full-length sequence must be longer than 50 bases and detected at least in the CLL used in ditag collection or at least in two additional CLL samples, 220 full-length sequences were generated from 100 unmapped ditags. Mapping the full-length sequences to hg18 identified different types of genomic aberrations caused by insertion, deletion and base change. Many of these aberrations created new HindIII restriction site that leads to the release of unmapped ditag, or the change of ditag sequence composition that prevents ditag mapping. These aberrations were observed in both coding and non-coding regions in CLL genome. For example, aberrations were detected in exons of NEK8, RUNX1 and MUC2 genes, and introns of 20 other genes (Table 2A , Additional file 2 : Supplementary table S2). NEK8 encodes a member of the serine/threonine protein kinase family, which plays a role in cell cycle progression from G2 to M phase and is over-expressed in breast cancer [ 12 ]. A 353-base sequence converted from the unmapped ditag AAGCTTACCCTCTGGACGCCTGTATGAAGCTT maps to the last exon (Exon 15) coding for the 3' UTR of NEK8. Two HindIII restriction sites were inserted in the sequence that are not present in the wild-type NEK8 gene. RUNX1 is a gene involved in AML through its involvement in the t(8;21) [ 13 ]. A 434-base full-length sequence from a ditag AAGCTTCGGCCTATAG/ACAACCTAACAAGCTT was detected in all eight CLL samples, and maps to intron 3 and exon 4 of RUNX1. Analyzing the mapped region shows a T to C single-base change between the sequence and exon 4 of RUNX1 gene. Searching dbSNP reveals that this is a SNP (rs1235270). Due to the uncertainty of RUNX1 protein coding sequence itself, it is not certain if this germline SNP causes a coding amino acid change. Several bases are also changed in the mapped intron 3 of RUNX1 gene. These base changes raise an interesting question whether RUNX1 could be involved in CLL. MUC2 is a member of the MUCIN family, which codes for high molecular weight glycoproteins. The abnormalities of MUC2 is linked with colorectal and pancreas cancer [ 14 ]. A 410-base sequence derived from an unmapped ditag AAGCTTCCGGTCGGCTTCGCAGTAGAAAGCTT covers intron 29, exon 30 and intron 30 of MUC2 gene. This sequence also contains two HindIII restriction sites AAGCTT inserted at both its ends that do not exist in the wild-type MUC2 gene. Only three aberrations were detected in the exon of three known genes. This could be attributed to the limited genome coverage of the study and the low percentage of the exon-coding sequences in the genome. With increased genome coverage, it would be possible to identify the aberrations affecting more exons. Aberrations also affect the introns of multiple genes. FHIT encodes diadenosine 5',5'''-P1,P3-triphosphate hydrolase involved in purine metabolism [ 15 ]. It is located in the common fragile site FRA3B on chromosome 3, where carcinogen-induced damage can lead to translocations in several cancers. A 283-base sequence maps to intron 8 of FHIT gene but its tag 1 contains GA to TG change. HYDIN encodes an axonemal protein; mutation of HYDIN is related to congenital hydrocephalus [ 16 ]. Two full-length sequences of 605-bp and 614-bp from two different unmapped ditags were obtained from seven CLL samples. Both sequences map to 21st intron of HYDIN. The 605-bp sequence contains CCTACGGCG in its tag 2 converted from wild-type gCcACaGCa (lowercase refers to the changed base), and the 614-bp sequence contains CGCC converted from wild-type tGCt in its tag 1 and an internal insertion. NCOR2 is a transcriptional regulator that recruits histone deacetylases to promoters [ 17 ]. A 582-base sequence maps to intron 1 of NCOR2, but its tag 1 contains an AAGC insertion, and tag 2 contains a C to T change, an AG deletion, and a T insertion. TYK2 is a member of the JAK family involving in IFN-g, IL-6, IL-10 and IL-12 signaling. Mutation in this gene is associated with hyperimmunoglobulin E syndrome [ 18 ]. A 268-base sequence maps to intron 14 of TYK2 but its tag 1 contains an AAGCTTA insertion and its tag 2 contains a TGAAGCTT insertion. Both insertions create HindIII restriction sites that lead to the generation of the unmapped ditag. A 197-base sequence was detected in seven CLL samples and two different sequences of 112-base and 170-base were generated from the CLL used in ditag collection. All three sequences map to UBAP2 located at 9p13.3, a gene involved in the ubiquitination pathway [ 19 ]. For the 197-base sequence, its 178 bases map to intron 6 of UBAP2 gene and the remaining 18 bases have no map, whereas the 112-base and 170-base sequences contain different insertions. Although the aberrations in many of these genes have been correlated with different types of cancer, most have not been linked with CLL. Non-coding regions contribute to the majority of the genome, and contain important functional elements involving DNA replication, genome stability, regulation of gene expression, and coding for non-coding transcripts etc. Extensive characterization of non-coding region could provide rich candidate markers for clinical applications and identify the hotspots of genomic aberrations involving cancer development. A total of 37 sequences generated from 30 unmapped ditags mapped to the non-coding regions in the genome with various types of abnormalities (Table 3 , Additional file 3 : Supplemental Table S3). Although these loci are not directly located in the coding regions, many genes are located nearby the mapped locations. Of the 26 loci specifically mapped by the sequences, 15 have genes located either upstream, downstream or both within 100 kb distance. For example, a 614 base sequence maps to 5q35.1 between169443856-169444467, where DOCK2 is located 27,836 base upstream and FOXI1 is located 21028 downstream. A 398-base sequence maps to 15q26.1 between 88110782 and 88111168, where two homologous transcriptional factor genes, MESP1 and MESP2, are located 16,678-base upstream and 9,425-base downstream correspondingly. microRNA gene MIR663 are located 20,580 base upstream of 20p11.1 between 26157494-26158252 mapped by a 920-base sequence detected in seven CLL samples. Another microRNA gene MIR663B is located 10,964-base upstream of 2q21.2 between 132742087-132742356 mapped by a 290-base sequence, of which a non-coding RNA gene NCRNA00164 is located in between. The aberrations could affect the nearby genes through influencing the regulation of gene expression. One hundred and forty seven full-length sequences converted from 57 unmapped tags map to the highly repetitive sequences in the non-coding regions. Of these sequences, 110 sequences map to the ALR/Alpha satellite sequences of the centromere, and chromosome 2, 10, and 17 are among the most frequent ones (Table 4 , Additional file 4 : Supplemental Table S4): 23 sequences converted from 13 unmapped tags map to the centromere of chromosome 2 at 2p11.1, 41 sequences converted from 16 ditags map to the centromere of chromosome 10 at 10q11.1, and 22 sequences converted from 6 unmapped ditags map to the centromere of chromosome 17 at 17p11.1. The presence of highly frequent aberrations in ALR/Alpha satellite sequences in these three chromosomes suggests that these could be the hot spot of genomic aberrations in CLL. Aberrations in repetitive sequences have been shown to contribute to cancer development [ 20 ]. However, it is difficult to analyze the aberrations in these highly repetitive regions using the hybridization-based approach due to the difficulty to designing specific probes. Our results show that restriction sequencing-based approach provides a useful tool to study the aberrations in these regions. Ten full-length sequences generated from eight unmapped ditags did not map to known human genome sequences (Table 2B . Additional file 5 : Supplementary table S5). For example, a 107-base full-length sequence converted from an unmapped ditag AAGCTTAGATAGAGCGCAGTCAACTGAAGCTT was detected in all eight CLL samples. However, it does not map to the reference genome sequences. These sequences represent the DNA contents present in CLL genomes but not in normal genomes. Through high-resolution scanning of three CLL genomes and verifying the results using full-length sequences and additional CLL genomes, our study provides evidence showing the wide presence of genomic aberrations in CLL, of which most are small lesions. Studies with increased number of CLL samples and at high genome coverage will be required to better understand the genetic aberrations in CLL. Although the study used multiple genomics databases to eliminate the changes from normal genomic polymorphism, further studies with normal DNA from the same patient will be required to fully distinguish somatic mutations from germline variations in CLL. Competing interests The authors declare that they have no competing interests. Authors' contributions YK, YCJ, JC, AHA, PK performed laboratory work. YZ, SM, SR provided clinical samples and data analysis, SR, SW designed the experiment. SW wrote the paper. All authors read and approved the final manuscript. Supplementary Material

Acknowledgements We wish to thank Dr. Janet Rowley for comments on the study. The study was supported by Mazza Foundation and Guglielmi Fidelity Charitable Fund (SW).

CC BY

no

2022-01-12 15:21:44

BMC Res Notes. 2010 Dec 20; 3:341

oa_package/25/a3/PMC3016268.tar.gz

PMC3016269

21176142

Background Shiga toxin-producing E. coli (STEC) O157 has emerged as a public health threat following its initial identification as a pathogen in a 1982 outbreak of illness associated with the consumption of undercooked ground beef [ 1 ]. Specifically, E. coli O157:H7 and O157:NM (nonmotile) are recognized as major etiologic agents in hemorrhagic colitis (HC) and hemolytic-uremic syndrome (HUS) in humans. This enteric organism is able to secrete Shiga toxin, a binary toxin that affects the endothelium of the kidney, gut and brain that can result in glomerular vascular damage, bloody diarrhea, and brain edema, distinguishing it from other pathogenic strains of E. coli . STEC has been implicated as the source of numerous outbreaks and sporadic cases stemming predominantly from consumption of bovine food products. Transmission has also been linked to unpasteurized milk and cider, contaminated drinking and swimming water, fresh vegetables and secondarily through person-to-person contact. Infection due to E. coli O157:H7 or enterohemorrhagic E. coli (EHEC) is typically characterized by diarrhea, abdominal cramping and hemorrhagic colitis. Hemolytic uremic syndrome and thrombotic thrombocytopenic purpura are less common, but severe sequelae of infection. The public health impact of EHEC infections is high because of systemic complications from infections, such as HUS (an important cause of acute renal failure in childhood), late post-infection sequelae[ 2 ] and the ability of STEC to cause large outbreaks. To date, EHEC O157:H7 has caused hundreds of outbreaks worldwide[ 3 ]. The largest known outbreak on record occurred in Japan, in Sakai City, in 1996[ 4 ], where thousands, mostly school children, were affected. The U.S. Centers for Disease Control and Prevention estimates that E. coli O157:H7 causes approximately 73,400 illnesses and 60 deaths each year in the United States[ 5 ]. Cattle are a major reservoir of E. coli O157:H7[ 6 ], but additional potential reservoirs of this pathogen include sheep, goats, pigs, horses, dogs, poultry, and deer[ 7 - 10 ]. The ability to identify accurately and track the strains of infectious agents that cause disease is central to epidemiological surveillance and public health decisions, but there are no wholly satisfactory methods of achieving this goal[ 11 ]. All of the numerous methods that are currently used suffer from one or more significant drawbacks, including inadequate discrimination, limited availability of reagents, poor reproducibility within and between laboratories, and ability to quantitatively define genetic relationships between isolates. However, perhaps the most important limitation of current typing methods is the difficulty of comparing the results achieved by different laboratories. Molecular typing methods are used to address two very different kinds of problems. First, for short-term or local epidemiology, the isolates recovered from a localized outbreak of disease need to be accurately grouped using previously characterized isolates as benchmarks. Second, for long-term or global epidemiology, pathogenic strains collected over relatively long time periods and/or from disparate geographic regions must be correctly classified as related or not to those isolated world-wide. Different methods may be appropriate for investigating local and global epidemiology, but in both cases they should be discriminatory enough so that isolates can be classified. High specificity levels of isolate strain identity can be achieved in two different ways. In one approach, individual loci, or uncharacterized and dispersed regions of the genome, that are highly variable within the bacterial population are identified. For bacterial pathogens, several methods based on this approach are currently popular, e.g., ribotyping, pulsed-field gel electrophoresis (PFGE), and PCR with repetitive element primers, or arbitrary primers [ 11 ]. For these methods, restriction enzymes (or PCR primers) are chosen that yield maximal variation within the population; consequently, the variation that is indexed is evolving very rapidly, usually for reasons that are not clear. The second approach, typified by multilocus enzyme electrophoresis (MLEE), is to use variation that accumulates very slowly in the population and is likely to be selectively neutral. Although only a small number of alleles can be identified within the population by using this type of variation, high levels of discrimination are achieved by analyzing multiple loci. Methods that index rapidly evolving variations are useful for short term epidemiology but may generate misleading results when applied in a global epidemiological survey. Several studies have shown that techniques such as PFGE resolve isolates that are indistinguishable by MLEE. For example, MLEE studies of populations of Salmonella enterica have shown that isolates of serovar Typhi from typhoid fever belong to one of two closely related electrophoretic types (ETs)[ 12 ]. In contrast, isolates of serovar Typhi are relatively diverse according to PFGE[ 13 ]. PFGE is therefore useful for studying individual outbreaks of typhoid fever because, unlike MLEE, it identifies the microvariation (some of which is caused by transient or unstable genetic elements or regions i.e. phages, transposons, genetic rearrangements, repetitive elements etc) that is needed to distinguish between strains circulating within a given geographic area. However, this technique is not well suited for long term epidemiology (and occasionally for short term epidemiology for reasons described above) because it does not indicate that isolates that cause typhoid fever are members of a single globally distributed clonal lineage of S. enterica . Metaphorically, PFGE and other similar methods may cause epidemiologists not to see the forest for the trees. The best current techniques for long term epidemiology, and for the identification of lineages that have an increased propensity to cause disease, is undoubtedly MLEE. This approach also has contributed greatly to our understanding of the global epidemiology and population structure of infectious agents. For many pathogens, MLEE successfully has identified clusters of closely related strains (clones or clonal complexes) that are particularly liable to cause disease[ 11 ]. However, a major problem with MLEE, and virtually all other current (gel-based) typing methods, is that the results obtained in different laboratories are difficult to compare. Multilocus sequence typing (MLST) uses nucleotide sequences of internal fragments of selected genes as the unit of comparison and, therefore, does not suffer from the drawbacks of gel-based fingerprinting methods. Sequence data are unambiguous, more easily comparable, and transferable between laboratories and are highly reproducible[ 11 ]. Furthermore, the digital format of MLST data has facilitated the establishment of global, web-accessible databases for a variety of organisms and is rapidly contributing to our understanding of the clonal distribution of infectious disease agents. The most commonly used MLST schemes index the nearly neutral genetic variation in housekeeping genes, which are believed to evolve slowly because they are for the most part, under stabilizing, selective pressure. MLST is thus a powerful tool for global and long-term surveillance. Various authors [ 13 - 16 ] have reported the use of MLST methodology for the study of E. coli , particularly the serogroup O157. These authors have reported use of various house keeping genes for analysis of E. coli O157 strains by MLST. The development of an effective MLST scheme for subtyping E. coli O157:H7 has been hindered in the past due to lack of sequence variation found within analyzed housekeeping and virulence genes. A recent study suggested that rhs genes are under strong positive selection pressure and therefore, may be useful markers for phylogenetic analysis of E. coli O157:H7 [ 15 - 17 ]. EHEC differ from EPEC in that they produce Shiga toxins but not bundle-forming pili. The same authors also reported that E. coli with the O157 O antigen are not always EHEC but may belong to other pathotypes. They also described two serogroups of O157 E. coli strains from Brazilian infants with diarrhea and shown with variety of assays that those strains belonged to the enteropathogenic, not the enterohaemorrhagic pathotype. The putative virulence factors of O157:H7 strains, the ability to produce Shiga-like toxins and adhere to epithelial cells, also exist in other groups of E. coli [ 18 ]. In this study we have performed MLST on a set of E. coli O157 clinical isolates obtained from two different locations of USA to determine whether (or not) any genetic diversity exists among the isolates.

Materials and methods Bacterial isolates A total of 33 Escherichia coli O157 isolates (23 from the Wadsworth Center, New York State Department of Health, Albany, New York, USA and 10 from Pennsylvania State, USA) were used in the study. The samples were collected from the patients with HUS, diarrhea or asymptomatic. Age group of the patients varied from 1 year to 42 years. The characteristics of the isolates used are shown in Table 3 . Reference strains used in this study were EDL933 (accession no. NC_002655 ) and Sakai (accession no. NC_002695 ). Examination of virulence genes The virulence genes were investigated by PCR as listed in Table 2 . Selection of genes The following seven housekeeping genes were included in the study: arcA (aerobic respiratory control protein), aroE (shikimate dehydrogenase), dnaE (DNA polymerase III, alpha subunit), mdh (melate dehydrogenase), gnd (6-phosphogluconate dehydrogenase), gapA (glyceraldehyde 3-phosphate dehydrogenase), and pgm (phosphoglucomutase). In addition to these genes, two membrane protein coding genes espA ( E. coli secreting protein A) and ompA (outer membrane protein A) were also included. The seven selected housekeeping genes were chosen for their potential sequence diversity. Three of the genes, aroE, arcA and mdh have been used to determine the evolution of pathogenic bacteria[ 28 ]. Two genes, dnaE and pgm were chosen because they were found to be informative for Salmonella and Vibrio cholerae [ 29 ]. The last two housekeeping genes, gapA and gnd were chosen because they were transferred into the O157 genome at different evolutionary times. Finally, the two membrane proteins were chosen as being potential targets of the immune system and presumably, under balancing selective pressure. The primers for these genes were from Noller et al[ 13 ] except for genes pgm and espA . The primers for these two genes were designed based on the published sequences from Genbank (Table 4 ). DNA Isolation E. coli isolates were obtained from a collection of strains stored in 15% glycerol at -80°C. Isolates were incubated at 37°C on LB agar plates overnight. Single colonies were picked and inoculated into 2 ml LB media and further incubated in a shaking incubator for 12-15 h at 37°C. A 1 ml suspension of bacteria was centrifuged, and DNA was extracted from the bacterial pellet with the QIAGEN DNeasy blood and tissue kit (QIAGEN, Valencia, CA). PCR Amplifications were carried out in a total volume of 50 μl with 2 μl of template DNA (50-100 ng), 1 μl of each 20 mM primer, 5 μl 10× Accuprime PCR buffer II (Invitrogen) and 1 μl of Accuprime Taq DNA Polymerase (Invitrogen). PCR was performed in a GeneAmp PCR System 9700 (Applied Biosystems, USA). Primers were synthesized by Integrated DNA Technologies, Inc. (Coralville, Iowa). PCR conditions depended on the different target genes and primer sequences used are depicted in Table 4 . Sequencing PCR products were purified for sequencing with the QIAGEN QIAquick PCR purification kit. Both the forward and reverse strands were sequenced with the PCR primer set. Sequencing was performed at Cornell University Life Sciences Core Laboratories Centre using the Applied Biosystems automated 3730 DNA analyzer using big dye terminator chemistry and AmpliTaq-FS DNA polymerase. Phylogenetic analysis Following sequencing of forward and reverse strands, the sequences were edited and aligned using Bioedit version 4.8.10[ 30 ] and converted into FASTA files. For each gene fragment, distinct alleles were identified and numbered by using the non-redundant databases (NRDB) program http://pubmlst.org/analysis/ . When combined, the allele numbers assigned to each of the nine loci constituted a strain's allelic profile or sequence type (ST). Thus, each distinct allelic profile was considered a unique sequence type. A dendrogram (based on Neighbor Joining Method) was constructed by using the web-based data analysis tool "Tree drawing" http://pubmlst.org/analysis/ which uses the PHYLIP suite of programmes to generate neighbor-joining and UPGMA trees from allelic profile data. eBurst, a noncommercial algorithm previously designed for MLST of bacterial pathogens, was used to divide the 33 E. coli isolates into clusters of genetically related strains. The eBurst algorithm groups strains according to their allelic profiles by employing a user-specified group definition, which is the number of alleles that the isolates need to have in common to belong to the same group http://www.mlst.net/ . The relatedness among the 33 E. coli isolates was also assessed by SplitsTree analysis[ 31 ], an alternative algorithm for the analysis and visualization of evolutionary data that is not always best represented by a standard tree. Nucleotide sequence accession numbers The nucleotide sequences obtained by sequencing of the PCR products of each allele of all genes have been entered into the GenBank databases under different accession numbers which are shown in Table 5 .

Results Multilocus sequence analysis Nucleotide sequence analysis of unlinked housekeeping genes (multilocus sequence typing) is widely used for evolutionary and population analysis, and also for epidemiological investigations. Since differences in the sequences of essential housekeeping genes are thought to display long-term genetic changes, we used this technique to investigate the phylogenetic relationships of E. coli O157 clinical isolates obtained from New York State and Pennsylvania State of USA. DNA sequences of corresponding housekeeping genes from the O157:H7 strains EDL933 (accession no. NC_002655 ) and Sakai (accession no. NC_002695 ) were compiled from the respective published genome sequences and used as reference strains for this study. Sequence data from each isolate, including the two published strains, was then garnered from each of the nine loci. Forward and reverse sequence reads from each locus were aligned and edited. Analogous to other MLST schemes, we assigned the allele numbers to each unique allele sequence for each of the nine loci investigated. The number of alleles identified for the 33 E. coli O157 isolates varied from 3 to 14 per gene; ompA generated the least number of alleles (3 alleles), while pgm generated the most (15 alleles). The combinations of allele numbers for all isolates are shown in Table 1 . Each unique combination of allele numbers represents one sequence type (profile). The allelic diversity found at the nine loci resulted in 26 unique ST for the 33 isolates. Five (14.28%) of these showed MLST profile 2, four isolates (11.42%) showed MLST profile 1 and the remaining 24 strains fell into MLST profile 3 to 26. Figure 1 shows a dendrogram drawn by using Tree drawing, a web-based software used for displaying phylogenies in the form of tree dendrograms. From the dendrogram it has been observed that the isolates in 12 profiles (ST1, ST2, ST5, ST6, ST7, ST10, ST13, ST14, ST15, ST16, ST25 and ST26) are found to be closely related to the reference strains EDL933 and Sakai in comparison to rest of the isolates in the study. Isolates no. 27 and 28 (ST21 and ST22) are also closely related to each other. Analysis by eBurst revealed two non-overlapping groups or clonal complexes, consisting of related isolates sharing identical alleles at eight out of the nine loci with at least one other member of the group (Figure 2 ). The relatedness among the 33 E. coli O157 isolates was also analyzed with SplitsTree, an alternative algorithm for the analysis and visualization of evolutionary data that is not always best represented by a standard tree. The SplitsTree graph (Figure 3 ) showed a clustering of strains highly similar to those obtained with both the eBurst algorithm and dendrogram (based on Neighbor Joining Method). Analysis of virulence markers All E. coli O157 isolates were investigated for the presence of virulence genes associated with EHEC (Table 2 ). All isolates were also screened for the presence of bfpA gene (which is associated with typical EPEC), because of the growing evidence that some EHEC share common genes with EPEC. All isolates except for two (isolate no. N6 and N9) were positive for at least one of the Shiga toxin genes (either stx1 or stx2 or combination of both). eaeA and hlyA genes were carried by 78.78% and 60.60% of the isolates, respectively. None of the isolates were found to be positive for bfpA .

Discussion Molecular epidemiology is largely descriptive and characterizes bacteria based on their natural genetic variation. A number of molecular methods have been employed to determine this variation, which may be due to mutation or horizontal gene transfer events. Basically two branches of molecular epidemiology can be distinguished, classification and typing. Classification can illustrate evolutionary relationships and groups species into clonal groups or complexes, whereas typing is used more often for differentiation of clinical or environmental isolates. Long-term epidemiology may require classification, whereas short-term local epidemiology is more often performed by high resolution typing methods. While the most prominent methods for molecular classification are MLST and multilocus enzyme electrophoresis[ 11 ], PFGE or PCR-based methods such as randomly amplified polymorphic DNA-PCR, AFLP or virulence gene characterization, have been used frequently for typing approaches[ 19 ]. In our study, we used MLST in combination with PCR detection of selective virulence genes for classification of pathotypes and for phylogenetic analysis of E. coli O157 isolates. The isolates were collected from patients over a period of several years (2000-2008) from two different states of the United States of America., Some of these isolates were associated with outbreak cases. In our study, MLST analysis of nine genes (seven housekeeping genes and two membrane protein coding genes) revealed 26 profiles from 33 E. coli O157 isolates. Analysis by eBurst revealed two clonal groups or clonal complexes (clonal group I and II). All the isolates in clonal group I were closely related to reference strains EDL933 and Sakai. It has also been observed that the leading markers of clonal group I are the presence of stx1 and/or stx2 genes, eaeA gene and hlyA gene, characterizing them as typical STEC. The only difference was that the isolate N29 and N32 were negative for the presence of the hlyA and eaeA genes, respectively. We also observed that, of the nine loci investigated, the pgm gene generated the highest number of alleles, which could be due to use of a larger DNA fragment for amplification of the pgm gene in the study. One surprising finding of our study is that although the isolates N6 and N9 were negative for stx genes still they belonged to the same clonal group along with the EDL933 and Sakai strains. It is possible that these strains might have lost the phage-borne stx genes upon subculture as reported for other STEC[ 20 , 21 ]. Spontaneous loss of both stx1 and stx2 genes in vitro has also been described in E. coli O157:H7 clinical isolate[ 22 ]. The timing of the stool sample collection is also very important for finding Stx producing strains in patients with HUS. Tarr et al. demonstrated that if the stool samples of patients with HUS were cultured within 6 days after onset of diarrhea for EHEC O157:H7, the recovery rate was nearly 100% [ 23 ]. This rate decreased to 33.3% in stool samples collected >6 days after the onset of diarrhea. Mellmann et al. also demonstrated that EHEC are difficult to identify in patient's feces at late stages in illness[ 24 ]. Friedrich et al. has proposed two possible explanations for the findings of stx -negative E. coli O157 strains related to sporadic cases of diarrhea and HUS and to outbreaks[ 25 ]. First, stx -negative E. coli O157 strains are thought to have evolved from stx -positive E. coli O157 of an original infection by loss of stx genes during the course of infection. Alternatively, E. coli O157 strains that are inherently stx -negative from the beginning might be responsible for the disease in these cases. Maintaining the Stx-encoding phage may adversely affect or be lethal to the bacterium when changes occur in normal environmental conditions or may become so when in the human host during the course of infection. Survival might be favored by loss of the phage, since stx - progeny of stx + progenitors are less prone to lysis. Mammalian host signals such as those initiated by exposure to hydrogen, can perhaps induce Stx-encoding prophages to become lytic[ 26 ]. By generating stx - mutants, a strain can survive without automatically lysing (because it no longer carries the lysogenic phage or the phage-encoded toxin). The loss of stx -encoding phage can thus offer a selective advantage. The finding of stx -negative E. coli O157 isolates has clinical significance and is important from a diagnostic standpoint because in such cases, stx - and Stx-independent procedures are required to detect strains that might have lost their stx genes. Our study indicates that routine testing of stools for either Stx by ELISA or the stx genes by PCR or merely relying on the results from culturing E. coli O157 on sorbitol MacConkey agar, may not identify all potentially important factors that are associated with diarrhea and HUS. The same observation was made during recent outbreaks of gastroenteritis in the United States. Recently Bielaszewska et al.[ 24 ] also reported that at the time of microbiological analysis, ~5% of HUS patients no longer shed the causative EHEC, but excrete stx -negative derivatives of EHEC that have lost stx during the course of infection. In such patients, the EHEC etiology of HUS is missed using current methods, which rely solely on detecting stx or Shiga toxin which can hamper epidemiological investigations and lead to inappropriate clinical management. Therefore, it is suggested that Stx- and stx gene-based detection methods should be complemented by additional methods for the identification of stx-negative E. coli O157 in microbiologic evaluations. We observed in our study that the clonal group II consisted of only two isolates (N27 and N28) both of which share a similar virulence profile yet their virulence profile was different from the majority of the isolates under clonal group I. We have also observed that 11 isolates did not belong to any clonal group and instead, each isolate exhibited a different sequence type (profile). These isolates are only distantly related phylogenetically to clonal group I. The reasons for this can not be explained with MLST data alone; this points to towards the need of whole genome sequencing of larger number O157 strains. However, our present findings are in agreement with the findings of Ogura et al. who observed a high level of genetic diversity among E. coli O157 strains from human isolates[ 27 ]. The greater genetic diversity observed among E. coli O157 isolates in our study could be due to several factors like (i) inter-species transfer of strains to humans from farm animals/pets, (ii) contaminated food sources (iii) increased international travel and (iv) longer span of sample collection period (2000-2008).

Conclusions Our study shows that all the methods compared here (dendrogram based on Neighbor Joining Method, eBurst algorithm and SplitsTree) exhibited a high degree of concordance showing a highly similar clustering pattern for the strains thereby indicating that the sequence data were reliable and were accurately represented by these algorithms. Infections with EHEC O157 may involve asymptomatic carriage or uncomplicated diarrhea, but other outcomes include haemorrhagic colitis and HUS. This poses considerable challenges both clinically and for disease control measures, because the disease is not yet treatable and colonization of humans can occur with only a few organisms. Although there are reports on clonal diversity and pathogenic properties of E. coli O157 strains, the present study extends our knowledge about the virulence spectrum and genetic relationships of E. coli O157 isolates that were collected only from one host (humans) over a period of several years (2000-2008). Phylogenetic analysis has demonstrated genetic diversity among E. coli O157 isolates collected from humans and thereby warrants further in-depth studies involving a larger number of samples so that more information about their diversity can be obtained.

Background Although many strain typing methods exist for pathogenic Escherichia coli , most have drawbacks in terms of resolving power, interpretability, or scalability. For this reason, multilocus sequence typing (MLST) is an appealing alternative especially when applied to the typing of temporal and spatially separated isolates. This method relies on an unambiguous DNA sequence analysis of nucleotide polymorphisms in housekeeping genes and has shown a high degree of intraspecies discriminatory power for bacterial and fungal pathogens. Results Here we used the MLST method to study the genetic diversity among E. coli O157 isolates collected from humans from two different locations of USA over a period of several years (2000-2008). MLST analysis of 33 E. coli O157 patient isolates using the eBurst algorithm distinguished 26 different sequence types (STs), which were clustered into two clonal groups and 11 singletons. The predominant ST was ST2, which consisted of 5 isolates (14.28%) followed by ST1 (11.42%). All the isolates under clonal group I exhibited a virtually similar virulence profile except for two strains, which tested negative for the presence of stx genes. The isolates that were assigned to clonal group II in addition to the 11 singletons were found to be phylogenetically distant from clonal group I. Furthermore, we observed a positive correlation between the virulence profile of the isolates and their clonal origin. Conclusions Our data suggests the presence of genetic diversity among E. coli O157 isolates from humans shows no measurable correlation to the geographic origin of the isolates.

Competing interests The authors declare that they have no competing interests. Authors' contributions SR and YFC conceived the study, experimental design and prepared the manuscript. SR was involved in all stages of the experiment. JS assisted in sequence data analysis and phylogeny estimation. DLG, KAM and BLA contributed E. coli strains. All authors read and approved the final manuscript.

Acknowledgements This project was supported with Federal funds from the National Institute of Allergy and Infectious Diseases, National Institute of Health, Department of Health and Human Services under contract, N01-AI-30054, Project No. ZC002-03.

CC BY

no

2022-01-12 15:21:44

BMC Res Notes. 2010 Dec 21; 3:343

oa_package/90/27/PMC3016269.tar.gz

PMC3016270

21138548

Background Pharmacological chaperone (PC) therapy has been recently proposed as a promising strategy for the treatment of some genetic diseases. PC therapy exploits small molecules which can be administered orally, reach difficult tissues such as the brain and have low cost. The new approach relies on an unexpected finding: some molecules that at high dosage inhibit specific proteins, can, at low dosage, restore their activities in cells. They act as life jackets or chaperones for proteins that, although retaining the essential residues needed for activity, become unstable upon mutation and are degraded. These proteins are able to fulfill their duty if they are given the chance to survive long enough and get to the site where they are needed. Pharmacological chaperone therapy, despite its novelty, has produced a few drugs which are already in clinical trials. The treatment of metabolic diseases with competitive inhibitors as chemical chaperons at sub-inhibitory intracellular concentrations was first proposed by Fan et al in 1999 [ 1 ]. They presented evidence that administration of Deoxy-galactonojirimycin (DGJ) at low concentration effectively enhanced mutant lysosomal alpha-galactosidase A [UNIPROT: AGAL_HUMAN] activities in lymphoblasts from Fabry patients with R301Q or Q279E mutations. Since then PC have been exploited for other lysosomal storage disorders such as Gaucher [ 2 ], Pompe [ 3 ], Tay-Sachs, Sandhoff [ 4 ], GM1 gangliosidosis [ 5 , 6 ], Niemann-Pick [ 7 , 8 ] and for the stabilization of a variety of non lysosomal proteins of medical interest such as the ATP binding cassette (ABC) family of transporters, G-protein-coupled receptors (GPCRs), tyrosinase, copper ATPase, p53 and carnitine transporters [ 9 , 10 ]. Fabry disease (FD) is X-linked and relatively frequent, 1-9 in 100000 [ORPHANET: orpha324, OMIM: 30150]. Different mutations of the gene encoding AGAL result in a wide phenotypic spectrum, with respect to age at onset, rate of disease progression, severity of clinical manifestations. Mutations with low or absent residual AGAL activity are generally observed in the classic infantile form of the disease. Patients with the late onset form of FD retain some AGAL activity and are asymptomatic until adult age when they develop cardiac and kidney problems. Nonetheless, as pointed out by Schaefer et al "clinical phenotype, age of onset and course of Fabry disease are very variable, even within the same family, which makes it difficult to define a genotype-phenotype relationship by analysing individual patients " [ 11 ]. Since the age of onset can be late and its complications, cardiac manifestations, stroke and chronic renal disease, are very similar to those of other very common disorders, FD could have been under diagnosed and an estimate as high as 1 in 3100 live births has been put forward [ 12 ]. FD offers an interesting case for studying the potentiality of PC because, since the pioneering work by Fan et al [ 1 ], responsiveness to PC has been assessed for a huge number of mutations, covering both early and late onset forms of Fabry disease [ 13 - 15 ]. A relatively large proportion of mutants, in particular among mutations associated with the late onset form of FD, recover activity when treated with DGJ. In this study, we correlated the responsiveness to DGJ with specific properties of the mutant AGAL sequences and developed a predictive protocol of general applicability to spot mutations which respond to PC.

Methods Construction of data set A list of AGAL mutations was created and to each one a label, 1 for responsive and 0 for non responsive ones, was added. Original data of enzymatic activity in the presence or in the absence of DGJ were taken from three different papers [ 13 - 15 ]. The data is redundant since responsiveness of some mutants have been analysed by more than one group and we obtained in total 96 different tested mutations. In additional file 1 and 2 we calculated the percentage recovered activity, that is activity of the mutant in the presence of DGJ divided by the activity of wild type AGAL multiplied by 100 (+DGJ/wild × 100), and the appropriate reference. The reference wild type activity, measured in the absence of DGJ was obtained for each mutation from the appropriate paper. Mutants are responsive if in the presence of DGJ recover at least 50%, of normal activity and non responsive in the other case. Construction of PSSMs We identified three sets of sequences: the first includes 163 homologs of AGAL with e-value < e-3 and identity ranging from 98% to 20% from GenBank CDS Translations, PDB, SwissProt, PIR, PRF; the second includes 13 homologs with e-value <e-50 and identity ranging from 95% to 30% from Uniprot/Swissprot; the third includes 15 orthologs with evalue e < e-150 and identity ranging from form 98%to 64% from translated GenBank, EMBL, DDBJ. BLASTPGP (BLASTP2.2.18) in PSI-BLAST mode was used on each set flagging on the output of a checkpoint file (PSSM matrix) and selecting one of the following scoring matrices: Blosum80, Blosum62, Blosum45, Pam70 or Pam30. Fifteen different PSSM were obtained and exploited in a second round of BLASTPGP to calculate net scores for AGAL mutants (mutant score PSSM(i)-Agal_wild score PSSM(i)) [ 30 ]. With any set of homologous proteins the best results (Table 1 ) were obtained selecting Blosum62 (BLASTPGP -d Blosum62); complete results for all PSSM are given in additional file 3 . The percentage of identical amino acids was also assessed aligning AGAL with 13 homologs with e-value <e-50 and identity ranging from 95% to 30% from Uniprot/Swissprot. At each position in the alignment we count the number of sequences sharing the same amino acid of AGAL and we divided by the total number of sequences in the alignment. Results in percentage are included in additional files 1 and 2 . Active site identification We detected active site residues with a procedure of general applicability inspired by the paper published by Dundas et al [ 17 ]. We detected cavities in AGAL structure, 3GXT, with the program CASTP [ 17 , 18 ] and we selected the pocket with the highest proportion of atoms belonging to conserved amino acids. To identify conserved residues we aligned 13 homologs with e-value <e-50 and identity ranging from 95% to 30% from Uniprot/Swissprot At each position in the alignment we count the number of sequences sharing the same amino acid of AGAL and we divided by the total number of sequences in the alignment. All the residues lining that pocket, irrespective of their conservation, were considered to potentially belong to the active site. Prediction of mutants stability MUPRO1.1 [ 27 ] was kindly provided by Dr J. Cheng: we run the program locally using as inputs only AGAL sequence, the position affected by the mutation, the original and the mutant amino-acid. Models of all AGAL mutants are constructed using as template 3GXT [ 21 ] with the program ANDANTE which predicts side-chain conformations by use of environment-specific substitution probabilities and a high-quality rotamer library [ 38 ]. The asymmetric unit in 3GXT [ 21 ] contain the dimeric enzyme and monomers are not fully super-imposable as expected. For this reason each mutation was built in both chains. SDM [ 25 , 26 ] uses a set of conformationally constrained environment-specific substitution tables and calculates the difference in the stability scores for the folded and unfolded state of each mutant and the wild-type protein. It uses as inputs the mutants generated with ANDANTE and the structure of the wild type enzyme. The programs SDM and ANDANTE were kindly provided by T. Blundell and collaborators. Accessibility and secondary structure Residue accessibility was calculated using PSA v2.0 (L.Cheng, unpublished) which uses the rolling probe algorithm [ 39 ]. For the threshold of solvent accessibility, we used a cut-off of 5.0% relative total side-chain or main chain accessibility. The numbers of responsive or non responsive mutations with accessible side chain was divided by the total number of mutations with accessible side chain and multiplied by 100. In a similar way it was calculated the percentage of responding and non responding mutations with non accessible side chain. The program PSA was kindly provided by T. Blundell and collaborators. We assigned each residue in the wild type AGAL structure 3GTX to alpha helices, beta sheets or other with the program SEGNO [ 40 ]. Supervised classification We tested all classification methods available in MATLAB-Arsenal developed by Rong Yan [ 29 ]The accuracy of the methods has been assessed using tenfold cross validation. This means that we have divided all mutants for which the responsiveness is known in ten folds. At each step, one fold has been sorted out and used as a test. The remaining mutants have been used to train the classifier. In total, tests were repeated thirty times and values have been taken on average. Results are expressed as mean accuracy ((TP+TN)/TOT), which represents the mean percentage of correctly classified instances, precision (TP/(TP+FP)) and recall (TP/(TP+FN)). When analysing structural features, mutations occurring on different chains had to be taken into account and therefore DecisionStump used 96 × 2 input data, when analysing sequence derived features, i.e. MUPRO, Blosum 62, PSSMs scores or PolyPhen PSIC score differences, DecisionStump used 96 input data. Statistical analysis All statistical analyses were carried out with MATLAB. To calculate the correlations and p values shown in additional files 1 and 2 we used corrcoef: this function returns a matrix R of correlation coefficients and a matrix of p-values for testing the hypothesis of no correlation. Each p-value is the probability of getting a correlation as large as the observed value by random chance, when the true correlation is zero. To analyse the data shown in Figure 2 and 3 (panel C,D) we carried out the Wilcoxon test with the function ranksum. This function performs a two-sided rank sum test of the null hypothesis that data are independent samples from identical continuous distributions with equal medians, against the alternative that they do not have equal medians. To evaluate the influence of solvent accessibility on PC responsiveness and to assess the statistical significance of the differences observed and differences expected by chance, we took a random sample of 30% in both exposed and non exposed population and counted the number of responding and non responding mutations. We repeated the randomization 100 times and calculated the differences between the percentage of responsive mutations in exposed and non exposed sites and the differences between the percentage of non responsive mutations in exposed and non exposed sites. To test the hypothesis that distribution of Mupro and SDM scores among responsive and non responsive mutations is different from what could be expected by chance (Figure 3 panel A and B), we run the chi2test. We ordered mutations by increasing SDM (or MUPRO) score and divided them into four equally populated bins. The function used as inputs the percentages observed and expected in each bin and returned chi2 values and the associated p values.

Results and Discussion Classification of mutants To try to correlate the properties of AGAL mutants to their responsiveness to PC, we needed a set of mutations as large as possible. DGJ has been tested on a large proportion of AGAL mutants and we collected data from three different papers which reported the enzymatic activity for each mutation in the presence and absence of DGJ and the activity of the wild-type protein [ 13 - 15 ]. We gathered 96 different mutations in total which had been tested for responsiveness to DGJ. Responsiveness to DGJ is variable among mutations and we needed a precise threshold to define a binary label. We calculated the ratio between the activity of the mutants in the presence of the drug and the reference wild type activity measured in the absence of DGJ, for each study. This data (+DGJ/wild × 100) and the appropriate references are reported in additional files 1 and 2 . We assigned responsiveness to two classes: 23 mutants were considered responsive because they recover at least 50% of normal activity in the presence of DGJ and 73 were considered non responsive. This conservative definition of responsiveness was adopted because the clinical indication of GLA mutations associated with FD is galactosidase activity less than 50% of the normal mean value in plasma [ 16 ]. Structural characterization of mutants Clinically important mutations can affect either the function or the structure of a protein: assessment of responsiveness to PC may depend critically on the correct classification of mutations. The first step in our analysis was the identification of the active site. For this purpose we detected pockets on the surface of AGAL, pdb code 3GXT, with the program CASTP [ 17 , 18 ]. Among these pockets we selected the one with the highest proportion of atoms belonging to conserved amino acids. We preferred this approach to direct identification of the residues in contact with galactose or DGJ in the crystal structures of the holo-enzyme [ 19 - 21 ].because galactose is a product of the enzyme obtained after the hydrolysis of a much larger substrate, globotriaosylceramide [ 22 ]. It might well be that residues in contact with galactose or with its analogue DGJ [ 19 - 21 ], represent only one part of the real active site. The pocket with the highest proportion of conserved amino acids includes four groups of amino acids: a) D92, D93, C142, D170, R227, D231 (red in Figure 1 ), residues completely conserved and associated with mutations not responding to DGJ; b) Y207 (blue in Figure 1 ), residue not conserved and associated with mutations not responding to DGJ; c) E203, L206, S297 (yellow in Figure 1 ) residues conserved and not associated with mutations tested with DGJ; d) W47, Y134, K168 (green in Figure 1 ), residues not conserved and not associated with mutations tested with DGJ. The pocket overlaps the residues which bind galactose or DGJ (D92, D93, K168, D170, E203, R227, D231) in the crystal structure of AGAL [ 21 ]. Mutations in the active site are inevitably non responsive: this is expected because DGJ acts only on the stability of the protein, and cannot restore the active site if this has been damaged. They represent 12% of the total number of non responsive mutations tested (Figure 1 and additional file 1 ). The distance from the active site was also measured. Not only mutations that occur in the active site pocket, but also those close to it, tend to be non responsive (additional file 1 ) whereas it is possible that mutation occurring quite far from the DGJ binding site can be responsive to the drug as observed for example for R363H and R363C (R363 is coloured in purple in Figure 1 ). Indeed, distance from active site correlates with the percentage of recovered activity (+DGJ/wild × 100) with an r-value 0.16 and a p-value 0.02 (additional file 1 ). AGAL has 5 intra-chain disulphide bonds and any tested mutation affecting a Cys involved in a bridge results in a non responsive protein. This result is also expected because disruption of a disulphide bond is usually highly destabilising and can prevent correct protein folding. It is worth noticing that not all Cys in AGAL are oxidised: Cys 90 and Cys 174 are not involved in a disulphide bridge, but they have not been found mutated in Fabry patients. We then analysed other structure-derived independent features. Accessibilities to solvent of AGAL residues was measured both for the main chain atoms and the side chain atoms of each residue and are reported in additional file 1 . Side chain accessibility correlates with recovered percentage activity (+DGJ/wild × 100) with an r-value 0.14 and a p-value 0.02 whereas main chain accessibility does not (additional file 1 ). We observe that mutations affecting exposed residues have higher chances of being rescued by PC (Figure 2 ), but, this holds only if we consider side chain accessibility. To assess the statistical significance of the differences observed in Figure 2 we performed the Wilcoxon rank sum test to reject the null hypothesis of equal medians at the default 5% significance level (P = 0.03). We assigned each residue in AGAL to alpha helices, beta sheets or others (data in additional file 1 ). By others we mean any secondary structure different from alpha helices and beta sheets including coils, turns, helices 3-10, poly-proline, etc. Although some of these elements might well be associated with functional roles or with protein protein association [ 23 , 24 ], they are too rare in AGAL structure to be useful for predictive purposes. The percentages of responding residues in alpha helices, beta sheets and others reflect the percentage of responding mutations in the total data set suggesting that the occurrence in a specific secondary structure element does not determine the possibility of a mutation to be sensitive to PC. Although residues in beta sheet tend to respond when their side chains are exposed, paucity of data in each class prevents a reliable statistical analysis (data not shown). We then used two programs, SDM [ 25 , 26 ] and MUPRO [ 27 ], to assign a stability score to the mutants, negative for unstable mutants and positive for stable ones, which is analogous to the free energy difference between a wild-type and mutant protein. The two programs provide independent assessment of protein stability because they rely on completely different approaches and do not reflect in any simple manner the structural features already analysed. SDM requires the knowledge of 3D-structure of AGAL and takes into account several structural features to predict the effect of mutations on protein stability. MUPRO does not require structural information on AGAL structure since it learns with a support vector machine method from the sequences deposited in ProTherm [ 28 ] database, a collection of numerical data of thermodynamic parameters for wild type and mutant proteins, and applies the derived rules to the sequence of an uncharacterised protein. SDM scores correlate with the percentage of recovered activity (+DGJ/wild × 100) with an r-value 0.23 and a p-value 0.2 10 -3 (additional file 1 ) whereas MUPRO scores do not (additional files 1 and 2 ). Twenty per cent of the mutations obtain a score lower than -3 with SDM and all of these, but R363C, are non responsive. R363C is responsive since it reaches 57% of wild type enzyme activity upon treatment with DGJ, but gets a negative score, -5.84, by SDM. On the other hand, a different responsive mutation occurring at the same site, R363H, gets a positive score, 0.11(additional file 1 ). Therefore, SDM recognizes that a mutation of R363 can have a small effect on protein stability and hence can be potentially recoverable by PC, but over-estimates the damage caused by the specific substitution with a Cys. We ordered mutations by increasing SDM score and divided them into four equally populated bins: Figure 3 panel A shows that the percentage of responding mutations is low at low SDM scores and increases progressively as SDM score increases. Similarly we sorted mutations by increasing MUPRO score and divided them into 4 bins: non responding mutations are more abundant in the bins associated with low MUPRO scores (Figure 3 panel B) but differences among bins are not as large as with SDM. The Pearson chi test confirms that the percentage of responsive mutations in the bin associated with low SDM scores (bin 1) is significantly different from expected (p = 0.01) whereas that in the bin associated with low MUPRO scores is not. On the other hand the percentage of responsive mutations in the bin associated with high MUPRO scores (bin 4) is significantly different from expected (p = 0.04) whereas that in the bin associated with high SDM scores is not. We built box plots for the results obtained with the two programs: differences between respondent and non respondent mutations are more evident comparing the first quartile for SDM (Figure 3 panel C) or the third quartile for MUPRO (Figure 3 panel D). The Wilcoxon rank sum test rejects the null hypothesis of equal medians for SDM boxplots at the default 5% significance level (P = 0.03). Regarding MUPRO, we cannot assess the normal distribution of the two samples with sufficient confidence. However, we observed that the first quartile of SDM scores for non responding mutations occurs at -3.14, a value very close to the threshold of -3 below which, as already discussed, mutations are non responsive. On the other hand we observed that the third quartile of MUPRO scores for responding mutations occurs at -0.14 a value above which mutations are mainly responsive (additional file 2 ). Prediction of responsiveness Analysis of AGAL structure reveals that three groups of mutations, those affecting the active site, those affecting disulphide bridges and those severely scored by SDM (< -3) are unlikely to be respondent to PC. However, a mutation which does not belong to any of these groups, can be either responsive or non responsive. In order to improve the usefulness of the model in patient therapy, we would like to predict responsiveness to DGJ for any AGAL mutation after having seen a number of training examples and this is, in statistical sense, a typical supervised classification task. Mutants have been labelled as respondent (23) or not respondent (73) and seven independent features derived from the analysis of 3D-structure of the enzyme and two features associated with their thermo-dynamic stability have been assigned to each of them as described before. Our goal is to find a mathematical function that given a set of features returns the correct class of the mutation. We tested all classification methods available in MATLAB-Arsenal developed by Rong Yan [ 29 ], which represent a large sample of all de facto standard classification algorithms. We compared the results obtained with MATLAB-Arsenal classifiers to those obtained with the ZeroR classifier. ZeroR classifier simply predicts the majority class in the training data. For example, if most of the training data are non respondent, ZeroR will predict all inputs as non respondent. Running ZeroR is necessary for determining a baseline performance as a benchmark for other learning schemes. We found that although many structural derived features, distance from active site, side chain accessibility, SDM scores, disruption of disulphide bonds, correlate with percentage recovered activity (+DGJ/wild × 100 please see additional file 1 ), they are not useful to predict responsive mutations with an accuracy, precision or recall higher than the baseline ZeroR (data not shown). The classifier which performed significantly better than ZeroR is DecisionStump on MUPRO results, with 81% accuracy versus 24% of the baseline. Prediction of responsive mutations can be achieved with MUPRO because it assigns the highest scores preferentially to mutations which recover activity upon treatment with PC (additional file 2 ). MUPRO predicts protein stability relying only on sequence information. This result encouraged us to rely on sequence alone and try and do better than MUPRO exploiting evolutionary conservation. In the first place we scored mutations using Blosum62, a matrix which considers only the pair of amino-acids involved regardless of the site where the mutation occurs. Performance of DecisionStump with Blosum62 scores (74% accuracy) is better than the baseline (24% accuracy), but much lower than that obtainable with MUPRO scores (81% accuracy). We decided to take into account the position where mutations occur in the protein using position specific substitution matrices (PSSM). PSSMs are built with PSI-BLAST [ 30 ], a program that uses as an input a set of homologous proteins, aligns the sequences and calculates amino acid substitution scores separately for each position in the multiple alignment. We carried out three independent experiments: we collected a first set of AGAL homologous proteins which includes very distant homologs with e-value as low as e-3, a second set which includes only close homologs with e-value higher than e-50 and a third set which includes only close orthologs, excluding the sequences derived by gene duplication. In fact, two paralogous lysosomal enzymes alpha-galactosidase A (AGAL_HUMAN; EC 3.2.1.22) and alpha-N-acetylgalactosaminidase (NAGAB_HUMAN; EC 3.2.1.49) exist in higher animals. We run PSI_BLAST independently on the three sets and obtained different PSSM. The scores obtained with the PSSM built with close homologs of AGAL as well as those obtained with far homologs and close orthologs correlate with the percentage of recovered activity upon treatment with DGJ (+DGJ/wild × 100) respectively with r-value 0.44 and p-value <0.1 10 -3 ; r-value 0.44 and p-value <0.1 10 -3 ; r-value 0.32 and p-value 0.2 10 -3 . The scores obtained with PSSMs were used as inputs for DecisionStump to test their ability to predict responsive mutations. The PSSM built aligning close homologs (e-value <e-50) led to the best results: its performance (87% accuracy) is better than that obtained with the scores of MUPRO and, in particular, recall is much higher (49% versus 30%) which means that less false negatives are predicted (Table 1 ). The approach we used to predict responsiveness to PC is similar to that used by some programs to predict disease-associated mutations: results obtained with PolyPhen [ 31 ], which are reported in additional file 2 were used as inputs of DecisionStump classifier and allowed prediction of AGAL PC responsive mutations with 75% accuracy. When developing a classifier, the numbers of responding mutations erroneously predicted as non responding (FN) as well as non responding mutations erroneously predicted as responding (FP) should be kept at a minimum. Therefore precision (TP/TP+FP) and recall (TP/TP+FN) are very useful to assess the performance of different methods. It is not surprising that precision and recall are lower than accuracy (TP+TN/TP+TN+FP+FN) in any case reported in Table 1 because correctly predicted non responding mutations (TN) increase accuracy but not precision and recall. Specific classes of non responding mutations are identified by structural analysis, and in general it is easier to predict TN than TP. DecisionStump based on the scores of the PSSM built with AGAL_HUMAN close homologs provides the highest precision and recall in comparison with the other tested methods (including PolyPhen). Nonetheless it should be emphasized that, regrettably, false positives and even more false negatives are predicted as implied by the fact that precision and recall values are well below 1 (Table 1 ). Results obtained with sequence information alone do not improve if structure derived information are added as features for the classifiers in MATLAB-Arsenal [ 29 ](data not shown). After this manuscript was submitted we became aware of three new AGAL mutations [ 32 , 33 ], C52Y, Y216C and G183A [ 32 ] which recover respectively <10%, 40% and 65% activity upon treatment with PC. This report offered us the opportunity for a test on mutations not included in the development of the method. C52Y disrupts a disulphide bonds and thus belongs to a class of mutations which are definitely unrecoverable. C52Y gets a negative score by SDM, -1.582, by MUPRO, -1.37 and by PSSM, -5, and is correctly predicted as non responsive. Y216C according to our conservative definition is non responsive since it does not reach 50% of wild type activity: it does not belong to the three groups of mutations which are definitely unrecoverable (i.e. it does not belong to the active site, it does not disrupt disulphide bonds, it is not severely destabilizing), but it gets a negative score -2.319 by SDM, -1.16 by MUPRO and -3 by PSSM and is correctly predicted as non responsive. G183A does not belong to the three groups of mutations which are definitely unrecoverable, it gets a positive score 2.33 by SDM and a negative score, -1.47, by MUPRO and is erroneously predicted as non responsive, because of a PSSM score -2. The PSSM score -2, is immediately below the threshold set by the classifier to minimize false negatives and false positives. Indeed, this score represents a twilight zone where the recovered activity (+DGJ/wild x100) of mutations can vary from 0% to 100% and for 19% of mutations is above 50%. We observed that with a few exceptions, responsive mutations erroneously predicted by the classifier (FN) either have a positive SDM score (as it is the case for G183A) or have a solvent exposed side chain.

Results and Discussion Classification of mutants To try to correlate the properties of AGAL mutants to their responsiveness to PC, we needed a set of mutations as large as possible. DGJ has been tested on a large proportion of AGAL mutants and we collected data from three different papers which reported the enzymatic activity for each mutation in the presence and absence of DGJ and the activity of the wild-type protein [ 13 - 15 ]. We gathered 96 different mutations in total which had been tested for responsiveness to DGJ. Responsiveness to DGJ is variable among mutations and we needed a precise threshold to define a binary label. We calculated the ratio between the activity of the mutants in the presence of the drug and the reference wild type activity measured in the absence of DGJ, for each study. This data (+DGJ/wild × 100) and the appropriate references are reported in additional files 1 and 2 . We assigned responsiveness to two classes: 23 mutants were considered responsive because they recover at least 50% of normal activity in the presence of DGJ and 73 were considered non responsive. This conservative definition of responsiveness was adopted because the clinical indication of GLA mutations associated with FD is galactosidase activity less than 50% of the normal mean value in plasma [ 16 ]. Structural characterization of mutants Clinically important mutations can affect either the function or the structure of a protein: assessment of responsiveness to PC may depend critically on the correct classification of mutations. The first step in our analysis was the identification of the active site. For this purpose we detected pockets on the surface of AGAL, pdb code 3GXT, with the program CASTP [ 17 , 18 ]. Among these pockets we selected the one with the highest proportion of atoms belonging to conserved amino acids. We preferred this approach to direct identification of the residues in contact with galactose or DGJ in the crystal structures of the holo-enzyme [ 19 - 21 ].because galactose is a product of the enzyme obtained after the hydrolysis of a much larger substrate, globotriaosylceramide [ 22 ]. It might well be that residues in contact with galactose or with its analogue DGJ [ 19 - 21 ], represent only one part of the real active site. The pocket with the highest proportion of conserved amino acids includes four groups of amino acids: a) D92, D93, C142, D170, R227, D231 (red in Figure 1 ), residues completely conserved and associated with mutations not responding to DGJ; b) Y207 (blue in Figure 1 ), residue not conserved and associated with mutations not responding to DGJ; c) E203, L206, S297 (yellow in Figure 1 ) residues conserved and not associated with mutations tested with DGJ; d) W47, Y134, K168 (green in Figure 1 ), residues not conserved and not associated with mutations tested with DGJ. The pocket overlaps the residues which bind galactose or DGJ (D92, D93, K168, D170, E203, R227, D231) in the crystal structure of AGAL [ 21 ]. Mutations in the active site are inevitably non responsive: this is expected because DGJ acts only on the stability of the protein, and cannot restore the active site if this has been damaged. They represent 12% of the total number of non responsive mutations tested (Figure 1 and additional file 1 ). The distance from the active site was also measured. Not only mutations that occur in the active site pocket, but also those close to it, tend to be non responsive (additional file 1 ) whereas it is possible that mutation occurring quite far from the DGJ binding site can be responsive to the drug as observed for example for R363H and R363C (R363 is coloured in purple in Figure 1 ). Indeed, distance from active site correlates with the percentage of recovered activity (+DGJ/wild × 100) with an r-value 0.16 and a p-value 0.02 (additional file 1 ). AGAL has 5 intra-chain disulphide bonds and any tested mutation affecting a Cys involved in a bridge results in a non responsive protein. This result is also expected because disruption of a disulphide bond is usually highly destabilising and can prevent correct protein folding. It is worth noticing that not all Cys in AGAL are oxidised: Cys 90 and Cys 174 are not involved in a disulphide bridge, but they have not been found mutated in Fabry patients. We then analysed other structure-derived independent features. Accessibilities to solvent of AGAL residues was measured both for the main chain atoms and the side chain atoms of each residue and are reported in additional file 1 . Side chain accessibility correlates with recovered percentage activity (+DGJ/wild × 100) with an r-value 0.14 and a p-value 0.02 whereas main chain accessibility does not (additional file 1 ). We observe that mutations affecting exposed residues have higher chances of being rescued by PC (Figure 2 ), but, this holds only if we consider side chain accessibility. To assess the statistical significance of the differences observed in Figure 2 we performed the Wilcoxon rank sum test to reject the null hypothesis of equal medians at the default 5% significance level (P = 0.03). We assigned each residue in AGAL to alpha helices, beta sheets or others (data in additional file 1 ). By others we mean any secondary structure different from alpha helices and beta sheets including coils, turns, helices 3-10, poly-proline, etc. Although some of these elements might well be associated with functional roles or with protein protein association [ 23 , 24 ], they are too rare in AGAL structure to be useful for predictive purposes. The percentages of responding residues in alpha helices, beta sheets and others reflect the percentage of responding mutations in the total data set suggesting that the occurrence in a specific secondary structure element does not determine the possibility of a mutation to be sensitive to PC. Although residues in beta sheet tend to respond when their side chains are exposed, paucity of data in each class prevents a reliable statistical analysis (data not shown). We then used two programs, SDM [ 25 , 26 ] and MUPRO [ 27 ], to assign a stability score to the mutants, negative for unstable mutants and positive for stable ones, which is analogous to the free energy difference between a wild-type and mutant protein. The two programs provide independent assessment of protein stability because they rely on completely different approaches and do not reflect in any simple manner the structural features already analysed. SDM requires the knowledge of 3D-structure of AGAL and takes into account several structural features to predict the effect of mutations on protein stability. MUPRO does not require structural information on AGAL structure since it learns with a support vector machine method from the sequences deposited in ProTherm [ 28 ] database, a collection of numerical data of thermodynamic parameters for wild type and mutant proteins, and applies the derived rules to the sequence of an uncharacterised protein. SDM scores correlate with the percentage of recovered activity (+DGJ/wild × 100) with an r-value 0.23 and a p-value 0.2 10 -3 (additional file 1 ) whereas MUPRO scores do not (additional files 1 and 2 ). Twenty per cent of the mutations obtain a score lower than -3 with SDM and all of these, but R363C, are non responsive. R363C is responsive since it reaches 57% of wild type enzyme activity upon treatment with DGJ, but gets a negative score, -5.84, by SDM. On the other hand, a different responsive mutation occurring at the same site, R363H, gets a positive score, 0.11(additional file 1 ). Therefore, SDM recognizes that a mutation of R363 can have a small effect on protein stability and hence can be potentially recoverable by PC, but over-estimates the damage caused by the specific substitution with a Cys. We ordered mutations by increasing SDM score and divided them into four equally populated bins: Figure 3 panel A shows that the percentage of responding mutations is low at low SDM scores and increases progressively as SDM score increases. Similarly we sorted mutations by increasing MUPRO score and divided them into 4 bins: non responding mutations are more abundant in the bins associated with low MUPRO scores (Figure 3 panel B) but differences among bins are not as large as with SDM. The Pearson chi test confirms that the percentage of responsive mutations in the bin associated with low SDM scores (bin 1) is significantly different from expected (p = 0.01) whereas that in the bin associated with low MUPRO scores is not. On the other hand the percentage of responsive mutations in the bin associated with high MUPRO scores (bin 4) is significantly different from expected (p = 0.04) whereas that in the bin associated with high SDM scores is not. We built box plots for the results obtained with the two programs: differences between respondent and non respondent mutations are more evident comparing the first quartile for SDM (Figure 3 panel C) or the third quartile for MUPRO (Figure 3 panel D). The Wilcoxon rank sum test rejects the null hypothesis of equal medians for SDM boxplots at the default 5% significance level (P = 0.03). Regarding MUPRO, we cannot assess the normal distribution of the two samples with sufficient confidence. However, we observed that the first quartile of SDM scores for non responding mutations occurs at -3.14, a value very close to the threshold of -3 below which, as already discussed, mutations are non responsive. On the other hand we observed that the third quartile of MUPRO scores for responding mutations occurs at -0.14 a value above which mutations are mainly responsive (additional file 2 ). Prediction of responsiveness Analysis of AGAL structure reveals that three groups of mutations, those affecting the active site, those affecting disulphide bridges and those severely scored by SDM (< -3) are unlikely to be respondent to PC. However, a mutation which does not belong to any of these groups, can be either responsive or non responsive. In order to improve the usefulness of the model in patient therapy, we would like to predict responsiveness to DGJ for any AGAL mutation after having seen a number of training examples and this is, in statistical sense, a typical supervised classification task. Mutants have been labelled as respondent (23) or not respondent (73) and seven independent features derived from the analysis of 3D-structure of the enzyme and two features associated with their thermo-dynamic stability have been assigned to each of them as described before. Our goal is to find a mathematical function that given a set of features returns the correct class of the mutation. We tested all classification methods available in MATLAB-Arsenal developed by Rong Yan [ 29 ], which represent a large sample of all de facto standard classification algorithms. We compared the results obtained with MATLAB-Arsenal classifiers to those obtained with the ZeroR classifier. ZeroR classifier simply predicts the majority class in the training data. For example, if most of the training data are non respondent, ZeroR will predict all inputs as non respondent. Running ZeroR is necessary for determining a baseline performance as a benchmark for other learning schemes. We found that although many structural derived features, distance from active site, side chain accessibility, SDM scores, disruption of disulphide bonds, correlate with percentage recovered activity (+DGJ/wild × 100 please see additional file 1 ), they are not useful to predict responsive mutations with an accuracy, precision or recall higher than the baseline ZeroR (data not shown). The classifier which performed significantly better than ZeroR is DecisionStump on MUPRO results, with 81% accuracy versus 24% of the baseline. Prediction of responsive mutations can be achieved with MUPRO because it assigns the highest scores preferentially to mutations which recover activity upon treatment with PC (additional file 2 ). MUPRO predicts protein stability relying only on sequence information. This result encouraged us to rely on sequence alone and try and do better than MUPRO exploiting evolutionary conservation. In the first place we scored mutations using Blosum62, a matrix which considers only the pair of amino-acids involved regardless of the site where the mutation occurs. Performance of DecisionStump with Blosum62 scores (74% accuracy) is better than the baseline (24% accuracy), but much lower than that obtainable with MUPRO scores (81% accuracy). We decided to take into account the position where mutations occur in the protein using position specific substitution matrices (PSSM). PSSMs are built with PSI-BLAST [ 30 ], a program that uses as an input a set of homologous proteins, aligns the sequences and calculates amino acid substitution scores separately for each position in the multiple alignment. We carried out three independent experiments: we collected a first set of AGAL homologous proteins which includes very distant homologs with e-value as low as e-3, a second set which includes only close homologs with e-value higher than e-50 and a third set which includes only close orthologs, excluding the sequences derived by gene duplication. In fact, two paralogous lysosomal enzymes alpha-galactosidase A (AGAL_HUMAN; EC 3.2.1.22) and alpha-N-acetylgalactosaminidase (NAGAB_HUMAN; EC 3.2.1.49) exist in higher animals. We run PSI_BLAST independently on the three sets and obtained different PSSM. The scores obtained with the PSSM built with close homologs of AGAL as well as those obtained with far homologs and close orthologs correlate with the percentage of recovered activity upon treatment with DGJ (+DGJ/wild × 100) respectively with r-value 0.44 and p-value <0.1 10 -3 ; r-value 0.44 and p-value <0.1 10 -3 ; r-value 0.32 and p-value 0.2 10 -3 . The scores obtained with PSSMs were used as inputs for DecisionStump to test their ability to predict responsive mutations. The PSSM built aligning close homologs (e-value <e-50) led to the best results: its performance (87% accuracy) is better than that obtained with the scores of MUPRO and, in particular, recall is much higher (49% versus 30%) which means that less false negatives are predicted (Table 1 ). The approach we used to predict responsiveness to PC is similar to that used by some programs to predict disease-associated mutations: results obtained with PolyPhen [ 31 ], which are reported in additional file 2 were used as inputs of DecisionStump classifier and allowed prediction of AGAL PC responsive mutations with 75% accuracy. When developing a classifier, the numbers of responding mutations erroneously predicted as non responding (FN) as well as non responding mutations erroneously predicted as responding (FP) should be kept at a minimum. Therefore precision (TP/TP+FP) and recall (TP/TP+FN) are very useful to assess the performance of different methods. It is not surprising that precision and recall are lower than accuracy (TP+TN/TP+TN+FP+FN) in any case reported in Table 1 because correctly predicted non responding mutations (TN) increase accuracy but not precision and recall. Specific classes of non responding mutations are identified by structural analysis, and in general it is easier to predict TN than TP. DecisionStump based on the scores of the PSSM built with AGAL_HUMAN close homologs provides the highest precision and recall in comparison with the other tested methods (including PolyPhen). Nonetheless it should be emphasized that, regrettably, false positives and even more false negatives are predicted as implied by the fact that precision and recall values are well below 1 (Table 1 ). Results obtained with sequence information alone do not improve if structure derived information are added as features for the classifiers in MATLAB-Arsenal [ 29 ](data not shown). After this manuscript was submitted we became aware of three new AGAL mutations [ 32 , 33 ], C52Y, Y216C and G183A [ 32 ] which recover respectively <10%, 40% and 65% activity upon treatment with PC. This report offered us the opportunity for a test on mutations not included in the development of the method. C52Y disrupts a disulphide bonds and thus belongs to a class of mutations which are definitely unrecoverable. C52Y gets a negative score by SDM, -1.582, by MUPRO, -1.37 and by PSSM, -5, and is correctly predicted as non responsive. Y216C according to our conservative definition is non responsive since it does not reach 50% of wild type activity: it does not belong to the three groups of mutations which are definitely unrecoverable (i.e. it does not belong to the active site, it does not disrupt disulphide bonds, it is not severely destabilizing), but it gets a negative score -2.319 by SDM, -1.16 by MUPRO and -3 by PSSM and is correctly predicted as non responsive. G183A does not belong to the three groups of mutations which are definitely unrecoverable, it gets a positive score 2.33 by SDM and a negative score, -1.47, by MUPRO and is erroneously predicted as non responsive, because of a PSSM score -2. The PSSM score -2, is immediately below the threshold set by the classifier to minimize false negatives and false positives. Indeed, this score represents a twilight zone where the recovered activity (+DGJ/wild x100) of mutations can vary from 0% to 100% and for 19% of mutations is above 50%. We observed that with a few exceptions, responsive mutations erroneously predicted by the classifier (FN) either have a positive SDM score (as it is the case for G183A) or have a solvent exposed side chain.

Conclusions Mutations which occur at non exposed sites or in the active pocket are likely to be non responsive to DGJ as well as those compromising stability, but the complex interplay between functional and structural features on responsiveness makes difficult their exploitation for predictive purposes. However, the same features strongly tune evolutionary pressure at different sites in a protein. For this reason methods based on conservation alone can be exploited to predict responsiveness to PC: addition of structural features does not improve prediction because by measuring conservation we have already, implicitly, taken them into account. It is generally accepted that disease causing mutations are more frequent at conserved sites whereas non disease polymorphisms prevail at variable site and we demonstrated that this holds also for non responsive and responsive mutations. The score assigned to mutation by a PSSM takes into account the degree of conservation of the wild type amino acid in homologous sequences and the specific substitution that is introduced: if the new amino-acid is present in homologous species a less negative score is obtained. We have shown that high PSSM scores are associated with respondent mutations whereas low PSSM scores are associated with non respondent mutations and that this property can be exploited for predictions. It is not trivial, however the effect of the inclusion of homologs with a low percentage of identity in the construction of the PSSMs for at least three reasons: 1) sites that are critical for the function of enzymes in mammals or in metazoa might have varied in evolutionary distant species and many compensative mutations might have arisen 2) the inclusion of proteins in the twilight zone necessarily raise the possibility of introducing false homologs. 3) The inclusion paralogs can hinder functional important sites. Our analysis proves that the choice of homologs for the construction of the PSSM is critical: paralogs can be included, but too distant homologs must be excluded. Our method can be applied to predict PC responsiveness of novel and as yet untested mutations of alpha galactosidase. This is feasible and useful because a very high number of natural missense mutations have been found for AGAL, a middle sized protein of 429 aa: 146 are listed in Uniprot/Swissprot [ 34 ], 256 in the public version of Hgmd [ 35 ]. This figure might well rise if Fabry disease has been under diagnosed as it seems to be the case. In particular it can be predicted that new mutations associated with the late on-set form of the diseases, that are the most likely be respondent to DGJ will be found. In order to make a projection of the results obtainable with our method, we tested 299 mutations of AGAL listed in UniprotSwissprot and HGMD and we estimated that 40 mutations as yet untested with PC are likely to respond. When a AGAL mutation not already tested in vitro for responsiveness to DGJ is encountered, clinicians can direct Fabry patients towards the therapy more likely to ameliorate their phenotype: a high PSSM score should suggest a beneficial effect of DGJ whereas a low PSSM score should suggest caution and direct towards enzymatic replacement [ 36 , 37 ]. DecisionStump classifier sets a threshold to minimize false negatives and false positives and allows, if required by clinicians, a clear cut indication for therapeutic intervention. Although PC therapy will not substitute other therapies for FD such as enzymatic replacement [ 36 , 37 ], our results reinforce the idea that, at least in principle, these novel drugs, affordable and suitable for oral administration, have vast applicability. Our approach to find candidates for PC therapy is not limited to alpha galactosidase, but can be extended on a genomic scale among proteins with unknown tertiary structures. Moreover due to the low cost associated with an in silico screening it meets regulators which look for ways to make it easier and cheaper for drug companies to develop treatments for rare diseases.

Background The pharmacological chaperones therapy is a promising approach to cure genetic diseases. It relies on substrate competitors used at sub-inhibitory concentration which can be administered orally, reach difficult tissues and have low cost. Clinical trials are currently carried out for Fabry disease, a lysosomal storage disorder caused by inherited genetic mutations of alpha-galactosidase. Regrettably, not all genotypes respond to these drugs. Results We collected the experimental data available in literature on the enzymatic activity of ninety-six missense mutants of lysosomal alpha-galactosidase measured in the presence of pharmacological chaperones. We associated with each mutation seven features derived from the analysis of 3D-structure of the enzyme, two features associated with their thermo-dynamic stability and four features derived from sequence alone. Structural and thermodynamic analysis explains why some mutants of human lysosomal alpha-galactosidase cannot be rescued by pharmacological chaperones: approximately forty per cent of the non responsive cases examined can be correctly associated with a negative prognostic feature. They include mutations occurring in the active site pocket, mutations preventing disulphide bridge formation and severely destabilising mutations. Despite this finding, prediction of mutations responsive to pharmacological chaperones cannot be achieved with high accuracy relying on combinations of structure- and thermodynamic-derived features even with the aid of classical and state of the art statistical learning methods. We developed a procedure to predict responsive mutations with an accuracy as high as 87%: the method scores the mutations by using a suitable position-specific substitution matrix. Our approach is of general applicability since it does not require the knowledge of 3D-structure but relies only on the sequence. Conclusions Responsiveness to pharmacological chaperones depends on the structural/functional features of the disease-associated protein, whose complex interplay is best reflected on sequence conservation by evolutionary pressure. We propose a predictive method which can be applied to screen novel mutations of alpha galactosidase. The same approach can be extended on a genomic scale to find candidates for therapy with pharmacological chaperones among proteins with unknown tertiary structures.

Abbreviations DGJ: Deoxy-galactonojirimycin; FD: Fabry disease; FN: false negative; FP: false positive; PC: pharmacological chaperone; PSSM: position specific substitution matrix; TP: true positive; TN: true negative. Competing interests The authors declare that they have no competing interests. Authors' contributions GA and MVC designed the study and wrote the paper. MRG carried out preliminary statistical analysis and advised on MATLAB usage. MC and AC carried out structure based and sequence based analysis. All authors read and approved the final manuscript. Supplementary Material

Acknowledgements This work was supported by a grant from MIUR PRIN 2007 (to MVC) and by the Università Federico II, programma Breve Mobilità (to MVC). MVC wishes to gratefully thank Prof. A Riccio for his encouragement and interest. The authors thank Dr Panos Pardalos for useful discussions and Antonio Federico for preparing figures.

CC BY

no

2022-01-12 15:21:44

Orphanet J Rare Dis. 2010 Dec 7; 5:36

oa_package/db/e6/PMC3016270.tar.gz

PMC3016271

21176172

Introduction Attention-Deficit/Hyperactivity Disorder (ADHD) is a common, long-lasting, treatable childhood psychiatric disorder, characterised by a pattern of developmentally inappropriate inattention, motor restlessness, and impulsivity that affects approximately 3-7% of school-aged children [ 1 ]. ADHD was first recognised 100 years ago as a childhood disorder found mainly in boys, and was initially described as "hyperactivity" or "hyperkinetic disorder of childhood". This abnormal behaviour was found to be the result of a biological condition rather than a result of poor parenting [ 2 ]. In the 1960's and 70's much of the focus on what is now called ADHD was on hyperactivity. The presence of excessive movements in children was proposed to result from bilateral cortical activity secondary to a lack of transcallosal fibre tract-mediated interhemispheric inhibition [ 3 ]. Attention Deficit Disorder with or without Hyperactivity first featured in DSM-III in 1980 [ 4 ], and the more recent DSM-IV-TR provided updated ADHD criteria [ 5 ]. For a diagnosis of ADHD, symptoms need to occur often, have persisted for the past six months, and be maladaptive and incongruent with the individual's developmental level. Additionally, an ADHD diagnosis is only given if at least some of the behavioural symptoms were present before the age of 7 years, occur in more than one setting, and cause significant impairment in social and school functioning. The renaming of the disorder, the subsequent focus on attention, and the clarification of three subtypes led to a range of neurocognitive and neurobiological hypotheses regarding the etiology and pathophysiology of ADHD within a more specific brain localisation. Furthermore, neurocognitive models of ADHD have become more refined, and one particular executive process, inhibition, is now considered to be a core deficit [ 6 ]. Current theories emphasise the central role of attentional and executive dysfunctions in children [ 7 , 8 ], as well as affective components involving emotional control and motivational processes [ 9 ]. A growing body of evidence supports a model in which multiple genetic and environmental factors interact during early development to create a neurobiological susceptibility to the disorder; the expression of which is mediated by alterations within different and diverse neural networks and deficits in the neuropsychological functions that these subserve [ 10 ]. Individuals with ADHD present difficulties in several domains of attentional and cognitive functions: problem solving, planning, orienting, alerting, cognitive flexibility, sustained attention, response inhibition, and working memory [ 7 , 11 ]. Other domains involving affective components, such as motivation and delay aversion, are also affected [ 8 , 9 ]. Motor difficulties, such as problems with sensory motor coordination, including poor handwriting, clumsiness, and marked delays in achieving motor milestones [ 12 ], have also been reported and the prevalence of motor impairment in the ADHD population has been estimated to be approximately 50% [ 13 ]. Motor problems might be partially related to abnormalities in structure and/or function of the cerebellum and basal ganglia found in ADHD [ 14 ]. Recently, neuroimaging has led to several important advances in the understanding of the neurobiology underlying the clinical picture of ADHD, and demonstrates a clear brain basis to the disorder in regions involved in attention, and executive and inhibitory control [ 15 , 16 ]. Furthermore, transcranial magnetic stimulation (TMS) has provided evidence that intracortical inhibition, as indexed by the immature ipsilateral motor cortex, normalises with psychostimulant treatment [ 17 ]. There is an exciting confluence between emerging studies in basic neurobiology and the genetic, neuroimaging, and neuropsychological analyses of ADHD. Knowledge of neurobiology can offer child neurologists, psychiatrists and other healthcare professionals a valuable framework for the interpretation of clinical findings of children meeting the criteria for diagnosis of ADHD. In this article we provide a brief overview of the salient neurological basis of the disorder. Etiology ADHD is not a single pathophysiological entity and appears to have a complex etiology. Multiple genetic and environmental factors act together to create a spectrum of neurobiological liability. The genetic basis for ADHD Genetic factors are implicated in ADHD, but the mechanism of action is not completely understood. Twin, family and adoption studies of ADHD have supported a strong genetic contribution to the disorder, with heritability ranging from 60-90% [ 18 , 19 ]. Genes regulating neurotransmitter systems have been implicated in ADHD. Candidate gene studies of ADHD have produced substantial evidence implicating several genes in the etiology of the disorder, with meta-analyses supportive of a role of the genes coding for DRD4, DRD5, SLC6A3, SNAP-25, and HTR1B [ 20 ]. Genome scan studies on potential alleles for ADHD have demonstrated linkage on chromosomes 5p13, 6q12, 16p13, 17p11 and 11q22-25 [ 21 , 22 ]. However, genome-wide association studies have failed to report any associations that are significant after correction for multiple testing [ 23 ]. Therefore, a plausible genetic hypothesis for ADHD is a mixture of dominant and recessive major genes that act with complex polygenic transmission patterns [ 18 ]. An increased rate of large, rare, chromosomal deletions and duplications known as copy number variants have been reported in individuals with ADHD [ 24 ]. However, genetic testing in an individual child is not currently practical in normal clinical practise. Sometimes ADHD-like symptoms are exhibited by patients with established neurogenetic disorders such as Tuberous Sclerosis Complex, Neurofibromatosis I, Turner Syndrome, Williams Syndrome, Velocardiofacial syndrome, Prader-Willy syndrome, and Fragile × Syndrome. Although each syndrome may arise from different genetic abnormalities with multiple molecular functions, the effects of these abnormalities may give rise to common effects downstream in the biological pathways or neural circuits, resulting in the presentation of ADHD symptoms [ 25 ]. The environmental basis of ADHD Pre-, peri- and postnatal environmental factors play an important role in the pathogenesis of ADHD. Prenatal factors are associated with maternal lifestyle during pregnancy. For example, prenatal alcohol exposure is known to induce brain structural anomalies, especially in the cerebellum [ 26 ]. Children exposed prenatally to alcohol can become hyperactive, disruptive, impulsive, and are at an increased risk of a range of psychiatric disorders [ 27 , 28 ]. Maternal smoking produces a 2.7-fold increased risk for ADHD [ 29 ], and a dose-response relationship between maternal smoking during pregnancy and hyperactivity has been reported [ 30 ]. This may be due to an effect on nicotinic receptors, which modulate dopaminergic activity. Dopaminergic disruption is believed to be involved in the pathophysiology of ADHD [ 31 , 32 ]. Peri-natal factors have also been implicated, with a two-fold increase in ADHD in very low-birthweight children and an increased rate of pregnancy and birth complications in mothers of children later diagnosed with ADHD [ 33 ]. Among postnatal factors, a role for malnutrition and dietary deficiency in ADHD has been proposed. An imbalance of essential fatty acid (omega-3 and omega-6) intake has been suggested to be potentially involved in the development of ADHD [ 34 ], although further evidence is required to establish a role. Iron deficiency has been implicated in some cases [ 35 ]. Early deprivation of social environment during the postnatal period may also have significant effects. Gene-environment interactions More complex models of the etiology of ADHD incorporating gene-environment interplay need to be considered. Recent studies have focused on the joint effects of gene variants (of DRD4 and DAT1) and prenatal substance exposures on subtypes of ADHD children, demonstrating that smoking during pregnancy is associated with the combined ADHD type in genetically susceptible children [ 36 ]. A significant interaction between DAT1 genotype and prenatal smoke exposure was found in males. Men homozygous for the DAT1 10-repeat allele had higher hyperactivity-impulsivity than males from all other groups [ 37 ]. Despite the heterogeneity of the etiology and pathophysiology of ADHD, abnormal DAT density seems to be common among subjects with ADHD [ 38 ]. Neuroimaging Growing evidence points to the involvement of the frontostriatal network as a likely contributor to the pathophysiology of ADHD. This network involves the lateral prefrontal cortex, the dorsal anterior cingulate cortex, and the caudate nucleus and putamen. In ADHD patients, reductions in volume have been observed in total cerebral volume, the prefrontal cortex, the basal ganglia (striatum), the dorsal anterior cingulate cortex, the corpus callosum and the cerebellum [ 39 ]. A developmental trajectories study in ADHD patients showed a delay in cortical maturation, and demonstrated that different clinical outcomes may be associated with different developmental trajectories in adolescence and beyond [ 40 ]. In studies of cortical development in children with ADHD, a marked delay in brain maturation was seen; the grey matter peaks were about 3 years later than in healthy controls [ 41 ]. The delay was most prominent in prefrontal regions important in the control of cognitive processes including attention and motor planning [ 41 , 42 ]. Compensatory networks including basal ganglia, insula and cerebellum have been implicated for relative lower cognitive load tasks in ADHD patients [ 43 ]. Neuroimaging studies have also reported reduced white matter (WM) volumes [ 43 ], midsagittal corpus callosum (CC) areas [ 44 ], and cortical thickness [ 43 ] in ADHD patients compared with controls. One of the most replicated alterations is a significantly smaller CC, but there are conflicting reports regarding the affected callosal segments [ 45 ]. Recent magnetic resonance imaging (MRI) structural investigations have shown that WM alterations are present in children, adolescents and adults with ADHD [ 46 ]. In 15 young males with ADHD, Silk et al . (2008) found WM abnormalities in several distinct regions underlying the inferior parietal, occipito-parietal, inferior frontal, and inferior temporal cortex [ 47 ]. Tractography methods showed that these regions form part of WM pathways connecting prefrontal and parieto-occipital areas with the striatum and the cerebellum. The authors also demonstrated anomalous WM development in ADHD in distinct cortical regions that they had previously shown to be dysfunctional or hypoactive in a functional MRI study of subjects with ADHD [ 47 ]. Diffusion tensor imaging (DTI) is an MRI modality that provides information about the direction and integrity of neural fibre tracks in the brain in vivo . DTI studies have revealed developmental changes in cortical WM pathways in prefrontal regions and in pathways surrounding the basal ganglia and cerebellum in patients with ADHD, which presumably reflect decreasing myelination of axons. It is believed that these changes cause a decrease in the speed of neuronal communication [ 48 ]. Moreover, the neural networks serving the corticostriatal and corticocerebellar circuits could represent putative biomarkers for ADHD. Indeed, in this disorder their quantification using DTI could be relevant for both diagnostic and therapeutic purposes [ 46 ]. As well as offering new data to map the brain systems involved in ADHD, and to integrate these findings with clinical symptoms, functional neuroimaging studies allow us to understand the mechanisms of treatment response [ 42 , 49 ]. Positron emission tomography (PET) studies have shown that methylphenidate hydrochloride (MPH) blocks dopamine active transporters (DAT) and that extracellular dopamine (DA) increases in proportion to the level of blockade and to the rate of DA release. This process is associated with an enhanced perception of the external stimulus as salient in subjects with ADHD [ 50 ]. Clinical diagnosis and comorbidities Clinical presentation of ADHD may vary according to age and stage of development and there are cultural differences in the level of activity and inattention that are regarded as a problem [ 51 ]. Diagnosis requires that there should be clear evidence of clinically significant impairment in social, academic, or occupational functioning [ 5 ]. The predominantly inattentive type is relatively more common in females. Children with the predominantly hyperactive-impulsive type are aggressive and impulsive, and tend to be highly rejected by their peers. The combined type causes more impairment in global functioning, in comparison with the other two types. Adolescents with ADHD often report low self-esteem and poor peer relationships; and are at high risk of smoking and substance abuse early in life [ 52 , 53 ]. Endophenotypes can be used as trait markers for disease susceptibility, to identify more genetically homogeneous subgroups, to highlight distinct pathophysiological mechanisms or etiological pathways, or to define "spectrum" phenotypes suitable for quantitative trait analyses [ 54 ]. Cognitive deficits and motor response inhibition are the prime endophenotype candidates in ADHD [ 55 ]. The co-existence of several other types of psychopathology along with ADHD, such as oppositional defiant disorder, mood and anxiety disorders, learning disorders, tics, and mental retardation, is very common [ 56 ]. Treatment Before starting treatment, it is important to identify the target outcomes to guide the therapy decision. Drug treatment should be based on a thorough assessment and should always be part of a comprehensive treatment plan that includes psychosocial, behavioural, and educational advice and interventions. Psychotherapy combined with medication may play a role in treating behavioural problems, organisational issues and psychiatric comorbidities [ 57 ]. In Italy, an ADHD diagnosis can only be made at a regional referral centre approved by the Italian Ministry of Health. Treatment guidelines put forward by the Ministry of Health and based on European guidelines, specify that pharmacological treatment can only be initiated after failure of cognitive behavioural therapy over a period of 6 months or longer has been demonstrated. Patients must first be enrolled in the ADHD medication registry before treatment with MPH or atomoxetine (ATX) can be prescribed. Behavioural therapy and pharmacological treatment have both been shown to benefit ADHD patients. A longitudinal study of the efficacy of different treatments (an intensively monitored medication program, behavioural therapy, combination of medication and behavioural therapy or treatment as usual by community care) showed after 8-year follow-up that all four of the original treatment groups had a similar outcome: all showed improvement in comparison with pretreatment baseline scores, but none demonstrated superiority [ 58 ]. The fronto-subcortical circuits (lateral prefrontal cortex, dorsal anterior cingulate cortex, caudate, and putamen) associated with ADHD are rich in catecholamines, which are involved in the mechanism of action of medications used to treat this disorder. Neuropharmacological studies have provided evidence that ADHD involves dysregulation of both noradrenaline (NE) and DA neurotransmitter systems [ 59 ]. MPH treatment causes an increase in DA signalling through multiple actions, including blockade of the DA reuptake transporter, amplification of DA response duration, disinhibition of the dopamine D2 receptor and amplification of DA tone [ 60 ]. MPH is also an inhibitor of NE re-uptake. ATX is a selective inhibitor of synaptic re-uptake, and in vivo , it specifically increases extracellular levels of DA in the prefrontal cortex but not in the striatum; probably by modulating cortical synaptic DA uptake via the NE transporter [ 61 ]. Dextroamphetamine increases the synaptic activity of DA and NE by increasing the release of the neurotransmitters into the synaptic cleft, decreasing reuptake back into the presynaptic neuron, and inhibiting their catabolism [ 62 ]. Strong evidence exists indicating that stimulant medications, such as MPH and dextroamphetamine, and the non-stimulant ATX, are effective in improving ADHD symptoms [ 63 ]. Guanfacine is a selective alpha2A adrenergic receptor agonist, which improves working memory by stimulating postsynaptic alpha2A adrenoceptors, strengthening the functional connectivity of prefrontal cortex networks [ 64 ]. Guanfacine has also been shown to be effective in reducing ADHD symptoms [ 65 , 66 ]. Table 1 summarises the most important characteristics of these pharmacological treatments for ADHD. Only ATX and immediate release MPH are currently approved for the treatment of ADHD in Italy. ADHD pharmacological therapies are generally well-tolerated (Table 1 ). However, concerns surrounding the cardiovascular safety of some of these drugs has prompted a recent examination of the effects of ATX and MPH on blood pressure (BP), heart rate (HR), and ECG parameters. MPH appears to cause minor increases in BP and HR, with no strong data to suggest that itincreases the QT interval. Limited data suggest that ATX may increase BP and HR in the short term; in the long term it appears to only increase BP. The effects of ATX on QT interval remain uncertain. Because the current evidence is based on research that has not been specifically designed to investigate the cardiovascular effects of these drugs, it is difficult to draw firm conclusions [ 67 ]. Both MPH and ATX significantly increase activation in key cortical and subcortical regions subserving attention and executive functions. Therefore, alterations in dopaminergic and noradrenergic function are apparently necessary for the clinical efficacy of pharmacological treatment of ADHD [ 68 ]. However MPH and ATX have both common and distinct neural effects, consistent with the observation that while many children respond well to both treatments, some respond preferentially to one or the other. Although pharmacotherapy for ADHD appears to prepare and facilitate the brain for learning, experiential programs need to elicit compensatory development in the brain. The clinical amelioration of some children after environmental experiential inputs and early cognitive/behavioural treatment could indicate outcome-associated plastic brain response [ 69 ]. One year of treatment with MPH may be beneficial to show enduring normalisation of neural correlates of attention. However, little is known about the long-term effects of stimulants on the functional organisation of the developing brain [ 70 ]. Recent findings have shown that chronic MPH use in drug-naive boys with ADHD enhanced neuropsychological functioning on "recognition memory" component tasks with modest executive demands [ 71 ]. Patients receiving pharmacological treatment for ADHD should always be closely monitored for both common and unusual potentially severe adverse effects.

Conclusions Convergent data from neuroimaging, neuropsychology, genetics and neurochemical studies consistently point to the involvement of the frontostriatal network as a likely contributor to the pathophysiology of ADHD. This network involves the lateral prefrontal cortex, the dorsal anterior cingulate cortex and the caudate nucleus and putamen [ 39 ]. Functional neuroimaging has provided new ways to examine the pathophysiology of ADHD, has shown widespread dysfunction in neural systems involving the prefrontal, striatal, and parietal brain regions, and has led to a brain model of deficits in multiple developmental pathways [ 72 ]. Molecular genetic studies support dysregulation of neurotransmitter systems as the basis of genetic susceptibility to the disorder, and it is becoming clear that the genotype may influence the response to medications [ 73 ]. Hopefully, advances in understanding the underlying neurobiology of ADHD will contribute to identifying more specific and targeted pharmacotherapies, and will help child neurologists to better manage their patients.

Attention-Deficit/Hyperactivity Disorder is not a single pathophysiological entity and appears to have a complex etiology. There are multiple genetic and environmental risk factors with small individual effect that act in concert to create a spectrum of neurobiological liability. Structural imaging studies show that brains of children with Attention-Deficit/Hyperactivity Disorder are significantly smaller than unaffected controls. The prefrontal cortex, basal ganglia and cerebellum are differentially affected and evidence indicating reduced connectivity in white matter tracts in key brain areas is emerging. Genetic, pharmacological, imaging, and animal models highlight the important role of dopamine dysregulation in the neurobiology of Attention-Deficit/Hyperactivity Disorder. To date, stimulants are the most effective psychopharmacological treatments available for Attention-Deficit/Hyperactivity Disorder. Currently only immediate release methylphenidate and atomoxetine are approved for the treatment of ADHD in Italy. Drug treatment should always be part of a comprehensive plan that includes psychosocial, behavioural and educational advice and interventions.

Competing interests Professor Paolo Curatolo has served as an Advisory Board Member for Eli Lilly and Shire, and has received research grants from Eli Lilly and Shire. The other authors have no conflicts of interest to declare. Authors' contributions PC proposed and designed the study, and revised the final draft. EDA and RM reviewed all the relevant articles on the literature, and prepared the first draft under the supervision of Prof. Paolo Curatolo. All authors contributed to the intellectual content and approved the final version.

Acknowledgements Editing of the manuscript and table preparation was performed by Rachel Wright of Fishawack Communications, supported by Shire Development Inc.

CC BY

no

2022-01-12 15:21:44

Ital J Pediatr. 2010 Dec 22; 36:79

oa_package/f6/42/PMC3016271.tar.gz

PMC3016272

21134289

Background Angiogenesis occurs in the central nervous system (CNS) not just during development [ 1 ], but also in pathological conditions, including cerebral ischemia [ 2 ], neoplasia [ 3 ], and neuroinflammation [ 4 , 5 ]. An improved understanding of the factors that control cerebral angiogenesis would be a big step forward in our attempts to regulate angiogenesis for therapeutic means, either to increase blood vessel growth during cerebral ischemia, or to inhibit vessel growth during neoplasia. Angiogenesis is regulated by a plethora of factors, including growth factors [ 6 ], cytokines [ 7 ], and extracellular matrix (ECM) molecules [ 8 ]. Within the CNS, it has been established that hypoxia promotes angiogenesis by at least two separate pathways. One involves hypoxia inducible factor-1α (HIF-1α)-dependent vascular endothelial growth factor (VEGF) release [ 9 ], and the other, that involves a HIF-1α-independent COX-2-dependent stimulation of PGE2, leading to angiopoietin-2 release [ 10 ]. In addition to soluble factors, ECM proteins also provide important instructional cues in angiogenesis [ 11 ], and recent work from our laboratory showing that fibronectin is strongly induced on angiogenic capillaries in the hypoxic CNS [ 12 ], as well as on angiogenic vessels in the developing CNS [ 13 ], suggests that this protein may also be important for cerebral angiogenesis. In the normal adult CNS, brain endothelial cells (BEC) occupy an angiostatic state, and have the impermeable, tight-barrier characteristics of mature cerebral endothelium [ 14 ]. During cerebral ischemia and other neuroinflammatory conditions, vessels in the adult CNS mount an angiogenic response in which BEC proliferate to form new capillary sprouts [ 15 , 16 ]. Interestingly, studies of cerebral ischemic tissue have demonstrated a strong association between new vessel formation and microglial recruitment and activation [ 17 , 18 ], raising the possibility that microglia, the principal immune effector cells in the CNS, may actively promote angiogenesis. As endothelial cell proliferation is a fundamental early step in the angiogenic response, the aim of this study was to test this hypothesis by examining the influence of microglial secreted factors on BEC proliferation.

Materials and methods Animals The studies described have been reviewed and approved by The Scripps Research Institute Institutional Animal Care and Use Committee. All animals were maintained under pathogen-free conditions in the closed breeding colony of The Scripps Research Institute (TSRI). Cell Culture Pure cultures of mouse brain endothelial cells (BEC) were prepared as previously described [ 19 ]. Briefly, brains were removed from 8 week-old C57Bl/6 mice, minced, dissociated for one hour in papain, centrifuged through 22% BSA to remove myelin, and then endothelial cells cultured in endothelial cell growth media (ECGM) consisting of Hams F12, supplemented with 10% FBS, Heparin, ascorbic acid, L-glutamine, penicillin/streptomycin (all from Sigma, St. Louis, MO) and endothelial cell growth supplement (ECGS) (Upstate Cell Signaling Solutions, Lake Placid, NY), on type I collagen (Sigma)-coated 6-well plates. Puromycin (4 μg/ml, Alexis GmbH, Grunberg, Germany) was included in culture media between days 1-3 to remove contaminating cell types. Endothelial cell purity was >99% as determined by CD31 in flow cytometry. For all experiments, BEC were used only for the first passage. Mixed glial cultures were prepared from 0-2 day old C57Bl/6 mouse pups, as previously described [ 19 ], and maintained in poly-D-lysine coated T75 flasks in DMEM containing 10% fetal bovine serum (FBS) (all from Sigma). After 7-10 days, flasks were mechanically shaken to yield microglia, which were plated into uncoated 6 well plates. Microglial purity was >99% as determined by Mac-1 in flow cytometry. Pure astrocyte cultures were prepared as previously described [ 20 ], by plating neurospheres into poly-D-lysine coated 6-well plates and maintained in DMEM containing 10% FBS. Astrocyte purity of these cultures was >99% as determined by GFAP immunocytochemistry. Microglia-conditioned media (MG-CM) Microglia were shaken off mixed glial cultures and plated in 6-well plates overnight. Media was then changed to serum-free DMEM containing N1-supplement, L-glutamine and penicillin/streptomycin (all from Sigma). Microglia were left unstimulated or activated with 1 μg/ml lipopolysaccharide (LPS, Sigma) to produce resting and activated MG-CM respectively. After 3 days, media was collected and filtered through a 0.22 μm filter before use in BEC studies. Astrocyte conditioned media (A-CM) was prepared in a similar manner. Brain endothelial cell proliferation assay Glass coverslips were coated with collagen I (10 μg/ml, Sigma) for 2 hours, then washed, and BEC plated and cultured until cells reached ~50% confluence. For investigation of the effect of MG-CM, BEC were cultured in 67% ECGM and 33% MG-CM. Control cultures were maintained in 67% ECGM and 33% N1-supplemented media. In the function-blocking experiments, the anti-TNF and anti-TGF-β1 blocking antibodies and control antibodies (R&D, Minneapolis, MN) were used at 2 μg/ml. For the investigation of the influence of cytokines on BEC proliferation, BEC were cultured in ECGM with factors added across a concentration range, with the maximum indicated: IFN-α (10 3 U/ml, PBL Biomedical Labs), IFN-γ (10 U/ml, R&D), IL-6 (20 ng/ml, R&D), TNF (20 ng/ml, Genentech, San Francisco, CA), and TGF-β1 (20 ng/ml, R&D). BEC were cultured for 16 hours in the presence of BrdU (Invitrogen, Carlsbad, CA), then fixed in acid/alcohol and processed for BrdU immunocytochemistry according to the manufacturer's instructions. BEC proliferation was assessed by quantifying the number of BrdU-positive cells as a percentage of the total number of cells (Hoechst staining), and the results expressed as the mean ± SEM of four experiments. Statistical significance was assessed by using Student's t test, in which p < 0.05 was defined as statistically significant. ELISA analysis of glial cytokine production Concentrations of TNF and TGF-β1 in microglial and astrocyte conditioned media were quantified by standard ELISA techniques using the Duoset ELISA system (R&D) according to the manufacturer's instructions. Results represent the mean ± SEM of 3 experiments, with each sample examined in triplicate within each experiment.

Results Microglial secreted factors regulate BEC proliferation in a biphasic manner Microglia were purified from mixed glial cultures of postnatal mice, and grown in serum free media for 3 days, in the absence or presence of the activating agent lipopolysaccharide (LPS), to produce microglia-conditioned media (MG-CM) from resting (unstimulated) or activated microglia, respectively. The influence of resting or activated MG-CM on BEC proliferation was examined using BrdU incorporation. As shown in Figure 1 , MG-CM had a clear biphasic influence on BEC proliferation rate. Compared to control conditions, resting MG-CM strongly inhibited BEC proliferation (6.4 ± 2.0% vs. 20.6 ± 1.6%, p < 0.001), while activated MG-CM had the opposite effect (30.1 ± 2.0% vs. 20.6 ± 1.6%, p < 0.002). The influence of astrocyte-conditioned media (A-CM) was also investigated. This revealed that resting and activated A-CM both inhibited BEC proliferation (resting A-CM 14.5 ± 1.8% vs. 20.6 ± 1.6%, p < 0.05, and activated A-CM 13.4 ± 1.0% vs. 20.6 ± 1.6%, p < 0.01). TNF and TGF-β1 have antagonistic effects on BEC proliferation Microglia produce a large number of cytokines, whose production is heavily-dependent on the state of microglial activation [ 21 ]. To identify which of these factors might be responsible for the microglial influence on BEC proliferation, we screened a panel of different cytokines for their ability to regulate BEC proliferation. This showed that TNF and TGF-β1 had the strongest, though opposing influences on BEC proliferation (Figure 2 ). TNF promoted BEC proliferation in a dose-dependent manner, which plateaued at a concentration of 10 ng/ml. Compared to control conditions (21.3 ± 1.4%), 10 ng/ml TNF increased the rate of BEC proliferation to 32.6 ± 4.5% (p < 0.01). In contrast, TGF-β1 reduced the rate of BEC proliferation, an effect which plateaued at a concentration of 10 ng/ml. Compared to control conditions, this concentration of TGF-β1 reduced BEC proliferation from 20.6 ± 1.9% to 13.8 ± 2.1% (p < 0.05). IFN-α, IFN-γ and IL-6 had no significant effect on BEC proliferation at any concentration tested. Microglia secrete TNF and TGF-β1, with the balance determined by microglial activation state We have demonstrated that BEC proliferation rate is regulated by MG-CM in a biphasic manner; inhibited by resting MG-CM, and promoted by activated MG-CM. Furthermore, we have shown that BEC proliferation is inhibited by TGF-β1, but promoted by TNF. While it is known from previous studies that microglia can produce these two cytokines [ 22 , 23 ], we next used ELISA to investigate in our own cultures, whether MG-CM contained TNF or TGF-β1, and how this expression is regulated by LPS. As shown in Figure 3A , TNF was present in resting MG-CM (27.7 ± 8.6 pg/ml), but strongly increased (50-fold) in activated MG-CM (1381.2 ± 82.7 pg/ml). In contrast, TGF-β1 levels were not significantly different in resting (196.5 ± 42.3 pg/ml) or activated MG-CM (177.6 ± 43.4 pg/ml) (Figure 3B ). A-CM contained no TNF, either before or after activation (Figure 3A ). A-CM did contain TGF-β1, though this expression was not regulated by LPS treatment, and was at lower levels than MG-CM (39.2 ± 23.6 pg/ml and 75 ± 20.5 pg/ml in resting and activated A-CM, respectively). Thus, microglial activation switches the TNF/TGF-β1 balance heavily in favor of TNF. Microglial supernatants regulate BEC proliferation via TNF and TGF-β1 To address whether the influence of MG-CM on BEC proliferation is mediated by TNF or TGF-β1, we examined the influence of resting or activated MG-CM on BEC proliferation in the presence of function-blocking antibodies directed against TNF or TGF-β1. As shown in Figure 4A , compared to control conditions (20.9 ± 1.6%), resting MG-CM (Con IgG) significantly inhibited BEC proliferation (7.2 ± 2.0%, p < 0.005). This effect was significantly released by the anti-TGF-β1 antibody (12.5 ± 1.0%, p < 0.01). In contrast, compared to control conditions (19.7 ± 1.4%), activated MG-CM (Con IgG) significantly increased BEC proliferation (29.2 ± 1.7%, p < 0.01; Figure 4B ), and this effect was significantly inhibited by the anti-TNF blocking antibody (18.9 ± 1.7%, p < 0.01). This data strongly suggests that TGF-β1 is in part responsible for the negative influence of resting MG-CM on BEC proliferation, and conversely, implicates TNF as a major factor responsible for the positive effect of activated MG-CM on BEC proliferation.

Discussion Cerebral angiogenesis occurs in the adult CNS in a number of conditions, including cerebral ischemia [ 2 ], neoplasia [ 3 ], and the neuroinflammatory conditions multiple sclerosis [ 5 ], and Alzheimer's disease [ 4 ]. In these conditions, angiogenesis is often associated with microglial accumulation and activation [ 17 , 18 ], raising the possibility that activated microglia may promote angiogenic events. In the current study, we directly examined this question in vitro, by exposing BEC to microglia-conditioned media. This revealed that microglia regulate BEC proliferation in a biphasic manner. Soluble factors from resting microglia inhibit, while those from activated microglia promote BEC proliferation, and these effects are largely attributable to the microglial cytokines TGF-β1 and TNF, respectively. This data suggests that in vivo, microglial activation state might be an important determinant of the earliest stage of cerebral angiogenesis, namely endothelial cell proliferation. This model predicts that in the normal CNS, tonic levels of TGF-β1 inhibit BEC proliferation, but that during cerebral ischemia or other neuroinflammatory processes, activated microglia ramp up TNF production, which promotes BEC proliferation. Microglial activation and angiogenesis Microglia, as the primary immune effector cells in the CNS, play important roles in the surveillance and response to pathological insults. As the gatekeeper of the BBB, microglia are well positioned to mount a rapid aggressive response to noxious stimuli that enter the CNS. Under normal conditions, microglia occupy a resting "on surveillance" phenotype, but after stimulation, switch to an activated highly migratory, mitogenic phenotype, with upregulated production of inflammatory cytokines such as TNF and IFN-γ [ 21 , 24 ]. Within days of an ischemic insult, angiogenic remodeling can be observed at the ischemic penumbra [ 15 , 16 ], and interestingly, angiogenic vessels are often surrounded by inflammatory microglia and macrophages [ 17 , 18 ], suggesting that activated microglia and/or macrophages may be instrumental in promoting the angiogenic response to cerebral ischemia. While this idea has not yet been directly tested in the ischemic CNS, traumatic CNS injury leads to activation of microglia and macrophages, and drugs that block activation and proliferation of these cells, also inhibit neovascularization [ 25 ]. In addition, studies from outside the CNS support the concept that activated cells of the macrophage lineage directly promote angiogenic events. In tumor models, depletion of monocytes, the blood precursors of tissue macrophages, leads to marked reduction in tumor-associated angiogenesis [ 26 ], and conversely, addition of tissue macrophages strongly promotes neovascularization in corneal and rabbit ear chamber models [ 27 , 28 ]. Consistent with this, and in agreement with our own finding, conditioned media taken from activated macrophages directly promotes endothelial cell proliferation in vitro [ 29 ]. The influence of TNF and TGF-β1 on angiogenic events Our studies suggest that the positive and negative influences of microglia on BEC proliferation is mediated, at least in part, by the antagonistic cytokines TNF and TGF-β1, respectively. The influence of these cytokines on angiogenic events is still a matter of debate, with different studies demonstrating either pro- or anti-angiogenic effects, depending on the source of endothelial cells and the concentration of cytokine employed. TNF has been shown to both promote [ 30 - 32 ] and inhibit [ 33 ] angiogenic events, with one report demonstrating a negative effect of TNF on endothelial cell proliferation in vitro, but a stimulation of neovascularization in vivo [ 34 ], while another showed both positive and negative effects on endothelial cell proliferation, depending on the cellular source and cytokine combinations [ 35 ]. Similar apparently conflicting data have also been described for TGF-β1 [ 7 , 36 ]. In an attempt to reconcile these data, it has been suggested that the effects of these factors may have biphasic effects on angiogenesis, depending on cytokine concentration [ 36 ]. To directly investigate this possibility, we examined BEC proliferation across a wide range of cytokine concentrations, but in each case, this demonstrated a clear dose-response effect in one direction only, namely TNF promoting and TGF-β1 inhibiting BEC proliferation. Significantly, recent data has shown that the biphasic effect of TNF on endothelial cells can be explained by antagonistic functions of the TNF receptors TNFR1 and TNFR2, with TNFR2 stimulating endothelial cell survival, migration, and angiogenesis, while TNFR1 inhibits these functions [ 37 , 38 ]. In the current study, close examination of the TGF-β1 concentrations that mediated an inhibitory effect on BEC proliferation revealed an apparent difference in the potency of the endogenous form (present in resting MG-CM; 0.2 ng/ml) when compared with the pure recombinant form (5 ng/ml). Two reasons may explain this discrepancy. First, while the effect of recombinant TGF-β1 did not reach statistical significance until a concentration of 5 ng/ml, it did have an inhibitory trend at lower concentrations (0.5-1 ng/ml). Second and perhaps more important, in this study we used the human recombinant form of TGF-β1, and it is entirely plausible that the biological activity of this form on mouse BEC may not be as high as the endogenous murine form. In summary, our data provides evidence that microglia, the principal immune effector cells in the CNS, can regulate BEC proliferation in a biphasic manner by altering the balance of TNF and TGF-β1. As BEC proliferation is an early stage of the angiogenic process, and new vessel formation leads to increased cerebral blood flow [ 39 ] and clinical outcome [ 40 , 41 ], our findings suggest that a controlled level of microglial activation and TNF release might prove beneficial in the treatment of stroke patients by promoting BEC proliferation and subsequent neovascularization. In light of the recent finding that microglia protect neurons from ischemia via a TNF-mediated mechanism [ 42 ], this approach has the potential to stimulate a positive outcome via two separate mechanisms.

Conclusions Our data demonstrate that microglia regulate BEC proliferation in a biphasic manner; microglia conditioned medium (MG-CM) from resting microglia inhibit, while that from activated microglia promote BEC proliferation. BEC proliferation was also inhibited by TGF-β1, but promoted by TNF. ELISA showed that TNF and TGF-β1 are both present in MG-CM, and that while TGF-β1 dominated in resting MG-CM, TNF levels were massively increased in activated MG-CM, shifting the balance in favor of TNF. Antibody-blocking studies revealed that the influence of MG-CM to inhibit or promote BEC proliferation was largely attributable to the cytokines TGF-β1 and TNF, respectively. Taken together, this data suggests that microglial activation state might be an important determinant of BEC proliferation, an early event in cerebral angiogenesis.

Background Studies of cerebral ischemia and other neuroinflammatory states have demonstrated a strong association between new vessel formation and microglial recruitment and activation, raising the possibility that microglia may be involved in promoting angiogenesis. As endothelial cell proliferation is a fundamental early step in angiogenesis, the aim of this study was to test this hypothesis by examining the influence of microglial secreted factors on brain endothelial cell (BEC) proliferation using BrdU incorporation. Methods Primary cultures of mouse BEC, microglia and astrocytes were used in this study. Proliferation of BEC was examined by BrdU incorporation. ELISA was used to quantify TNF and TGF-β1 levels within cell culture supernatants. Results Microglia regulated BEC proliferation in a biphasic manner; microglia conditioned medium (MG-CM) from resting microglia inhibited, while that from activated microglia promoted BEC proliferation. A screen of microglial cytokines revealed that BEC proliferation was inhibited by TGF-β1, but promoted by TNF. ELISA showed that TNF and TGF-β1 were both present in MG-CM, and that while TGF-β1 dominated in resting MG-CM, TNF levels were massively increased in activated MG-CM, shifting the balance in favor of TNF. Antibody-blocking studies revealed that the influence of MG-CM to inhibit or promote BEC proliferation was largely attributable to the cytokines TGF-β1 and TNF, respectively. Conclusion This data suggests that microglial activation state might be an important determinant of cerebral angiogenesis; inhibiting BEC proliferation and neovascularization in the normal central nervous system (CNS), but stimulating the growth of new capillaries under neuroinflammatory conditions.

Competing interests The authors declare that they have no competing interests. Authors' contributions JVW prepared the cell cultures, carried out the biochemical analysis and contributed to drafting the manuscript. LL participated in the design of the study and also assisted in manuscript preparation. RM conceived of the study, performed the proliferation studies, and helped to draft the manuscript. All authors read and approved the final manuscript.

Acknowledgements This work was supported in part by a Harry Weaver Neuroscience Scholar Award (JF 2125A1/1) from the National Multiple Sclerosis Society (RM), and by NIH grant RO1 NS060770. This is manuscript number 20532 from The Scripps Research Institute.

CC BY

no

2022-01-12 15:21:44

J Neuroinflammation. 2010 Dec 6; 7:89

oa_package/93/27/PMC3016272.tar.gz

PMC3016273

21159187

Although infiltration of peripheral monocytes/macrophages is implicated in stroke pathology, in vivo data regarding the deployment of monocytes and their mobilization to the infarct area is scarce. Recent literature showed that mouse monocytes exhibit two distinct populations that represent pro-inflammatory (Ly-6C hi /CCR2+) and anti-inflammatory (Ly-6C low /CCR2-) subsets and that spleen is a major source for monocyte deployment upon injury. By reducing post-ischemic infection with antibacterial moxifloxacin (MFX) treatment, the present study investigates the effect of the treatment on Ly-6C and CCR2 expression in the spleen following ischemia and the extent to which the effect is associated with attenuation of post-ischemic inflammation and injury. Mice subjected to a middle cerebral artery occlusion (MCAO) showed a significant reduction in their spleen weights compared to sham animals. Compared to vehicle controls, splenocytes obtained from daily MFX-treated mice 7 days after ischemia exhibited significantly reduced mean Ly-6C expression within pro-inflammatory subsets, whereas the distribution of pro- and anti-inflammatory subsets was not different between the treatment groups. Additionally, MFX treatment significantly reduced CCR2 expression in the spleen tissue and in the post-ischemic brain and attenuated infarct size. The study suggests a potential contributing role of spleen monocytes in post-ischemic inflammation and injury. The influence of peripheral inflammatory status on the primary injury in the CNS further implies that the attenuation of post-stroke infection may be beneficial in mitigating stroke-induced brain injury.

Findings Ischemia-reperfusion causes inflammation that attracts monocyte/macrophage cells to infarct [ 1 - 3 ]. Monocytes are circulating antigen-presenting leukocytes that play an important role in inflammation, T-cell differentiation, phagocytosis, and innate immunity [ 4 , 5 ]. It has been shown that circulating and spleen monocytes are similar in their morphology, phagocytic capability, and gene expression profiles [ 6 ]. The study also identified the spleen as a monocyte reservoir and their numbers in the spleen are several folds higher than in circulation [ 6 ]. In addition, the number of monocytes that migrate to the infarct area after a myocardial infarction well exceeds the number in circulation under homeostatic conditions [ 4 ]. These studies suggest a potential role of the spleen in deploying monocytes upon cerebral ischemia. Human and mouse monocytes exhibit distinct subsets that are reminiscent of macrophage phenotypes [ 5 , 7 , 8 ]. In mice, the subset that expresses a high level of the hematopoietic cell differentiation antigen Ly-6C (Ly-6C hi ) also expresses the G-protein linked membrane protein, CCR2. The Ly-6C hi /CCR2+ monocyte subset is specifically recruited to an injury site by monocyte chemoattractant protein-1 (MCP-1), which is produced by the inflamed tissue, and become classically activated M1 macrophages. In contrast, the Ly-6C low monocyte subset expresses CX 3 CR1, a receptor for the chemokine CX 3 CL1 (fractalkine), but is devoid of CCR2 expression. This anti-inflammatory Ly-6C low /CCR2-/CX 3 CR1+ subset is recruited to normal tissue and develops into resident M2 macrophages that function in host defense and repair after injury [ 9 , 10 ]. Recruitment of the pro-inflammatory Ly-6C hi /CCR2+ subset to inflammatory sites is believed to be CCR2-dependent, since monocytes from CCR2-null mice do not traffic as efficiently into a myocardial infarct as CCR2+ monocytes [ 6 ]. Furthermore, CCR2-null mice were protective against cerebral inflammation following ischemia [ 11 ], suggesting that CCR2 is a contributing factor for stroke-induced injury. Studies suggest a potential influence of peripheral inflammatory status on primary injury. Fever and systemic infections are frequently observed conditions in patients suffering from stroke and are associated with increased mortality and poorer outcome [ 12 , 13 ]. Treatment with antibacterial agents such as moxifloxacin (MFX) and minocycline was shown to reduce infarct in experimental animal models of stroke [ 14 , 15 ]. In addition, MFX treatment also reduced peripheral infection in patients who have suffered an ischemic stroke and in animal models of stroke [ 16 ]. The present study investigates whether improving peripheral infection by treatment with MFX shifts spleen monocytes to a less pro-inflammatory state and if the effect is associated with attenuation of post-ischemic inflammation and injury. Here, we report a potential influence of peripheral inflammatory status on stroke-induced inflammation and injury. All experimental procedures on animals were approved by the Institutional Animal Care and Use Committee of Weill Medical College of Cornell University. C57BL/6 male mice obtained from Jackson Laboratory (Bar Harbor, ME) were subjected to a 40 min middle cerebral artery occlusion (MCAO) as described previously [ 17 , 18 ]. The cerebral blood flow (CBF) in the center of the ischemic territory was monitored by laser-Doppler flowmetry (Periflux System 5010; Perimed, Jarfalla, Sweden). Moxifloxacin (MFX; Bayer, Wayne, NJ) solution (10 mg/ml) was prepared in a mixture of saline and 1 mol/L HCl (10:1) and adjusted to pH 7 with NaOH. Mice were treated with either vehicle or MFX solution (100 mg/kg) immediately after reperfusion, then once a day for 7 days following MCAO and sacrificed 7 days after ischemia. Sham-operated mice served as controls. Brains were removed, frozen, and sectioned (thickness, 20 μm) in a cryostat as previously described [ 17 , 18 ]. Brain sections were collected serially at 600 μm intervals, and stained with Cresyl Violet. Infarct volume was determined using Axiovision (Zeiss, Germany) and the contribution due to swelling was corrected. The distribution of monocyte subsets and expression of Ly-6C were analyzed by flow cytometry/FACS according to a published method [ 6 ]. After removal of RBC using a lysis buffer (Sigma-Aldrich, St. Louis, MO), single splenocyte suspension was incubated with a cocktail of phycoerythrin (PE)-conjugated antibodies (BD Biosciences, San Jose, California) against T cells (CD90.2-PE, Clone 53-2.1), B cells (CD45R/B220-PE, Clone RA3-6B2), NK cells (CD49b/Pan-NK cells-PE, Clone DX5,; NK1.1-PE, Clone PK136), granulocytes (Ly-6G-PE, Clone 1A8), allophycocyanin (APC)-conjugated antibody against myeloid cells including monocytes/macrophages (CD11b-APC, Clone M1/70), and fluorescein isothiocyanate (FITC)-conjugated antibody to monocyte subsets (Ly-6C-FITC, Clone AL-21). After washing with Dulbecco's phosphate buffer saline (DPBS, VWR, West Chester, PA), 100,000 cells were analyzed by FACS. The selected gate (low PE/high APC which were identified as monocytes) was further analyzed for Ly-6C-FITC. The procedures for RNA isolation and real time reverse transcriptase (RT)-PCR for gene expression were previously described [ 17 ]. Relative mRNA levels were quantified with real-time quantitative RT-PCR (qPCR) using fluorescent TaqMan technology. PCR primers specific for CCR2, MCP-1 and β-actin (internal control) were obtained as TaqMan pre-developed optimized assay reagents for gene expression (Applied Biosystems, Foster City, CA). The PCR reaction was performed using FastStart Universal Probe Master (Roche), according to the manufacturer's instructions. Reactions were performed in 20 μl total volumes and incubated at 95°C for 10 min, followed by 45 cycles of 15 s at 95°C and 1 min at 60°C. The results were analyzed by 7500 Fast Real-Time PCR System software. For CCR2 protein levels, spleen was homogenized in radioimmunoprecipitation assay buffer (RIPA buffer, Sigma-Aldrich) with freshly added protease inhibitors (Sigma-Aldrich) and centrifuged at 10,000 rpm for 10 min at 4°C. The protocol for SDS-PAGE and western blot were described previously [ 17 ]. After blocking the membrane for 1 h, anti-CCR2 monoclonal antibody (1:1000, ab32144, Abcam, Cambridge, MA) and anti-α-tubulin antibody as internal control (1:1000, T6199, Sigma-Aldrich) were added, followed by IRDye 680 Goat anti-rabbit IgG and IRDye 800 Donkey anti-mouse IgG for CCR2 and for α-tubulin respectively (1:10,000, LI-COR). The bands were visualized using the Odyssey Imaging System (LI-COR). To investigate involvement of the spleen in stroke pathology, spleen weights were determined after MCAO. Ischemia caused significant reductions in spleen weight at 3 and 7 days post-ischemia with a larger reduction at 7 days (Figure 1A ). Since body and spleen weights are significantly correlated in normal mice (R 2 = 0.85, p < 0.001, n = 13), we consider the possibility that the reduced spleen weights are due to body weight loss following stroke. Normalization of spleen weight by body weight still revealed significant weight reductions in the spleen (Figure 1B ), indicating a potential involvement of spleen in stroke pathology. Spleen monocytes were shown to resemble circulating monocytes in their morphology, phagocytic capability and transcriptome [ 6 ]. To reflect peripheral inflammatory state, we treated mice with vehicle or MFX after MCAO and analyzed the spleen weights and monocyte subsets in the spleen. Compared to pre-ischemic spleen weight (76.2 ± 3.6 mg, n = 9), stroke caused similar reductions in spleen weight between vehicle- and MFX-treated mice at 7 days post-ischemia (49.86 ± 7.92 mg vs 47.95 ± 5.66 mg, ns, n = 10-12). The lack of difference in weight at this time point may be due to immediate deployment of spleen monocytes upon injury as the spleen is an immediate source for monocytes [ 6 ]. Further FACS analyses in the selected gate for monocytes (Figure 2A , R1) revealed no differences in subset distribution between the treatment groups. However, MFX-treated mice showed reduction in mean Ly-6C expression within pro-inflammatory Ly-6C hi subset (Figure 2 ). The reduced mean Ly-6C expression in MFX-treated mice was also associated with attenuated CCR2 gene and protein expression (Figure 3A & 3B ). The findings are consistent with other reports showing increased Ly-6C expression in response to pro-inflammatory stimuli [ 19 , 20 ] and reduced expression of cytokines and Ly-6C in CCR2-null mice [ 6 , 21 , 22 ]. Collectively, our findings indicate that MFX treatment following stroke shifts the spleen toward a less inflammatory state. We further investigated whether the MFX treatment influences the expression of MCP-1 and CCR2, a primary inflammatory axis and injury in the ischemic brain. Stroke increased MCP-1 mRNA levels in the infarct similarly in both groups (Figure 4A ). Interestingly, MFX treated group showed a significant reduction in CCR2 mRNA levels in the post-ischemic brain, suggesting either less CCR2+ cell infiltration or reduced CCR2 expression in infiltrated cells to the injury site (Figure 4B ). This was accompanied by a modest but significant reduction in infarct size (Figure 4C ). Despite a neurocentric view of stroke-induced brain injury, infiltration of peripheral monocytes/macrophages into the infarct territory indicates the influence of these immune cells in developing brain injury [ 23 - 26 ]. Although controversial, approximately 10-20% of infiltrated cells express markers for monocytes/macrophages [ 27 , 28 ]. With reported peripheral studies showing that prolonged systemic inflammation following stroke exacerbates brain injury and MFX treatment reduces the peripheral infection and improves ischemic outcome [ 14 , 16 ], the present study addresses the biological events that underlie the peripheral influence on stroke injury. Although recruited monocytes are thought to arise from systemic circulation and bone marrow, the unexpected finding that identifies the spleen as an immediate reservoir to deploy a large quantity of monocytes upon injury [ 6 ], argues for a novel role of spleen monocytes in stroke pathology. The MFX-induced shift to less inflammatory status in the periphery was accompanied by reduced CCR2 expression in the post-ischemic brain (Figure 4B ). CCR2 is a chemokine receptor for MCP-1, and the CCR2/MCP-1 axis plays a critical role in a host of inflammatory conditions [ 11 , 29 , 30 ]. Although it has been shown that other CNS cells, such as neurons and microglia, also express CCR2 in pathological conditions [ 31 ], it is believed that the major source of CCR2 in the post-ischemic brain is from the periphery. The reduced CCR2 expression, thereby attenuating post-ischemic inflammation in MFX-treated mice, may be due to reduced trafficking of monocytes to the injury site and/or attenuated CCR2 expression among recruited cells. Since we found that MFX treatment reduces Ly-6C expression in the pro-inflammatory subset (Figure 2 ), these findings support the view that a shift of spleen Ly-6C hi subset to a less inflammatory phenotype may account for the reduced CCR2 expression in MFX-treated mice. Consistent with literatures showing that MFX improves ischemic outcome and reduces peripheral infection both in animal models and patients [ 14 , 16 , 32 , 33 ], we also observed that MFX treatment resulted in a moderate but significant reduction in injury size. The MFX-treated mice also exhibited a smaller interanimal variability in injury size (Figure 4C ). This raises an intriguing notion that the neuroprotection effects may not be caused by a primary ischemia-reperfusion event but by a secondary event such as reduced expression of inflammatory mediator in peripheral immune cells. The degree of ischemic severity was comparable between the treatment groups indicated by no noticeable differences in the degree of cerebral blood flow reduction during (MFX 83.8 ± 2.3% vs vehicle 83.6 ± 2.0%) and reperfusion 10 min after ischemia (MFX 122.6 ± 7.2% vs vehicle 134.7 ± 9.2%) and in MCP-1 mRNA levels produced in the post-ischemic brain (Figure 4A ). MFX significantly decreased the WBC counts in circulation (2264 ± 259 vs 960 ± 60 cells/μl, n = 7-10, p < 0.01). Further FACS analysis in the spleen revealed the reduced percentage of monocyte population in the MFX-treated group (4.1 ± 1.0% vs 2.9 ± 0.4%, n = 5, p < 0.05). Since elevated systemic inflammation induces monocytosis, the decreased population of spleen monocytes by MFX treatment indicates attenuation of peripheral inflammation following stroke. Therefore, the protective mechanism may lie in limiting the expansion of secondary injury associated with the reduced inflammatory state in monocytes. In summary, antibacterial MFX treatment after stroke shifted inflammatory status in the spleen toward a less inflammatory state and attenuated post-ischemic inflammation and injury. The tight association between MFX-induced neuroprotection and inflammatory status in the brain and spleen suggests that reducing peripheral inflammation may be beneficial in ameliorating post-ischemic inflammation and injury. The potential influence of inflammatory status in peripheral monocytes on CNS injury suggests a possible cell-based strategy to limit the expansion of secondary injuries during post-ischemic periods. Competing interests The authors declare that they have no competing interests. Authors' contributions YB contributed to design of the study, performed the western blot, Real-Time PCR and FACS analysis, reviewed and organized the data and wrote the manuscript; EK contributed to developing the concept of the involvement of spleen in stroke pathology, the spleen weight measurement and reviewed the manuscript; SB performed the MCAO surgery, white blood cell counting and reviewed the manuscript; HM performed tissue preparation and infarct analysis and reviewed the manuscript; SC directed the overall study, analysis of the data, and reviewed the manuscript. All authors have read and approved the final manuscript.

Acknowledgements This work is supported by NIH grants HL82511 and HL82511-04S1 (SC).

CC BY

no

2022-01-12 15:21:44

J Neuroinflammation. 2010 Dec 15; 7:92

oa_package/2f/6a/PMC3016273.tar.gz

PMC3016274

21122159

Background Protein secretion is one of the main means by which bacteria interact with their environment. The interaction may take place in a variety of manners: bacteria secrete enzymes, toxins and other virulence factors, excrete metabolic waste products, and export binding proteins into the periplasm for import of nutrients or export of toxic compounds. Bacteria also use different secretion systems to assemble on their surface organelles for motility, adhesion and injection of effector molecules into host cells [ 1 , 2 ]. Bacterial protein secretion systems are of great importance from a virulence-associated viewpoint as potential targets for novel antibacterial drugs [ 3 ] but are significant also commercially and therapeutically due to the use of bacteria as protein factories for the delivery of polypeptides into bacterial growth medium [ 4 , 5 ]. In bacteria, secretion is regarded as active transport of proteins from the cytoplasm to the exterior of the cell [ 6 ]. In Gram-negative bacteria, the proteins to be exported have to first cross the cytoplasmic membrane (CM), usually at the cost of ATP hydrolysis, and further the outer membrane (OM). Six distinct protein secretion pathways have currently been described in Gram-negative bacteria, some of these to atomic detail [ 7 - 9 ]. Some of the secretion pathways, i. e. the type 2 and type 5 secretion systems consist of separate protein complexes on the two membranes [ 10 - 12 ], whereas others, i. e. the type 1, type 3, type 4 and type 6 pathways form continuous pores or tunnels crossing all the way from the cytoplasm to the exterior [ 13 , 14 ]. In addition to the complexes specifically devoted to protein secretion, bacteria possess so-called ATP-binding cassette (ABC) secretion systems, which carry a wide variety of substrates across the CM. Of these, the ABC importers catalyze the uptake of nutrients that are delivered to them by specific periplasmic substrate-binding proteins (PBPs), whereas ABC exporters are involved in the transport of e.g. drugs, lipopolysaccharides, toxins as well as proteins from the cytoplasm [ 13 , 15 ]. The food-borne human gastrointestinal pathogen Campylobacter jejuni expresses the protein Peb1 [ 16 ], which acts in distinct roles in separate compartments of the bacterial cell. First, Peb1 is present on the bacterial surface of all C. jejuni isolates analyzed, which makes it a putative target for diagnostics, and is regarded as a significant colonization and virulence factor. Antibodies against Peb1 are found in sera derived from the majority of patients convalescent from C. jejuni infection [ 16 ]. The protein mediates C. jejuni adherence in vitro to HeLa cells and is required for C. jejuni colonization of mice intestine [ 17 , 18 ]. Second, the major fraction of Peb1 is present in the periplasm of C. jejuni and acts as an aspartate/glutamate -binding PBP essential for optimal microaerobic growth of C. jejuni on these amino acids [ 19 , 20 ]. Finally, Peb1 is also found in minor amounts in the culture supernatant, but it is not present in the CM and OM [ 19 ]. For fulfilling each of its roles, Peb1 must be efficiently exported to the correct cell compartment. In the cytoplasm of C. jejuni , Peb1 carries an N-terminal signal sequence, typical of proteins secreted by the SecA-YEG pathway, which directs its export across the CM. A processing site typical for signal peptidase II has additionally been predicted in the N-terminus of Peb1, but the biological relevance of this processing site as well as the mechanism for the further transport of Peb1 from the periplasm across the OM in C. jejuni are still unknown [ 20 ]. We have previously shown that Peb1 expressed without its native SecA-YEG -dependent N-terminal secretion sequence and without the putative processing site for signal peptidase II is exported very efficiently into the growth medium at a high concentration and purity in an Escherichia coli flagellar mutant [ 21 ]. The reader is referred to Chevance and Hughes for a more detailed account on flagellar regulation and assembly [ 22 ]. Here we further analyze and document this phenomenon in order to increase the knowledge on how a heterologous model protein may be transported across the cell wall into the growth medium of E. coli and to broaden the understanding of factors that influence heterologous protein secretion in E. coli . We report that Peb1 is actively secreted in a flagellar-independent manner by an unknown two-step process and that the secretion is affected by mutagenesis of two distinct transport-affecting regions in Peb1.

Methods Bacterial strains and growth conditions Bacterial strains and plasmids used in this work are described in Table 1 . E. coli strains were cultivated by shaking in Luria broth [ 54 ] or on Luria agar plates for 16 h at 37°C supplemented with antibiotics (tetracycline 12.5 μg/ml, streptomycin 100 μg/ml, ampicillin 100 μg/ml) when appropriate. For TCA-precipitation and periplasmic isolation, the bacteria were further inoculated (1:30) into 3 ml and 12 ml, respectively, of N-minimal medium [ 54 ] supplemented with 0,2% casamino acids and appropriate antibiotics, and cultivated statically for 16-24 h at 37°C for optimization of protein secretion as we have described previously [ 21 ]. For intracellular solubility assays, the strains were cultivated by shaking in Luria broth supplemented with appropriate antibiotics for 18 h at 37°C followed by dilution 1:100, and cultivated without agitation for 20 hrs at 37°C in 20 ml Luria broth with antibiotics. For analysis of the effect of subinhibitory concentrations of NaN 3 on Peb1 export, the strains were grown as described but in the presence of 3 mM NaN 3 . Antisera and reagents Rabbit anti-Peb1 antiserum (diluted 1:50 000 for immunoblot analysis) and anti-flagellum antiserum (diluted 1:2000) were available from the previous work [ 21 ]. For control of cell lysis, polyclonal anti-GroEL antibody was used at a dilution of 1:10 000 (Sigma-Aldrich) and mouse monoclonal anti-RNAP at a dilution of 1:10 000 (Neoclone). Alkaline phosphatase-conjugated secondary antibodies were obtained from DakoCytomation and diluted to a concentration of 1.5 μg/ml (1:1000). All antibodies were diluted in 1% BSA/PBS. Reagents for recombinant DNA techniques and PCR were mostly obtained from Promega. For PCR cloning and mutagenesis Pfu DNA polymerase (Promega) was used. Colony PCR was performed using Dynazyme II DNA polymerase (Finnzymes). Primers used in this study were purchased from Medprobe and are shown in Table 3 . Construction of bacterial strains Bacterial strains used in this work are described in Table 1 . The E. coli strains MKS111, MKS111ΔHBB and MKS12P lac peb1 were constructed from E. coli strain MKS12 [ 21 ] by site-specific mutagenesis using the pir -dependent suicide vector pCVD442 as described earlier [ 55 ]. For the construction of MKS111, a DNA fragment covering 2348 bp (nucleotides1128017-1130365) was amplified by PCR to contain an in-frame deletion of flgM (367 bp, nucleotides1129426-1129059). Chromosomal deletions of flgDEFGHIJ and fliFGH were performed to create the E. coli strain MKS111ΔHBB using the same principle as above. The genes flgDEFGHIJ were first deleted from the chromosome of MKS111 using a PCR-amplified fragment covering 6782 bp (nucleotides 1130748-1137529) containing a deletion of flgDEFGHIJ (4905 bp, nucleotides 1131696-1136600). From this strain fliFGH was then chromosomally deleted by the same principles; the DNA fragment of 5589 bp (nucleotides 2010111-2015699) contained the deletion of fliFGH (3329 bp, nucleotides 2011212-2014541). For the construction of strain MKS12P lac peb1 , a 3066 bp DNA fragment was amplified by PCR to contain fliC flanking regions (nucleotides 199936-2000133 and nucleotides 2001784-2003814 with the exception of the fliD deletion, nucleotides 2002082-2003109), P lac , fliC 5' UTR and peb1 lacking its signal sequence. This fragment was used for substitution, by homologous recombination, of fliC in strain MKS1 [ 56 ]. Primers used in mutagenesis were designed on the basis of the nucleotide sequence of E. coli MG1655 and, for the P lac promoter, E. coli DH1 (GenBank Accession numbers NC 000913 and CP001637.1 ) [ 57 ]. Chromosomal DNA of E. coli MG1655 Δ fimA-H , also called AAEC072A, and plasmid pTSF12 [ 58 ], were used as templates for PCR amplifications. The genotypes of the mutants were confirmed by PCR and Southern blot hybridization. Construction of plasmids Plasmids used in this work are described in Table 1 . The authenticity of the construct has been verified for each plasmid by sequencing. Plasmids pMCS3'UTR and pPeb15'3' were available from the previous work [ 21 ]. Our former results indicated that the 5' UTR of fliC is important for secretion of Peb1 in MKS12, whereas expression of peb1 from e.g. the promoter and 5'UTR of the ß-lactamase-encoding gene rendered Peb1 in the cytoplasm of MKS12 [ 21 ]. Therefore, in the current study all the peb1 constructs were generated to maintain the sequence corresponding to the 5' region of fliC mRNA and thereby facilitate secretion of the gene product. To construct pLA25 a DNA fragment containing the peb1 gene, without the sequence encoding the leader peptide, fused to the 173 bp 5'UTR of fliC that includes P fliC , was constructed by PCR amplification from pPeb15'3 with primers containing Bgl II and Xba I restrictions sites (primers 5'UTRfwBglIl and Peb1RXbaI). This DNA fragment was ligated into the Bgl II- Xba I digested pMCS3'UTR and the ligation mixture was introduced into E. coli MKS12 cells by electroporation. To construct the fliC 5'UTR- peb1 target for the in vitro transposition reaction, the plasmid pLA26 was generated. The 3'UTR of fliC in pLA25 was removed by Sca I- EcoR I digestion and the cohesive ends were subjected to a blunting reaction with Mung Bean Nuclease at a concentration of 5 units per μg DNA for 10 min at 37°C. The blunt ends were ligated, and the ligation mixture was introduced by electroporation into E. coli DH10B cells. To construct plasmid pLA31, a fragment containing P fliC and the sequence encoding the N-terminal 183 residues of FliC was amplified from chromosomal DNA of E. coli AAEC072A (primers 5'UTRfwBglIl and FliC183XbaIR, respectively). The Bgl II- Xba I-digested PCR fragment was cloned into pMCS3'UTR and subsequent ligation mixture was introduced into MKS12 by electroporation. To create pLA32 a PCR-generated DNA fragment containing the fliC 5'UTR and 60 nucleotides from the 5' end of the fliC gene (p20FliCGfp5'3' [ 21 ] as template, primers 5'UTRfw Bgl II and peb1/pLA25RIN) was fused to a PCR-generated peb1 fragment (pPeb15'3' as template, primers peb1/pLA25FIN and Peb1R Xba I). This fusion fragment was cloned into pMCS3'UTR using appropriate restriction sites ( BglI I and Xba I) and the ligated plasmid pLA32 was introduced into E. coli MKS12. For pTAR constructs, see Table 1 . Precipitation of protein in growth media To analyze the presence of Peb1 and mutant proteins in the growth medium, we precipitated culture supernatants with TCA as described previously [ 21 ]: 1,5 ml of culture in N-minimal medium was centrifuged in a tabletop centrifuge (Biofuge 13, Heraeus Instruments) at 13 000 rpm for 15 min at 4°C. The supernatant was recentrifuged, and 1ml of the secondary supernatant was precipitated with 10% TCA on ice for 30 min. Precipitates were pelleted (as above) and washed twice with ice cold ethanol/ether (1:1). Precipitated protein was allowed to dry before resuspension into loading dye and boiled for 5 min. An amount corresponding to 1 ml original cultivation when analyzed by SDS-PAGE and to 100 μl cultivation when analyzed by western blotting, respectively, was loaded onto the gel. Cell samples were generated by pelleting the cultivation followed by a wash with PBS. Cells were then resuspended into appropriate volume of loading dye for the analysis of the amount corresponding to bacterial cells from 100 μl culture, and boiled. Isolation of periplasmic contents To assess the presence of Peb1 and mutant proteins in the periplasmic space, contents of the periplasm were isolated. Bacteria were grown overnight in Luria broth and diluted in N-minimal medium (1:30) for further cultivation for 20 hrs without agitation at 37°C. Cells were then pelleted using a F21-rotor (FiberLite ® , Piramoon Technologies, Inc.) in a Sorvall RC-5B centrifuge (Du Pont Instruments) at 10 000 g (15 min, 4°C). Cells were resuspended in 4 ml 20% glucose/g wet cells and incubated 5 minutes at RT. Next, the sample was centrifuged for 5 min at 13.000 rpm (RT) in tabletop centrifuge (PICO 17, Thermo Scientific). The pellet was further resuspended in 1 ml ice cold 10 mM Tris/20 Mm EDTA (pH 7.5)/g cells, incubated on ice 10 min and centrifuged additionally for 5 min (13 000 rpm, 4°C). The remaining supernatant represented the periplasmic content. All SDS-PAGE and Western blot analyses contained periplasmic contents corresponding to 100 μl original culture as to be comparable with cell and TCA-precipitation samples. Intracellular solubility assay To separate the soluble and insoluble protein fractions of the bacterial cytoplasm, we performed fractionation by ultracentrifugation. The expression sample (100 μl) was removed after 18 h cultivation and centrifuged for 10 min at 10 000 × g. The pellet was resuspended in SDS-PAGE sample buffer and boiled for 5 min prior to loading onto gel. The remaining portion of the overnight culture was centrifuged for 10 min 3500 rpm in a tabletop centrifuge and washed once with 10 ml PBS. The pellet was resuspended in 5 ml PBS and sonicated (in a Vibracell apparatus, Sonics & Materials Inc., tip model CV26) 10 × 10 sec at 20% amplitude to lyse the cells. The cytoplasmic content was cleared by centrifugation in a tabletop centrifuge (Biofuge 13, Heraeus Instruments) for 15 min at 6000 rpm (4°C). The supernatants were subsequently fractionated by ultracentrifugation for 1 h at 4°C with a Ti 50 rotor for 100 000 × g (Beckman L-60 Ultracentrifuge). The supernatant consisting of the soluble fraction of the cytosolic content was removed and the pellet, representing the insoluble fraction, was resuspended in PBS. Samples for SDS-PAGE analysis, corresponding to 100 μl cultivation, were prepared by adding SDS-PAGE sample buffer and boiling for 5 min before loading onto gel. In Western blot analysis of soluble and insoluble fractions the samples corresponded to 30 μl cultivation. Western blotting and SDS-PAGE analysis Western blotting and SDS-PAGE analyses were carried out essentially as described using 15% SDS-polyacrylamide gels [ 21 ]. In Western blotting, nitrocellulose membrane with pore size 0.2 μm or 0.45 μm (Protran Whatman ® , Schleicher&Schuell), was used. Blocking with 2% BSA/PBS was done for 18 h at RT, and detection employed appropriate antibodies. Generation of the insertion mutant library Insertion mutagenesis of the DNA fragment fliC 5'UTR- peb1 was accomplished using the Mutation Generation System (Finnzymes) that employs MuA transposase to insert an artificial transposon randomly into a target plasmid [ 59 ]. Upon transposon elimination by restriction digestion and plasmid recirculation by ligation, the system yields fifteen base pair insertions in target plasmids [ 60 , 61 ]. The in vitro transposition reaction was essentially done according to instructions supplied by the manufacturer. The reaction contained 100 ng M1-CamR transposon DNA as a donor, 480 ng plasmid pLA26 as a target, 0.2 μg MuA transposase, 25 mM Tris-Hcl, pH 8.0, 100μg/ml BSA, 15% (w/v) glycerol, 100 mM NaCl, 0, 05% Triton X-100 and 10 mM MgCl 2 in a total volume of 25 μl. The reaction was conducted at 30°C for 3 h. The DNA was purified, precipitated, and transformed into E. coli DH10B cells by electroporation. The plasmid DNA was isolated from a pool of 90 000 colonies using the Plasmid Maxi Kit (Qiagen). Subsequently, the fliC 5'UTR- peb1 target fragment was removed by Bgl II- Xba I digestion and ligated with equally digested pMCS3'UTR. The body of the transposon was removed by Not I digestion after which the plasmids were recirculated by ligation. This pool of mutated plasmids represented the insertion mutant library. Screening of the mutant library For screening purposes, plasmids of the insertion mutant library were introduced into E. coli MKS12 cells by electroporation and transformants were randomly picked for screening from selection plates. Primary screening of the mutant library was based on dot blotting of the cleared supernatant from cultures of library clones. In total 2037 transformants were cultured without agitation on 96-well microtiter plates in 200 μl LB supplemented with tetracycline at 12.5 μg/ml for 16 h at 37°C and 5 μl overnight culture was subsequently transferred into 200 μl N-minimal medium supplemented with 0.2% casamino acids and tetracycline (12.5 μg/ml) and cultured statically for 20 hours. The cells were pelleted by centrifugation of the microtiter plate for 15 minutes at approximately 2000 × g (Heraeus Megafuge 1.0, rotor model 2705, microtiter carrier model 2708) and the remaining supernatants were cleared by centrifuging 150 μl supernatant for an additional 15 min at 2000 × g. Next, 100 μl cleared supernatant was dot blotted onto nitrocellulose membrane (pore size 0.45 μm; Protran Whatman ® , Schleicher&Schuell) using the Bio-Dot ® Microfiltration Apparatus (Bio-Rad Laboratories) according to manufacturers' instructions. The membrane was subsequently incubated in 2% BSA/PBS for 18 h at 20°C and thereafter the presence of Peb1 protein was detected with anti-Peb1 antibodies and alkaline phosphatase -conjugated secondary antibodies. In order to verify that Peb1 was expressed intracellularly, the cell pellets from the microtiter plates were washed with 100 μl PBS, repelleted (10 min for 1000 rpm, approx. 170 × g) and lysed by resuspension in SDS-PAGE sample buffer (30 μl) and boiling for 5 min. Samples equivalent to 30 μl of bacterial culture were analysed by SDS-PAGE. Secondary screening of clones expressing Peb1 intracellularly at the same level as wt Peb1 but showing decreased Peb1 export was performed by precipitation of cleared supernatants with 10% TCA and Western blotting with anti-Peb1 antibodies as described elsewhere [ 21 ]. Localization of the insertion site in the mutated peb1 genes was done by colony PCR using either forward (pBRAhdIF) or reverse (pLA21R) primer together with the Not I-miniprimer (Finnzymes, Table 3 .) that recognizes the 15 bp insert originating from the transposition event. Ultimately the insertions' location was specified by sequencing (Institute of Biotechnology, University of Helsinki, Finland) using the appropriate primers (pBRAhdIF or 028R). Bioinformatic analysis The amino acid sequence of mature Peb1 was analyzed for the presence of a secretion signal using the softwares SignalP [ 26 ], TatP [ 27 ], SecretomeP [ 30 ], LipoP [ 29 ] and Type-III effector prediction [ 28 ]. The protein structures were visualized with the program VMD [ 62 ]. The sequences of bacterial periplasmic amino acid binding proteins were aligned by ClustalX 2.0.5 [ 63 ] and the solvent exposed surface area per residue was calculated with the program GetArea [ 64 ]. Accession numbers The accession numbers [ 65 ] of the cited atomic coordinates are Peb1 [PDB: 2v25 ], glutamine binding protein GlnBP from E. coli [PDB: 1ggg ], histidine binding protein HBP from E. coli [PDB: 1hsl ], and glutamate/aspartate binding protein Debp from S. flexnerii [PDB: 2vha ]. The cited amino acid sequence is Peb1 [GenBank: AAA02919.1 ]; the genome sequences are from E. coli MG1655 [GenBank: U00096 ], the parental strain of MKS12, and E. coli DH1 [GenBank: CP001637.1 ] [ 57 ].

Results Peb1 is transported from E. coli MKS12 also independently of the flagellar secretion system In a previous report, we showed that the E. coli strain MKS12 (Δ fliC Δ fliD ), lacking the flagellar filament, is able to secrete recombinant flagellin (FliC), FliC fragments, FliC fusion proteins as well as heterologous proteins, like the model Peb1 protein, into the growth medium [ 21 ]. We also demonstrated that the extracellular location of these recombinant polypeptides was not a result of leakage from the periplasm or cytoplasm or lysis of the cells by showing the lack of the periplasmic maltose binding protein as well as the cytoplasmic proteins chloramphenicol acetyl transferase and GroEL in the growth medium. To generate a more detailed picture of the Peb1 transport event in E. coli , we here assessed the role of the flagellar export gate on Peb1 secretion. To construct a host strain lacking the flagellar export gate, the following modifications in strain MKS12 were made. We constructed the E. coli strains MKS111 and MKS111ΔHBB (see Table 1 for strains used), which are isogenic with strain MKS12; MKS111 carries additionally a deletion in the chromosomal gene encoding the transcriptional regulator FlgM. The flgM deletion releases the flagellar-specific sigma factor FliA, which is required for transcription from the P fliC promotor used in this study [ 23 ]. In strain MKS111ΔHBB, genes encoding flagellar basal body structures, i.e. the rings, the rod, the hook as well as components of the secretion apparatus, have additionally been deleted from the chromosome. Consequently, FliA-mediated transcription can proceed but flagellar filament components are not secreted from the cytoplasm across the membranes in strain MKS111ΔHBB. The Δ fliC gene of the strains MKS12, MKS111 and MKS111ΔHBB was complemented in trans with plasmids pLA25 and pLA31 (see Table 1 for plasmids used), which encode Peb1 without the sequences encoding its native N-terminal leader peptide and putative processing site for signal peptidase II, and the FliC1-183 peptide, respectively. Western blot analyses with anti-Peb1 and anti-flagella antibodies showed that Peb1 was present in cell-free culture supernatant of all the strains tested, whereas the FliC1-183 peptide, expressed similarly and used as a flagellum-secreted control [ 21 ], remained intracellular in MKS111ΔHBB (Figure 1A ). The results showed that Peb1 is secreted to the extracellular milieu also in the strain lacking the flagellar export gate, whereas the C-terminally truncated flagellin and the full-length FliC (not shown) remain intracellular in the same strain. This finding indicated that the secretion of Peb1 in E. coli also can proceed independently of the flagellum. Peb1 is present in the periplasmic space of E. coli The finding that Peb1 is present in the growth medium also in a mutant lacking the flagellar transport complex prompted us to further analyze the phenomenon. To compare the localization of Peb1 expressed from a multicopy plasmid and from a single-copy gene from the chromosome, we first constructed the strain MKS12P lac peb1 , which carries peb1 without signal sequences under a constitutive lac promoter (P lac ) that lacks the lac operator region of P lac in the fliC deletion site on the chromosome of MKS12. We then prepared periplasmic fractions and cell samples of the E. coli strains MKS12 (pLA25) and MKS12P lac peb1 , both expressing Peb1 without its native N-terminal signal sequences. SDS-PAGE and Western blot analysis with anti-Peb1 antibodies showed that the cellular concentration of Peb1 was approximately 50% lower in MKS12P lac peb1 than in MKS12 (pLA25), but the protein was clearly present in the periplasm in both strains. We estimated on the basis of SDS-PAGE analysis that the percentage of intracellular Peb1 found in the periplasm was 50% in MKS12 (pLA25) and 25% in MKS12P lac peb1 . Peb1 was not detected in the control strains MKS12 and MKS12 (pMCS3'UTR) (Figure 1B ). Western blot analysis with antibodies against the intracellular RNA polymerase β subunit (RNAP) indicated that the periplasmic fractionation was successful and that the presence of Peb1 in the periplasmic samples was not due to cytoplasmic contamination (bottom panel in Figure 1B ). The results thus indicated that mature Peb1 is exported to the periplasm in E. coli and that the localization of Peb1 in the periplasm is apparently not related to leakage of significantly overexpressed protein. To study the impact of the SecA-YEG pathway in Peb1 export, we analyzed periplasmic and cell samples of MKS12 (pLA25) grown in the presence and absence of subinhibitory concentration of azide, which is known to arrest SecA-YEG dependent export across the CM [ 24 ]. Results from SDS-PAGE analysis and Western blotting with anti-Peb1 antibodies revealed that the presence of Peb1 in the cytoplasm and the periplasm was not affected by growth in the presence of 3 mM NaN 3 (data not shown). The data thus indicate that the SecA-YEG pathway may not be involved in the export of Peb1 across the CM. Fusion of a FliC fragment to the N-terminus of Peb1 does not affect secretion We next assessed whether the N-terminus of Peb1 is involved in the secretion process, as has been reported for flagellar proteins and effector proteins in type 3 secretion systems [ 22 , 25 ]. The plasmid pLA32 encoding the fusion protein FliC 1-20 Peb1 was constructed and the protein expressed in the E. coli strains MKS12 and MKS111ΔHBB. SDS-PAGE and Western blot analysis with anti-Peb1 antibodies showed that the intracellular expression and extracellular secretion of FliC 1-20 Peb1 was at the same level as those of Peb1. FliC 1-20 Peb1 and Peb1 were also present in almost equal quantities in the periplasm of all the strains studied (Figure 1C ). The periplasmic location of Peb1 and FliC 1-20 Peb1 was not due to cytoplasmic contamination as indicated by the Western blot analysis with antibodies against the intracellular protein RNAP (Figure 1C ). The results thus revealed that fusion of a heterologous protein fragment to the processed N-terminus of Peb1 does not interfere with Peb1 secretion in MKS12 and MKS111ΔHBB, and that this fusion protein is secreted equally efficiently in both strains. Exhaustive search for transport-affecting regions in Peb1 To identify regions in Peb1 putatively important for its transport, we constructed an exhaustive library of 5 amino acid insertion mutants. The target for the mutagenesis was the DNA fragment fliC 5'UTR- peb1 , in which peb1 is expressed from the P fliC promoter without the sequences encoding the N-terminal leader peptide and the putative processing site for signal peptidase II of Peb1. The coverage of the library was determined by DNA sequencing of the fliC 5'UTR- peb1 fragment in all the 2037 library clones. The sequence of peb1 was successfully determined in 1827 clones and 1518 of these carried the pentapeptide inserts in the open reading frame of peb1 . We found unique inserts at 178 sites in total along the entire amino acid sequence of Peb1, which is 233 amino acid residues in length. The longest region lacking mutations was only four amino acid residues in length (Figure 2A ). To identify secretion-deficient insertion mutants, we screened the 2037 library clones for Peb1 secretion by dot blotting of cell-free culture supernatants with anti-Peb1 antibodies. In the majority of Peb1-expressing mutants analyzed, the pentapeptide insertion in the amino acid sequence of Peb1 had no significant effect on secretion of Peb1 into the growth medium. Interestingly, we found 14 clones that showed impaired extracellular export of Peb1. When we analyzed the deduced amino acid sequence of these mutants, we noted that the pentapeptide insertions occurred at two confined regions in the Peb1 sequence: D42-V43 and S173-L180. The two regions were named Transport Affecting Regions 1 and 2 (TAR1 and TAR2; see Figure 2A ). Six of the transport-affecting insertions were localized to the TAR1 region and eight to the TAR2 region. We verified by SDS-PAGE and Western blot analysis using anti-Peb1 antibodies that the absence of the TAR mutants in the extracellular milieu was not due to impaired Peb1 expression or degradation of the protein. Results from the analyses of cells and cell-free culture supernatants of mutants representative of the entire data set are shown in Figure 2B . According to our analysis, the TAR1 mutant proteins were not degraded but expressed intracellularly at lower quantities than wt Peb1 (see top panel of Figure 2B ), which compromised the judgement of the effect of these mutations on Peb1 secretion. The intracellular expression of TAR2 mutant proteins was equal to that of wt Peb1. Cell lysis was analyzed using antibodies against intracellular marker proteins as described previously [ 21 ] and was judged to be minimal (bottom panels of Figure 2B ). The finding that the well-expressed TAR2 mutants were not found in the growth medium was additionally considered as an internal control for lack of leakage from or lysis of the cells. When we compared the amino acid sequences of the inserted pentapeptides in TAR1, TAR2 and neutral mutants, we found that the neutral mutants carried insertions with similar residues as the secretion-affected mutants, but in locations outside the identified TAR regions. We deduced that the secretion-deficient phenotype of TAR1 and TAR2 mutants was not dependent on the type of sequence inserted and that the lack of extracellular localization by the TAR mutants was apparently due to the site of insertion (See Table 2 for examples). Taken together, the exhaustive search for transport-affecting domains in Peb1 revealed two insertion-sensitive regions in the open reading frame of Peb1, TAR1 and TAR2, which affect its localization in the growth medium of E. coli MKS12. Effect of targeted mutations in TAR regions on Peb1 secretion The results obtained from the mutant library of Peb1 indicated the two TAR areas in the protein as critical for its extracellular localization. To analyze these regions of Peb1 in more detail, we substituted all the charged and polar residues in the vicinity of and within the TAR1 region (Glu40, Asp42, Lys45 and Lys49) (encoded by plasmid pTAR1P) and in the TAR2 region (Ser173, Asp175, Lys176 and Ser177 (encoded by plasmid pTAR2P) with alanine (Figure 3 ). In addition, we generated an insertion of five Gly between Asp42 and Val43 in the TAR1 region (encoded by pTAR15G) (Figure 3 ). The Ala point mutations in TAR2 did not significantly influence the expression and secretion of the mutated Peb1 proteins, whereas in TAR1 the cellular concentration and, as a consequence, secretion levels of the Ala point mutants were slightly decreased. The cellular concentration of pentaglycine insertion mutant was also diminished and its transport was abolished (Figure 3 ). We next assessed the impact of the TAR1 and TAR2 regions on Peb1 secretion by creating long Ala substitutions (Glu40 to Leu46 and Val174 to Tyr182) and for comparison, identical Gly substitutions were constructed (Figure 3 ). The mutant Peb1 proteins with nine residues long Ala or Gly substitutions in region TAR2 (encoded by pTAR2A and pTAR2G) were synthesized at the level of wt Peb1 but were not secreted, whereas the Peb1 mutants carrying seven-residue long Ala or Gly substitutions in TAR1 (pTAR1A and pTAR1G) showed a lower amount of Peb1 in cells compared to the wild type strain. The mutant with a long Gly substitution was not secreted and the mutant carrying a long Ala substitution was present in barely detectable amounts in the growth medium (Figure 3 ). The results thus showed that selected Ala point substitutions in TAR1 and TAR2 regions do not affect Peb1 secretion significantly, whereas longer Ala substitutions in TAR1 and especially in TAR2 markedly impair Peb1 secretion. Also long Gly substitutions in the TAR regions and the insertion of five Gly in TAR1 affected transport of Peb1. In addition, the results showed that all manipulations performed in the TAR1 region, i.e. insertions, selected point substitutions as well as long substitutions, affected the amount of intracellular Peb1 whereas manipulations of the TAR2 region did not. Cytosolic Peb1 proteins are soluble and not subjected to aggregation To assess whether intracellular aggregation of mutant proteins was the reason for the lack of their presence in the growth medium, the cytoplasmic content of E. coli MKS12 expressing Peb1 and its mutants was fractionated by ultracentrifugation, separated by SDS-PAGE, and analyzed by Western blotting using anti-Peb1 antibodies (Figure 4 ). The experiment included wt Peb1, Peb1 with Ala- and Gly-substituted TAR regions, Peb1 carrying a pentaglycine insertion in TAR1, and one of the Peb1 mutants from the transposon library with the pentapeptide insertion in the TAR1 region (encoded by plasmid p9D5; see Table 2 ). The results showed that all the proteins that were expressed well in the cells were also found in the soluble fraction. The Peb1 mutant with a long Gly substitution in TAR1 (encoded by pTAR1G) was expressed very poorly and was therefore detected neither in the soluble nor the insoluble fraction, a fact hampering the analysis of its solubility. Conclusively, the results indicated that the lack of extracellular localization by Peb1 mutants in the analyses was not due to aggregation or proteolysis of the proteins in the cytoplasm. Subcellular localization of Peb1 mutants To analyze the localization of Peb1 mutants, we prepared periplasmic fractions and cell samples of selected mutants which expressed Peb1 well but showed decreased Peb1 secretion. SDS-PAGE and Western blot analysis with anti-Peb1 antibodies showed that mutant Peb1 proteins with pentapeptide insertions or long Ala substitutions in the TAR1 and TAR2 regions were present in the periplasm at quantities almost equal to those of wt Peb1 (Figure 5 ). Western blot analysis with antibodies against the intracellular protein RNAP indicated that the presence of Peb1 and its mutants in the periplasmic samples was not due to cytoplasmic contamination (bottom panel in Figure 5 ). The results thus revealed that Peb1 mutants, which are well expressed in the cytoplasm but not secreted, are also transported almost as efficiently as wt Peb1 into the periplasmic space. Bioinformatics analysis of Peb1 and TAR regions We performed sequence analysis of Peb1 in order to identify sequences important for its transport. The software programs SignalP, TatP, LipoP and Type-III effector prediction were used for prediction of putative signal sequences in the amino acid sequence of mature Peb1 (shown in Figure 3 ) and thus lacking the SecA-YEG -dependent signal sequence and the putative signal peptidase II recognition site [ 26 - 29 ]. We found no indication of the presence of a signal sequence related to SecA-YEG, Tat, lipoprotein or type 3 secretion system pathways in the mature Peb1. Interestingly, the SecretomeP [ 30 ] software predicted Peb1 to be a non-classically secreted protein. Since the TAR regions were found to have a major impact on Peb1 secretion efficiency, we analyzed the sequence and the structure of Peb1 with emphasis on the TAR regions in order to identify possible structural aspects important for secretion. Mature Peb1 is a globular protein of 233 amino acids that folds into two domains both comprising of a five-stranded beta sheet surrounded by alpha helices, and it binds two Zn 2+ cations (Figure 6A ). The TAR1 and TAR2 regions coincided with the first helix of the N-terminal core structure (Nα1) (Figure 6A, B ) and with the end of the third C-terminal beta strand and the following helix (Cβ3-Cα3) (Figure 6A, C ), respectively. The structural analysis here considered only the Ala substitutions since the pentapeptide insertions generated by the transposon mutagenesis are challenging to analyze in structural terms and interpretation of the effects of the Gly substitution in TAR1 was not reliable due to the decreased Peb1 expression (See Figure 3 ). The helix Nα1, that covers the TAR1 region, is a 14-residue long, straight and strongly amphiphilic helix that interacts via Ser50 with one zinc cation but is not part of the ligand-binding site of Peb1. The polar residues of Nα1 form strong interactions with the two terminal helices of Peb1, Zα1 and Zα2, and the large apolar residues pack tightly to Zα2. The helix structure is stabilized by several hydrogen bonds (see Figure 6B ). By the number of polar and apolar helix-helix interactions, the interface between Nα1 and Zα1 and Zα2 is the strongest in the whole protein. Substitution of all polar residues in and close to TAR1 by alanines (pTAR1P) did not significantly interfere with Peb1 transport, whereas a seven-residue run of alanines covering TAR1 (pTAR1A) did so, analogously to the results from the six amino acid transposon insertions (as shown in Figure 2 and 3 ). The TAR2 region of Peb1 is located in the very center of the protein covering the end of Cβ3 and the beginning of Cα3 (see Figure 6A and 6C ). It forms the ligand-binding pocket (residues Val174, Asp175, Ile178) along with numerous other residues from both domains. The results from Ala substitutions in TAR2 were similar to those in TAR1: the substitution by nine consecutive alanines compromised secretion, while changes of only the polar residues did not (see Figure 3 ). Sequence comparison between the secretion-deficient Peb1 constructs and those exhibiting a wt secretion level pointed towards an importance of the bulky hydrophobic residues Val174, Ile178, Leu179 and Leu180 for Peb1 secretion. Except for Leu180, these are fully buried residues, with solvent exposed surface areas smaller than 3% of their total area. Taken together, the structural analysis of TAR regions showed that the TAR1 region comprises a densely packed and highly stable part of Peb1 and the TAR2 region is located in the center of protein forming the ligand-binding pocket along with several other residues. We suggest that the unifying feature of the TARs is their crucial role for proper folding of Peb1, which apparently is important for the export across the OM.

Discussion When the Peb1 protein of C. jejuni is expressed without its native N-terminal signal sequence and without the putative signal peptidase II cleavage site in E. coli MKS12 (Δ fliCfliD ), it is still efficiently secreted into the growth medium. The extracellular presence of Peb1 in MKS12 is not due to leakage from or lysis of the cells, as we have demonstrated previously [ 21 ]. Here, our aims were to analyze the molecular mechanisms behind the secretion competency of the heterologous model protein Peb1 in the E. coli host in order to broaden our understanding of factors that may affect secretion of proteins in E. coli . A detailed knowledge of mechanisms of extracellular transport of heterologous proteins in E. coli is of great importance in the development of secretion systems for biotechnological applications [ 4 ]. The present work shows that secretion of mature Peb1 into the growth medium of E. coli is facilitated also in the absence of the flagellar machinery as shown in the MKS12 derivative MKS111ΔHBB, a strain which completely lacks the flagellar secretion complex. The hypothesis of a flagellar-independent secretion mechanism was further supported by the fact that fusion of an N-terminal FliC fragment to the N-terminus of Peb1 did not interfere with its secretion. Our finding that Peb1 is also present in the periplasm strongly indicates that the protein is secreted by a mechanism that involves at least two export steps, one across the CM to the periplasm and another across the OM. This excludes secretion dependent only on the flagellar machinery, which is known to proceed from the cytoplasm to the extracellular milieu without involvement of a periplasmic step [ 22 ]. A comprehensive pentapeptide scanning mutagenesis of Peb1 was performed to decipher transport affecting regions in the molecule itself, and the results were confirmed by site-specific mutations. Importantly, modifications of only two distinct areas, named TAR1 and TAR2, in Peb1 affected secretion. This phenomenon was not due to insolubility or aggregation of the mutant proteins in the bacterial cytoplasm as shown by cell fractionation analysis. However, manipulations in the TAR1 region affected the intracellular amount of Peb1, whereas no such effect was observed after mutagenesis of the TAR2 region. We anticipate that the reduced amount of Peb1 seen in the cells of TAR1 mutants is due to decreased expression levels. As no degradation products of Peb1 were detected, we assume that significant proteolysis of Peb1 mutants did not occur. The results showed that also well-expressed, secretion-deficient Peb1 mutants were present in the periplasmic space. This finding strongly indicated that the export of the TAR1 and TAR2 mutants across the CM was unencumbered, whereas their further export across the OM was compromised and led to the observed lack of Peb1 mutants in the culture supernatant. The secretion-deficiency of TAR mutants was studied also by analyzing the Peb1 structure. The effect of TAR1 mutations could lie in the importance of the hydrophobic surface packing toward the rest of the protein. Native-like amphiphilicity of Nα1 may be needed for the folding of Peb1, and inability to fold may be the basic cause for secretion defects. The importance of TAR2 in secretion is apparently related to the stability of the protein, as TAR2 is located in the center of Peb1 and forms the ligand-binding pocket along with several other residues. We suggest that the unifying feature of the TARs is their crucial role in proper folding. We speculate that Peb1 remains unfolded or is only partially folded during its export across the CM to the periplasm since mutagenesis of TAR regions did not affect this step. Correct or at least partially correct folding of Peb1 may, on the contrary, be important for its export across the OM into the growth medium. Searches in the literature and publically available databases revealed that the host strain MKS12 used in this work, a Δ fliCfliD -derivative of E. coli MG1655, expresses at the culture conditions used only three characterized protein transport systems [ 31 - 34 ]. These are the SecA-YEG and Tat systems, which export proteins with cognate cleavable N-terminal signal sequences across the CM [ 35 ], and the flagellum-specific secretion system responsible for assembly of the flagellar filament [ 36 ]. As indicated by the bioinformatics analysis of the Peb1 amino acid sequence, none of these are apparently involved in E. coli -based Peb1 secretion. The SecA-YEG -independent secretion was here experimentally studied by analysis of periplasmic and cell samples of Peb1-expressing E. coli grown with or without subinhibitory concentrations of azide, which is known to inhibit protein export by the pathway [ 24 ]. Other well-characterized pathways dedicated for protein secretion in Gram-negative bacteria are normally not expressed, or the genes are fragmented, deleted or totally lacking in MKS12 as described below. The type 1 secretion system of E. coli , which is responsible for secretion of the toxin hemolysin in pathogenic E. coli , is not found in MKS12. The strain carries the genes encoding a hemolysin-like protein named HlyE/SheA/ClyA secreted into the medium by a so far unidentified mechanism in H-NS mutants [ 37 ], which MKS12 is not. The genes encoding a putative type 2 secretion system are present in MKS12, but the transcription of the two operons encoding the majority of the type 2 secretion system components is repressed by the H-NS protein expressed in MKS12 [ 38 ]. The genes encoding a pathogenesis-related type 3 secretion system are not present in MKS12 [ 32 , 39 ]. The alternative cryptic type 3 secretion system called ETT2 present in some pathogenic and commensal E. coli strains as well as the ancestral gene cluster encoding the alternative flagellar system Flag-2 in 20% of E. coli strains are both nonfunctional in MKS12 [ 39 , 40 ]. The strain possesses at least 16 chromosomal genes related to the type 4 secretion system, but the genes are not expressed in the culture conditions used in the study [ 41 ]. The type 5 secretion system is represented in E. coli K-12 strains only by the SecA-YEG -dependent autotransporter protein called antigen 43, which facilitates its own transport to the surface of the cell [ 42 ]. Finally, no orthologs of the type 6 secretion system proteins are present [ 43 ], the type 1 fimbria genes have been site-specifically deleted in MKS12 [ 44 ], and the curli-encoding operons are not expressed in MKS12 at the growth conditions used [ 45 ]. ABC transporters are CM-spanning proteins that energize the transport of their substrates by binding and hydrolysis of ATP. In bacteria, the ABC exporters facilitate the export of various compounds from the cytoplasm and the more common ABC importers import essential nutrients that are delivered by specific PBPs [ 15 ]. In E. coli , almost 80 ABC importer and exporter systems have been reported. These frequently transport only small molecules [ 13 ] and are probably not responsible for the efficient secretion of Peb1, which is a full-length globular protein of 233 amino acids. The transport mechanism of the aspartate/glutamate -binding PBP Peb1 appears to differ markedly in E. coli from that of its structurally known homologs HBP and GlnBP from E. coli [ 46 , 47 ] and Debp from Shigella flexneri [ 48 ]. These ABC-transporter associated PBPs are all exported by the SecA-YEG pathway involving a cleavable N-terminal signal sequence. The sequence identity between Peb1, HBP, GlnBP and Debp is low (27% and 22%) and at the structural level, root mean square deviation values calculated between mature Peb1, HBP and Debp are only 2.2 Å since the orientation of the domains relative to each other varies between structures. Hence, the overall structural similarity of Peb1 with E. coli PBPs is probably not the reason for its efficient secretion in the heterologous E. coli host. Our results show that in E. coli , mature Peb1 is present in the extracellular milieu but cannot be visualized on the surface of strain MKS12 (pLA25) with indirect immunofluorecence microscopy using anti-Peb1 antibodies or by electron microscopy (unpublished data) indicating that blebbing of OM vesicles [ 49 ] may not be involved in the secretion. Our results also show that Peb1 can be exported to the periplasm independently of the flagellar machinery, without a SecA-YEG-dependent N-terminal signal sequence independently of the Sec pathway, and without a signal peptidase II recognition site. Taken together, we conclude that Peb1 may be secreted by a still unknown secretion route, a conclusion supported by our result from software analysis of the Peb1 sequence indicating that the protein is secreted by a non-classical pathway. Our findings are indirectly supported by a line of independent reports. According to Hu and colleagues [ 31 ] one third of the 4225 protein-encoding genes of E. coli K-12 are functionally unannotated, a fact which makes it possible to reveal novel mechanisms even in this well-studied organism. Proteins denoted as non-classically secreted are found extracellularly in various bacteria despite the absence of a defined secretion signal or motif [ 50 , 51 ]. Frequently, these proteins called moonlighting proteins have novel extracellular functions in addition to their well-known roles in the cytoplasm [ 52 ]. Comparison of the extracellular proteomes of E. coli K-12 and BL21 revealed a large number of proteins, both cytoplasmic and periplasmic, to be released by unknown mechanisms into the growth medium when cells were grown at high cell density [ 53 ]. We assume that Peb1 is secreted in E. coli by a still uncharacterized two-step pathway and that the structurally significant TAR- regions affect the secretion. We are currently exploring this possibility further.

Conclusions We have previously reported that the diagnostically interesting model protein Peb1 is efficiently secreted, without the sequences for SecA-YEG dependent secretion and signal peptidase II, into the growth medium of the secretion-competent strain E. coli MKS12 (Δ fliCfliD ). Here, we analyzed the secretion mechanism further and noticed that Peb1 was also transported independently of the flagellar secretion machinery. We found, by construction and analysis of an exhaustive library of Peb1 insertion mutants, that mutagenesis of two segments called TAR1 (residues 42 and 43) and TAR2 (residues 173 to 180) prevented Peb1 secretion individually. Site specific mutagenesis of the TAR regions confirmed the importance of these regions for Peb1 secretion, and we verified that the secretion deficiency of Peb1 mutants was not due to insolubility or aggregation of the proteins in the cytoplasm. The mutant proteins were present in the periplasm as well as in the cells of MKS12 and we deduced that mutagenesis of TAR regions did not affect export of Peb1 across the CM, whereas its export over the OM was markedly impaired. We conclude that the localization of the mature model protein Peb1 in the growth medium of E. coli is due to an active secretion process, which involves two export steps, one across the CM into the periplasmic space and another across the OM. The export across the OM apparently requires proper folding of the TAR regions since mutagenesis of the regions affected Peb1 export across the OM and thereby presence of the mutant proteins in the extracellular milieu. Bioinformatics analysis of the Peb1 sequence indicated that the protein is transported by an unknown, non-classical pathway, which is a topic for further studies.

Background Escherichia coli is frequently the first-choice host organism in expression of heterologous recombinant proteins in basic research as well as in production of commercial, therapeutic polypeptides. Especially the secretion of proteins into the culture medium of E. coli is advantageous compared to intracellular production due to the ease in recovery of the recombinant protein. Since E. coli naturally is a poor secretor of proteins, a few strategies for optimization of extracellular secretion have been described. We have previously reported efficient secretion of the diagnostically interesting model protein Peb1 of Campylobacter jejuni into the growth medium of Escherichia coli strain MKS12 (Δ fliCfliD ). To generate a more detailed understanding of the molecular mechanisms behind this interesting heterologous secretion system with biotechnological implications, we here analyzed further the transport of Peb1 in the E. coli host. Results When mature Peb1 was expressed without its SecA-YEG -dependent signal sequence and without the putative signal peptidase II recognition sequence in E. coli MKS111ΔHBB lacking the flagellar secretion complex, the protein was found in the periplasm and growth medium which indicated a flagellum-independent translocation. We assessed the Peb1 secretion proficiency by an exhaustive search for transport-affecting regions using a transposition-based scanning mutagenesis strategy. Strikingly, insertion mutagenesis of only two segments, called TAR1 (residues 42 and 43) and TAR2 (residues 173 to 180), prevented Peb1 secretion individually. We confirmed the importance of TAR regions by subsequent site-specific mutagenesis and verified that the secretion deficiency of Peb1 mutants was not due to insolubility or aggregation of the proteins in the cytoplasm. We found by cell fractionation that the mutant proteins were present in the periplasm as well as in the cytoplasm of MKS12. Hence, mutagenesis of TAR regions did not affect export of Peb1 across the cytoplasmic membrane, whereas its export over the outer membrane was markedly impaired. Conclusions We propose that the localization of the model protein Peb1 in the growth medium of E. coli is due to active secretion by a still unknown pathway of E. coli . The secretion apparently is a two-step process involving a periplasmic step and the TAR regions.

List of abbreviations used (3'UTR): 3' untranslated region; (5'UTR): 5' untranslated region; (ABC): ATP-binding cassette; (CM): cytoplasmic membrane; (P fliC ): fliC promotor; (OM): outer membrane; (PBP): periplasmic binding protein; (PBS): phosphate-buffered saline, pH 7,1; (BSA/PBS): bovine serum albumin in PBS; (RNAP): RNA polymerase β subunit; (SDS-PAGE): sodium dodecylsulphate polyacrylamide gel electrophoresis; (TCA): trichloro acetic acid; (wt): wild-type. Competing interests The authors declare that they have no competing interests. Authors' contributions LA carried out the generation and screening of the insertion mutant library, the precipitation of proteins, the isolation of periplasmic contents, the intracellular solubility assays as well as the SDS-PAGE and Western blot analyses. She also constructed the plasmids, participated in the study design and interpretation of data, and in drafting of the manuscript. KM carried out construction of the E. coli host strains and participated in revising the manuscript critically. LL carried out the structural analysis, participated in the study design and in revising the manuscript critically for important intellectual content. HS participated in the design and generation of the insertion mutant library and participated in revising the manuscript critically. BWW had the main responsibility for the study design, data interpretation and manuscript writing. All authors read and approved the final manuscript.

Acknowledgements Raili Lameranta and Lotta Happonen are acknowledged for skilled technical assistance and expert advice in the technical details of the transposition mutagenesis, respectively. Elina Hienonen is thanked for providing the plasmid pTSF12 and for fruitful discussions, Lars Paulin for excellent assistance in DNA sequencing, and Janne Ravantti for advice in computer related questions, and for computer resources. This work was supported by the Academy of Finland (EraNet Pathogenomics grant number 118982 and the General research grant numbers 110716, 123900 and 80690) and the European Network of Excellence in EuroPathoGenomics EPG (number LSHB-CT-2005-51206).

CC BY

no

2022-01-12 15:21:44

Microb Cell Fact. 2010 Dec 2; 9:97

oa_package/d8/ad/PMC3016274.tar.gz

PMC3016276

21129207

Background The scientific value of multilevel, multiscale, computational, biomedical models will be greatly enhanced by making them broadly available and sufficiently manipulable to address a variety of scientific questions at reasonable costs, regardless of the hardware at the researcher's disposal. The availability of cloud technology opens the door to that eventuality. However, such models are analogous to an entire, specialized, wet laboratory. As with wet laboratories, insuring scientific equivalency of duplicate experimental systems in different laboratories is a necessary precondition for placing confidence in the results of experiments arising from those laboratories. A goal of this project was to test the scientific equivalence of experiments conducted using multilevel, multiscale, In Silico Livers (ISLs) executed on a local cluster with those executed in the Amazon EC2 cloud platform. Equivalence demonstrates that validation results obtained on the local cluster still hold in the cloud. Further scientific work on ISLs executing in the cloud can build upon that work. By executing ISLs in a cloud, we gain more computational power to develop more detailed, larger scale, larger scope ISLs, while outsourcing systems administration and hardware maintenance costs. The ISL illustrated in Figure 1 is not intended to be a finalized model having a fixed structure. It is designed to facilitate exploration of plausible mechanistic explanations for observations related to xenobiotic disposition [ 1 - 4 ]. The validation aspect for these experiments is the outflow profile from a single-pass rat liver perfusion experiment [ 5 ]. However, concrete instances of multiscale validation by adding finer grained aspects have been reported [ 6 , 7 ]. An ISL is a biomimetic analogue designed to help formulate and challenge mechanistic hypotheses about the hepatic disposition of xenobiotics in health and disease. It is an assembly of abstracted components representing aspects of hepatic form, space, and organization interacting with compounds (a simulated compound; when referring to an ISL feature or component that has a wet-lab counterpart, we use small caps). It has its own unique phenotype, which is necessarily simpler than that of the entire set of lobule mechanisms. Because of the stochastic nature of ISL simulations, each in silico experiment-constructed from several Monte Carlo (MC) lobule samples-generates a slightly different outflow profile. The modeling and simulation methods used [ 8 , 9 ], described in Methods, are intended for representing large-scale biological systems in silico. They require the computational ability to represent any layer in a hierarchical biological system well enough to falsify (or not) its outputs against corresponding wet-lab observations, and to extend the representation, at will, to any other phenomena necessary to achieve falsification or validation. An ISL cannot be constrained by the use cases or aspects by which it was originally constructed. It must be arbitrarily extensible. It is usually impractical to expect ISLs and similar models to execute in a sequential computing context, e.g., using a single CPU. That is because the depth and scale of most biological systems leads to models that are computationally and analytically insoluble. Inductively derived, equation-based models are more amenable to execution on sequential machines because of their high degree of abstraction from the particulars of the referent mechanisms. At the other end of the computational spectrum, massively parallel machines are also impractical because the solutions that can be implemented are very tightly engineered to match exacting specifications; hence a massively parallel solution is specialized and limited in the extent to which it can propagate [ 6 , 10 ]. In addition, as is evident in Figure 1 , the behavior of each lobule is tightly coupled to the behavior of all lobule components. Hence, distribution of the lobule sub-components over multiple, parallel processors would likely result in a very high communication-to-computation ratio, making it inefficient for that type of fine-grained parallelism. In contrast, the salient interactions between lobule s, for outflow profile validation data, can be abstracted parsimoniously to lobule input/output: at the portal vein tracts (triads) and the central vein. It is a natural fit to run a single lobule on a single cluster node. For other analogues, we can, for example, envision running single cells on single cluster nodes. Moreover, as demonstrated in [ 4 ], highly specific intra- lobule mechanisms can be studied in a very concrete and particular way on a single cluster node. Cluster type machines meet ISL parallelism needs fairly well and have provided the ISL's infrastructure [ 1 - 4 ]. The combination of the liver's homogeneity at a coarse grain and the heterogeneity that is evident at fine grains makes it a good referent system for ISLs built atop cluster parallelism within a cloud. There are three model- and referent-independent reasons for exploring using a cloud infrastructure for cluster style parallelism: 1) reduced systems administration costs, 2) model expansion, and 3) model sharing. ISL uses provide two additional motivations: 4) ISL experiments need many lobule samples, and 5) micromechanistic details need to be assessed. Although significantly lower cost than biological modeling projects relying on massively parallel machines, those based on local clusters can still present a laboratory with relatively high administrative costs. There are temperature, power, maintenance, and manpower considerations. The resulting constraints limit the extent to which methods suited to cluster computing can be propagated, repeated, and leveraged by others. Local constraints also place practical limits to expanding an analogue's scale and scope. The advent of publicly accessible, on demand, platform level, cloud computing infrastructures, like Amazon's EC2 raises the accessibility of large scale, expandable, biological models so that anyone with a broadband Internet connection can build virtual clusters and execute such models on demand. They also greatly expand the diversity of in silico, biomimetic devices that can be constructed. The EC2 platform is an interface to an executive web service that allows the on-demand instantiation and management of virtual computers running any of a suite of operating systems that Amazon supports. It is distinguishable from other "cloud" technologies in that platform resources are customizable. As an extreme contrast, Google Documents is also a cloud technology. However, it only allows the user to create particular types of documents and host them remotely on pre-configured servers. At the other extreme, EC2 allows one to instantiate entire virtual machines, running Windows or Linux. There is no limitation on the type of software you can run, unlike with Google Documents. The spectrum of cloud technologies is covered in some depth in [ 11 ]. To date, business, collaboration, and data processing dominate the use cases for cloud infrastructures. Although desktop grids have contributed to some scientific domains like molecular dynamics [ 12 , 13 ], recruiting volunteers to allow an application to execute on their computers constrains the researcher in ways that on-demand services do not. The third and possibly more powerful justification for exploring cloud computing is the potential for sharing models more easily. Often, a simulation is dependent on a particular technology stack that is not widely available or easily configured. With a platform level service like the EC2, virtual machine instances containing the entire technology stack can be shared. When researchers want to examine more than just the published information, they can instantiate a grid of virtual machines, design and execute their own experiments, or simply repeat those of the original laboratory. We can anticipate that, as models and their uses grow and the computational platforms upon which they run scale upward, issues of reproducibility will grow. Methodological details of virtual laboratory assembly and in silico experimental protocols will become as important for computational scientists as they are for scientists experimenting in different wet-labs. Because each out-flow profile exhibits high variation, it takes averaging many lobule samples to approach the smooth outflow profile exhibited by liver perfusion experiments. A benefit of having an ever-larger number of lobule samples when needed is that the lobular variety within a liver can be approached: an adult rat liver is comprised of a few hundred lobules. Finally, when using the local eight-node cluster (seven slaves and one master), there is a trade-off between abstraction layer for the fine-grained micromechanisms and MC sample size: assessment of the micromechanistic details of the type described in [ 4 ] requires larger sample sizes-more lobule samples-to clearly identify event trends. ISLs are designed for systematic, iterative revision and easy alteration to improve realism and enhance heuristic value. Following revision, the new ISL must be re-parameterized and re-validated using the same criteria used for its predecessor. Addition of new features must not compromise any previously validated behaviors. Planned ISL uses required adding to the most recent ISLs [ 4 ] a biliary elimination and flow space, an improved metabolism mechanism, and enabling an alternative method of dose input. Those features were not needed by earlier ISLs. We implemented an ISL that included those features (Figure 1 ). Observations made during the initial re-validation experiments indicated that we needed to scale experiments to include larger numbers of MC samples. That situation provided an immediate, convenient impetus to move ISL simulations to a Cloud. Results documenting the benefits of such scaling are presented herein. In order to migrate successfully to a cloud, we must demonstrate experimental indistinguishability (defined in Methods) between results from similar size cloud and local cluster experiments. Results of such experiments are presented for the re-validation experiments for ISLs that included the preceding new features.

Methods Middle-out and synthetic modeling and simulation methods ISLs use middle-out [ 8 ] and synthetic [ 9 ] modeling and simulation methods. They are specifically intended for representing large-scale biological systems in silico, and are well suited for simulation executions using cloud technology. Middle-out models are a compromise between the long-standing dichotomy between bottom-up and top-down models. Top-down models begin with a high level construct, usually a phenomenon to be generated, and specify its constituents purely from the perspective set by that high level construct. For example, the Navier-Stokes equations for fluid flow are a top-down model because they describe, directly, the aggregate fluid properties of the material being studied, and do not delve down into describing the properties or behavior of the molecules that generate the behavior. Bottom-up models, in contrast, leave the high level aggregate properties of the system unspecified and directly describe the properties of the bottom-most, atomic (devoid of structural information), constituents of the system. For example, a bottom-up model of a fluid would merely describe the molecules and their local interactions without making any attempt to describe the fluid, gas, or solid phenomena at the higher level. Both top-down and bottom-up approaches to modeling implicitly assign an aspect or perspective to the system. That perspective is not necessarily an objective, independently correct perspective. There are layers above any top-down model and layers below any bottom-up model. The modeler chooses where to "pin" the base perspective of the model. Middle-out models accept this a priori perspective and force the modeler to make the model explicit about the aspect chosen as the base. Extensions to the model are then explicitly allowed to go "up" or "down", depending on the requirements set for the uses to which the model will be put. When those layers need to be accessed, it will be easy to do so when using a cloud cluster. Another spectrum that can be used to describe models and that influences choice of simulation framework is that between inductive and synthetic modeling. Inductive models are derived by abstracting generic properties from many particular situations or objects. Synthetic models, by contrast, are constructed or pieced together from objects available to the modeler. A simple example would be a semblance of an automobile assembled from Lego pieces. These differences are elaborated in [ 9 ]. Inductively derived models are more amenable to execution on sequential machines because of their high degree of abstraction from the particulars of the referent mechanisms; but the referent biological systems are typically highly nonlinear and particular. A consequence is that those inductive models are only applicable to more generic contexts and are inapplicable to the specific contexts that will eventually be required to advance pharmaceutical research and support clinical relevance. ISLs are executed within a co-simulation [ 3 ] framework alongside an equation-based, extended, convection dispersion model [ 14 ] and an agent that interpolates and outputs the referent wet-lab data. The ExperAgent manages all three simulations. Running them in co-simulation allows similarities in behavior to be compared on the fly or after the execution terminates. Comparisons are made via a quantitative Similarity Measure (SM) [ 1 - 4 ]. When ISL observations fail to achieve a prespecified SM, one or more ISL micromechanisms are considered falsified and components or parameters must be modified to generate a new set of micromechanistic hypotheses. A failure to falsify means the mechanisms as implemented represent a plausible hypothesis and we say that ISL has achieved a degree of validation: it validates against the particular aspect quantified by that SM. Cluster architecture The architecture is diagrammed in Figure 2 . The local cluster consists of eight Intel dual core Pentium 4 based computers, one of which serves as the master node providing the Message Passing Interface (MPI) executive that dispatches jobs to the other seven slave nodes. Each node is connected to a gigabit switch over which it receives commands and data from the master node. The master node contains two network interface cards, one of which provides access to the local area network and Internet, and the other to the Gigabit backplane. The Network File System provides file data and executable to the slave nodes. Execution control and shared memory data are provided via the MPI. The cluster is a Grid in the sense that all the nodes have installed the same operating system, compilers, and libraries. Each ISL is launched as an executive program on the master node. Event sequences are diagrammed in Figure 3 . The executive node instantiates the Experiment Agent ( ExperAgent ). It partitions the simulation among the slave nodes based on command line parameters and the contents of parameter files. For the experiments reported herein, one MC lobule sample is executed per slave node and the results are collected and maintained on the master node until all lobule executions finish. However, the framework also supports a coarser parallelism where whole experiments are executed on a single node. The coarser mode is useful for parameter sweeping and behavior space searching, restricted to experiments with few MC samples. Only the ExperAgent and its utility components execute on the master node. The lobule and all the modeling components having wet-lab experiment counterparts, execute on the slaves. Each lobule sample has a proxy for the ExperAgent and the Data Management Module that handle execution control and data I/O. It is important to note that agency is orthogonal to the cluster architecture. The other agents within an ISL ( cells and Sinusoidal Segments, SSs) execute serially with interleaved discrete event schedules using the Swarm ( http://swarm.org ) scheduling engine. Similarity Measures and co-simulation As a lobule executes, the observations taken in compliance with the co-simulation SM are sent to and logged by the ExperAgent on the master node. Co-simulation and the SM are described in detail in [ 1 - 4 ]. Briefly, the co-simulation framework requires that any model be executed in tandem with another, a mechanistically different but behaviorally similar model, where the measurement protocol is identical for each co-simulation model. For ISLs, we use two other models, a two compartment ODE model and one interpolated from wet-lab experimental data. The SM provides a ± coefficient of variation band around a nominal profile derived from the wet-lab data. The SM score is the percentage of observations from the simulation that fall within that band. Results of one independent wet-lab experiment are experimentally indistinguishable from another when a prespecified degree of similarity is achieved. Yan et al. [ 1 , 2 ] and Park et al. [ 3 , 4 ] specified SM scores (e.g., ≥ 80%) that must be achieved to establish experimental indistinguishability between results from independent ISL experiments. When a lobule execution finishes, the ExperAgent shuts it down and dispatches the next one. When all samples are finished, the ExperAgent sends the raw observations to the Statistics Module. It calculates the derived measures that will be used to compare the results for validation or falsification. In the experiments reported herein, the Statistics Module merely averages the results from individual lobule executions and sends that to the Data Management Module to write to the disk for offline analysis. Replicating cluster architecture in the cloud In order to demonstrate experimental indistinguishability between cluster and cloud experiments, the experimental apparatus must be replicated in the cloud. Because an exact replication of the Grid and computation would be degenerate and uninteresting, decisions had to be made about where the cloud architecture would be allowed to differ from the cluster. Amazon's EC2 uses virtual machines running within a farm of actual machines. This means that some or even all the virtual machines, though unlikely, might be executing on the same actual machine. Hence, it is impractical to guarantee the same inter-node network as used on the cluster. Likewise, even though the generic specification of the virtual machines, like amount of memory and compute capacity, are specifiable, as virtual machines, their behavior will be somewhat dependent on the other virtual machines with which they share the real hardware. There are also constraints placed on virtual machine specification by Amazon's infrastructure. For example, a 32-bit virtual machine has a maximum memory of 1.7 gigabytes, whereas the 32-bit machines used in the cluster have 2 gigabytes of memory. As such, the virtual machines differ from the local cluster machines in almost every way at the hardware layer. Thirty-two bit virtual machine images were used in the cloud to limit precision and register size differences that might be introduced. Networking is handled the same as in the local cluster. However, as mentioned above, we cannot be sure that the maximum potential data flow from one virtual node to another is a Gigabit, given the allocation of hardware by the EC2 system. The master node acts as the network address translator, gateway, and firewall to the slave nodes. On both the cluster and the cloud, the simulation executive invokes MPI facilities that query a file on the operating system listing all network reachable nodes available. The script that starts and sets up the slave instances automatically assembles this file. The machines listed could be anywhere on the Internet. Nevertheless, these experiments were only run using virtual instances in the EC2 cloud. Via MPI, the ExperAgent is shared between the master and slave nodes, executing only the appropriate code for master or slave. The ExperAgent proxy on each slave makes live observations for its lobule until that execution is terminated. When all lobules have terminated, the remaining ExperAgent on the master node aggregates the data. The technology stack for the cloud exactly mimicked that of the cluster. The operating system was Ubuntu 8.04.2, the compiler GNU C/C++/Objective-C 4.2.4, the communications layer MPICH 1.2.7, and the simulation toolkit was the Swarm Subversion trunk from 2009-02-03. Additional file 1 contains instructions for setting up an ISL using EC2. Additional file 2 contains the ISL source code. The local cluster maintains the state of its hard disk every time it is shut down and restarted. An EC2 machine image, however, does not. The image is a snapshot of a running machine and when virtual machines are instantiated from that image, the state of the hard disk is as it was when the image was created. Hence, although a running instance of an image maintains changes made to the hard disk as long as that instance is running, when it is terminated, all the changes made since it was started are lost. That reality slightly changes the methods for running the ISL in an EC2 cloud. On the local cluster, if the mechanisms being tested are already in the ISL, then all that is required to run a new experiment is to start the cluster, write the new parameters and/or input data, and execute the simulation. Data written to the disk can be downloaded for analysis immediately, or left on the hard disk for later analysis. The only difference when executing in an EC2 is that the resulting data must be downloaded immediately or it will be lost. When new or additional mechanisms are to be tested, which was the case for this report, then on the cluster, those changes are made to the ISL source code. Canonical "experiments" are executed as regression tests to verify that the changes work as expected, and the changes are checked into the repository. Thereafter, the experiment proceeds as before. In an EC2 instance, however, because the instance comes up in the state with which it was originally created, it will have no ISL testing capabilities on its disk. If new mechanisms are to be tested, the latest version of the simulation source code must be checked out of the repository to the running virtual machine instance each time a new instance is invoked. The local cluster could not robustly handle experiments of more than 84 MC lobule samples. So, although the model scales to experiments requiring larger sample sizes, to scale the local cluster, we would have to add more nodes. Because the cloud platform is easy to scale simply by invoking more virtual machines, the 152 sample experiments were possible. In Silico Liver structure and components An ISL during execution maps to a mammalian liver undergoing perfusion as in [ 14 ]. An ISL represents a liver as a large, parallel collection of similar lobules, the functional units of a liver. The earlier version, absent BileCanal and the new metabolism mechanism, is detailed in [ 2 , 3 ]. The following is an abridged description. Components mimic essential hepatic form and function features. Flow networks are represented by an interconnected, directed graph, the structure of which is Monte Carlo determined for each lobule . Graph edges specify flow connections between Sinusoidal Segment (SS) placed at each node. There are multiple, different flow paths from portal vein tracts (PV) to CV. Most functions reside within SSs. A SS is a discretized, tube-like structure comprised of a blood "Core" surrounded by three identically sized 2D grids (Spaces A-C), which together simulate a 3D structure. Two SS classes (Table 1 ) are specified to provide sufficient variety of compound travel paths. Compounds are represented using mobile objects that move through the lobule and interact with encountered SS features. A typical compound maps to many drug molecules. A compound' s behavior is determined by the physicochemical properties of its referent compound, along with the lobule and SS features encountered. Multiple, different compounds can percolate through SS features during the same experiment. Objects called cells (two types: endothelial cells and hepatocytes ) map to an unspecified number of cells. They function as containers for other objects. Different colored grid locations in Figure 1 illustrate that the features at any location can be unique. Cells contain a stochastic, parameter-controlled number of binders in a well-stirred space. Binders map to transporters, enzymes, lysosomes, and other cellular material that binds or sequesters drug molecules. A binder within an endothelial cell only binds and later releases a compound . B inders within hepatocytes are called enzymes because they can bind compounds and later either release or metabolize it. Adjacent to Space C is the BileCanal (Space D). It is configured similar to the blood "Core." Objects placed in the BileCanal are collected in the bile duct (BD). Experiments Park et al. [ 3 ] reported an ISL and parameterization that validated against diltiazem outflow profiles from single pass, perfused, normal rat livers. The same wet-lab data serves as referent for the experiments reported herein. The simulation methods and iterative refinement protocol are also the same. However, for the experiments reported herein, two mechanisms were added and one mechanism modified. The former were discrete time metabolism and the bile canaliculi (BileCanal) illustrated in Figure 1 . Previous studies used a discrete event model for metabolism where an enzyme tested a pseudo-random number draw only when SoluteBindingCycles had passed. Although this micromechanism cannot be falsified against the available, coarse validation data, it has been pointed out that it is overly discrete and may produce abiotic artifacts. The experiments reported herein used a discrete time model for metabolism such that after binding , an enzyme determines if metabolism has occurred (or not) every simulation cycle up to and including the last before it releases the compound . This micromechanism provides more variation in the event history of the compound and more opportunity for metabolism . Consequently, these change require re-parameterization and re-validation. Previous studies did not include the creation of a metabolite because having it was not necessary for validation; when a metabolic event occurred, the compound was simply destroyed. However, it has been pointed out that the omission of this micromechanism detracts from the believability of the model. Because believability is closely tied to falsification and validation, the addition of this micromechanism adds heuristic value. The new micromechanism constructs a new metabolite when an enzyme successfully metabolizes a compound . A pseudo-random number draw is tested against the BileRatio parameter specific to the compound (diltiazem in this case). The hepatocyte decides whether the metabolite goes into the BileCanal or into the intracellular space. The dynamics of the BileCanal are identical to the Core. Once inside the BileCanal, solute steps forward each simulation cycle the number of spaces designated by CoreFlowRate . When a metabolite in the BileCanal reaches the outlet of a SS it moves to the BileCanal inside the next SS. If the outlet goes to the CV, the metabolite is moved, logged, and destroyed. That process maps to metabolite in bile entering the bile duct. The BileCanal is specified with a "circumference" as if it, like Space A, had a two dimensional grid wrapped around it. This geometry is used solely as a constraint to estimate how many metabolites may flow from a predecessor SS to a successor SS. It is not used when metabolite enters the BileCanal or when all metabolites are pushed forward within the data structure. So doing provides a minimal model for the constraints on bile flow and its output. A manual search of the parameter space for a parameter vector that satisfies the SM produced the parameters in Table 1 . Of particular importance are BileCanalCirc , BileRatio , SoluteBindingCycles , MetabolismProb , BindersPerCell , DosageParams , and the various values for JumpProb . A value of 1 for BileCanalCirc means that only a single compound can flow from a predecessor SS to a successor SS in a single step. Note that BileRatio is compound specific and is 0.5 for metabolizable diltiazem but 0.0 for both sucrose and metabolite ; because sucrose does not enter cells and enzymes cannot bind metabolites (unless configured to do so), their BileRatio values are irrelevant. The DosageParam values for these experiments are set so that dosage is an impulse and all compounds are created and injected into the PV in a single simulation cycle. It is also important to note that the JumpProb values in Table 1 are biased significantly outward so that the tendency is for compound to move from Space A to Space C. Manual searches were performed on the cluster using only 28 MC samples, which produced outflow profiles that were smooth enough for decision-making, but were not smooth enough to satisfy the SM. When a 28-sample result showed promise, experiments having 49 or more samples were undertaken to actually test the hypothesis that the parameterized mechanisms will produce outflow profiles that achieve the SM. In the latter experiments, the point was only secondarily to validate against the wet-lab data given the new mechanisms. The primary objective was to test the equivalence of the cluster and EC2 platforms. In order to do so, the parameter vector would have to produce a pair of results that validate in both the cluster and the EC2. If either pair of profiles, from the cluster or cloud, failed to validate, then there was a significant experimental distinction between the local cluster and EC2. In principle the simulation could be configured to produce identical outputs when run with identical pseudo-random number seeds on both platforms. However, it is important to note that such identical experiments should be explicitly avoided. Determinism is critical for software engineering and verification, where the computational system is not the subject of experimental exploration and hypothesis falsification. As laid out in [ 9 ], in order to synthesize heuristically valuable computational analogues that are useful for mechanism discovery and falsification, the system must be suitable for experimentation, just as in vitro systems are the objects of experimentation for wet-lab studies. Further, for simulations where the coupling is tight between the context (or use-case) and the internal mechanisms, as it is when identical deterministic results are expected of a simulation, the model tends to be too abstract, inductive, and very fragile to context [ 9 ]. The method of repeating stochastic experiments where some elements are tightly controlled and others are completely uncontrolled is more biomimetic and it mimics wet-lab methods more closely while enhancing scientific impact and usefulness. For these reasons, the experiments reported herein use the full stochasticity of the ISLs over and above the underlying platform differences and place all emphasis on experimental distinguishability or lack thereof.

Results Validation of an enhanced ISL To demonstrate methodological scientific equivalence, we must use the same model scaling and validation/falsification method on both platforms. To improve realism and heuristic value, we altered the ISL phenotype by making three changes: a biliary elimination and flow space, an improved metabolism mechanism, and an alternative method of dose input. Changes to in vitro models can have unintended consequences. The same is true with multiscale models like ISLs. Because the character of an ISL outflow profile is a consequence of networked micromechanisms, change can alter previously validated behaviors. Thus, new validation evidence is required. The enhanced ISLs that were executed on the local cluster showed a potentially abiotic behavior that had two possible causes. 1) It was a consequence of micromechanisms that were unintentionally altered by the enhancements, or 2) it was an artifact of a small lobule sample size. To test the latter, we needed the option to scale the execution platform to handle a larger number of lobule samples. Rather than buying more hardware to add to the local cluster, we chose to replicate the cluster architecture in the cloud. Validation experiments: local and cloud clusters Experiment results are shown for the local cluster in Figure 4 and for the cloud cluster in Figure 5 . The outflow profile for diltiazem is superimposed on the target zone, which designates the SM band derived from wet-lab data, as described above. As with the initial experiments in [ 3 ], when SM ≥ 80% the ISL and wet-lab profiles are designated experimentally indistinguishable. We specified that acceptable demonstration platform equivalence would be when SM ≥ 80% for two independent sets of experiments on both cloud and local cluster. Outflow profiles for diltiazem, d, in normal livers [ 15 ] provide the center for the SM band in Figures 2 and 3 . The width of the band is derived from the coefficient of variation of the sucrose outflow profiles from [ 5 ]. The upper and lower edges of the target zone are {d ± coefficient of variation}. To show the smoothness and asymptoticity of the profiles, the finite differences {d n - d n -1 } are also plotted. The outflow profiles for the 49- and 70-sample experiments failed to satisfy the SM in two important ways: the presence of an apparent large period oscillation and high variance. The high variance can be smoothed to show a much more well behaved profile; but for the purposes of this cross validation exercise, the raw data provide more insight into events occurring during the simulation experiments. The variance in wet-lab outflow profiles is typically rather constant. Deviations about the trend line are typically random, and those observations are the basis for SM target bands in Figures 4 and 5 . However, the patterns in Figures 5A, B and 5A, B are different: all four, 49-sample studies showed significant oscillations in variance prior to about 60 seconds that appeared somewhat abiotic. The apparent variance oscillations caused fluctuations in the outflow profiles. One of the runs in cluster and cloud (Figure 4B and Figure 5B ) exhibited strikingly large variances. The variance oscillations are obvious in the finite difference scatter plot in Figure 5A . The outflow peak is reinforced by the first peak of the oscillation, making the leading edge of the profile seem markedly false in comparison to the peak shape of the validation band, which is sharper. The apparent oscillation in variance reaches minima at approximately 20, 35, and 50 seconds in Figure 4A ; it reaches minima at approximately 30 and 50 seconds in Figure 5A . Variance oscillations were still evident in the 70-sample local cluster and cloud studies, but overall variance decreased. Because of the ISL's structure, as the number of lobule samples increased, the variance decreased and the asymptoticity increased. That expected trend is evident in the finite difference plots. Both pairs of 84-sample studies (Figures 4A, B and Figure 5A, B ) achieved validation. Although hints of the variance oscillation remain in the 84-sample outflow plots, they satisfied the SM. Further, the variance oscillation is faint or absent in the 152 sample experiments executed in the cloud (Figure 5G, H ).

Discussion Arguably, the more important methodological impact of validating these ISLs in both the local cluster and EC2 cloud platforms is on model sharing and computational experiment repeatability amongst various laboratories and collaborators. It is a simple matter to acquire and read the source code for a simulation like the ISL. For an organization with ample resources, the appropriate systems administration, and programming skills, it is straightforward to set up a serial computer platform on which to execute an experiment. However, because synthetic models like ISLs are very concrete and implement fine grained, multilevel and multiscale networked mechanisms, experiments on serial computers are infeasible. Similar models will face the same problem. Hence, in order for a laboratory to repeat a given experiment or, more importantly, engage in its own exploration of the analogue, it must have access to a cluster and enough privileged user access to configure the tool-chain on that cluster. This obstacle is higher than it seems, for it requires a well-organized laboratory and the motivation and resources to do the work. In effect, this obstacle contributes to the dominance of engineering oriented modeling and simulation (typically relying on proprietary software), and limits scientific, exploratory modeling to a privileged elite with large budgets. Although Amazon's EC2 infrastructure costs some money, the semi-automated methods used to instantiate virtual machine images, terminate those instances, and pay for the service with a credit card allow anyone to construct and use their own cluster at minimal cost. During the exploration phase of this work, an individual 49-sample experiment cost approximately $30, representing what a curious scientist might pay simply to instantiate and run a single experiment. When compared with the cost of building, purchasing, or maintaining a cluster, or acquiring time on another laboratory's cluster and then arranging for it to be configured properly for these experiments, the EC2 costs seem quite low. However, this cost estimate is informal. Kondo et al. [ 11 ] reported a detailed treatment of costs for the EC2. Costs have also been reported for voluntary computing grid platforms like BOINC [ 16 ]. Even though we attempt to replicate as much of the environment as possible, our objective, detailed in Methods, was scientific repeatability, not data duplication. Previous experiments [ 2 - 4 ] only required 49 samples to show experimental indistinguishability, in part because prolonged rather than bolus dosing was used. However, some anomalous results from previous experiments had shown periodicity [ 17 ]. Nevertheless, the 49- and 70-sample experiments in both the cluster and cloud failed to validate because of large variance oscillations. We conjectured that perhaps because the diltiazem outflow profile was relatively flat, the ISL parameterizations needed to validate against it would show underlying oscillations that are not evident in profiles that decline more steeply. Something causes ISL outflow profile variance to fluctuate. It could be: a) the parameters, code, or input data differ between each experiment or b) the discretized nature of each lobule is sufficiently unique so that it takes a large lobule sample to produce an acceptable approximation of a continuous liver outflow profile. If it is a), then the apparent artifact would still be evident after a large number of samples; it would not be steadily washed out with additional samples. If it is b), then having experiments comprised of enough samples will make the oscillations cancel each other. The latter is what we observed; we needed at least 84 lobule samples (Figs. 4 and 5) to achieve validation. One way to conceptualize a discretized ISL during operation is as a large set of oscillators. Each space within an ISL acts, to some extent, like a timed storage device. A compound comes into a space (e.g., Space C), is effectively sequestered there for a number of cycles as it moves about, and it then exits that space. It may or may not be sequestered again later along its path to the CV. Hence each compound goes through a series of store-release processes, except that the store and release events are stochastic. The delay SoluteBindingCycles , however, is not stochastic. Each lobule execution starts with 5,000 compounds pulsed into the PV. Each one encounters a large array of "inductors" that can retain a compound for an interval and then release it. When a large number of compounds get caught, released, caught, and released all at roughly the same intervals, we see what looks like a stochastic oscillator with a random signal riding atop a larger oscillation. Such a signal is made more obvious by the impulse bolus and a small number of compounds . The only elements of the model interfering with the predictability of the combined oscillation are: 1) the stochasticity surrounding when each compound is bound and rebound and 2) that some compounds are metabolized and their product moved into the BileCanal, and no longer contribute to the profile. Hence, as the number of compounds increases, the variance oscillations should disappear. Indeed, the high variance that showed up in the 49-sample cluster run in Figure 4B , disappeared when the number of samples were increased to 70. Furthermore, the variance swings diminished in each of the 70 then 84-sample studies on both the local cluster and in the cloud. Finally, for the 152-sample studies in the cloud, the variance swings have disappeared completely and the outflow profile decays more smoothly. The similarity between the cluster and cloud results for 84 and fewer MC lobules demonstrates the equivalence of the models executing on the two platforms. Further, confirming the low sample size conjecture with the 152-sample experiments demonstrates the value of having available the scalable cloud platform to extend the mechanism validation and falsification methods beyond the capabilities of the local cluster. The results demonstrate experimental indistinguishability of ISL outflow profiles from the local cluster and Amazon EC2 cloud platforms for experiments consisting of 84 or more lobule samples. The results also demonstrated that mechanistic changes confined ISL behaviors to a smaller, more constrained region of parameter space. The modifications provide an even more concrete and specific, falsifiable hypothesis about the referent mechanisms. Making these changes and validating them against the coarse aspect of outflow profiles have made more specific the regions of the parameter and behavior spaces that are biomimetic without losing the generality of ISL parameterization.

Conclusion The results contribute evidence of silico-to-wet-lab phenotype similarity, specifically that ISL behaviors can cover some of the behavior space of referent liver experiments. So doing allows for concrete representations of multilevel hepatic mechanisms in many distinct experimental contexts, without degrading the ability to parameterize an ISL so as to accurately represent a particular context. To date, the scientific benefits of experimenting with multiscale biomedical models has been limited to small numbers of researchers with access to computer clusters. Cloud technology coupled with the evidence of scientific equivalency has lowered the barrier and will greatly facilitate model sharing while providing straightforward tools for scaling simulations to encompass greater detail with no extra investment in hardware. The flexibility and dynamic expandability of a platform cloud infrastructure is important for open-ended, exploratory, mechanism-focused research.

Background In Silico Livers (ISLs) are works in progress. They are used to challenge multilevel, multi-attribute, mechanistic hypotheses about the hepatic disposition of xenobiotics coupled with hepatic responses. To enhance ISL-to-liver mappings, we added discrete time metabolism, biliary elimination, and bolus dosing features to a previously validated ISL and initiated re-validated experiments that required scaling experiments to use more simulated lobules than previously, more than could be achieved using the local cluster technology. Rather than dramatically increasing the size of our local cluster we undertook the re-validation experiments using the Amazon EC2 cloud platform. So doing required demonstrating the efficacy of scaling a simulation to use more cluster nodes and assessing the scientific equivalence of local cluster validation experiments with those executed using the cloud platform. Results The local cluster technology was duplicated in the Amazon EC2 cloud platform. Synthetic modeling protocols were followed to identify a successful parameterization. Experiment sample sizes (number of simulated lobules) on both platforms were 49, 70, 84, and 152 (cloud only). Experimental indistinguishability was demonstrated for ISL outflow profiles of diltiazem using both platforms for experiments consisting of 84 or more samples. The process was analogous to demonstration of results equivalency from two different wet-labs. Conclusions The results provide additional evidence that disposition simulations using ISLs can cover the behavior space of liver experiments in distinct experimental contexts (there is in silico-to-wet-lab phenotype similarity). The scientific value of experimenting with multiscale biomedical models has been limited to research groups with access to computer clusters. The availability of cloud technology coupled with the evidence of scientific equivalency has lowered the barrier and will greatly facilitate model sharing as well as provide straightforward tools for scaling simulations to encompass greater detail with no extra investment in hardware.

Abbreviations CV: central vein; ISL: In Silico Liver; MC: Monte Carlo; MPI: Message Passing Interface; M&S: modeling and simulation; PV: portal vein tracts; SM: similarity measure; SS: Sinusoidal Segment Authors' contributions GEPR replicated the cluster architecture in the cloud. GEPR designed and conducted validation experiments. GEPR and CAH evaluated validation experiments. GEPR and CAH wrote, read, and approved the final manuscript. Supplementary Material

Acknowledgements This work was supported in part by the CDH Research Foundation and the Alternatives Research & Development Foundation.

CC BY

no

2022-01-12 15:21:44

BMC Syst Biol. 2010 Dec 3; 4:168

oa_package/13/ab/PMC3016276.tar.gz

PMC3016277

21144052

Background The formation of an embryo and its organ systems requires the coordination of diverse cellular behaviors to achieve proper development of form and function. Cells must migrate, often collectively, and proliferate in a regulated way, while simultaneously carrying out specific developmental programs. A full characterization of these events requires a description of the cellular histories (lineage) and knowledge about the molecules that regulate these processes. Active cell movements take place not only during the development of organisms, but also in processes that occur during adult life. For example, in the immune system, an effective immune response depends on the regulated traffic of its cellular components. Disrupted cell migration also contributes to several important pathological processes, including cancer and chronic inflammatory diseases such as rheumatoid arthritis and multiple sclerosis. To date, most knowledge about cell migration is based on in vitro studies of single cells in two-dimensional cultures. These studies have allowed great progress on the intracellular events that take place during cell motility, elucidating the details of the cellular machinery driving migration. However, in vivo models of collective cell migration have been less intensely studied because of the inherent difficulty in undertaking such analyses. Recently, the migrating primordium of the zebrafish posterior lateral line has emerged as an attractive system for genetic analysis of cell migration and tissue organization and for understanding how these processes are controlled [ 1 - 9 ]. The lateral line is a mechanosensory system present in fish and amphibians that responds to water movements and is involved in a large variety of behaviors such as predator avoidance, prey detection, or swimming in schools [ 10 , 11 ]. This sensory system is formed by a number of discrete sense organs, the neuromasts, distributed over the body in species-specific patterns. The neuromasts of the head form the anterior lateral line (ALL), while the neuromasts of the trunk and tail, including those on the caudal fin, form the posterior lateral line (PLL). In zebrafish, the embryonic PLL comprises 7-8 neuromasts and its development begins at 20 hours post-fertilization (hpf) when a group of about 120 cranial placodal cells delaminates and begins to migrate collectively along the horizontal myoseptum towards the tail of the developing larva [ 12 ]. During its journey, the PLL primordium deposits groups of 15-20 cells (proneuromasts) at regular intervals and, a few hours after deposition, each proneuromast differentiates into a functional neuromast. They contain at least three cell types, including the mechanosensory hair cells and, therefore, the primordium contains multipotent progenitors for different cell types. Progenitors are also set aside within mature neuromasts, as damage to these organs is followed by functional recovery due to robust regeneration of hair cells and other structural elements [ 13 - 19 ]. One of the essential features of the migrating lateral line primordium is that its cells must become organized in order for them to migrate coherently and produce a functional organ. Today, it is known that the primordium of the LLP contains mesenchymal-like cells at its leading edge and, towards the trailing edge, 2 to 3 groups of rosette-shaped polarized cell clusters are formed, each corresponding to a proneuromast. Moreover, recent studies have shown that ligands of the FGF family, fgf3 and fgf10 , are essential for the internal organization, patterning and migration of the primordium and therefore for correct neuromast deposition [ 4 , 5 ]. These ligands, together with Wnt pathway activation, are responsible for the restricted expression of the chemokine receptors CXCR4b and CXCR7b in the different compartments of the primordium [ 6 ]. Both these chemokine receptors, and the SDF1a/CCL12 ligand, are critical for proper directional guidance of the primordium cells as inactivation of these molecules produces alterations in the migration pattern of the primordium [ 1 - 3 , 20 , 21 ]. While directionality of migration is affected under these conditions, the migratory behavior of the cells themselves is not impaired, suggesting that independent pathways may regulate directionality and movement of the cells. Thus, there are likely to be numerous additional molecules involved in the process. As the PLL primordium migrates, cells are simultaneously dividing and undergoing developmental programs that will build the functional lateral line. Hair cells and accessory cells derive from migratory primordium cells and their fates become apparent prior to neuromast deposition. The proneural master control gene for hair cell fate, atonal homolog 1a ( atoh1a ), is expressed in the center of the rosettes that form successively within the primordium [ 22 , 23 ]. Likewise, the neurogenic genes delta and Notch participate in cell type specification in neuromasts and are detected in presumptive neural and accessory cell progenitors, respectively [ 22 , 24 ]. The identification of new molecules that control the directional movement of the primordium, cell proliferation and subsequent differentiation of hair cells and other cell types will contribute to a better understanding of the processes leading to development of the mechanosensory lateral line, which shares many physiological properties with the mammalian auditory system. More importantly, formation of this organ shares molecular mechanisms with processes involved in cancer metastasis [ 8 ] and immune system cell migration [ 25 ]. In this study, we carried out a global survey of genes expressed in the PLL primordium and in maturing neuromasts. We have taken advantage of the ClaudinB:GFP transgenic zebrafish line ( cldnb:gfp ) [ 1 ], in which primordium and neuromast cells specifically express the membrane-bound enhanced Green Fluorescent Protein (eGFP). We have isolated GFP expressing cells as well as non-labeled cells at two different developmental stages (36 and 48 hpf) using Fluorescence Activated Cell Sorting (FACS). We show that comparison of expression profiles from GFP-positive and GFP-negative cells by microarray analysis allows identification of genes highly expressed in the migrating primordium (36 hpf set) and in deposited neuromasts (48 hpf set). Thus, the expression profiles of lateral line precursors during and after the migratory phase could be compared. We further concentrated on the 36 hpf set of data, to find molecules belonging to diverse cellular and developmental processes. Using this approach, we demonstrated its utility for the identification of genes involved in morphogenesis, migration and cell type specification within the migrating PLL primordium in the zebrafish. Importantly, some of the identified gene products can also open new insights into a more global comprehension of cell migration, which may be crucial to create efficient therapies to treat human diseases such as cancer and autoimmune disease.

Methods Zebrafish Fish and embryos (AB and T/AB5 strains) were maintained in our facility according to standard procedures [ 47 ]. The cldnb:gfp transgenic line was obtained from Darren Gilmour [ 1 ]. All embryos were collected by natural spawning, staged according to Kimmel et al. [ 48 ] and raised at 28.5°C in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, and 0.1% Methylene Blue) in petri dishes, as described [ 49 ]. Embryonic and larval ages are expressed in hours post-fertilization (hpf). Embryo dissociation and fluorescence activated cell sorting (FACS) Cldnb:gfp embryos were dechorionated with Pronase (P8811, Sigma) and transferred to glass dishes. We used the posterior third of the transgenic embryos in order to avoid recovering GFP-expressing cells in tissues other than the lateral line. Fish were anaesthetized and the tail was transected with a sharp scalpel adjacent to the end of the yolk extension. Tails of embryos were recovered from 36 hpf (for isolating primordium cells) or from 48 hpf (for neuromast cells) embryos, then transferred to Hank's medium containing 0.25% trypsin and 1 mM EDTA and incubated for 15 to 30 min at room temperature during which they were dissociated mechanically by pipetting them up and down every 5 minutes. The digestion was stopped by adding fetal bovine serum to 10% and cell suspensions were then filtered through 40 μm nylon mesh, washed twice with PBS, pelleted by centrifugation at 600Xg for 2 minutes and resuspended in L-15 medium (Sigma) supplemented with 10% fetal bovine serum. FACS of single cell suspensions was performed at room temperature using a FACSAria Sorter (Becton Dickinson) with a Coherent Innova 70 laser at 488 nm and 200 mW power. GFP+ and GFP- cells were sorted directly into Trizol Reagent (Invitrogen) and stored at -80°C. RNA isolation and quantitative RT-PCR Total RNA was isolated from equal numbers of GFP+ and GFP- cells by extraction with Trizol Reagent, according to the manufacturer's instructions. RNA pellets were resuspended in nuclease-free water (Ambion). RNA integrity was confirmed by visualization of the ribosomal RNA on picochips for the Bioanalyzer (Agilent). Approximately 500 ng to 800 ng RNA was linearly amplified by using the Amino Allyl MessageAmp II aRNA Amplification kit (Ambion) with yields ranging from 12 to 30 μg of aRNA. aRNA samples were split and labeled, half with Cy3 mono NHS ester and half with Cy5 mono NHS ester (CyDyes from GE Healthcare; post-labeling reagents from the MessageAmp II kit). Quantitative PCR was performed using SuperScriptTM III Platinum ® SYBR ® Green One-Step qRT-PCR (Invitrogen) using 20 ng of RNA and 5 μM gene specific-primers in a 25 μL reaction, according to the manufacturer's instructions. PCR primers were as follows: atoh2a 5'-GCACGAGCTGTTCACCTGCTAAAAC and 5'-GACGGGATCCTCAAAGGGAAGAG β-actin 5'-TGGCCCCTAGCACAATGAAG and 5'- GCCTCCGATCCAGACAGAGTAT; cldnb 5'-GAAGGAATTTGGATGAGCTGCGTGG and 5'-CGACAGCATGATTCCCATCAGTCCG; cd9b 5'-GTGATTGGAGGAGTCGGCATCG and 5'-CACTCGCGCACATTCACAAACG; dlx6 5'-CCAGCAGACTCAATACCTGGC and 5'- CCGCCTTGTTTCAACAGCTTC egfp 5'-GACCCTGAAGTTCATCTGCACCACC and 5'-GCGGGTCTTGTAGTTGCCGTCGTCC; epcam 5'-ATGGCCAGCCATTTGAGGTTGATG and 5'-GCCGTGCAAGAAAGAACAGAACCA. klf3 5'-CCACAGCCAAGAGAAATCGGTC and 5'-GTGTGCGTCTATGGGCTTTCAG; lect1 5'-GAGTGAAAGCTGAGGTGGCGTC and 5'-CTGGCGTCTCATTCGTGTCGC; myod 5'-ATGACACACCAAATGCTGACGCAC and 5'-TCTCTGTGGAAATTCGCTCCACGA; pecaml 5'-CGTTGGCCGGATGGTTAAATCC and 5'-GCAGCGATGGGAACTTTCACTC; ptgs2 5'-CATGTTTGCTTTCTTCGCCCAGC and 5'-GTGTTGAACCTCCAGCGTCTCTC 3; ptgds 5'-CCTGAAGAGTGATGGTTCCTGCTG and 5'-GTCAACCACGCGCATGTCATTG; ptgisl 5'-GAACAACCTCCGCCTGCTTATG and 5'-GCTTGTCAAACCGGCGAAACTC; scyba 5'-GATGAATCGCTGTAGTACGGCCGC and 5'-CCGGATCTTCGGCCCTTTTCTTG; sema3 d 5'-CAGGTGTCACTGCACAGGTGCC and 5'-GTGTCCCAACAATGGCTGGATG; similar hcd205 5'-GTGCGTCAGATCTCAGTAGTTTGA G and 5'-GCTAAAGCAGTGCACCGAGTGCTG; similar hcxcr6 5'-GAAGAATGGCTGGAACCTGCAAC and 5'-CCGACTATGCCAGTCACAACC; sox8 5'-GAACGCGTTTATGGTTTGGGCTC and 5'-CCTCCACGAACGGCCTCTTCTC; sox17 5'-CCCAGACCTGCACAATGCGG and 5'-GGGCTCCAATCGCTTGTTTCG Expression levels were determined by comparison to a standard curve from total RNA isolated from whole embryos. Values from GFP- and GFP+ samples were then normalized to β-actin to obtain relative expression levels. Standard deviations were calculated from triplicate measurements. Microarrays analysis We used "in house" 34 K zebrafish oligo microarray chips, which was derived from three sources: Compugen (16,512 × 60 mers), MWG (14,240 × 50 mers) and Operon (3,479 × 70 mers). The whole set contains replicates of several positive (known housekeeping genes) and negative control oligos (random sequences) to control for the homogeneity and specificity of the hybridization. Two biological replicates of GFP+ and GFP- cells were cohybridized, each with one dye swap, for a total of eight hybridizations. Hybridizations were performed overnight at 45°C in Maui Mixer FL hybridization chambers (BioMicro Systems). Microarray slide post-hybridization processing and scanning were done as previously described [ 50 ]. Data analysis to identify genes enriched in GFP+ vs. GFP- cells in pair-wise comparisons was carried out with FileMaker Pro 9 (FileMaker, Inc.) and GeneSifter http://www.genesifter.net/ software. Data points with average quality values below 0.7 were eliminated and the datasets were Lowess normalized. Normalized data were log-transformed and GFP+ and GFP- values were separately averaged over the eight experiments. Fold differences were calculated from log averages and Student's t-test with Benjamini and Hochberg correction to generate p-values that were used to determined statistical significance. The microarray data is available at NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE25617. In situ hybridization Whole mount in situ hybridizations (ISH) were carried out essentially as described in [ 51 ], with post-hybridization washes performed by an automated liquid exchanger (Biolane HTI, Hölle & Hüttner). Plasmids containing zebrafish cDNAs and EST clones were obtained from OpenBiosystems (Huntsville, AL). Inserts from plasmids were PCR amplified using specific primers containing RNA polymerase binding sites and purified using Qiaquick columns (Qiagen). Antisense digoxigenin-labeled riboprobes were synthesized from PCR products using the appropriate bacteriophage RNA polymerase. Pathway analysis Pathway analysis of the genes enriched in the primordium was done with MetaCore software (GeneGo). The known relationships between the human orthologs of the zebrafish genes were used to generate the interaction network. The network was then manually examined and curated. In addition, an extensive search of the "GeneGo processes" was done for genes known to be involved in cell migration/adhesion and cancer pathways. The primordium-enriched genes involved in two or more of these processes were identified.

Results Isolation of primordium cells (36 hpf set) The cldnb:gfp transgenic zebrafish line expresses membrane-bound GFP in the PLL primordium and neuromasts, in the olfactory system, the ear, branchial arches, and a lower amount of expression in the pronephros and skin in embryos and larvae [ 1 ]. As we wished to isolate mostly primordium and neuromast cells -and avoid other labeled cell types- we proceeded to dissect 36 hpf and 48 hpf embryo tails by sectioning the animals at the end of the yolk extension (posterior to the cloaca) (Figure 1 ). At 36 hpf the PLL primordium has migrated beyond the yolk extension (Figure 1A-B ) and, therefore, GFP-labeled cells in the tail sections include almost exclusively lateral line precursors. To confirm proper recovery of the migrating primordium at 36 hpf, isolated tail fragments were observed under fluorescence (Figure 1C ). We then processed them for RT-PCR using primers specific for genes with known expression patterns in the PLL such as cxcr4b [ 20 ], epcam [ 26 ] and cldnb [ 1 ] (Figure 1D ). Additionally, we carried out in situ hybridization to detect expression of eya1 , a gene specific to the primordium at this stage [ 27 ] (data not shown). These controls all showed that, under our experimental conditions, the primordium was present in the entire population of dissected tail fragments. Embryo tails were immediately processed for cell isolation. Cells were passed through a flow cytometer and two cell populations were separated: GFP+ and GFP- (Figure 2A-B ). The GFP window selected for flow cytometry indicates that purity was higher than 95%. To confirm that we were able to achieve enrichment of primordium cells from the 36 hpf fraction by selection on the basis of GFP labeling, we prepared mRNA from both sorted samples of cells and carried out quantitative RT-PCR (qRT-PCR) using primers specific for genes known to be highly expressed in this tissue ( epcam , cldnb ) and a muscle-specific gene ( myod ), which should be enriched in the GFP- fraction. As expected, the epcam and cldnb mRNAs are enriched in the GFP+ population of cells while myod is overrepresented in the unlabeled collection of cells (Figure 2C ). Similarly, qRT-PCR detection of GFP mRNA showed high enrichment (16 fold) in the GFP+ group of cells. These results indicate that we were able to purify lateral line primordium cells from zebrafish embryo tails at this stage. Isolation of neuromast cells (48 hpf set) A parallel experiment was carried out using 48 hpf cldnb:gfp embryos to isolate cells belonging to deposited neuromasts and interneuromastic cells, lying between neuromasts (data not shown). This preparation was carried out in order to have a sample of RNA obtained from the same tissue but at a different developmental stage. Given that these cells are beginning to differentiate and they are no longer migratory, comparison with the primordium sample would allow us to discriminate between genes involved in the different cellular processes occurring during lateral line development. As in the earlier stage, RNA was prepared from GFP+ and GFP- cells obtained from 48 hpf embryos. Microarrays and data analysis Identification of the PLL primordium and neuromast transcriptomes For isolating lateral line-specific transcripts, we reasoned that we could subtract the transcriptome obtained from the sorted GFP- cells to the one obtained from GFP+ cells by carrying out a co-hybridization experiment using microarrays. The sorted GFP- sample contains unlabeled cell types belonging to many different tissues of the same embryos and should thus allow us to remove most housekeeping transcripts from the lateral line transcriptome. We obtained RNA from GFP labeled and unlabeled cells from two independent tail sectioning experiments for each stage analyzed. From each of these samples, cDNA synthesis was carried out and four rounds of dye labeling (including one dye swap) were done for probe generation. "In house" microarray slides printed with 34,000 oligonucleotides that correspond to 19,000 zebrafish-expressed sequences were co-hybridized with probes prepared from GFP+ and GFP- cDNA from each stage analyzed. Differential expression was considered significant when the average hybridization signal from all 4 hybridizations differed by a factor of more than 1.5 (p < 0.05) in either direction (Figure 3A results for the 36 hpf set, 48 hpf not shown). We have grouped the data obtained from our analysis of differentially detected sequences in a number of tables (Supplementary data). We identified 4,449 (36 hpf set) sequences enriched in the primordium, that represent 3,505 known genes and 944 ESTs that do not map uniquely to a single gene; and 3,425 genes and 575 ESTs (48 hpf set) in the neuromasts (Additional files 1 and 2 , respectively). Using the GeneSifter software, our data for the 36 hpf set shows representation of a wide variety of biological functions among the genes identified in this screen (Figure 3B ). Gene ontology classification shows that, compared to the relative proportion of cellular functions found in the entire collection of genes represented on the array, some categories are over-represented in primordium cell mRNAs. There is an enrichment in specific biological roles that include, cell motility, adhesion, cell communication and cellular developmental processes. These results are in accordance with known biological processes occurring during migration of the lateral line primordium [ 8 ]. Given the diversity of cellular functions involved in development of the lateral line, we combined the results from our two data sets (36 and 48 hpf) and grouped some of the sequences obtained according to known biological processes or developmental pathways (Table 1 ). Genes implicated, for example, in cancer progression, immune function, cell adhesion and cell signaling were found in this study, many of them described previously in lateral line cells, others being novel. The data provided in this work should help to pinpoint some of the molecular players in lateral line development and provides further parallels with diverse biological functions in normal and pathological situations. Validation of the microarray data In order to estimate the degree of success of our strategy for identifying specifically expressed sequences, we collected information from publicly available sources (ZFIN and PubMed) to identify all genes previously known to be expressed in the lateral line system (primordium and/or neuromasts). We excluded genes expressed only at the placodal stage (PLL precursors), in neural components (PLL ganglion, nerve) and in glial cells. These cell types do not express GFP in the cldnb:gfp transgenic line and thus would not have been purified during the cell sorting procedure. In our database survey, we found 372 genes in which specific expression in the PLL primordium or neuromasts has been convincingly described (Additional file 3 ). We then compiled how many of those genes where actually represented on our chips, we found 267 genes. Out of those, 148 genes where overrepresented in our two data sets (from 36 hpf and 48 hpf embryos) meaning that we were able to detect more than half of the previously described genes and had a 68% success rate if the genes were represented on the chip. Some genes will escape this analysis, particularly genes that have a function in both the migrating primordium and in other tissues of the tail, but the initial capture of genes enriched in the migrating primordia marked by cldnb expression appears to be primarily limited by representation on the array. Expression and functional analysis of newly identified genes in the migrating primordium To provide an independent verification of our results, we randomly selected 15 different cDNAs from sequences showing higher than 1.5-fold enrichment in the GFP+ cells in our 36 hpf set. We performed qRT-PCR to confirm the presence of these transcripts in the purified primordium cells (Figure 4A ). For 12 of these genes, quantitative RT-PCR reveled significant enrichment in GFP+ cells. We wished to provide additional evidence for the success of the purification and selection strategy for identification of genes expressed highly in the lateral line system. We selected differentially expressed genes and the corresponding cDNAs were used as templates for synthesis of digoxigenin-labeled antisense RNA probes for in situ hybridization. Specific label was detected in 36 hpf embryos with seven of these probes in the PLL primordium and/or neuromasts (6 of them are shown in Figure 4B-G ). The remaining hybridized embryos did not show specific labeling patterns or provided inconclusive results. Two of the genes, f11r and cd9b , which displayed enriched expression in the primordium, were selected for loss-of-function analysis, based on their known roles in a large variety of physiological and pathological processes such as the immune response, reproduction and development, infectious and genetic diseases as well as metastasis, where both promote cell adhesion and motility [ 28 - 32 ] and also because of their unknown function in the lateral line development. F11r (JAM-A) is a transmembrane adhesive glycoprotein that participates in numerous cellular adhesive process including leukocyte migration, intracellular junction assembly and the regulation of cell morphology and motility. In endothelia and epithelia, JAM-A localizes at cell-cell contacts with the tight junction proteins occludin and claudins [ 28 ]. In cancer, the tumor progression has been correlated with high JAM-A gene expression levels and patients were significantly more likely to develop metastasis [ 33 ]. As tumor cell motility is required early in the metastasis process, is has been speculated that increased JAM-A expression in cancer promote motility, through unregulated β1 integrin protein expression [ 33 ]. CD9 in turn, is an integral membrane protein that belongs to the family of tetraspanins and has been implicated in various biological functions, including cell adhesion, motility, metastasis, growth, signal transduction, differentiation, and sperm-egg fusion [ 31 , 32 ]. We inhibited translation of these genes through microinjection of antisense morpholinos (MOs) directed against the ATG region of each transcript. We injected cldnb:gfp transgenic zygotes and examined the development of the PLL system from 24 to 48 hpf. Neither of the two MOs elicited gross developmental abnormalities, as was the case for the corresponding control (a nonsense morpholino, not shown). In early LL development, the overall migration process seemed unaffected in either f11rMO or in cd9bMO injected embryos (Figure 4H-J , white arrows). However, in 50% of cd9b morphants (Figure 4L-M ) and in 30% of fr11 MO injected larvae (Figure 4N-O ) at 36hpf, we observed differences in the shape and in the internal organization of the primordium compared to control primordia (Figure 4K ). Differences were also seen in the neuromast deposition pattern at 48 hpf (Additional file 4 ). Control embryos show the stereotypical neuromast distribution (Additional file 4 : Supplementary Figure 1A) while, in cd9b morphants, the neuromast number in the PLL is reduced (Additional file 4 : Supplementary Figure 1B-C, arrows). In contrast, in f11r MO embryos we observed alterations in the timing and spacing of the deposited neuromasts (Additional file 4 : Supplementary Figure 1D-E). These results support previous work [ 4 - 6 ] that the internal organization of the primordium is critical for morphogenesis and patterning of the organ, which is in turn needed for correct neuromast deposition. Pathway analysis In order to determine possible pathways and biological relationships in our 36 hpf data set, we searched for functional interactions among enriched genes using the GeneGo Bioinformatics software. Of the primordium-enriched set of 4,451 zebrafish genes, we identified 1, 911 human orthologs. We focused on the Gene Ontology (GO) terms for the following 5 categories: cell-cell adhesion, guidance, collective cell polarization, EMT remodeling and signaling (Figure 5A ). All of these 5 categories have been previously implicated in collective cell migration [ 8 ]. As expected, by examining the microarray data from the migrating primordium, we detected an enrichment in genes known to be involved in cell movement, such as cxcr4 , cxcr7 , fgfr , egfr and matrix metalloproteases [ 34 ]. We also reconstituted the functional network of candidate genes using the known interactions between human orthologs (with GeneGo), which revealed new potential crosstalk between different pathways that are essential for cell migration. Interestingly, all the enriched genes captured in these categories have demonstrated roles in cancer progression. Furthermore, the many interactions found between these groups of genes, corroborate that collective cell migration results from a complex molecular cascade where dynamic processes occur in a coordinated fashion. Additionally, we compiled the complete list of genes with known functions in cell adhesion, migration and cancer metastasis provided by GeneGo (Figure 5B ) to identify how many of those genes where represented in our dataset. Of the 1,911 primordium-enriched genes (light purple circle in 5B) we founded 807 cancer-related genes (pink circle in 5B) and 224 that are cell adhesion and/or migration-related (blue circle in 5B). Moreover, we found that 165 genes are shared by all events, confirming that cancer progression and normal cell migration in the developing zebrafish involve common mechanisms and pathways. Thus, we propose that the PLL migrating primordium is a relevant model in which to study the diverse processes that occur during collective cell migration in normal and pathological conditions.

Discussion Our aim was to uncover a set of genes expressed in the lateral line primordium and in developing neuromasts, the principal components that form the mechanosensory lateral line system in the zebrafish. We took advantage of the cldnb:gfp transgenic line, which expresses GFP in all of the cells derived from the posterior lateral line primordium. GFP positive cells were isolated through cell sorting and the mRNA prepared from these cells was reverse transcribed and probes generated; GFP negative cells were treated similarly to generate probes for subtraction in microarray hybridization experiments. Our reasoning was that this strategy would reveal a collection of genes whose expression was specific to lateral line cells. While this experimental design allowed us to detect genes highly expressed in primordium cells in relation to other tissues -and to discard housekeeping genes-, it was obviously limited by the fact that genes expressed both in the lateral line cells and at the same time in one of the GFP negative cell types, would go undetected. Indeed, as our results show (Additional file 5 ), not all genes previously described as expressed in the lateral line primordium or neuromasts were recovered in our screen. To date, about 372 genes have been characterized as having high levels of mRNA expression (detectable by in situ hybridization) in these organs and thus may have specific roles in development of the lateral line. Our estimates, comparing publicly available data with our results, show that we have detected between half and two thirds of the genes expressed specifically in the lateral line primordium and neuromasts. We speculate that the remaining genes are either expressed simultaneously in other tissues, their expression levels are only slightly elevated in the primordium and neuromasts vs. other cell types, or the appropriate oligonucleotide is absent from the gene array that was used in this study. Despite these caveats, we believe that we have successfully identified a significant fraction of the genes expressed at two different stages of lateral line development: primordium migration and neuromast formation. Moreover, the recovery of genes previously known to be expressed in the LL indicates that the sorting process itself has relatively modest effects on their normal expression profile. Again, this result indicates that the approach followed in this work is able to correctly identify lateral line-specific genes. Despite the evidence described above, it is also likely that a number of genes identified in our screen are not expressed in the lateral line system and represent false positive identifications. These genes could be expressed in some of the other cell types that also express GFP in the cldnb:gfp transgenic line. In the trunk and tail of the transgenic embryos there is detectable expression of the reporter in the pronephros and in skin cells and, given the wide selection window used in the cell sorting procedure and the sensitivity of this technique, there is likely to be contamination of our sorted cell sample with these other cell types. Further exploration of the identified gene collection by in situ hybridization will determine the extent of false positive sequences incorporated in our sample. One strategy we followed for increasing the likelihood of identifying lateral line-specific genes was to consider only those overrepresented in both the primordium and neuromast cell selection experiments (36 hpf and 48 hpf; Supplementary Table 4). Additionally, it may be possible to obtain a superior level of enrichment if a transgenic GFP line with higher expression specificity becomes available. Moreover, a different gene identification approach might be warranted, such as deep sequencing of cDNA tags ( i.e ., SAGE) [ 35 ], which does not rely on arrayed sequences that are inherently biased. This study provides a novel set of candidate genes potentially having roles in primordium organization, migration and lateral line differentiation. Formation of the lateral line system is a highly complex and coordinated process in which cell migration, morphogenesis, neural specification and cell differentiation occur simultaneously. Thus, this system offers a unique opportunity to analyze these biological processes in a simple and accessible manner, and reaffirms the conviction that the study of the general principles governing cell migration and fate determination can be studied using the lateral line of the zebrafish. Our study further opens the door for molecular dissection of these processes by increasing the repertoire of genes that are known to be expressed during lateral line development and relate these genes to diverse biological functions (Table 1 ). For example, genes implicated in cancer progression and metastatic behavior such as matrix metalloproteases ( e.g ., mmp9 and mmp14 ) [ 36 ], the ADAM metalloproteinases ( e.g ., adam9 and adam23 ) [ 37 ] and the tyrosine kinase receptor c-met , were detected in this study. C-met itself is strongly expressed in the migrating primordium [ 38 ] and we also find increased expression of downstream effectors of this pathway such as gab1 , shp2, stat3, pi3k, pak1 and Mitogen-activated protein kinases [ 39 ]. Cell adhesion molecules such as f11r , cd9 , cd99 and members of the tight junction Claudin family genes were also amply represented in our collection. These molecules may be important for maintaining internal tissue organization, coherence of the primordium, pro-neuromast deposition, and/or allowing coordinated multicellular movements [ 8 ]. The collective migration of primordium cells has been the focus of intense study and others have noted the parallels of this cellular behavior with that of invasive cancer cells [ 6 , 8 , 26 ]. In both events, cells undergo an epithelial to mesenchymal transition, acquire migratory behavior, and progress through extracellular matrix to reach a target. Given the cellular and molecular equivalences between both processes (Figure 5 ), it should be possible to analyze in further detail some of the mechanisms implicated in cancer progression using the migratory lateral line primordium. And most importantly, we envision that it should also be possible to carry out screens for drugs that inhibit this process using zebrafish embryos, a model system used widely for in vivo small molecule drug screens [ 40 ]. Another biological process that shows molecular parallels with the migrating primordium is immune cell migration. The involvement of chemokines and their receptors in regulating guidance of the primordium cells is well documented. These molecules were first identified in mammals due to their involvement in homing events in the immune system though lymphocytes migrate singly rather than as ensembles of cells. In addition to the chemokine receptors cxcr4 and cxcr7 , we detected lect1 ( Leukocyte Cell Derived Chemotaxin 1 ), scyba and a sequence with homology to human CXCR6 (Gene ID: AW232037), suggesting that the high level of complexity required for directional migration of the PLL primordium may involve additional molecular components. Interestingly, a number of proteins that regulate the innate immune response were detected in our experiment. Downstream components of the transcription factor NFkB, key regulator of innate immune response, were also expressed in the primordium and/or neuromasts such as Inhibitor of NFkappab ( nfkbiaa ), Interferon Regulatory Factor 5 ( irf5 ), 7 ( irf7 ) and 11 ( irf11 ), and ticam1 (toll-like receptor adaptor molecule 1, previously known as trif ). Likewise, we found members of the prostaglandin family, such as ptgs1, ptgs2 , ptgds and ptgisl , that regulate diverse functions of many cell types of the immune system, including T and B lymphocytes and dendritic cells [ 25 ]. It is not clear why these genes should be expressed specifically in the lateral line. One possibility is that neuromasts require a highly active innate immune system given their exposure on the body surface, and neuromast cells must respond quickly to biological hazards from the environment. Interestingly, several authors have noted a high concentration of macrophages surrounding the neuromasts, possibly part of the same defense mechanism [ 41 ]. In addition, we have detected the expression of members of well known developmental and cellular signaling pathways. Among these is the Wnt pathway, which has been shown to be crucial for primordium organization [ 9 ]. While expression of apc , axin2 , dickopf , frizzled 7a , lef1 and sfrp1a have been described [ 6 , 42 , 43 ], we found new components of this pathway in our survey. Importantly, wnt4b , a Wnt ligand, could be implicated in segregation of the different compartments of this group of cells. Likewise, Fgf signaling has been shown as a key regulator during lateral line development [ 9 ]. The Fgf ligands, fgf3 and fgf10 , guide the tissue compartmentalization in the primordium for proper rosette formation, required for efficient migration [ 4 , 5 ]. Here, we also found new members of the Fgf pathway, such as fgf4, fgf8 and fgf13 , which could be involved in primordium morphogenesis and collective cell movement. Finally, in our survey we found two genes belonging to the Nodal pathway, squint and sox17 , both highly expressed in forerunner cells (FCs) of the zebrafish gastrula [ 44 ]. Nodal/TGFβ signaling has a role in the migration of the FCs and in the subsequent rearrangement into rosette-like epithelial structures during embryogenesis. Therefore, these two genes may be involved in the rosette-like organization of cells to form discrete radial clusters. Notably, the migrating primordium and the deposited proneuromasts contain precursor cells for the diverse cell types that later on differentiate to form a functional sensory organ. Therefore, it is expected that these cells will express neural progenitor cell markers, genes related to neurogenesis, and genes involved in differentiation of the mechanoreceptor hair cells. Most of the known neural progenitor cell markers expressed in the PLL primordium and neuromasts are also expressed in other tissues, including the CNS [ 19 ]. Accordingly, these went undetected in our survey and they include sox2 , sox3 , oct4, pou4f2 and prox1 . However, we did see overexpression of shha and b , a set of genes of the dlx family and members of the msx family, among others. The analysis of genes expressed by neurons that innervate neuromasts or by glial cells that accompany the nerve and neuromasts were beyond the scope of this study. However, they are integral components of the functioning lateral line and contribute to its development and may provide important cues for its migration and survival. For example, glial cells appear to be responsible for inhibiting neuromast formation from cells lying between neuromasts [ 45 ] by producing an inhibitory factor. The sensory neurons located in the PLL ganglion, provide afferent innervation to hair cells and the pioneering growth cone of the lateral line nerve travels together with the primordium as it migrates. The PLL neurons establish somatotopic innervation of hair cells according to their antero-posterior positions through unknown mechanisms. Clearly, there is coordination between the different components and regulatory relationships involving signaling molecules of diverse types must be playing a role in this process. We have found expression of many secreted factors in the primordium and neuromasts, which could play a role in regulation of glial or neuronal behavior. Among these, brain derived neurotrophic factor, bdnf , is expressed in neuromasts and may be crucial in maintaining innervation of hair cells [ 46 ].

Conclusions In this study, we catalogue a variety of biological process and a wide number of signaling pathways involved in lateral line development. Molecules belonging to diverse cellular and developmental processes were uncovered using this approach, demonstrating its utility for the identification of new genes involved in the morphogenesis, migration and cell type specification within the migrating PLL primordium in the zebrafish. Importantly, some of the identified gene products can also open new insights into the understanding of cell migration, which may be crucial to create efficient therapies for a variety of human diseases.

Background Development of the posterior lateral line (PLL) system in zebrafish involves cell migration, proliferation and differentiation of mechanosensory cells. The PLL forms when cranial placodal cells delaminate and become a coherent, migratory primordium that traverses the length of the fish to form this sensory system. As it migrates, the primordium deposits groups of cells called neuromasts, the specialized organs that contain the mechanosensory hair cells. Therefore the primordium provides both a model for studying collective directional cell migration and the differentiation of sensory cells from multipotent progenitor cells. Results Through the combined use of transgenic fish, Fluorescence Activated Cell Sorting and microarray analysis we identified a repertoire of key genes expressed in the migrating primordium and in differentiated neuromasts. We validated the specific expression in the primordium of a subset of the identified sequences by quantitative RT-PCR, and by in situ hybridization. We also show that interfering with the function of two genes, f11r and cd9b , defects in primordium migration are induced. Finally, pathway construction revealed functional relationships among the genes enriched in the migrating cell population. Conclusions Our results demonstrate that this is a robust approach to globally analyze tissue-specific expression and we predict that many of the genes identified in this study will show critical functions in developmental events involving collective cell migration and possibly in pathological situations such as tumor metastasis.

Competing interests The authors declare that they have no competing interests. Authors' contributions VEG designed and performed all the experiments. JL participated in the microarray and pathways analysis. AE carry out the microarray experiments. MB, EJV and VR participated in the design of the study and established the cell sorting conditions. MLA and SMB designed the experiments, contributed to the conclusions and drafted the manuscript. All authors read and approved the final manuscript. Supplementary Material

Acknowledgements We thank Martha Kirby for help with FACS analysis, Catalina Lafourcade and A. Davis for expert fish care and Florencio Espinoza for technical help. The CldnB:lynGFP strain was kindly provided by Darren Gilmour. This work was supported by grants to M.A. from Fondecyt (1070867), ICM (P06-039F) and FONDAP (15090007). V.G. was supported by a Conicyt fellowship and travel fellowships from the Vicerrectoría de Asuntos Académicos, Departamento de Postgrado y Postítulo, Universidad de Chile and MECESUP. This research was supported by the Intramural Research Program of the National Human Genome Research Institute, National Institutes of Health (S.B.).

CC BY

no

2022-01-12 15:21:44

BMC Dev Biol. 2010 Dec 13; 10:120

oa_package/5d/01/PMC3016277.tar.gz

PMC3016278

21144016

Background Mesenchymal stem cells (MSCs) constitute a cell population, which features self-renewal and differentiation into adipocytes, chondrocytes, and osteocytes. Human MSCs have been isolated from various tissues and organs, such as muscle, cartilage, synovium, dental pulp, bone marrow, tonsils, adipose tissues, placenta, umbilical cord, and thymus (reviewed by [ 1 ]). The biological roles of MSCs were initially described by Friedenstein and colleagues in 1970s. They observed bone formation and reconstitution of the hematopoietic microenvironment in rodents with subcutaneously transplanted MSCs (reviewed by [ 2 ]). In addition to providing support for the early stage of hematopoiesis, MSCs have also been reported to suppress the proliferation of CD3 + T-cells [ 3 ], which led to the utilization of MSCs in the management of various pathologic conditions, such as graft-versus-host disease (GvHD) after allogeneic bone marrow transplantation (reviewed by [ 4 - 6 ]). Recent studies have successfully isolated cancer-initiating cells with properties similar to those of MSCs from cases with some neoplasms, such as osteosarcoma [ 7 ], Ewing's sarcoma [ 8 ], and chondrosarcoma [ 9 ]. Furthermore, the characteristics of MSCs isolated from cases with hematopoietic neoplasms have also been investigated. Shalapour et al. [ 10 ] and Menendez et al. [ 11 ] identified the presence of oncogenic fusion transcripts, such as TEL - AML1 , E2A - PBX1 , and MLL rearrangements, in MSCs isolated from cases with B-lineage acute lymphoblastic leukemia (B-ALL). These reports suggested that some leukemias may be derived from the common precursors of both MSCs and hematopoietic stem cells (HSCs). HPB-AML-I has been considered a unique cell line. In spite of its establishment from the peripheral blood mononuclear cells (PBMCs) of a case with acute myeloid leukemia (AML)-M1, this cell line reportedly has the features of spindle-like morphology and plastic adherence [ 12 ]. The detached HPB-AML-I cells were surprisingly capable of proliferating and adhering to plastic surfaces after passage. Immunophenotypic analysis of HPB-AML-I demonstrated the absence of hematopoietic cell-surface antigens and showed that this cell line resembles marrow stromal cells [ 12 ]. Moreover, in the presence of methylisobutylxanthine, hydrocortisone, and indomethacin, but not troglitazone, an increase in the number of lipid droplets was observed in these cells [ 12 ]. In view of these features, we further investigated the possibility of HPB-AML-I being a neoplasm of MSC origin. Recently, some human MSC lines have been established from the bone marrow [ 13 , 14 ] and umbilical cord blood [ 15 ] cells of healthy donors. To establish a stable cell line, genes encoding the human telomerase reverse transcriptase (hTERT), bmi-1, E6, and E7 proteins were transduced to prolong the life span of the healthy donor-originated MSCs [ 13 - 15 ]. However, there have been no reports of the establishment of MSC lines from human bone marrow cells without in vitro gene transduction. Since a number of the characteristics of HPB-AML-I are not typically observed in leukemic cells, we wondered whether HPB-AML-I cells are neoplastic cells originating from the non-hematopoietic compartments of bone marrow, such as MSCs.

Methods Cell lines and cell culture HPB-AML-I cells were kindly provided by Dr. K. Morikawa (Sagami Women's University, Sagamihara, Japan) and 5 × 10 5 of these cells were cultured in 10 ml of RPMI-1640 medium supplemented with L-glutamine (Gibco, Carlsbad, CA), 10% fetal bovine serum (FBS) (Clontech, Mountain View, CA), 50 U/ml of penicillin (Lonza, Walkersville, MD), and 50 μg/ml of streptomycin (Lonza). Cell culture was performed in a T-25 flask and was maintained in a 37°C incubator humidified with 5% CO 2 . Cell passage was performed twice a week. UCBTERT-21, the hTERT -transduced umbilical cord blood mesenchymal stem cell (MSC) line [ 15 ], was obtained from the Japanese Collection of Research Bioresources (JCRB, Osaka, Japan) and propagated in a T-75 flask in a total number of 1.5 × 10 5 cells. Cell culture was maintained in 15 ml of Plusoid-M medium (Med Shirotori, Tokyo, Japan) containing 5 μg/ml of gentamicin (Gibco). The culture medium was replaced twice a week and cell passage was performed when the cultured cells reached 80-90% of confluence. Cytochemical analysis The following cytochemical staining was performed according to the manufacturer's instructions: May Grünwald-Giemsa (Sysmex, Kobe, Japan), myeloperoxidase-Giemsa, toluidine blue, alkaline phosphatase-Safranin O (Muto, Tokyo, Japan), Sudan Black B-hematoxylin, oil red O-hematoxylin (Sigma-Aldrich, St. Louis, MO), and von Kossa-nuclear fast red (Diagnostic BioSystems, Pleasanton, CA). Cytogenetic analysis Cytogenetic analysis was performed according to the standard protocols. The karyotype was determined by G-banding using trypsin and Giemsa (GTG) [ 16 ] to examine 50 cells. The best metaphase was then photographed to determine the karyotype. The specimen was also submitted to spectral karyotyping (SKY)-fluorescence in situ hybridization (FISH) assay according to Ried's method using whole chromosome painting (WCP) libraries (cytocell for WCP) and α-satellite DNA probes [ 17 ]. Cell-surface antigen analysis Flow cytometric analysis was performed by using the following monoclonal antibodies recommended by the International Society for Cellular Therapy (ISCT) (reviewed by [ 2 ]) and monoclonal antibodies used in the study of Wang et al. [ 18 ]: MφP9 (CD14), SJ25C1 (CD19), MAR4 (CD29), 8G12 (CD34), 515 (CD44), 2D1 (CD45), IA10 (CD55), p282 (CD59), AD2 (CD73), 5E10 (CD90), SN6 (CD105), 104D2 (CD117), and L243 (HLA-DR). All of these monoclonal antibodies were obtained from BD Biosciences (San Jose, CA), except for SN6 from Invitrogen (Carlsbad, CA). Cells were resuspended in a total number of 2 × 10 5 in 50 μl of phosphate-buffered saline (PBS) supplemented with 4% FBS, then incubated with 20 μl of monoclonal antibodies, except for 5E10 (2 μl) and SN6 (5 μl), for 45 min at 4°C, and the conjugated cells fixed with 1 ml of 4% paraformaldehyde solution (Wako, Osaka, Japan). Flow cytometric analysis was performed with Cell Quest software and the FACSCalibur device (BD Biosciences) to examine 20,000 events. In vitro differentiation toward adipocytes, chondrocytes, and osteocytes To induce adipogenesis and osteogenesis, 1 × 10 3 cells were cultured in 500 μl of medium in a four-well chamber slide. Three days after propagation, the culture medium was replaced with 500 μl of StemPro adipogenesis or osteogenesis differentiation medium (Gibco) containing 5 μg/ml of gentamicin. Chondrogenesis was induced with a micromass culture system [ 19 , 20 ], in which 5 × 10 2 of the cells were resuspended in 10 μl of culture medium and applied to the center of a culture well. A 96-well culture plate was used in our study. Two hours after propagation, 100 μl of StemPro chondrogenesis differentiation medium containing 5 μg/ml of gentamicin was added. The differentiation medium was replaced twice a week. Mixed lymphocyte culture assay PBMCs were separated from the heparinized peripheral blood of a healthy donor by means of Ficoll-Paque density gradient centrifugation (Amersham Biosciences, Uppsala, Sweden). CD3 + T-cells were purified from PBMCs by magnetic-activated cell sorting (MACS) positive selection (Miltenyi Biotec, Auburn, CA) and 1 × 10 6 of these cells were cultured for 48 h in a 96-well culture plate in the presence of 12.5 μg/ml of phytohemagglutinin (Wako) with or without irradiated (25 Gy) HPB-AML-I and UCBTERT-21 (0, 1 × 10 3 , 1 × 10 4 , and 1 × 10 5 cells/well) cells. From each culture well, 100 μl of cell suspension was pulsed with 10 μl of Cell Counting Kit-8 solution (Dojindo, Tokyo, Japan) at 37°C for 4 h. The optical density at 450 nm was measured to determine cell viability in each of the culture wells.

Results HPB-AML-I shows plastic adherence, negative myeloperoxidase expression, and complex chromosomal abnormalities Inverted microscopic examination (Figure 1A ) and May Grünwald-Giemsa staining (Figure 1B ) of HPB-AML-I cells revealed that this cell line is composed of round-polygonal and spindle-like cells. Unlike the round-polygonal cells, HPB-AML-I cells with the spindle-like morphology attached to plastic surfaces. Since HPB-AML-I was established from a case with AML, we examined this cell line for the presence of myeloperoxidase expression. The human acute promyelocytic leukemia (APL) NB4 cell line was used as positive control in this examination (Figure 1C ). We found that HPB-AML-I was negative for myeloperoxidase expression (Figure 1D ). HPB-AML-I was also subjected to cytogenetic analysis, which demonstrated the presence of a complex karyotype with a modal chromosome number of 64 (range: 57-65; Figure 2A ). A single X chromosome and a number of other abnormalities, mainly consisting of chromosome gains, chromosome losses, translocations, and deletions, were detected by SKY-FISH assay (Figure 2B ). There were no reciprocal chromosomal translocations, which are frequently observed in AML cases. HPB-AML-I expresses cell-surface antigens characteristic for MSCs HPB-AML-I was examined by means of flow cytometric analysis for cell-surface antigens, which are widely used to identify the presence of MSCs. HPB-AML-I expressed CD29, CD44, CD55, CD59, and CD73, but no cell-surface expression of CD14, CD19, CD34, CD90, CD105, CD117, or HLA-DR was detected (Figure 3A ). The cell-surface antigen expression patterns of UCBTERT-21 [ 15 ] and F6 [ 21 ] cell lines and human MSCs isolated from aorta-gonad-mesonephros, yolk sac [ 18 ], bone marrow [ 22 ], and umbilical cord blood [ 23 ] are presented in Table 1 for comparison, showing that there are phenotypic similarities between HPB-AML-I and UCBTERT-21, which was established from human umbilical cord blood and transduced with hTERT . Flow cytometric analysis showed that 11.9% of HPB-AML-I cells expressed CD45 (Figure 3A ). We postulated that the presence of two morphological phases of HPB-AML-I cell line may be related to CD45 expression. For addressing this hypothesis, we performed a prolonged cell culture to increase the confluence, resulting in a morphological change of spindle-like HPB-AML-I cells toward round-polygonal. The round-polygonal cells, which were harvested from a confluent culture with gently washing, but no trypsinization, were positive for CD45 in 25.7% of cells (Figure 3B ). Interestingly, the CD45 expression returned to low positivity (10.1%) after the round-polygonal cells were cultivated for another three days, when they became adherent and spindle-like (Figure 3B ). HPB-AML-I cells are capable of acquiring the properties of adipocytes, chondrocytes, and osteocytes To investigate the multipotency of HPB-AML-I cells, we induced them to differentiate toward adipocytes, chondrocytes, and osteocytes. For comparison, the results of examination of undifferentiated HPB-AML-I cells with an inverted microscope are also shown (Figure 4A ). Two weeks after the induction of adipogenesis, morphological changes were observed in HPB-AML-I cells. The differentiated cells retained the spindle-like morphology or appeared as large polygonal cells. In addition, cytoplasmic vacuoles of various sizes were observed and inverted microscopic examination showed that these vacuoles occurred in solitary or aggregated formations (Figure 4B ). While Sudan Black B and oil red O did not stain the cytoplasm of undifferentiated cells (Figure 4C and 4E ), the cytoplasmic vacuoles of differentiated HPB-AML-I cells were positive for these cytochemical staining (Figure 4D and 4F ), suggesting the presence of lipid accumulation in the adipogenic-differentiated HPB-AML-I cells. Two weeks after the induction of chondrogenesis, the differentiated HPB-AML-I cells showed polygonal morphology, which made them distinct from the undifferentiated cells. Inverted microscopic examination demonstrated the presence of a number of vacuoles in the cytoplasm of differentiated HPB-AML-I cells (Figure 4G ). In contrast to the undifferentiated cells (Figure 4H ), the differentiated HPB-AML-I cells formed lacunae. The proteoglycan-rich extracellular matrix, as indicated by positive toluidine blue staining, surrounded the lacunae (Figure 4I ). The presence of lacunae, as well as extracellular proteoglycan accumulation, suggested that the micromass of chondrogenic-differentiated HPB-AML-I cells acquires the properties of a cartilage. Inverted microscopic examination three weeks after the induction of osteogenesis demonstrated the presence of a number of cell processes and an eccentrically located nucleus in the differentiated HPB-AML-I cells (Figure 4J ). The undifferentiated cells did not express alkaline phosphatase as shown by negative cytochemical staining for this protein (Figure 4K ). On the other hand, cytochemical staining resulted in positive staining for alkaline phosphatase in the cytoplasm of differentiated HPB-AML-I cells (Figure 4L ). Moreover, the differentiated HPB-AML-I cells also secreted calcium, which constitutes the extracellular matrix of the bone, as shown by von Kossa staining (Figure 4M and 4N ). These two findings suggested the acquisition of osteogenic characteristics by HPB-AML-I cells following the induction of osteogenesis. Inhibition of CD3 + T-cell proliferation in the presence of HPB-AML-I cells CD3 + T-cells obtained from peripheral blood were cultured with or without HPB-AML-I cells. The XTT absorbance levels at 450 nm, which show the viability of CD3 + T-cells, decreased in a dose-dependent manner similar to those of UCBTERT-21 (Figure 5 ). These findings suggested that HPB-AML-I cells dose-dependently suppress the antigen-driven proliferation of CD3 + T-cells, which is also characteristic of MSCs.

Discussion Even though HPB-AML-I was established from the PBMCs of an AML-M1 case [ 12 ], this cell line presents distinctive morphological features from AML. In terms of cell-surface antigen expression, multilineage differentiation, and CD3 + T-cell suppression, the characteristics of HPB-AML-I were found to be similar to those of MSCs. Our findings presented here suggest that HPB-AML-I may be a neoplastic cell line with MSC properties. Few reports have dealt with the establishment of human neoplastic MSC lines. A previous study established F6, a human neoplastic MSC line, from embryonic bone marrow MSCs. Transplantation of F6 cells into the SCID-nude mice resulted in fibrosarcoma formation and tissue metastasis [ 21 , 24 ]. To the best of our knowledge, however, HPB-AML-I is the first neoplastic MSC line derived from a leukemic case. The appearance of HPB-AML-I cells in suspension phase with their round-polygonal morphology intrigued us. We observed that an increase in the population of HPB-AML-I cells with such morphological patterns occurs in conjunction with the increased confluence of cultured cells. Morphological changes during culturing have previously been described in the case of bone marrow MSCs. Choi et al. [ 25 ] reported that the morphology of bone marrow MSCs changed from small spindle-like in the first passage to large polygonal in the later passages. In contrast to many other adherent cell lines, HPB-AML-I cells with their round-polygonal morphology were viable and capable of proliferating and adhering to plastic surfaces following cell passage. Similar findings have been reported for the F6 cell line [ 21 ]. While the exact mechanisms remain to be elucidated, we speculate that the loss of adherent capacity after confluent condition may be a pivotal property to neoplasms originated from mesenchymal stem cells. Flow cytometric analysis of HPB-AML-I disclosed that, based on ISCT criteria, the cell-surface antigen expression patterns of this cell line were similar to those of human MSCs (reviewed by [ 2 ]) with positive CD73 and negative CD14, CD19, CD34, CD45 and HLA-DR expression. However, contrary to those criteria (reviewed by [ 2 ]), HPB-AML-I did not express CD90 and CD105. Absence of CD90 expression has also been observed in UCBTERT-21 [ 15 ] and in human MSCs obtained from umbilical cord blood [ 15 , 26 ]. MSCs lacking CD105 expression have been reported by Jiang et al. [ 27 ] and Ishimura et al. [ 28 ], who isolated MSCs from the subcutaneous adipose tissue, and by Lopez-Villar et al. [ 29 ], who extracted MSCs from the bone marrow of a myelodysplastic syndrome case. These reports suggested that the absence of CD90 and CD105 expression in HPB-AML-I does not necessarily exclude the possibility that this cell line is derived from MSCs. The differentiation capability of MSCs with a negative CD105 expression has been investigated by Jiang et al. [ 27 ] and Ishimura et al. [ 28 ]. They found that this population of MSCs, while showing adipogenic differentiation, lacked chondrogenic and osteogenic differentiation. It is interesting that HPB-AML-I could differentiate into three lineages despite of CD105 negativity. In addition, a subpopulation of HPB-AML-I expressed CD45, even though most of HPB-AML-I cells were negative for CD45. Generally, CD45 is negative in MSCs, but CD45 expression has been detected in bone marrow MSCs from cases with multiple myeloma [ 30 , 31 ]. It is therefore not surprising that neoplastic MSC line, such as HPB-AML-I, shows the aberrant expression of this antigen. Interestingly, CD45 expression in HPB-AML-I cells is likely to be transient, as the expression levels of CD45 increased in round-polygonal cells in the confluent cell culture and they decreased after passage of round-polygonal cells. Normal cells are known to have the property of contact inhibition, which is lost in transformed cells. Therefore, cell-to-cell contact might induce the aberrant expression of CD45 with an unknown reason in HPB-AML-I cells. By using inverted microscopic examination and cytochemical staining, we demonstrated that HPB-AML-I cells are able to acquire the properties of adipocytes, chondrocytes, and osteocytes. The capability of MSCs to differentiate toward mesenchymal lineage cells reportedly correlates with their morphological and cell-surface antigen expression patterns. Chang et al. [ 26 ] demonstrated that MSCs isolated from human umbilical cord blood consisted of cells with a flattened or spindle-like morphology and that the capability of differentiating toward adipocytes of the spindle-like MSCs was superior than that of the flattened cells. Since such heterogeneous morphology is shared by HPB-AML-I, further analyses are needed to characterize the difference between the round-polygonal and spindle-like cells. As also reported by previous studies of the immunomodulatory effects on MSCs [ 18 , 32 ], we demonstrated that HPB-AML-I cells are capable of suppressing CD3 + T-cell proliferation. Similar studies have been performed on MSCs isolated from cases with various hematopoietic neoplasms, such as ALL, Hodgkin's disease, non-Hodgkin's lymphoma, myelodysplastic syndrome, AML [ 33 ], and chronic myeloid leukemia (CML) [ 34 ]. In contrast to our results, Zhi-Gang et al . reported that bone marrow MSCs isolated from AML cases did not inhibit the proliferation of CD3 + T-cells [ 33 ]. These findings suggest that bone marrow MSCs from cases with hematopoietic neoplasms may or may not be capable of inhibiting CD3 + T-cell proliferation as a consequence of the secretion of humoral factors by neoplastic cells or the direct interaction with them. It is therefore very interesting that HPB-AML-I, regardless of its HSC or MSC origin, maintains the capability of inhibiting T-cell proliferation even after neoplastic transformation. The cytogenetic analysis revealed the presence of complex chromosomal abnormalities in HPB-AML-I, although these were not the same as the frequently observed chromosomal alterations in AML cases. While it is not fully understood whether MSCs isolated from leukemic cases carry the cytogenetic characteristics common to leukemic cells, previous studies reported the absence of t(9;22)(q34;q11) chromosomal translocation or BCR - ABL rearrangement in bone marrow MSCs obtained from cases with Philadelphia (Ph) chromosome-positive CML [ 35 , 36 ]. On the other hand, a recent study demonstrated the presence of leukemic reciprocal translocation and fusion gene expression in bone marrow MSCs of MLL - AF4 -positive B-ALL cases [ 11 ]. However, monoclonal Ig gene rearrangements, uncontrolled cell proliferation, diminished cell apoptosis, and cell-cycle arrest characteristic of leukemic cells were not observed in the bone marrow MSCs of those cases [ 11 ]. Unfortunately, we could not obtain the karyotype of the original leukemic cells. Therefore, the complex karyotype in HPB-AML-I may not correspond to the cytogenetic status of the primary cells. It is possible that the complex karyotype of HPB-AML-I may include the additional genetic changes, which occurred in vitro during and after the establishment of the cell line. Nevertheless, the MSC-like properties of HPB-AML-I, as shown in this study, suggest the possibility that the first genetic event might have occurred at the stage of MSC.

Conclusions In summary, we were able to demonstrate that HPB-AML-I has morphological, cytochemical, and phenotypic features, as well as the capability of differentiating toward mesenchymal lineage cells and of suppressing CD3 + T-cell proliferation, which are all characteristic of MSCs. Our findings suggest that HPB-AML-I cells may represent a unique neoplastic cell line derived from bone marrow MSCs. We believe that this cell line will make an important contribution to a better understanding of the neoplastic transformation of bone marrow-derived constituents.

Background In spite of its establishment from the peripheral blood of a case with acute myeloid leukemia (AML)-M1, HPB-AML-I shows plastic adherence with spindle-like morphology. In addition, lipid droplets can be induced in HPB-AML-I cells by methylisobutylxanthine, hydrocortisone, and indomethacin. These findings suggest that HPB-AML-I is similar to mesenchymal stem cells (MSCs) or mesenchymal stromal cells rather than to hematopoietic cells. Methods To examine this possibility, we characterized HPB-AML-I by performing cytochemical, cytogenetic, and phenotypic analyses, induction of differentiation toward mesenchymal lineage cells, and mixed lymphocyte culture analysis. Results HPB-AML-I proved to be negative for myeloperoxidase, while surface antigen analysis disclosed that it was positive for MSC-related antigens, such as CD29, CD44, CD55, CD59, and CD73, but not for CD14, CD19, CD34, CD45, CD90, CD105, CD117, and HLA-DR. Karyotypic analysis showed the presence of complicated abnormalities, but no reciprocal translocations typically detected in AML cases. Following the induction of differentiation toward adipocytes, chondrocytes, and osteocytes, HPB-AML-I cells showed, in conjunction with extracellular matrix formation, lipid accumulation, proteoglycan synthesis, and alkaline phosphatase expression. Mixed lymphocyte culture demonstrated that CD3 + T-cell proliferation was suppressed in the presence of HPB-AML-I cells. Conclusions We conclude that HPB-AML-I cells appear to be unique neoplastic cells, which may be derived from MSCs, but are not hematopoietic progenitor cells.

List of abbreviations ALL: acute lymphoblastic leukemia; AML: acute myeloid leukemia; APL: acute promyelocytic leukemia; CML: chronic myeloid leukemia; GvHD: graft-versus-host disease; FBS: fetal bovine serum; FISH: fluorescence in situ hybridization; GTG: G-banding using trypsin and Giemsa; HSC(s): hematopoietic stem cell(s); hTERT: human telomerase reverse transcriptase; ISCT: International Society for Cellular Therapy; MACS: magnetic-activated cell sorting, MSC(s): mesenchymal stem cell(s); PBMC(s): peripheral blood mononuclear cell(s); PBS: phosphate-buffered saline; SKY: spectral karyotyping; WCP: whole chromosome painting. Competing interests The authors declare that they have no competing interests. Authors' contributions BA, TS, and SK1 contributed to the experimental design, data acquisition and analyses, and manuscript preparation. SK2 contributed to the mixed lymphocyte culture analyses. SNAJ and CK contributed to the differentiation asssay. ET and KM contributed to the karyotypic analyses. SK3 and YH contributed to the data analysis and discussion. All authors read and approved the final manuscript.

Acknowledgements The authors wish to thank Ms. Shino Tanaka for her technical assistance and Mr. Jan K Visscher for proofreading and editing the manuscript. Bambang Ardianto is supported by a Japanese Government Scholarship for Graduate Students under the supervision of Professor Yoshitake Hayashi.

CC BY

no

2022-01-12 15:21:44

J Exp Clin Cancer Res. 2010 Dec 13; 29(1):163

oa_package/b3/9c/PMC3016278.tar.gz

PMC3016280

21129202

Introduction Homologous recombination (HR) is a conserved biological process in which DNA strands are physically exchanged between DNA molecules of identical or nearly identical sequence (Figure 1 ). The DNA strand exchange mechanism in HR allows gene conversion events to occur, which is important for maintaining genetic diversity within populations of organisms. The DNA strand exchange mechanism in HR is also essential for the high-fidelity repair of DNA double-strand breaks (DSBs) and daughter-strand gaps, which is important for maintaining genome stability [ 1 - 3 ]. These homology-directed DNA repair (HDR) processes require the coordination of activities between HR and DNA replication machineries. Homologous recombination in bacteriophage T4 The bacteriophage T4 recombination system provides an important model for understanding recombination transactions including DNA strand exchange, recombination-dependent replication (RDR), and homology-directed DNA repair (HDR) [ 4 - 6 ]. The relatively simple, but functionally conserved, core recombination machinery of T4 facilitates detailed mechanistic studies of DNA strand exchange reactions and intermediates. The T4 paradigm for presynaptic filament assembly is widely used as a basis for studying presynaptic filaments in many cellular organisms including humans. At the same time, because of the close linkages between its DNA recombination, replication, and repair pathways, bacteriophage T4 has yielded novel insights on the cross-talk that occurs between recombination and replication proteins. This is especially true in the case of T4 DNA helicases, which are seen as critical for the channeling of recombination intermediates into RDR and HDR pathways. Single-stranded DNA and presynaptic filaments The generation of single-stranded DNA is a common early step of HR pathways [ 7 , 8 ]. ssDNA production typically occurs as a result of nucleolytic resection of DSBs (Figure 1 ), or due to replication fork stalling or collapse. In T4 recombination, exonuclease activities of a Gp46/Gp47 complex (orthologous to eukaryotic Mre11/Rad50) appear to be critical for DSB resection [ 9 ]. In addition to DNA damage-linked production of ssDNA, bacteriophage T4 routinely generates ssDNA during replication of its linear chromosome ends. The production of ssDNA tails or gaps in otherwise duplex DNA allows the assembly of core recombination machinery including presynaptic filaments on ssDNA. Presynaptic filaments are helical nucleoprotein filaments consisting of a recombinase enzyme and its accessory proteins bound cooperatively to ssDNA (Figure 2 ). Presynaptic filament assembly activates the enzymatic activities of the recombinase including ATPase and DNA strand exchange activities. Filament dynamics controls DNA strand exchange and its coupling to the downstream, replicative steps of HDR. These processes require the timely assembly of presynaptic filaments on recombinagenic ssDNA. Equally important is the coordinated disassembly or translocation of filaments, which appears to be required to make way for the assembly of replication enzymes on recombination intermediates [ 10 , 11 ]. The transition from recombination to DNA replication and repair The transition from recombination intermediate to replication fork occurs very efficiently in bacteriophage T4, which has evolved to use this as a major mode of DNA replication initiation. The transition likely involves not only the built-in dynamics of the presynaptic filament but also the coordinated activities of DNA helicases. In the following sections of this chapter, we will review what is known about presynaptic filament dynamics in the T4 system, as well as what is known about the influences of DNA helicases on recombination, and how these two ATP-driven machines may cooperate with each other to successfully couple HR to recombination-dependent replication and repair.

Conclusions Studies of the T4 recombination system have provided insights on recombination mechanisms that are highly relevant to HR and HDR processes in cellular organisms including eukaryotes. Work with T4 UvsY protein has helped to define the roles that recombination mediator proteins play in promoting presynaptic filament assembly and in the trafficking of recombination proteins (SSB, RMP, and recombinase) on ssDNA that occurs during the early stages of recombination and homology-directed DNA repair processes. It is clear that the UvsY model for assembly of recombinase filaments on ssDNA covered with ssDNA-binding protein is highly conserved [ 24 ], including in human beings where at least three classes of proteins with UvsY-like mediator activity participate in genome stability pathways. These include Rad52, the human Rad51 paralogs Rad51B, Rad51C, Rad51D, Xrcc2, and Xrcc3, and breast cancer susceptibility gene Brca2 [ 97 - 100 ]. Details of T4 presynaptic filament assembly and dynamics, such as ssDNA hand-offs and dynamic instability, suggest mechanisms that may be used by recombination machineries in many organisms to capture recombinagenic ssDNA, perform strand exchange, and pass the intermediates on to other repair enzymes such as the replicative components of HDR pathways. Recent biochemical and structural studies of UvsX recombinase shed light on its mechanism and relationship to other recombinases of the RecA/Rad51 superfamily. The observation that ssDNA-binding by UvsX allosterically regulates the enzyme's affinity for homologous vs. non-homologous dsDNA at a second site is an important breakthrough [ 66 ]. The sensitive fluorescence assay developed for this study represents an excellent opportunity to explore how micro-heterology affects homologous pairing, as well as the similarities and differences between pairing mechanisms used by recombinases from various organisms. The X-ray crystal structure of UvsX and its modeling in EM filament structures shows that UvsX shares the same extended filament structure in its active form as do E. coli and yeast filament structures (Gajewski, S., M.R. Webb, V. Galkin, E.H. Egelman, K.N. Kreuzer, and S.W. White: Crystal structure of the phage T4 recombinase UvsX and its functional interation with the T4 SF2 helicase UvsW, unpublished). The observation that UvsX appears to share the lysine bridges found at the active site of E. coli RecA-DNA places UvsX mechanistically closer to prokaryotic than to eukaryotic recombinases, at least in this detail. Opportunities for structure-driven mutagenesis and mechanistic studies, as well as for evolutionary studies, of UvsX will surely follow from this important structure. The T4 field pioneered studies of helicases in recombination, which are now known to be pervasive regulators of recombination and HDR metabolism in all organisms [ 100 ]. The biochemistry of T4 helicases demonstrates the diverse ways that these enzymes can influence recombination outcomes, including both positive and negative regulation of homologous pairing and strand exchange. It is noteworthy that T4 encodes threes different helicases on its phage genome that appear to have both unique and overlapping functions in recombination. Of particular relevance is the role of helicases in channeling strand exchange reactions toward the formation of intermediates that can serve as initiators of recombination-dependent DNA replication [ 4 , 6 , 96 ]. T4 RDR requires either Dda (for bubble-migration DNA synthesis) or Gp41/Gp59 (for semi-conservative DNA synthesis) to initiate replication via a recombination event. The biochemical role of UvsW in the RDR machine remains to be elucidated but is likely to be central given its ability to promote extensive branch migration. The coupling of recombination to replication is fundamental for DNA repair and genome stability in all organisms. Eukaryotic DNA helicases/translocases such as Rad54, Srs2 and others are known to play important roles in processing recombination intermediates, either for regulatory purposes or to facilitate access of downstream DNA replication and repair enzymes to the products of strand exchange [ 10 , 11 , 72 - 74 , 100 ]. The T4 helicases offer an excellent opportunity to study more about the mechanism of recombination/replication coupling, the findings of which will directly inform studies of genome stability mechanisms in cellular organisms including humans.

Homologous recombination (HR), a process involving the physical exchange of strands between homologous or nearly homologous DNA molecules, is critical for maintaining the genetic diversity and genome stability of species. Bacteriophage T4 is one of the classic systems for studies of homologous recombination. T4 uses HR for high-frequency genetic exchanges, for homology-directed DNA repair (HDR) processes including DNA double-strand break repair, and for the initiation of DNA replication (RDR). T4 recombination proteins are expressed at high levels during T4 infection in E. coli , and share strong sequence, structural, and/or functional conservation with their counterparts in cellular organisms. Biochemical studies of T4 recombination have provided key insights on DNA strand exchange mechanisms, on the structure and function of recombination proteins, and on the coordination of recombination and DNA synthesis activities during RDR and HDR. Recent years have seen the development of detailed biochemical models for the assembly and dynamics of presynaptic filaments in the T4 recombination system, for the atomic structure of T4 UvsX recombinase, and for the roles of DNA helicases in T4 recombination. The goal of this chapter is to review these recent advances and their implications for HR and HDR mechanisms in all organisms.

Properties of the T4 Core Recombination Machinery Although relatively simple, the core activities of the T4 recombination system are highly conserved. Three core protein components are required for T4 presynaptic filament assembly and for DNA strand exchange under physiological conditions: UvsX, the phage recombinase (orthologous to bacterial RecA and eukaryotic Rad51); Gp32, the phage ssDNA-binding protein (equivalent to bacterial SSB and eukaryotic RPA); and UvsY, the phage recombination mediator protein (equivalent to bacterial RecOR, eukaryotic Rad52, Brca2, and others) [ 4 , 5 ]. The DNA binding properties of UvsX, Gp32, and UvsY are presented below in context with their physical and enzymatic properties. UvsX recombinase UvsX protein (44 kDa) is a member of the RecA/Rad51 recombinase family and shares 28% sequence identity and 51% sequence similarity with the catalytic core domain of E. coli RecA [ 12 ]. UvsX catalyzes DNA strand exchange reactions that play central roles in T4 HR, RDR, and HDR pathways [ 4 , 6 ]. UvsX binds sequence-non-specifically to both ssDNA and dsDNA and can bind to both lattices simultaneously via two different binding sites (Maher, R.L. and S.W. Morrical: Coordinated binding of ssDNA and dsDNA substrates by UvsX recombinase and its regulation by ATP, unpublished). UvsX has higher affinity for dsDNA in the absence of other factors, but simultaneous ssDNA binding lowers UvsX-dsDNA binding affinity unless the duplex sequence is homologous to the bound ssDNA (Maher, R.L. and S.W. Morrical: Coordinated binding of ssDNA and dsDNA substrates by UvsX recombinase and its regulation by ATP, unpublished). At the same time, UvsX-ssDNA interactions are selectively stabilized by nucleoside triphosphates ATP, dATP, or their non-hydrolyzable analogs, and by UvsY protein [ 13 , 14 ]. These combined factors help to target UvsX filament assembly onto recombinagenic ssDNA even in the presence of excess dsDNA as would normally be found in the T4-infected cell. Binding of UvsX to ssDNA, not dsDNA, specifically activates catalysis by UvsX including ATPase and DNA strand exchange activities. Quantitative binding studies established the intrinsic ssDNA-binding parameters of UvsX [ 13 ]. Its average binding site size on ssDNA is 4 nucleotide residues per protomer. UvsX exhibits moderate affinity and cooperativity for ssDNA with K obs = K ω ≈ 10 6 M -1 at physiological ionic strength, where the cooperativity parameter ω ≈ 100 [ 13 ]. The observed cooperativity of UvsX is consistent with the formation of long filaments on ssDNA at high binding density. The ATPase activity of UvsX is strongly ssDNA-dependent under normal solution conditions [ 15 ], although very high salt concentrations can also stimulate ATP hydrolysis by UvsX in the absence of ssDNA. Double-stranded DNA does not activate UvsX ATPase activity. UvsX ATPase activity is also highly unusual in that it generates both ADP and AMP as products [ 15 , 16 ]. The two products appear to be generated independently by two different classes of active sites within UvsX-ssDNA presynaptic filaments, as indicated by results of steady-state kinetics studies [ 16 ]. These sites have different K m and k cat / K m values for the ATP and ssDNA substrates. One type of active site appears to produce ADP exclusively, while the other appears to generate AMP via a sequential mechanism (ATP → ADP → AMP) without releasing the ADP intermediate from the active site [ 16 ]. Thus UvsX presynaptic filaments exhibit active site asymmetry (Figure 2 ). This asymmetry may be important for UvsX-catalyzed DNA strand exchange reactions, since increases in ADP/AMP product ratio observed in UvsX site-directed mutants correlate inversely with strand exchange activity [ 16 ]. Active site asymmetry may be a general property of presynaptic filaments in many species, since evidence exists for two classes of active sites in filaments of E. coli RecA and S. cerevisiae Rad51 recombinases [ 17 , 18 ]. UvsX-ssDNA filaments rapidly search for homology in dsDNA substrates, leading to efficient homologous pairing and strand exchange. ATP binding (not hydrolysis) is required for homologous pairing, however ATP hydrolysis is needed to drive extensive polar (5' → 3') branch migration during strand exchange [ 19 - 21 ]. There is a strong requirement for Gp32 to stimulate UvsX-catalyzed strand exchange at normal concentrations of the recombinase [ 15 , 22 , 23 ]. In vitro, this Gp32 requirement can be circumvented by raising the UvsX concentration to super-saturating levels with respect to ssDNA binding sites. Stimulation of strand exchange by Gp32 requires the correct order of protein addition: Adding Gp32 to ssDNA prior to the addition of UvsX typically inhibits strand exchange. This ssDNA-binding protein/recombinase order of addition effect is a characteristic of all well-characterized recombination systems [ 24 ], and is reflective of the competition between the two proteins for binding sites on ssDNA. Similar inhibition of UvsX-catalyzed strand exchange is seen at high concentrations of Gp32 and/or at elevated salt concentrations, i.e. conditions that favor Gp32-ssDNA over UvsX-ssDNA interactions. Under conditions such as these there is an absolute requirement for the UvsY recombination mediator protein for strand exchange reactions in vitro [ 23 , 25 ]. This mimics the in vivo situation in which T4 recombination transactions are equally dependent on UvsX and UvsY [ 26 - 28 ]. Branched networks of single- and double-stranded DNA are the major products of UvsX-catalyzed DNA strand exchange, indicating that each DNA substrate molecule participates in many homologous pairing events [ 15 , 29 ]. One plausible explanation for this behavior is that UvsX appears to catalyze homologous pairing much more rapidly than branch migration. Therefore it is possible for different regions of one long ssDNA substrate to pair with homologous regions of different dsDNA substrates before any of the resulting D-loop intermediates can be completely extended into heteroduplex DNA. Rapid homologous pairing by UvsX may be an evolutionary adaptation for efficiently capturing 3' ssDNA tails and using them to prime recombination-dependent replication. Furthermore, branch migration appears to be dependent on T4-encoded DNA helicases, as we discuss in a later section. Gp32 ssDNA-binding protein Gp32 (34 kDa) is the prototype ssDNA-binding protein and a key component of the T4 replisome. Gp32 also plays important roles in homologous recombination and DNA repair. The biochemical properties of Gp32 have been thoroughly characterized [ 30 - 45 ], and the atomic structure of its central DNA-binding domain (DBD) has been solved [ 32 ]. The DBD contains an oligonucleotide/oligosaccharide-binding (OB)-fold motif plus a structural Zn ++ atom. An N-terminal domain (so-called basic or "B-domain") is required for self-association and cooperativity, whereas a C-terminal domain (so-called acidic or "A-domain") is the site for protein-protein interactions with various recombination and replication enzymes including UvsX and UvsY. Gp32 binds sequence-non-specifically to polynucleotides, with the highest observed affinity for ssDNA ( K obs ≈ 10 9 M -1 at physiological ionic strength), moderate affinity for single-stranded RNA, and very low affinity for dsDNA. The binding site size of Gp32 on ssDNA is approximately 7 nucleotide residues. Binding to ssDNA is highly cooperative (ω ≈ 1000), meaning that Gp32 exists almost exclusively in clusters or long filaments on ssDNA at protein concentrations normally encountered in in vitro DNA strand exchange assays as well as in vivo. Gp32 affects both pre- and post-synaptic steps of UvsX-catalyzed DNA strand exchange reactions [ 15 , 22 , 23 , 25 , 46 , 47 ]. An important function of Gp32 in presynapsis is to denature secondary structure in the ssDNA substrate, which eventually allows UvsX to saturate the ssDNA by forming long presynaptic filaments. Paradoxically, the immediate effect of Gp32 on UvsX-ssDNA filament formation is negative under physiological conditions, because Gp32 competes effectively with UvsX for binding sites [ 13 ]. Overcoming Gp32 inhibition requires either pre-incubation of UvsX with ssDNA in the presence of ATP (the previously mentioned order of addition effect), or the inclusion of UvsY in reaction mixtures (see below) [ 4 , 24 ]. Gp32 has also been shown to play a post-synaptic role in strand exchange, stimulating the reaction by sequestering the outgoing ssDNA strand that is displaced during D-loop formation and subsequent branch migration [ 47 ]. UvsY recombination mediator protein UvsY is the prototype recombination mediator protein or RMP [ 24 ]. By definition, RMPs are proteins that load recombinases of the RecA/Rad51 family onto ssDNA molecules that are pre-saturated with cognate ssDNA-binding protein. UvsY is absolutely required for UvsX-catalyzed DNA strand exchange in the presence of Gp32 under physiological or high-salt conditions [ 22 , 48 , 49 ]. In vivo, UvsY is also absolutely required for UvsX-dependent recombination since mutations knocking out either gene product have equivalent recombination-deficient phenotypes including the small-plaque phenotype associated with defective RDR [ 26 - 28 ]. UvsY is the only member of the core T4 recombination machinery that forms a discreet oligomeric structure: It exists as a stable hexamer of identical 15.8 kDa subunits in solution, and binds to ssDNA in this form [ 50 ]. UvsY binds to both ssDNA and dsDNA, but has a much higher affinity for the former under relaxed DNA conditions [ 51 ]. The preference of UvsY for ssDNA may be an important factor in directing UvsX filament assembly onto ssDNA in the presence of excess dsDNA, since UvsX itself has a relatively high affinity for non-homologous dsDNA (Maher, R.L. and S.W. Morrical: Coordinated binding of ssDNA and dsDNA substrates by UvsX recombinase and its regulation by ATP, unpublished). UvsY has a binding site size on ssDNA of 4 nucleotide residues per protomer, or 24 nucleotide residues per hexamer [ 52 ]. The protomeric binding site sizes of UvsY and UvsX are identical. UvsY binds to ssDNA with high affinity ( K - obs ≈ 10 7 M -1 at physiological ionic strength), but with little or no cooperativity (ω ≈ 1). Therefore UvsY has higher intrinsic affinity, but lower cooperativity, for ssDNA than either UvsX or Gp32 under conditions that are relevant for strand exchange reactions in vitro and in vivo. UvsY-ssDNA interactions are weakened by mutations at residues Lys-58 and Arg-60, which form part of a conserved LKARLDY motif (so-called 'KARL' motif) found in the N-terminal domain of UvsY, which is thought to comprise part of its DNA binding surface [ 14 , 48 , 51 , 53 , 54 ]. The KARL motif is also found in certain DNA helicases, however no helicase activity has ever been associated with UvsY, which lacks a motor domain. The C-terminal domain of UvsY is essential for hexamerization. Deletion of this domain drastically reduces the affinity of UvsY-ssDNA interactions, demonstrating the importance of UvsY hexamers as the relevant ssDNA-binding unit [ 55 ]. Several lines of evidence indicate that UvsY hexamers have the ability to wrap ssDNA strands around themselves, and that wrapping is responsible for the high affinity of UvsY-ssDNA interactions. Evidence includes the observation that a C-terminally deleted, monomeric form of UvsY has 10 4 -fold lower affinity for ssDNA than wild-type [ 55 ]. The wrapping hypothesis is supported by the finding that mutiple subunits within each UvsY hexamer are in contact with ssDNA [ 51 ]. Other evidence comes from results of single-molecule DNA stretching studies, which showed that the ssDNA that is created by the treatment of individual stretched dsDNA molecules with glyoxal is strongly wrapped by UvsY [ 54 ]. Wrapping of ssDNA occurs at low stretching forces where the DNA is relatively relaxed. At high stretching forces, where the DNA is under tension, wrapping is suppressed. The tension-dependent suppression of wrapping leads to the loss of preferential binding to ssDNA as shown by the fact that UvsY binds tighter to stretched dsDNA than to stretched ssDNA [ 54 ]. This contrasts with the observation that UvsY has ~1000-fold higher affinity for ssDNA than for dsDNA under relaxed conditions [ 51 ]. Therefore high-affinity binding of UvsY to ssDNA requires wrapping, which also imposes a preference for binding to ssDNA over dsDNA. Presumably UvsY cannot wrap dsDNA because its persistence length is much higher than that of ssDNA [ 56 ]. The surprising observation that UvsY binds tightly to stretched dsDNA could have important implications for presynaptic filament assembly. The binding of Gp32 to ssDNA creates an extended or "stiff" DNA conformation that might be recognized by UvsY in an unwrapped mode similar to its interaction with stretched dsDNA. Converting this extended ssDNA structure into a wrapped one might be an important step in the recruitment of UvsX recombinase, as we discuss in a later section. UvsY is absolutely required for UvsX-catalyzed DNA strand exchange assays performed under physiological conditions of Gp32 and salt [ 4 , 24 ], consistent with the co-dependency of recombination on UvsX and UvsY in vivo [ 26 - 28 ]. In vitro, UvsY lowers the critical concentration of UvsX for RDR and other recombination reactions [ 46 , 57 ]. UvsY stimulates the ssDNA-dependent ATPase activity of UvsX, possibly by acting as a nucleotide exchange factor for the recombinase [ 58 ]. The greatest stimulation of ATPase activity is seen when UvsY and Gp32 act together synergistically on the reaction [ 23 , 49 ]. UvsY stimulates the catalytic activities of UvsX mainly by promoting presynaptic filament assembly. The mechanism of UvsY's recombination mediator activity will be explored in greater detail below. Assembly and Dynamics of the T4 Presynaptic Filament Regulation of UvsX-ssDNA interactions by the ATPase cycle Like all RecA/Rad51 recombinases, UvsX is a member of the AAA + ATPase super-family and its interactions with ssDNA are regulated by ATP binding and hydrolysis. The analog ATPγS, which is tightly bound but slowly hydrolyzed by UvsX, induces a stable, high-affinity ssDNA binding state of the enzyme [ 13 , 14 ]. ATP itself transiently induces high-affinity ssDNA binding by UvsX until it is hydrolyzed to ADP or AMP [ 15 , 16 ]. Both of these hydrolytic products are associated with decreased ssDNA-binding affinity states of UvsX under steady-state conditions [ 16 ]. Regulation of protein-ssDNA interactions by UvsY Most evidence indicates that UvsX and Gp32 undergo mutually exclusive binding to ssDNA [ 48 , 59 , 60 ]. On the other hand there is overwhelming evidence that UvsY can co-occupy ssDNA binding sites simultaneously with either UvsX or Gp32 [ 14 , 19 , 25 , 60 - 62 ]. The interaction of UvsY with either Gp32-ssDNA or UvsX-ssDNA complexes alters the properties of both in ways that favor presynaptic filament formation and the activation of UvsX catalytic activities. UvsY forms a stable tripartite complex with Gp32 and ssDNA at physiologically relevant salt conditions [ 61 ]. These complexes contain stoichiometric amounts of both UvsY and Gp32 with respect to their normal binding site sizes on ssDNA (Figure 2 ). Gp32-ssDNA interactions are destabilized within the UvsY-Gp32-ssDNA complex as shown by their increased sensitivity to disruption by salt compared to Gp32-ssDNA complexes in the absence of UvsY [ 61 ]. Results of single-molecule DNA stretching studies confirm that UvsY destabilizes Gp32-DNA interactions [ 54 ]. It has been proposed that, since cooperativity is such a large component of K obs for Gp32-ssDNA interactions, UvsY could destabilize Gp32-ssDNA by lowering Gp32's cooperativity parameter [ 61 ]. This is probably the major pathway for destabilizing Gp32-ssDNA under physiological or high-salt conditions. It has also been proposed, based on results of single-molecule DNA stretching experiments, that UvsY directly displaces Gp32 from ssDNA under low-salt conditions [ 54 ]. In either case, the destabilization of Gp32-ssDNA interactions by UvsY lowers the energy barrier necessary for UvsX to displace Gp32 from ssDNA, which is necessary for nucleation and propagation of presynaptic filaments on ssDNA that is pre-saturated with Gp32 (as is likely to be the case in vivo). Biochemical studies demonstrate that UvsY stabilizes UvsX-ssDNA interactions [ 14 ]. UvsY, UvsX, and ssDNA form a tripartite complex with a stoichiometry of ~1 UvsY hexamer per 6 UvsX protomers, consistent with their equivalent binding site sizes (4 nucleotide residues/protomer). The increased stability of UvsX-ssDNA interactions within these complexes is demonstrated by their higher resistance to salt compared to filaments formed in the absence of UvsY. The most stable complex is formed when UvsY and ATPγS are both present, indicating that the RMP and nucleoside triphosphate act synergistically to stabilize UvsX-ssDNA [ 14 ]. UvsY also stabilizes UvsX-ssDNA in the presence of ADP or no nucleotide, so its effects are global. Results of recent kinetics studies are consistent with the idea that UvsY acts as a nucleotide exchange factor for UvsX, promoting the release of hydrolytic products so that new ATP substrate can bind to the active sites [ 58 ]. It is postulated that UvsY-enhanced nucleotide exchange allows UvsX to remain longer in its ATP-bound form with higher affinity for ssDNA, which would tend to stabilize presynaptic filaments and increase their catalytic activites activity. Through its dual activities in destabilizing Gp32-ssDNA and stabilizing UvsX-ssDNA interactions, UvsY allows UvsX filaments to nucleate and propagate on Gp32-covered ssDNA (Figure 2 ). ssDNA hand-offs govern filament assembly UvsX and UvsY interact specifically with the C-terminal "A-domain" of Gp32, and with each other [ 35 , 36 , 49 , 60 ]. Protein-protein interactions play a significant role in the overall DNA strand exchange reaction. Nevertheless, studies of UvsY have shown that its ability to destabilize Gp32-ssDNA complexes is independent of UvsY-Gp32 interactions [ 54 , 61 ], indicating that the ssDNA-binding activity of UvsY is responsible for destabilizing Gp32-ssDNA interactions. Results of in vitro complementation assays between UvsX and UvsY mutants further suggest that UvsY-ssDNA interactions create an optimal ssDNA conformation for high-affinity binding by UvsX [ 58 ]. Studies showed that UvsY KARL-motif mutants K58A and K58A/R60A have reduced affinities for ssDNA compared to wild-type [ 53 ]. Similarly UvsX missense mutants H195Q and H195A exhibit reduced affinities for ssDNA as well as altered enzymatic activities compared to wild-type [ 16 ]. Unlike wild-type UvsX, the ssDNA-dependent ATPase activities of UvsX-H195Q/A are strongly inhibited by wild-type UvsY at both low and high concentrations of the mediator. The UvsY KARL-motif mutants partially relieve this inhibition [ 58 ]. Furthermore the UvsX-H195Q mutant has weak DNA strand exchange activity that is inhibited by wild-type UvsY, but stimulated by the UvsY KARL-motif mutants [ 58 ]. These and other results support a mechanism in which presynaptic filament assembly involves a hand-off of ssDNA from UvsY to UvsX, with the efficiency of the hand-off controlled by the relative ssDNA-binding affinities of the two proteins. Evidence increasingly supports the notion that DNA and RNA pathways channel their substrates through series of hand-off transactions in which intermediate nucleic acid structures are passed directly from one protein in the pathway to the next [ 63 ]. This strategy avoids potential cytotoxic effects of the free nucleic acid structure and protects it from unprogrammed side reactions or degradation. The available data suggest that T4 presynaptic filament assembly is also governed by a sequence of hand-off events involving intermediate ssDNA structures generated by Gp32 and UvsY (Figure 3 ). Initially, Gp32 binding converts ssDNA into an extended conformation that resembles the mechanically stretched DNA created in force-spectroscopy experiments. In the first hand-off event, a UvsY hexamer binds to the extended ssDNA and converts it into a wrapped conformation that destabilizes Gp32-ssDNA interactions. The wrapped UvsY-ssDNA complex is thought to be in equilibrium between "closed" and "open" states. The "closed" state destabilizes Gp32-ssDNA interactions but is inaccessible to UvsX, whereas the "open" state favors high-affinity UvsX-ssDNA interactions. In the second hand-off event, ATP-bound UvsX binds to the "open" form of the wrapped UvsY-ssDNA structure, allowing nucleation of a UvsX-ssDNA filament while displacing Gp32 from the ssDNA. Other ssDNA hand-off transactions may occur as the filament transitions from the nucleation to the propagation phase, or as UvsY performs its nucleotide exchange factor function. In addition, the linkage of the UvsX ATPase cycle to the sequential hand-off mechanism creates opportunities for dynamic instability in presynaptic filaments, which we will address in a later section. UvsX-Gp32 exchanges on ssDNA Gp32F is a fluorescein-conjugated form of Gp32 that is useful as a fluorescence probe for Gp32 displacement from ssDNA and to study the kinetics of presynaptic filament assembly in real time [ 48 ]. As UvsX filaments assemble on Gp32F-covered ssDNA, Gp32F is displaced and the fluorescence of its fluorescein moiety decreases. This assay was used to study presynaptic filament assembly both in the absence of UvsY (low-salt conditions only) and in the presence of UvsY (physiological or high-salt conditions). The salt-dependence of the UvsY requirement for Gp32 displacement is a consequence of differential salt effects on the intrinsic association constants ( K parameters) of UvsX and Gp32 for ssDNA [ 13 , 41 , 44 , 45 , 64 ]. Under low-salt conditions (≤ 50 mM NaCl), the ATP or ATPγS-bound forms of UvsX possess sufficient affinity for ssDNA to compete with Gp32 and displace it from the lattice, causing a time-dependent decrease in the fluorescence of the Gp32F probe [ 48 ]. ADP-bound, AMP-bound, or apo forms of UvsX cannot displace Gp32 from ssDNA under any conditions. At higher, more physiologically relevant salt concentrations, all forms of UvsX lack the ability to displace Gp32 from the ssDNA. Under these conditions, the addition of UvsY restores UvsX-ssDNA filament formation and Gp32 displacement, as measured by the decrease in Gp32F fluorescence [ 48 ]. The UvsY-dependent reactions still require ATP or ATPγS as a prerequisite for filament assembly; ADP-, AMP-, and apo -UvsX conditions do not support Gp32 displacement. This observation is consistent with the previous finding that UvsY and ATPγS-binding stabilize UvsX-ssDNA filaments synergistically [ 14 ], which implies the cooperation of these two factors during filament nucleation and/or propagation steps. Following timecourses of Gp32F displacement from ssDNA allows detailed analyses of the kinetics of presynaptic filament assembly in a fully-reconstituted in vitro T4 recombination system (UvsX, UvsY, and Gp32). This has led to important new discoveries about filament dynamics and about the mechanism of UvsY in recombination mediation (Liu, J., C. Berger, and S.W. Morrical: Kinetics of Presynaptic Filament Assembly in the Presence of SSB and Mediator Proteins, unpublished). Under low-salt conditions, the ATP-dependent, UvsY-independent nucleation of UvsX filaments on Gp32F-covered ssDNA is highly salt-sensitive. Nevertheless nucleation rates are faster than propagation rates, suggesting that UvsX nucleates rapidly at many different sites. Under high-salt conditions, UvsY appears to specifically enhance the nucleation step to overcome the salt-sensitivity of UvsX filament assembly (Liu, J., C. Berger, and S.W. Morrical: Kinetics of Presynaptic Filament Assembly in the Presence of SSB and Mediator Proteins, unpublished). Rapid, salt-sensitive nucleation may be a general property of recombinase-DNA interactions, since similar behavior is observed for human Rad51 filament assembly on dsDNA [ 65 ]. It will be interesting to learn whether human RMPs such as Rad52, Brca2, or Rad51 paralogs also work by decreasing the salt-sensitivity of Rad51 filament nucleation. A simplified kinetic scheme for T4 presynaptic filament assembly is shown in Figure 4 , based on data derived from analysis of Gp32F displacement timecourses (Liu, J., C. Berger, and S.W. Morrical: Kinetics of Presynaptic Filament Assembly in the Presence of SSB and Mediator Proteins, unpublished). Results are consistent with a two-phase model, nucleation and propagation, both of which include a fast and reversible binding step ( K 1 or K 3 ) followed by a slow isomerization step ( k 2 or k 4 ) that is essentially irreversible under pre-steady-state conditions. We found that UvsY specifically enhances K 1 , thereby stabilizing the product of the reversible binding step during the filament nucleation phase. This product may be thought of as a "pre-nucleation complex". Therefore UvsY overcomes the salt-sensitivity of filament nucleation by stabilizing the pre-nucleation complex at high salt concentrations. We also found that k 4 , the rate constant for the isomerization step of filament propagation, is rate-limiting under all conditions (Liu, J., C. Berger, and S.W. Morrical: Kinetics of Presynaptic Filament Assembly in the Presence of SSB and Mediator Proteins, unpublished). This suggests that long presynaptic filaments are likely to be assembled from many shorter filaments that arise at multiple nucleation centers. In accord with this idea, human Rad51 assembles on dsDNA from many rapidly-formed nucleation sites and the cluster growth from each site is limited in length [ 65 ]. The requirement for many filament nucleation events may explain the observation that an apparent 1:1 stoichiometry between UvsX and UvsY has to be maintained for optimal recombination activity [ 22 , 46 , 60 ]. Dynamic instability in presynaptic filaments Presynaptic filaments are predicted to exhibit dynamic instability, or vectorial growth and collapse, due to the coupling of the recombinase ATPase cycle to changes in ssDNA binding affinity [ 15 , 19 , 47 , 60 ]. The Gp32F probe provides an indirect readout of the dynamic instability of UvsX-ssDNA filaments [ 49 ]. Results demonstrate that the dynamic instability of T4 presynaptic filaments depends not only on UvsX-catalyzed ATP hydrolysis, but also on competition between UvsX and Gp32 for binding sites on ssDNA (Figure 5 ). Experiments were designed in which UvsX and Gp32 undergo a pre-steady-state competition for a limited number of binding sites on ssDNA at physiological ionic strength [ 48 ]. The order of addition is controlled so that ssDNA is added to a pre-existing mixture of recombination proteins, which mimics the most likely pathway for filament assembly/disassembly in vivo. Filament assembly/disassembly is then monitored by following Gp32F dissociation/association using fluorescence. The data show that presynaptic filaments formed in the presence of Gp32 undergo constant assembly and collapse that is closely linked to the ATPase cycle of UvsX [ 48 ]. The reactions occur in three sequential phases (Figure 5 ): Phase 1--preparing the lattice . Gp32 rapidly binds and saturates all of the available ssDNA (rapid Gp32F fluorescence increase). Phase 2--filament growth . ATP-bound UvsX is loaded by UvsY and gradually displaces Gp32 (slow Gp32F fluorescence decrease). There is a stringent requirement for UvsY and either ATP or ATPγS in this phase, and the rate is optimal when UvsY stoichiometry is 1:1 with respect to UvsX and ssDNA binding sites. Phase 3--filament collapse . Depletion of ATP allows Gp32 to slowly re-occupy the ssDNA and drive off UvsX, which is now mainly in the low-affinity ADP/AMP forms [ 16 , 48 ] (slow Gp32F fluorescence increase). This collapse phase is sensitive to the nucleotide substrate/product ratio and does not occur if ATP is regenerated or if ATPγS is substituted. These observations are consistent with a dynamically unstable T4 presynaptic filament. Dynamic instability could take the form of treadmilling as shown in Figure 5 , in which UvsX-ssDNA filaments simultaneously grow at an ATP-capped end and contract at an ADP- or AMP-capped end. The vectorial motion would be reinforced by Gp32 which would out-compete UvsX for ssDNA binding sites preferentially at the ADP/AMP-capped filament end. Atomic Structure of T4 UvsX Recombinase A recently solved, high-resolution UvsX crystal structure provides important new information on the mechanism of the T4 recombinase [ 66 ]. The crystal was obtained from a truncation mutant UvsX 30-358 (full-length UvsX = 391 amino acid residues), which lacks the N-terminal protein-protein association domain and the extreme C-terminal region. The crystal has a P6 1 space group and the asymmetric unit is composed of dimer of identical subunits with a two-fold axis. In the crystal lattice these dimers are arranged as a right-handed helical filament, with one subunit of each dimer forming the filament while the opposite subunit in each dimer decorates the surface of the filament without interacting with its symmetry partners. The dimer interface in the asymmetric unit occludes the ATP binding site, therefore no bound ATP is observed in the structure. The DNA binding loops L1 and L2 of UvsX are disordered as is the case for all RecA family proteins crystallized in the absence of DNA. As expected, UvsX shares high similarity with E. coli RecA protein in overall architecture and protein folding, in spite of the remote sequence homology [ 67 ]. Compared to RecA, UvsX contains a larger N-terminal α/β motif, and a smaller C-terminal domain filled with helices and a small three-stranded β-sheet. The α/β ATPase core is highly conserved between UvsX and RecA in terms of structural motifs, locations, and amino acid compositions. The two nucleotide-binding motifs of UvsX, the Walker A and Walker B boxes, are located at similar positions compared to RecA structures. For example, the aromatic ring of Tyr99 in UvsX stacks with the adenine ring of ATP, similar to Tyr103 in RecA [ 66 ]. Docking of the UvsX structure into models of extended and compressed filament forms reconstituted from EM studies revealed additional details about the active site (Figure 6 )[ 66 ]. Docking into the high-pitch "active" filament (ADP-AlF 4 form) indicated that the ATPase site spans the filament interface, as is the case for high-pitch filaments of E. coli RecA and S. cerevisiae Rad51 [ 17 , 68 , 69 ]. Conserved residue Glu92 is positioned to activate a water molecule for nucleophilic attack on ATP γ-phosphate. Significantly, residues Lys246' and Arg248' reach across the filament interface and form salt bridges with the phosphates of ATP and with Glu92. These residues are structurally equivalent to the Lys248' and Lys250' bridges and to catalytic residue Glu96 in E. coli RecA. The lysine bridges are thought to promote catalysis by stabilizing the transition state during ATP hydrolysis [ 69 ]. This strategy is apparently conserved between RecA and UvsX. Interestingly, eukaryotic Rad51 and Dmc1 recombinases lack the entire motif containing the basic bridge residues, and no other basic residues take their places in the Rad51 crystal structures [ 17 , 68 ]. Thus there is a divergence of active site structure and function between the prokaryotic and eukaryotic recombinases, with UvsX more closely aligned to the prokaryotic mechanism. Docking of the UvsX structure into the low-pitch "inactive" filament (ADP form) indicates that residues Lys246' to Lys254' move by about 4 Å so that the ATP binding site no longer spans the filament interface. These observations indicate that changes in filament pitch observed at different stages of the ATPase cycle are accompanied by extensive remodeling of the active site itself. Overall, the high-resolution structure of UvsX [ 66 ] provides exciting new opportunities to investigate its catalytic and allosteric mechanisms. Actions of Helicases in DNA Strand Exchange Reactions The bacteriophage T4 recombination system provided one of the earliest demonstrations that a DNA helicase, Dda protein, can stimulate a recombinase-catalyzed DNA strand exchange reaction [ 70 ]. Subsequent work has shown that at least three T4-encoded helicases (Dda, Gp41, and UvsW) are capable of influencing recombination and/or recombination-dependent replication transactions in vitro, and probably in vivo as well. In this section we will focus on the impacts of Dda, Gp41, and UvsW on reconstituted strand exchange reactions in vitro . Helicase processing of recombination intermediates After a UvsX-catalyzed homology search and strand pairing, a joint molecule is formed between the invading 3' single-stranded DNA (ssDNA) tail and the homologous double-stranded DNA (dsDNA) template in the form of a displacement-loop (D-loop) (Figure 1 ). ssDNA regions of the D-loop are potential targets for helicase assembly. Depending on which strand the helicase translocates on, and on the polarity of the helicase, processing of the D-loop could have three different outcomes: extension of the heteroduplex by branch migration, unwinding of the heteroduplex by branch or bubble migration, or conversion of the D-loop into a nascent replication fork. In addition, certain helicases may use their translocase activity to remove presynaptic filaments from ssDNA. It appears likely that all four of these processes occur at some point during T4 DNA metabolism. It has been shown that all three T4 helicases, Dda, Gp41, and UvsW, are capable of catalyzing branch migration in vitro [ 29 , 70 , 71 ]. However, the biological functions of these helicases are distinctive, in spite of the overlapping branch migration activities. Dda helicase Dda is a unique helicase compared to Gp41 and UvsW, since it may regulate recombination both positively and negatively at two different stages: presynaptic filament formation and branch migration. E. coli UvrD and yeast Srs2 proteins are two translocases/helicases functioning to remove recombinases from ssDNA and to prevent improper presynaptic filament formation and illegitimate recombination events [ 72 - 74 ]. To date, no T4 helicase has been identified as a direct functional homolog of UvrD or Srs2. Dda may share some properties of these helicases though, since the phenotypes of certain dda mutants are consistent with a role in anti-recombination [ 75 ], and since Dda inhibits UvsX-mediated homologous strand pairing reactions in vitro [ 76 ]. It is speculated that destabilizing UvsX-ssDNA filaments through its translocase activity is one factor contributing to the observed inhibition of homologous pairing. Similarly, Dda might apply this translocation activity to DNA replication by allowing the fork to bypass DNA-bound proteins on the template in vitro [ 77 - 79 ]. If Dda protein does disrupt presynaptic filaments then its mechanism must differ somewhat from Srs2 and UvrD, since the latter two have 3' to 5' polarity while Dda has 5' to 3' polarity [ 80 - 82 ]. The strand exchange assay routinely uses a circular M13 ssDNA and a linearized M13 dsDNA as substrates. The extent of branch migration after initial synapsis can be monitored by the restriction endonuclease digestion pattern of the end-radiolabeled dsDNA [ 70 ]. This nicely-designed assay system allowed Kodadek and Alberts to monitor and measure the rate of branch migration of UvsX-catalyzed strand exchange in the presence and absence of Dda. The late addition of Dda after synapsis stimulates the rate of branch migration more than four-fold, from ~15 bp/sec to ~70 bp/sec [ 70 ]. Dda was the first helicase documented to stimulate strand exchange reactions by stimulating branch migration, on the premise that it is added late into the reconstituted reaction after synapsis has occurred. Furthermore, the specific protein-protein interaction between Dda and UvsX might be important for this stimulation, since Dda cannot stimulate RecA-catalyzed strand exchange reactions. In vitro , Dda's inhibition of homologous pairing and stimulation of branch migration can be separated by manipulating the addition sequence of Dda into the reconstituted reaction, either simultaneously with UvsX during presynapsis, or after the initiation of synapsis. How Dda balances these opposite activities and cooperates with UvsX in vivo remains largely unknown, however. It is observed that UvsX and Dda act synergistically in template switching to allow DNA lesion bypass and to rescue stalled replication forks [ 4 , 83 ]. Furthermore, protein-protein interactions between Dda and the C-terminal domain of Gp32 are required for the DNA replication activities of Dda [ 37 ]. These observations suggest that interactions with UvsX or with Gp32 could recruit Dda onto different nucleoprotein intermediates at different stages of the strand exchange process, perhaps regulating the recombination vs. anti-recombination functions of Dda. Gp41 helicase and Gp59 helicase loading protein Gp41, the essential replicative helicase in T4, facilitates both leading strand DNA synthesis catalyzed by the T4 DNA polymerase holoenzyme (Gp43, Gp44/Gp62, and Gp45 proteins), and lagging strand DNA synthesis by recruiting primase Gp61 to reconstitute the T4 primosome [ 4 ]. The Gp41 helicase translocates processively on the displaced strand in a 5' → 3' direction, as an asymmetric hexagonal ring on the DNA [ 84 , 85 ]. Gp59 has been classified as a replication mediator protein or helicase loading protein, based on the observation that it is required to load Gp41 onto Gp32-covered ssDNA [ 4 , 38 , 77 , 86 ]. Gp59 acts as an adapter protein by interacting with Gp32 at the N-terminus and with Gp41 at the C-terminus [ 86 - 88 ]. It is the key factor for the strand-specific recruitment of primosome onto the displaced strand of a D-loop to covert it into a replication fork during RDR, and to initiate new lagging-strand DNA synthesis during RDR. Gp41 cannot stimulate UvsX-dependent strand exchange unless Gp59 is present, and this stimulation occurs through branch migration [ 70 ]. UvsY stimulates homologous pairing, but strongly inhibits branch migration. The branch migration activity can only be recovered by adding Gp41 and Gp59. The protein-protein interaction between Gp59 and the C-terminal acidic domain of Gp32 is important for this rescue [ 70 ]. Interestingly, the formation and stability of Gp32-ssDNA clusters is a key factor for strand- and structure-specific loading of Gp41 helicase by Gp59. Gp59 targets Gp41 helicase assembly onto Gp32-ssDNA clusters [ 4 , 37 , 38 ]. The interplay between Gp32 and Gp59 is complicated. The formation of a tripartite Gp59-Gp32-ssDNA complex decreases the stability of Gp32-ssDNA interaction, but Gp32 also helps modulate the strand specificity of Gp59 [ 4 , 38 ]. Gp59-mediated primosome assembly is precluded from ssDNA that is saturated with UvsX and UvsY, but allowed when a few Gp32 clusters interrupt the presynaptic filament. In DNA strand exchange, the invading strand is typically saturated with UvsX and UvsY and therefore resistant to Gp41/Gp59 loading. However, Gp32 rapidly sequesters the displaced strand of the D-loop [ 19 , 47 ], forming a target for Gp41/Gp59. Thus UvsX/UvsY and Gp32/Gp59 enforce strand specific loading of Gp41 onto the displaced strand, where it is poised to catalyze branch migration using its 5' to 3' helicase activity (Figure 7 ). UvsX/UvsY prevent D-loop resolution (anti-recombination) by Gp41/Gp59 by preventing their assembly on the invading ssDNA strand. An identical partitioning mechanism is used during RDR to ensure primosome assembly on the displaced strand of the D-loop, assuring complete reconstitution of semi-conservative DNA synthesis beginning with a recombination event [ 4 ]. In the absence of UvsX and UvsY, the sole presence of excessive amount of Gp32 can produce joint molecules from M13 dsDNA with a 3' single-stranded termini of about 100 nucleotides and a circular M13 ssDNA [ 89 ]. The initial binding of Gp32 onto the single-stranded tail is probably sufficient to destabilize the double-stranded helix, starting from the junction point, and to promote spontaneous joint molecule formation. When coupled with Gp59 and Gp41, the polar branch migration mediated by Gp41 can drive the formation of nicked circle, the final product of standard three-strand exchange reactions [ 89 ]. This synergism between Gp32 and Gp41/Gp59 is also crucial for extensive strand displacement synthesis by the T4 DNA polymerase holoenzyme [ 39 , 90 ]. UvsW helicase UvsW plays a central role in T4 recombination and in the transition from origin to recombination-dependent replication. UvsW mutations cause hypersensitivity to UV and hydroxyurea, and a decreased frequency of recombination [ 91 , 92 ]. UvsW is a 3' to 5' RNA/DNA and DNA/DNA helicase with specificity for branched-DNA substrates such as X-shaped Holliday junctions and Y-shaped replication forks [ 71 , 93 , 94 ]. It does not unwind linear duplex substrates with either blunt ends or single-stranded tails. Substrate recognition may occur through a small but highly electropositive N-terminal domain and an arginine/aromatic-rich loop, as revealed by its crystal structure [ 95 ]. The mutant phenotype and substrate specificity lead to the hypothesis that UvsW might drive branch migration to resolve recombination intermediates during strand invasion and transfer. Indeed, purified UvsW protein can catalyze Holliday junction branch migration through more than 1 kb of DNA sequence, using a plasmid-based Holliday junction-containing substrate [ 71 ]. Recent data show that UvsW promotes branch migration in UvsX-catalyzed DNA strand exchange reactions [ 66 ]. In the classic three-strand exchange reaction with M13 circular ssDNA and linear dsDNA substrates, UvsW promotes resolution of the branched ssDNA/dsDNA networks formed by UvsX, leading to the robust generation of nicked circular heteroduplex product. Reactions occur in the presence of Gp32 and in either the presence or absence of UvsY. Thus UvsW appears to provide a "missing link" in the biochemistry of T4 recombination, since it can provide physiologically reasonable mechanisms for generating extensive heteroduplex DNA, involving the translocation of either 3- or 4-strand junctions. In summary, Dda, Gp41, and UvsW are three helicases all capable of stimulating branch migration, but with clearly different biological roles in T4 recombination. Dda may act as a negative regulator of homologous pairing, but may also be used to accelerate branch migration or to couple recombination to bubble migration DNA synthesis [ 70 , 75 , 76 , 96 ]. The major role of Gp41/Gp59 in recombination is likely to be the channeling of recombination intermediates into structures that can support RDR, and then launching lagging strand synthesis in the semi-conservative RDR mechanism [ 4 ]. UvsW on the other hand optimizes strand exchange and the formation of long heteroduplex DNA [ 66 ]. Complex interplays between the three different helicase activities are likely to modulate many aspects of T4 recombination metabolism. Abbreviations HR: homologous recombination; HDR: homology-directed repair; RDR: recombination-dependent replication; DSB: double-strand break; ssDNA: single-stranded DNA; dsDNA: double-stranded DNA; SSB: single-stranded DNA binding protein; RMP: recombination mediator protein; ATPγS: adenosine 5'- O -(3-thio)triphosphate; Gp32F: fluorescein-labeled bacteriophage T4 gene 32 protein (Gp32). Competing interests The authors declare that they have no competing interests. Authors' contributions JL and SM made equal intellectual contributions to this review and participated equally in writing the manuscript.

CC BY

no

2022-01-12 15:21:44

Virol J. 2010 Dec 3; 7:357

oa_package/6f/27/PMC3016280.tar.gz

PMC3016281

21129203

Introduction Studies during the last 15 years have provided strong evidence that T4 DNA replication initiates from specialized structures, namely R-loops for origin-dependent replication and D-loops for recombination-dependent replication (RDR). The roles of many of the T4 replication and recombination proteins in these processes are now understood in detail, and the transition from origin-dependent replication to RDR has been ascribed to both down-regulation of origin transcripts and activation of the UvsW helicase, which unwinds origin R-loops. One of the interesting themes that emerged in studies of T4 DNA metabolism is the extensive overlap between different modes of replication initiation and the processes of DNA repair, recombination, and replication fork restart. As discussed in more detail below, the distinction between origin-dependent and recombination-dependent replication is blurred by the involvement of recombination proteins in certain aspects of origin replication. Another example of overlap is the finding that repair of double-strand breaks (DSBs) in phage T4 infections occurs by a mechanism that is very closely related to the process of RDR. The close interconnections between recombination and replication are not unique to phage T4 - it has become obvious that the process of homologous recombination and particular recombination proteins play critical roles in cellular DNA replication and the maintenance of genomic stability [ 1 - 4 ]. Origin-dependent replication Most chromosomes that have been studied include defined loci where DNA synthesis is initiated. Such origins of replication have unique physical attributes that contribute to the assembly of processive replisomes, facilitate biochemical transactions by the replisome proteins to initiate DNA synthesis, and serve as key sites for the regulation of replication timing. While the actual determinants of origin activity remain ill defined in many systems, all origins must somehow promote the priming of DNA synthesis. Bacteriophage T4 contains several replication origins that are capable of supporting multiple rounds of DNA synthesis [ 5 , 6 ] and has very well-defined replication proteins [ 7 ], making this bacteriophage an ideal model to study origin activation and maintenance. Localization of T4 origins throughout the genome Clear evidence for defined T4 origin sequences began to emerge about 30 years ago when the Kozinski and Mosig groups demonstrated that nascent DNA produced early during infection originated from specific regions within the 169 kb phage genome [ 8 - 10 ]. The race was on, and several groups spent the better part of two decades trying to define the T4 origins of replication. These early efforts brought a battery of techniques to bear, including electron microscopy and tritium labeling of nascent viral DNA, localizing origins to particular regions of the genome. The first direct evidence for the DNA sequence elements that constitute a T4 origin emerged from studies of Kreuzer and Alberts [ 11 , 12 ], who isolated small DNA fragments that were capable of driving autonomous replication of plasmids during a T4 infection. Later approaches using two-dimensional gel electrophoresis confirmed that these two origins, oriF and oriG [also called ori(uvsY) and o ri(34) , respectively], were indeed active in the context of the phage genome [ 13 , 14 ]. All told, at least seven putative origins (termed oriA through oriG ) were identified by these various efforts, yet no strong consensus emerged as whether all seven were bona fide origins and how the multiple origins were utilized during infection. Recent work by Brister and Nossal [ 5 , 15 ] has helped to clarify many issues regarding T4 origin usage. Using an array of PCR fragments, they monitored the accumulation of nascent DNA across the entire viral genome over the course of infection, allowing both the origins and breadth of DNA synthesis to be monitored in real time. This whole-genome approach revealed that at least 5 origins of replication are active early during infection, oriA , oriC , oriE , oriF , and oriG (see Figure 1 ). Though all of these origins had been independently identified to some extent in previous studies, this was the first observation of concurrent activity from each within a population of infected cells. There do not appear to be any local sequence motifs shared among all the T4 origins. However, one origin, oriE , does include a cluster of evenly spaced, 12-nt direct repeats [ 16 ]. Similar "iterons" are also found within syntenic regions of closely related bacteriophage genomes, implying conserved function [ 17 ]. Indeed, this arrangement of direct repeats is reminiscent of some plasmid origins, such as the RK6 gamma origin, where replication initiator proteins bind to direct repeats and promote assembly of replisomes [ 18 ]. Despite this circumstantial evidence, no association has been established between the T4 iterons and oriE replication activity, and to this date their role during T4 infection remains ill defined. There is some indication that global genome constraints influence the position of T4 origins. Three of the more active T4 origins, oriE , oriF , and oriG are located near chromosomal regions where the template for viral transcription switches from predominately one strand to predominately the complementary strand [ 5 , 19 ] (see Figure 1 ). These regions of transcriptional divergence coincide with shifts in nucleotide compositional bias (predominance of particular nucleotides on a particular strand), a hallmark of replication origins in other systems [ 20 ]. That said, at least two origins ( oriA and oriC ) are well outside regions of intrastrand nucleotide skews and transcriptional divergence, so it is not clear what, if any, physical properties of the T4 chromosome contribute to origin location. Moreover, the T4 genome is circularly permuted with no defined telomeres, so the actual position of a given locus relative to the chromosome ends is variable in a population of replicating virus. The undulating T4 transcription pattern reflects the modular nature of the viral genome. T4 genes are arranged in functionally related clusters, and diversity among T4-related viruses appears to arise through the horizontal transfer of gene clusters [ 17 , 21 ]. The spacing of T4 origins over the length of the viral genome coincides with some of these clusters and may reflect genome mechanics. Most early T4 DNA synthesis originates from regions within the genome that are dominated by late-mode viral transcription [ 5 , 19 ]. This arrangement suggests an intimate relationship between T4 replication and transcription of late genes, like those encoding viral capsid components. It has been known for some time that late-mode transcription is dependent on gp45 clamp protein, which is a component of both the T4 replisome and late-mode transcription complexes (reviewed by Miller et al . [ 22 ]), but there is also evidence that the amount of replication directly influences the amount of transcription [ 23 ] (Brister, unpublished data). Molecular mechanism of origin initiation Though few obvious sequence characteristics are shared between them, all of the T4 origins are thought to facilitate formation of RNA primers used to initiate leading strand DNA synthesis. Most of what is known about the detailed mechanism of T4 replication initiation comes from studies of the two origins ( oriF and oriG ) that support autonomous replication of plasmids in T4-infected cells (see above). Origin plasmid replication requires the expected T4-encoded replisome proteins, and like phage genomic DNA replication, is substantially reduced and/or delayed by mutations in the replicative helicase, primase and topoisomerase [ 24 , 25 ]. The DNA sequences required for oriF and oriG function on recombinant plasmids have been defined by deletion and point mutation studies [ 26 ] (Menkens and Kreuzer, unpublished data). A minimal sequence of about 100 bp from each origin was shown to be necessary for autonomous replication, and though there is little homology between oriF and oriG , both minimal sequences include a middle-mode promoter and an A + T-rich downstream unwinding element (DUE) [ 26 , 27 ]. Middle-mode promoters consist of a binding site for the viral transcription factor MotA in the -30 region, along with a -10 sequence motif that is indistinguishable from the typical E. coli σ 70 -10 motif [ 28 , 29 ]. Transcripts initiated from the oriF MotA-dependent promoter were shown to form persistent R-loops within the DUE region, leaving the non-template strand hypersensitive to ssDNA cleavage. Formation of these R-loops is not dependent on specific sequences and the endogenous DUE can be substituted with heterologous unwinding elements [ 13 , 27 ]. The oriF R-loops are very likely processed by viral RNase H to generate free 3'-OH ends that are used to prime leading strand DNA synthesis [ 13 , 27 ]. Furthermore, the presence of an R-loop presumably holds the origin duplex in an open conformation, giving the gp41/61 primosome complex access to the unpaired non-template strand to allow extensive parental DNA unwinding and priming on the lagging strand. Less is known about replication priming at the other T4 origins [ 30 ]. Presumably, oriG uses the same mechanism as oriF [ 13 , 27 ], and there is some evidence that a transcript from a nearby MotA-dependent promoter is used to initiate replication at oriA [ 30 ]. Yet, MotA mutations do not fully prevent viral replication [ 16 , 31 ], and other types of viral promoters also appear important to origin function. For example, there are no middle-mode promoters near oriE ; instead this origin apparently depends on an early-mode promoter, which does not require viral transcription factors for activity [ 16 ]. Moreover, mutations that prevent late-mode viral transcription alter replication from T4 oriC , without affecting activity from the other origins (Brister, unpublished), raising the possibility that a late-mode promoter is required for activity from this origin. Discontinuous lagging strand replication is normally primed by the T4-encoded gp61 primase [ 32 - 34 ]. Even though T4 primase is required only for lagging strand synthesis in vitro , the in vivo results are more complex. First, mutants deficient in primase show a severe DNA-delay phenotype, with very little DNA synthesis occurring early during infection [ 24 , 30 , 35 , 36 ]. This implies that primase activity contributes directly to early steps of T4 DNA replication. Either leading strand synthesis at some T4 origins is primed by primase, or normal viral replication requires the coupling of leading strand synthesis with primase-dependent lagging strand synthesis. Second, T4 DNA replication eventually reaches a remarkably vigorous level in primase-deficient infections, even when using a complete primase deletion mutant [ 24 ] (also see [ 37 ]). One published report suggested that the primase-independent replication was abolished by mutational inactivation of T4 endonuclease VII, leading to a model in which endonuclease VII cleavage of recombination intermediates provides primers for DNA synthesis [ 38 ]. However, repetition of this experiment revealed little or no decrease in endonuclease-deficient infections [ 39 ], and the strain used in the Mosig study was later found to contain an additional mutation that was contributing to the reduced replication (G. Mosig, personal communication to KNK). The mechanism of extensive DNA replication late in a primase-deficient infection remains unclear, but could possibly result from extensive priming by mRNA transcripts (perhaps in combination with endonuclease cleavage as suggested by Mosig [ 38 ]). In other systems, there are examples of both primase- and transcript-mediated initiation of leading strand DNA synthesis from origins. A transcript is used to prime replication from the ColE1 plasmid origin, as well as mitochondrial DNA origins [ 40 , 41 ], yet primase is used to initiate replication from the major E. coli origin, oriC [ 42 , 43 ]. Indeed, there are even systems where both mechanisms of initiation are used within a single chromosome. For example, unlike oriC , R-loops are apparently used to initiate DNA synthesis at the oriK sites in E. coli (reviewed in [ 44 ]). The molecular mechanism of T4 replication initiation has been investigated in vitro using R-loop substrates constructed by annealing an RNA oligonucleotide to supercoiled oriF plasmids [ 45 ]. Efficient replication of these preformed R-loop substrates does not require a promoter sequence, but a DUE is necessary. In fact, non-origin plasmids are efficiently replicated in vitro by the T4 replisome as long as they have a preformed R-loop within a DUE region, implying that the R-loop itself is the signal for replisome assembly on these substrates. Experiments using radioactively labeled R-loop RNA directly demonstrated that the RNA is used as the primer for DNA synthesis. Several viral proteins are required for significant replication of these R-loop substrates: DNA polymerase (gp43), polymerase clamp (gp45), clamp loader (gp44/62), and single-stranded DNA binding protein (gp32). In addition, without the replicative helicase (gp41), leading-strand synthesis is limited to a relatively short region (about 2.5 kb) and lagging strand synthesis is abolished. While gp41 can load without the helicase loading protein (gp59), the presence of gp59 greatly accelerates the process. Finally, replication on these covalently closed substrates is severely limited when the T4-encoded type II topoisomerase (gp39/52/60) is withheld, as expected due to the accumulation of positive supercoiling ahead of the fork. Normal viral replication also requires gp59 protein, and though gene 59 mutants make some DNA early, this synthesis is arrested as the infection progresses [ 5 , 46 , 47 ]. This deficiency was initially thought to reflect a unique requirement for gp59 in recombination-dependent replication (i.e., no requirement in origin-dependent replication). However, gp59 mutations also affect origin activity, reducing the total amount of origin-mediated DNA synthesis, mirroring the in vitro studies mentioned above [ 5 ]. Further defects are clearly visible at oriG , where gene 59 mutations cause problems in the coupling of leading and lagging strand synthesis (but do not prevent replication initiation) [ 48 ]. The deleterious effects of gene 59 mutations could reflect several biochemical activities that have been characterized in vitro . A major function of gp59 is loading of the replicative helicase gp41 [ 49 ]. Gp59 is a branch-specific DNA binding protein with a novel alpha-helical two-domain fold [ 50 ]. The gp59 protein is capable of binding a totally duplex fork, but requires a single-stranded gap of more than 5 nucleotides (on the arm corresponding to the lagging strand template) to load gp41 [ 51 ]. As expected from this loading activity, gp59 stimulates gp41 helicase activity on branched DNA substrates (e.g. Holliday junction-like molecules). Interestingly, gp59 has another function in the coordination of leading- and lagging-strand synthesis and in this context has been called a "gatekeeper". When gp59 binds to replication fork-like structures in the absence of gp41, it blocks extension by T4 DNA polymerase [ 45 , 48 , 52 ]. This inhibitory activity of gp59 presumably acts to prevent the generation of excessive single-stranded DNA and allow coordinated and coupled leading and lagging strand synthesis. Unlike gp59, the viral gp41 helicase is required for extended replication of R-loop substrates in vitro (see above) and any appreciable replication during infection [ 15 , 45 , 53 ]. Yet, some viral replication is observed in gp59-deficient infections (see above), indicating that gp41 helicase can load onto origins at some rate through another means. T4 encodes at least two other helicases, UvsW and Dda, and earlier studies demonstrated that one of them, Dda, stimulates gp41-mediated replication in vitro [ 49 ]. It was therefore suggested that either gp59 or Dda was sufficient to load gp41 helicase at the T4 origins [ 49 ]. Consistent with this notion, dda mutants have a DNA delay phenotype and are deficient in early, presumably origin-mediated DNA synthesis, though replication rebounds at later times when it is dependent on viral recombination [ 15 , 46 ]. Moreover, dda 59 double mutants have a greater defect than either single mutant, essentially showing no replication (either early or late) and indicating a cumulative effect on origin activity [ 46 ]. Though there may be some functional overlap between Dda and gp59, DNA replication patterns indicate that each has distinct activities at the T4 origins [ 15 ]. Unlike dda mutations, which cause a generalized reduction in DNA synthesis that is particularly evident at oriE , gene 59 mutations have little effect on replication from this origin [ 15 ]. This difference may indicate that oriE uses a different mechanism to initiate replication, one less dependent on gp59. This idea has been expressed before and may simply reflect the difference in sequence elements at oriE compared to the other origins. One protein in particular, RepEB, has also been implicated in oriE activity [ 16 ], but repEB mutations have a more generalized effect, reducing replication from all origins [ 15 ]. Inactivation of origins at late times The regulation of origin usage has been studied directly for oriF and oriG , the two origins known to function via an R-loop intermediate. One level of control is exerted by the change in the transcriptional program. The RNA within the oriF and oriG R-loops are initiated from MotA-dependent middle mode promoters, which are shut off as RNA polymerase is converted into the form for late transcription [ 28 , 29 ]. A second level of control is exerted when the UvsW helicase is expressed from its late promoter [ 54 ]. UvsW is a helicase with fairly broad specificity for various branched nucleic acids, including the R-loops that occur at oriF and oriG [ 55 - 57 ]. Thus, any existing R-loops at these origins are unwound when UvsW is synthesized. While not yet studied directly, R-loops may also occur at one or more other T4 origins (e.g. oriE ), and thus the mechanisms of regulation could be identical to that of oriF and oriG . Further work is clearly needed to understand the regulation of other T4 origins. As will be discussed in more detail below, mutational inactivation of T4 recombination proteins leads to the DNA arrest phenotype, characterized by a paucity of late DNA replication. The additional inactivation of UvsW suppresses this DNA arrest phenotype and allows high levels of DNA synthesis at late times [ 58 - 61 ]. The simplest explanation is that R-loop replication becomes dominant in these double-mutant infections at late times. If true, it seems likely that much of this late replication is initiated at R-loops formed at late promoters, but these "cryptic origin" locations have not yet been experimentally defined. Recombination-dependent replication The tight coupling of homologous genetic recombination and DNA replication was first recognized in the phage T4 system when it was found that mutational inactivation of recombination proteins leads to the DNA-arrest phenotype characterized by defective late replication [ 62 ]. Based on this and other data, Gisela Mosig proposed that genomic DNA replication can be initiated on the invading 3' ends of D-loop structures generated by the recombination machinery (Figure 2A ) [ 63 ]. There is now abundant in vivo and in vitro evidence supporting this model for phage T4 DNA replication. T4 RDR is an important model for the linkage of recombination and replication, because it has become clear that recombination provides a backup method for restarting DNA replication in both prokaryotes and eukaryotes (see below). RDR on the phage genome The infecting T4 DNA is a linear molecule, and early genetic results showed that the (randomly located) DNA ends are preferential sites for homologous genetic recombination [ 64 - 66 ]. When an origin-initiated replication fork reaches one of the DNA ends, one of the two daughter molecules should contain a single-stranded 3' end that is competent for strand invasion and D-loop formation; the other daughter molecule is also presumably competent for strand invasion after processing to generate a 3' end. The complementary sequence that is invaded could be at the other end of the same DNA molecule, since the infecting T4 DNA is terminally redundant, or it may be within the interior region of a co-infecting T4 DNA molecule, since T4 DNA is also circularly permuted. In this way, the process of RDR can in principle initiate soon after an origin-initiated fork reaches a genomic end. As will be described below, RDR or some variant thereof might be needed to continue replication well before origin-initiated forks reach the genome ends. The overall role of RDR in genome replication and the relationship of RDR to the eventual packaging of phage DNA are discussed in detail elsewhere [ 6 , 67 ]. RDR of the phage genome is abolished or greatly reduced by mutational inactivation of most T4-encoded recombination proteins (see [ 68 ] for review on the biochemistry of T4 recombination proteins). The strongest DNA arrest phenotypes are caused by inactivation of gp46/47 or gp59, and correspondingly, these are essential proteins. Inactivation of the non-essential UvsX and UvsY proteins eliminate most but not all late DNA replication. These two proteins catalyze the strand invasion reaction that generates D-loops, and so one might expect RDR to be totally abolished. However, a significant amount of T4 genetic recombination still occurs in the absence of UvsX or UvsY, and this has been ascribed to a single-strand annealing pathway [ 69 , 70 ]. Single-strand annealing intermediates may also be used to initiate RDR, which could explain the residual late DNA replication in UvsX or UvsY knockout mutants. The uvsW gene is in the same recombinational repair pathway as uvsX and uvsY [ 71 ]. However, the uvsW gene product was not originally implicated in the process of RDR because uvsW knockout mutations do not block late DNA replication [ 71 ]. This inference was probably misleading - as described above, the UvsW helicase apparently unwinds R-loops that could otherwise trigger replication at late times. Thus, inactivation of UvsW could simultaneously reduce or eliminate RDR and activate an R-loop dependent mechanism of late replication, resulting in no net decrease in late DNA replication [ 54 ]. Consistent with this model, a uvsW mutant has reduced recombination and was shown to be defective in generating phage DNA longer than unit length (in alkaline sucrose gradients) [ 71 ]. In addition, UvsW is required for a plasmid-based model for RDR [ 55 ] (see below). The one T4 recombination function that is not required for RDR is endonuclease VII, which resolves Holliday junctions and other branched DNA structures [ 72 , 73 ]. The major function of endonuclease VII during infection is to resolve DNA branches during DNA packaging [ 74 , 75 ]. Because this is a very late step in genetic recombination, the lack of a role in RDR is unsurprising. Plasmid model systems for RDR Plasmid model systems have been productive for analyzing the mechanism of RDR in vivo , and have revealed a very close relationship between repair of DSBs and the process of RDR. Plasmids with homology to the T4 genome but no T4 replication origin are replicated during a phage T4 infection, as long as T4-induced host DNA breakdown is prevented [ 76 - 78 ]. This plasmid replication is not dependent on particular T4 sequences, because even plasmid pBR322 can be replicated when the infecting T4 carries an integrated copy of the plasmid [ 76 ]. Plasmid replication requires T4 recombination proteins, arguing that it occurs by RDR [ 77 ]. The products of plasmid replication in a T4 infection consist mostly of long plasmid concatamers, arguing that rolling circle replication is induced, but the mechanism of rolling circle formation is unknown [ 79 ]. The remarkable discovery of mobile group I introns in T4 [ 80 ] led to a simple way to introduce site-specific DSBs during a T4 infection, which has been valuable for in vivo studies of T4 RDR. These introns encode site-specific DNA endonucleases, such as the endonuclease I- Tev I from the intron of the T4 td gene (see below for discussion of the intron mobility/DSB repair events; also see [ 81 ]). The recognition site for I- Tev I (or another intron-encoded nuclease, SegC) has been introduced into recombinant plasmids and also into ectopic locations in the T4 genome, and in either case, the site is cleaved efficiently during a normal T4 infection when the endonuclease is expressed [ 76 , 82 - 86 ]. If the regions adjacent to the cut site have a homologous DNA target, either in the T4 genome or another segment of a plasmid residing in the same cell, coupled recombination/replication reactions are efficiently induced [ 76 , 79 , 87 ]. Using such model systems for RDR, it was shown that T4 recombination proteins UvsX, UvsY, UvsW, gp46/47, and gp59 are required for extensive DSB-directed replication, as are the expected T4 replication fork proteins (gp43, gp44/62, gp45, gp32, gp41, gp61; delayed replication of the plasmid occurs in the gp61-deficient infection, similar to the delayed replication of chromosomal DNA) [ 24 , 55 , 77 ]. In addition, by limiting the homology to just one side of the break, a single double-strand end was shown to be sufficient to induce RDR, as predicted by the Mosig model [ 76 , 86 ]. Molecular mechanism of RDR The heart of the RDR process is the strand-invasion reaction that creates D-loops, which is described in more detail in the review on T4 recombination [ 68 ]. Briefly, DNA ends are prepared for strand invasion by the gp46/47 helicase/nuclease complex, transient regions of ssDNA are coated by the single-strand binding protein gp32, UvsY acts as a mediator protein in loading UvsX onto gp32-coated ssDNA, and UvsX is the strand-invasion protein (RecA and Rad51 homolog). Recent evidence argues that the UvsW helicase also plays a direct role in strand invasion, promoting 3-strand branch migration to stabilize the D-loop [ 88 ]. As described in more detail by Kreuzer and Morrical [ 6 ], early reconstitution of a T4 RDR reaction in vitro generated a conservative replication reaction called bubble-migration synthesis [ 89 ]. In bubble-migration synthesis, the 3' invading end in the D-loop is extended by DNA polymerase as the junction at the back of the D-loop undergoes branch migration in the same direction (Figure 2B ). The net result is that a newly synthesized single-strand copy is created and then quickly extruded from its template, and lagging-strand synthesis does not occur within the D-loop. In the RDR reactions analyzed by Formosa and Alberts [ 90 ], the T4 DNA polymerase holoenzyme complex (polymerase gp43, clamp gp45 and clamp loader gp44/62) catalyzed synthesis in reactions containing only UvsX and gp32. Interestingly, synthesis did not occur if the host RecA protein was substituted for UvsX (even if host SSB protein was added), suggesting that the T4 polymerase complex has specific interactions with the phage-encoded strand-exchange protein. The extent of synthesis was limited unless a helicase was added to facilitate parental DNA unwinding - Dda was used in these initial experiments and allowed extensive bubble-migration synthesis [ 90 ]. Since the publication of Molecular Biology of Bacteriophage T4 in 1994 [ 91 ], much progress has been made in understanding the mechanism of loading of the helicase/primase complex onto D-loops. When T4 RDR reactions are supplemented with gp59, gp41 and gp61, lagging-strand synthesis is efficiently reconstituted on the displaced strand of the D-loop, and a conventional semi-conservative replication fork is established (Figure 2A ) (see [ 6 ]). As described above, gp59 is a branch-specific DNA binding protein that loads gp41, and gp59 interacts specifically with both gp41 and gp32 in the loading reaction [ 50 , 51 , 92 - 97 ]. Jones et al. [ 94 ] showed that gp59 can load helicase onto a structure that closely resembles a D-loop, reflecting its role in RDR. Once the replicative helicase is loaded onto the displaced strand of the D-loop (which becomes the lagging-strand template), leading strand synthesis by T4 DNA polymerase (gp43) is activated. Because the T4 primase gp61 binds to and functions with gp41 (see [ 7 ]), loading of gp41 is critical to begin lagging-strand synthesis as well. Overlap between origin- and recombination-dependent mechanisms The transition between origin- and recombination-dependent replication is not entirely clear cut during T4 infection, and there is significant interplay between the two replication modes. Moreover, the relationship between origin- and recombination-dependent replication is dynamic, which is clearly seen in experiments with varying multiplicities of infection. In singly infected cells, there is a prolonged period early during infection when the recombination protein UvsX is not required for replication. Yet, when cells are infected with an average of five viruses, the timing changes, and even very early replication is dependent on UvsX [ 5 ]. Though the mechanism of this regulation is not clear, it is evident that the infection program can somehow sense the amount of infecting viral DNA and switch replication modes under conditions where there are ample templates for RDR. Recombination proteins also appear to be more important to replication from some origins compared to others. As mentioned earlier, genetic requirements vary among the multiple T4 replication origins that are active within a single population of infected cells. At least one origin, oriA , appears more active later during infection, when replication is dependent on the viral recombination machinery. Moreover, replication from this origin is significantly reduced when the viral recombination protein UvsX is mutated [ 5 ]. Though these observations underscore a role for T4 recombination machinery at oriA , it is not clear whether RDR is preferentially initiated near oriA or if normal oriA -mediated replication is partially dependent on UvsX. One hint to the role of UvsX during origin-mediated replication comes from the apparently slow movement of replication forks across the T4 chromosome. Once initiated, T4 replication forks do not simply progress from an origin to the ends of the chromosome at the 30-45 kb per minute rate observed in vitro [ 5 ]. Rather, replication forks appear to move more slowly than expected, resulting in the accumulation of sub-genomic length DNAs early during infection. Only later are these short DNAs efficiently elongated into full-length genomes. This behavior was initially noticed by Cunningham and Berger [ 58 ], who analyzed the length of newly replicated single-stranded DNA using alkaline sucrose gradients. They also showed that efficient maturation of nascent DNAs into full genome length products requires the viral replication proteins UvsX or UvsY. A similar effect was observed during array studies where the elongation of nascent DNAs was greatly delayed in uvsX mutant infections compared to normal infections [ 5 ]. So why is there a delay in the elongation of T4 nascent DNAs? One possibility is that physical factors (e.g. tightly bound proteins) impede the progress of the replication forks across the T4 chromosome, causing replisomes to stall or disassociate from the DNA template. Rescue of model stalled forks in vitro can be catalyzed by UvsX and either gp41 helicase (with gp59) or Dda helicase [ 98 ]. Thus, one model is that UvsX is required in vivo to restart origin-initiated forks that have stalled before completing replication, and so the elongation of nascent DNAs is compromised during uvsX mutant infections. Several factors may impede the progress of replication forks (also see below). T4 replication occurs concurrently with transcription during infection [ 19 ] (Brister, unpublished results), so replisomes must compete with the transcriptional machinery for template. Head-on collisions with RNA polymerase cause pausing of T4 replisomes in vitro [ 99 ], and undulating patterns of T4 transcription imply that replication forks must eventually pass through regions of head-on transcription. Furthermore, if multiple origins are active on a single chromosome, then replication forks initiated at different origins would speed towards one another, plowing through the duplex template. In this scenario intervening sequences would be wound into impassable torsion springs, and T4 topoisomerase (gp39/52/60) would be necessary to relax the duplex and allow progression. Indeed, gene 52 mutants produce shorter than normal DNA replication products early during infection, similar to uvsX mutants [ 100 ]. Interrelationship between replication, recombination and repair Studies in many different biological systems have uncovered key roles of recombination proteins in the replication of damaged DNA [ 1 - 4 ]. One major set of pathways involves the repair of DSBs and broken replication forks. In addition, recombination proteins are involved in multiple pathways proposed for replication fork restart after blockage by non-coding lesions, some pathways coupled to repair of the DNA damage and others that result in bypass of the damage. Here, we briefly review unique contributions to this field that emerged from the phage T4 system. Tight linkage of DSB repair and RDR As indicated above, DSB repair in phage T4 is closely related to the process of RDR. Studies of DSB repair were greatly accelerated by the discovery of the mobile group I introns and their associated endonucleases. Intron mobility involves the generation of a DSB within the recipient (initially intron-free) DNA by an intron endonuclease, followed by a DSB repair reaction that introduces a copy of the intron from the donor DNA, such that both recipient and donor end up with a copy of the intron [ 80 , 81 , 101 ]. A variety of approaches have been used to study the detailed mechanism of DSB repair in vivo using intron endonuclease-mediated DSBs. One series of studies using a plasmid model system indicated that the DSBs are repaired by a pathway called synthesis-dependent strand annealing (SDSA), in which the induced DNA replication is limited to the region near the DSB (Figure 3A ) [ 102 , 103 ]. The SDSA repair mechanism is closely related to the bubble-migration reaction described above, and has been implicated in DSB repair in eukaryotic systems such as Drosophila [ 104 , 105 ]. Other studies, however, argue that the DSB leads to the generation of fully functional replication forks in a process that is very closely related to the RDR pathway that occurs in the phage genome [ 79 , 85 , 87 , 106 ]. This so-called extensive chromosomal replication (ECR) model leads to bona fide DSB repair, even though the two broken ends of the DSB can end up in different molecules (Figure 3B ). A major difference between the SDSA and ECR models for DSB repair is that SDSA does not involve primase (gp61)-dependent lagging strand synthesis, while the ECR model does. Perhaps either repair model can occur when a DSB occurs on the phage genome, but the choice of pathways depends on whether the helicase/primase complex is successfully loaded onto the displaced strand of the initial D-loop. Considering that gp59 efficiently inhibits polymerase and loads helicase/primase (see above), it is difficult to see how the bubble-migration pathway and SDSA could occur in vivo , unless there is some additional level of regulation that has not yet been uncovered. Shcherbakov et al. [ 106 ] have presented additional evidence that DSBs trigger normal replication like that postulated in the ECR model, and provided arguments against a major role for the SDSA pathway during wild-type T4 infections. If a DNA end can trigger a new replication fork by invading homologous DNA, there would seem to be no need to coordinate the processing of the two ends of a DSB - each could simply start a new replication fork on any homologous DNA molecule. Indeed, if the two DNA segments flanking a DSB are homologous to two different plasmid molecules, the DSB is repaired by inducing replication of both plasmids [ 86 ]. While this result clearly shows that the two ends can act independently when forced to do so, other experiments demonstrate that the two broken ends of a DSB are often repaired in a coordinated fashion, using the same template molecule [ 86 , 106 ]. Moreover, Shcherbakov et al. [ 106 ] presented striking evidence that the end coordination is dependent on the gp46/47 complex. The eukaryotic homolog, Rad50/Mre11, has also been implicated in end coordination in DSB repair by a mechanism involving tethering of the two ends via a protein bridge [ 107 , 108 ]. How does end tethering relate to the extensive replication triggered by the broken ends? The simplest explanation is that one end of the DSB triggers a new replication fork on a homolog, and then the second broken end invades one of the two newly-replicated products from that first replication event and triggers a second replication fork in the opposite direction, as diagrammed in Figure 3B [ 86 , 106 ]. Replication fork blockage and restart Replication forks can be blocked or stalled by template lesions, lack of nucleotide substrates, or problems with the replication apparatus. In addition to the natural blockage that appears to occur in normal infections (see above), the consequences of fork blockage and possible pathways for fork restart have been studied using two different inhibitors. First, hydroxyurea (HU) inhibits the reduction of ribonucleotides to deoxyribonucleotides and thereby depletes the nucleotide precursors for replication [ 109 ]. Second, the topoisomerase inhibitor 4'-(9-acridinylamino)-methanesulfon- m -anisidide ( m -AMSA) stabilizes covalent topoisomerase-DNA complexes and thereby physically blocks T4 replication forks [ 110 ]. Wild-type T4 induces breakdown of host DNA, providing a significant source of deoxynucleotide precursors for phage replication and thereby making the phage relatively resistant to HU. One class of HU hypersensitive mutants consists of those defective in the breakdown of host DNA (e.g., denA which encodes DNA endonuclease II) [ 111 , 112 ]. A second well-studied HU hypersensitive mutant class consists of those with knockouts of the uvsW gene [ 71 , 113 ]. These mutants are not defective in host DNA breakdown, and the HU hypersensitivity of uvsW mutants was shown to result from a different genetic pathway than that of denA mutants. We will suggest below that the UvsW protein plays a special role in processing blocked replication forks, namely that it catalyzes a process called replication fork regression. We also suggest that fork regression might somehow lead to efficient replication fork restart, although the details are unclear. Interestingly, the HU hypersensitivity of uvsW knockout mutants can be eliminated by additional knockout of uvsX or uvsY [ 58 ]. This result suggests that the UvsXY homologous recombination system creates some kind of toxic intermediate/product from stalled replication forks when the UvsW protein is unavailable - the nature of this toxic structure is currently unknown. The phage T4 type II DNA topoisomerase is sensitive to anticancer agents, including m -AMSA, that inhibit mammalian type II topoisomerases [ 114 ]. For both enzymes, the drugs stabilize an otherwise transient intermediate in which the enzyme is covalently attached to DNA with a latent enzyme-induced DNA break at the site of linkage. Treatment of phage T4 infections with m -AMSA thereby leads to replication fork blockage at the sites of topoisomerase action [ 110 ]. Interestingly, the blocked replication fork does not immediately resume synthesis when the topoisomerase dissociates from its site of action (and reseals the latent DNA break in the process). This result strongly suggests that key components of the replisome had been disassembled upon fork blockage, so that a fork restart pathway must be used to resume DNA replication. Mutations in genes 46/47 , 59 , uvsX , uvsY , and uvsW each lead to hypersensitivity to m -AMSA, arguing that the RDR pathway or some close variant is required to survive damage caused by m -AMSA [ 115 , 116 ]. Consistent with this model, continued replication of an origin-containing plasmid in the presence of the drug (but not in its absence) was shown to be inhibited in a 46 uvsX double knockout mutant [ 110 ]. The simplest interpretation is that the T4 RDR system allows the restart of replication after the fork blockage event. One plausible scenario is that the blocked replication forks are especially prone to cleavage, for example by a recombination nuclease such as endonuclease VII (gp49), and that the RDR pathway provides a mechanism to restart the broken forks. Evidence supporting this view was obtained when it was found that endonuclease VII can indeed cleave blocked replication forks in vitro , and that blocked forks accumulate to a higher level during infections with a gene 49 knockout mutant [ 117 ]. These latter results led the authors to propose a "collateral damage" model, in which cytotoxic DNA damage from these anticancer agents results from endonuclease-mediated cleavage of stalled replication forks. It should be noted that the processing of forks stalled by HU and m -AMSA must differ significantly, because inactivation of uvsX or uvsY causes hypersensitivity to m -AMSA but not to HU, suggesting that only m -AMSA leads to high levels of broken forks. In an attempt to further study the restart pathway(s) of blocked replication forks in T4 infections, Long and Kreuzer [ 118 , 119 ] analyzed the fork-shaped intermediates ("origin forks") that accumulate at oriG after one replication fork has left the origin region. Novel intermediates were detected by two-dimensional gel electrophoresis at a relatively low abundance in wild-type infections, and these were ascribed to replication fork regression [ 118 ]. Replication fork regression is a process in which the two newly synthesized strands of a replication fork are unwound from their complementary partners and rewound together, backing up (regressing) the location of the fork along the DNA. Many years ago, Higgins et al. [ 120 ] proposed this general model as a step in the accurate replication of damaged DNA in mammalian cells (Figure 4 ). What is the significance of the fork regression at oriG ? A clue was uncovered when the amount of regressed fork was found to be substantially increased when either gp46/47 or gp49 (endonuclease VII) was mutationally inactivated [ 118 ]. The authors therefore proposed that gp46/47 normally processes the extruded duplex of the regressed fork, and that endonuclease VII can cleave the regressed fork (which resembles a Holliday junction). Either of these steps could initiate a fork restart pathway, and it is possible that they normally function together as a single fork reactivation pathway (see [ 118 ]). One possible model for the fork restart pathway is that the extruded duplex in the regressed fork undergoes a strand invasion reaction ahead of the position of the fork, and thereby initiates replication by an RDR reaction (also see above). In a subsequent study, the UvsW helicase was shown to be required for detection of the regressed forks in vivo , and also that purified UvsW helicase can catalyze fork regression in vitro with forked DNA isolated from a T4 infection [ 119 ]. These results strongly suggest that fork regression is an active process that contributes to survival after DNA damage, since uvsW knockout mutants are hypersensitive to DNA damaging agents [ 71 , 113 , 116 ]. Is fork regression required for restart of stalled forks (or other fork-shaped structures) in a T4 infection? The above experiments do not provide a clear answer to this question, because the regressed fork intermediates were only a relatively small subset of the blocked forks and because some of the RDR proteins could themselves be involved in loading a replisome onto a simple fork structure. Indeed, gp59 can bind to simple fork structures in vitro and load the gp41 helicase (see above), supporting the possibility of a simple direct loading pathway. Perhaps multiple processing pathways compete for access to blocked/stalled forks in T4, and the different pathways have unique capabilities to resolve different kinds of problems (e.g., different forms of DNA damage). Replication of damaged DNA by strand switching The general involvement of recombination in the successful replication of damaged DNA was first uncovered in the pioneering experiments of Luria [ 121 ], who discovered the phenomenon of multiplicity reactivation (MR). In MR, co-infection with multiple phages, each of which has extensive DNA damage, results in viable progeny, while single infections with the same phage particles result in no burst. Subsequent studies clearly showed the involvement of both replication and recombination functions in MR (reviewed by [ 122 , 123 ]). A favored model to explain MR involves DNA polymerase strand switching upon encounter with DNA damage, but the molecular details of MR have yet to be elucidated (see [ 123 ] for discussion of this and other models). While further experiments are needed to test the strand-switching model for MR, studies over the last 15 years have provided direct evidence for strand switching in the T4 system. Strand switching may also play important roles in the process of post-replication recombination repair (PRRR) and a pathway called replication repair (see [ 123 ]). Strand switching events can promote the accurate replication of damaged DNA when the second template is a bona fide homolog, either from the opposite daughter duplex behind a replication fork or from another homologous DNA molecule. An in vitro model for this process was established by Kadyrov and Drake [ 98 , 124 ], who engineered replication-fork like substrates with a blocking lesion in the leading strand and a pre-existing lagging strand product that extended past the site of blockage. They were able to demonstrate that the leading-strand product can be extended past the site of blockage by a strand switching event that allows extension using the longer lagging-strand product as template. The simplest way to model the strand switching event is by replication fork regression, followed by polymerase extension on the extruded duplex [ 98 ] (Figure 4 ). To complete the error-free bypass of the DNA damage, a second strand switching event is needed, and this can occur by reversal of the fork regression process. This event was also detected in the studies of Kadyrov and Drake [ 98 ]. The in vitro strand switching analyzed by Kadyrov and Drake had several properties that resemble the replication of damaged DNA during T4 infections. Certain alleles of gene 32 and 41 compromise a process called replication repair in vivo , and these same alleles greatly reduced the strand switching process in vitro [ 124 ]. Furthermore, the UvsX recombinase is centrally important in survival after DNA damage in vivo , and greatly stimulated the strand switching reaction in vitro [ 98 ]. The Dda helicase was also shown to stimulate the in vitro strand switching, but Kadyrov and Drake [ 98 ] suggested that the UvsW helicase was more likely to promote this role in vivo based on the phenotypes of dda and uvsW mutants. Consistent with this suggestion, the UvsW protein was subsequently shown to catalyze fork regression (see above), and very recent evidence has directly demonstrated in vitro strand switching promoted by UvsW [ 125 ]. Even earlier evidence for strand switching in vitro came from reactions in which the polymerase changed templates, presumably at inverted repeat sequences [ 37 , 126 ]. In this reaction, the 3' end of a newly replicated strand base pairs with a short complementary sequence that happens to be on the same strand, resulting in a replication event in which the same strand is used as both template and primer (also see [ 127 ]). This reaction is genetically aberrant and would create genome rearrangements rather than assist in the replication of damaged DNA. Indeed, Schultz et al. [ 128 ] presented evidence that a similar kind of aberrant strand switching can lead to "templated" mutations during a T4 infection. These mutations apparently arise from sequential strand switching events in which DNA polymerase copies an imperfect repeat elsewhere in the template and then returns to the correct initial location on the template. Interestingly, these templated mutations became more frequent with certain mutations in genes 32 , 41 and uvsX , arguing that these proteins normally help to accurately direct the template switching events, e.g. to the opposite daughter strand rather than to ectopic locations elsewhere in the genome.

Conclusions and perspectives Phage T4 has continued to provide an important model system for studies of the mechanisms of DNA replication, recombination and repair, and in several cases has led the way in illuminating the interconnections between these processes. A major example is RDR, a process that was first studied in detail in phage T4, that was originally thought to be an odd peculiarity of the phage's life cycle, but that is now appreciated as central in the completion of cellular genomic replication and the repair of DSB's in prokaryotic and eukaryotic chromosomes (for reviews, see [ 1 - 4 ]). Recombination-related pathways, including RDR, strand-switching and replication fork regression, are now appreciated to be critical in the maintenance of genome stability in mammalian systems and thereby important in cancer biology. There seems to be particularly strong parallels between DNA metabolism in phage T4 and in eukaryotic mitochondrial DNA and mitochondrial plasmid DNA. In these systems, evidence has been obtained for both R-loop-mediated replication and RDR, as well as EM data showing branched concatameric DNA similar to that of intracellular replicating T4 DNA [ 129 - 135 ]. While we have learned much about how T4 initiates replication at both origins and from recombination structures, many important questions remain to be answered. We will close with a few of the most interesting, which suggest that important principles and lessons remain to be uncovered using the T4 model system: (i) What are the rules governing R-loop formation at oriF and oriG ? (ii) Do other T4 origins use an R-loop mechanism or some other initiation process? (iii) What are the factors that govern origin usage and change the pattern of origin function? (iv) What are the precise roles of recombination proteins, gp59 and Dda in origin usage? (v) Why do replication forks often fail to complete replication, or move very slowly, at early times of infection? (vi) Does T4 use a "direct restart" pathway in vivo , in which the replisome is loaded directly onto a fork structure? (vii) What are the detailed roles of the gp46/47 complex, the homolog of eukaryotic Mre11/Rad50 complex? (viii) How does replication fork regression contribute to replication fork restart? (ix) How frequently does T4 use strand switching mechanisms in vivo , which proteins are required, and how is the process regulated?

Bacteriophage T4 initiates DNA replication from specialized structures that form in its genome. Immediately after infection, RNA-DNA hybrids (R-loops) occur on (at least some) replication origins, with the annealed RNA serving as a primer for leading-strand synthesis in one direction. As the infection progresses, replication initiation becomes dependent on recombination proteins in a process called recombination-dependent replication (RDR). RDR occurs when the replication machinery is assembled onto D-loop recombination intermediates, and in this case, the invading 3' DNA end is used as a primer for leading strand synthesis. Over the last 15 years, these two modes of T4 DNA replication initiation have been studied in vivo using a variety of approaches, including replication of plasmids with segments of the T4 genome, analysis of replication intermediates by two-dimensional gel electrophoresis, and genomic approaches that measure DNA copy number as the infection progresses. In addition, biochemical approaches have reconstituted replication from origin R-loop structures and have clarified some detailed roles of both replication and recombination proteins in the process of RDR and related pathways. We will also discuss the parallels between T4 DNA replication modes and similar events in cellular and eukaryotic organelle DNA replication, and close with some current questions of interest concerning the mechanisms of replication, recombination and repair in phage T4.

Competing interests The authors declare that they have no competing interests. Authors' contributions KK wrote the first drafts of the "Introduction", the section on "Recombination-dependent replication", the section on "Interrelationship between replication, recombination and repair", and the "Conclusions and perspectives"; JRB wrote the first draft of the section on "Overlap between origin- and recombination-dependent mechanisms"; both authors contributed to the first draft of the section on "Origin-dependent replication"; both authors revised all sections and read and approved the final draft.

CC BY

no

2022-01-12 15:21:44

Virol J. 2010 Dec 3; 7:358

oa_package/66/45/PMC3016281.tar.gz

PMC3016282

21134275

Background Dengue is a vector borne disease rapidly spreading in urban areas in tropical and subtropical countries. It is estimated that at least 10% of dengue fever cases evolve to severe and eventually lethal forms of the disease. The clinical and laboratory findings in dengue are very similar to those of other febrile diseases that are prevalent in the same geographical regions [ 1 ]. Therefore, a dengue diagnostic test is required for adequate case management and to reduce misclassification in the dengue surveillance system. However, dengue diagnosis in the first days of fever is yet problematic. There are three main laboratory methods to diagnose dengue infection: viral isolation in culture, detection of viral RNA, and specific IgM/IgG antibodies in paired sera. The gold standard is usually a combination of these methods [ 1 , 2 ]. Viral isolation is costly, the results are usually available after 6 to 10 days and it is only obtainable in laboratories with the appropriate infrastructure for cell culture or mosquito colonies. The RT-PCR and other PCR-based techniques give results within 24 hours but they are also costly and they are not available for most clinicians. On the contrary, there are commercially available immunochromatographic and ELISA tests for the detection of IgM/IgG antibodies which give results within minutes or few hours. However, the detection of antibodies in a dengue infected person is only possible after 4-5 days of disease onset. Moreover, a single positive IgM or IgG result suggests recent infection but paired sera samples showing seroconversion or a fourfold titer increase are required to confirm diagnosis [ 1 ]. Recently, several dengue diagnostic tests based on the detection of NS1 (Non-structural Protein 1) have become commercially available. NS1 is a highly conserved glycoprotein of flaviviruses including Dengue, Japanese encephalitis, Yellow fever and tick-borne encephalitis virus [ 3 ]. The specificity of the NS1-based Dengue tests is reported to be between 86.1% and 100% and false positives are considered rare [ 4 , 5 ]. Higher variability (between 37% and 98.9%) has been reported in the sensitivity of these tests (Table 1 ) [ 6 - 24 ]. This variability could be partly explained by the fact that sensitivity has been found to decrease with time after fever onset and in secondary infections [ 12 , 18 , 21 ]. The addition of IgM and IgG specific antibodies detection to NS1-based tests in a single kit has been suggested [ 25 ] may improve the assessment of dengue infection status and one such test (SD BIOLINETM Dengue Duo) has become commercially available. With all these options in the market, it is necessary to identify which of the current NS1-based diagnostic tests would be potentially more useful in the clinical setting. We sought to compare the performance of the current commercially available NS1-based assays for the early diagnosis (within 7 days since fever onset) of dengue infections. The objectives of this study were: 1) To identify differences in sensitivity, specificity, and likelihood ratios between all the diagnostic assays, 2) To describe the effect of duration of symptoms, type of infection, viral serotype, and severity of the disease on the sensitivity of the tests, and 3) to determine the reproducibility of each diagnostic test.

Methods Type of study and sample size calculation The study was a cross sectional case-reference design to assess diagnostic tests [ 26 ]. A paired analysis of samples from febrile subjects with and without dengue was done using viral isolation, RT-PCR or IgM seroconversion as gold standard. Sample size for dengue (n = 210) and non-dengue (n = 100) was estimated based on an expected 90% sensitivity and 100% specificity for the PlateliaTM test versus 80% sensitivity and 90% specificity for the other assays. The Conner method for the paired McNemar test was used for sample size calculation with a 5% alfa and 20% beta errors [ 27 ]. Half dengue and no dengue samples were used to assess reproducibility. Clinical samples Stored serum (229, 73.9%) or plasma (81, 26.1%) samples from febrile subjects with clinically-suspected dengue infection who took part in studies carried out by Universidad del Valle and Universidad Industrial de Santander in Colombia between 2004 and 2008 were selected randomly. The following criteria were considered: 1) dengue status known as a result of one or more of the following: viral isolation, RT-PCR or IgM seroconversion, 2) sample taken between day 0 and 7 of onset of fever, and 3) a minimum of 1 mL volume available. Day 0 was defined as the same day of fever onset. To avoid the spectrum bias, samples representing subjects who had been previously classified as dengue fever and hemorrhagic dengue were included and further classified as non-severe and severe dengue, respectively [ 1 ]. Gold standard tests (viral culture, nested RT-PCR or paired IgM) had been done during the previous studies at the virology laboratory of Universidad del Valle. Briefly, for viral isolation sera samples had been cultured in the mosquito cell line clone C6/36 HT and incubated at 33°C for 10 to 14 days. Viruses were detected and identified by immunofluoresce with serotype specific monoclonal antibodies 15F3 (DENV1), 3H5 (DENV2), 5D4 (DENV3) and 1H10 (DENV4) (Chemicon International, Inc. Temecula, California) and fluorescein isothiocyanate-conjugated goat anti-mouse antibody [ 28 ]. For RT-PCR, viral RNA had been extracted with trizol (Gibco-BRL, Gaithersburg, MD) and cDNA obtained with the reverse transcriptase of the Avian myeloblastosis virus (Promega, Madison, WI) and a dengue universal antisense primer targeting the C/prM region of the genome. cDNA amplification was performed with a nested PCR using the same universal dengue primers in a first round of amplification and viral serotype specific primers in a second round of PCR [ 29 ]. Finally, IgM MAC-ELISA in paired samples had been done using affinity-purified goat anti-human IgM as a capture antibody (KPL; Gaitersburg, Maryland 1 μg/ml), followed by addition of 1:40 dilution of serum samples duplicates. Assay antigen was home-made and consisted of a mixture of 4 HA U (hemoagglutinating units) each of the four dengue serotypes obtained by i.c. inoculation of suckling mice and antigen extraction by a sucrose/acetone gradient. Detection was performed using 1:10,000 dilution of a peroxidase-conjugated dengue-complex specific monoclonal antibody MAB 6B6C-1 (kindly provided by CDC, San Juan de Puerto Rico) and substrate p-nitrophenyl-phosphate [ 30 , 31 ]. Positivity was defined as having an assay absorbance of ≥2.0 (405 nm) after subtracting the background value (negative sample). Because up to 30% of secondary dengue infections do not have detectable IgM [ 32 ], and most non-dengue samples had been analyzed only by paired IgM, samples classified as non-dengue were further analyzed using an algorithm of RT-PCR plus IgG and IH. All non-dengue samples (except for four samples with insufficient volume left) were processed with RT-PCR as described elsewhere [ 29 ]. RT-PCR positives were considered as dengue. To discard secondary infections which do not increased IgM, all RT-PCR negative non-dengue samples underwent IgG detection in acute sera using Dengue Duo (IgM/IgG) Cassette (Inverness - Brisbane, Australia). Non-dengue samples with negative IgG were considered as true negatives. Those samples with positive IgG were processed with haemagglutination-inhibition test (HI) and considered as true negatives if not increased (>2560) titers were detected in convalescent sera (Figure 1 ). HI was done at the virology laboratory of Universidad del Valle using goose red blood cells and sucrose/acetone extracted antigens obtained in suckling mice brains following Kuno et al. 1991 [ 33 ]. This study was approved by the Universidad del Valle Ethics Review Board. Diagnostic tests and procedures All 5 diagnostic NS1-based tests commercially available at the time of the study were analyzed. These included: PlateliaTM Dengue NS1 Ag Test (Bio-Rad Laboratories - Marnes La Coquette, France), second generation Pan-ETM Dengue Early ELISA (Inverness - Brisbane, Australia), Dengue NS1 Ag ELISA (Standard diagnostic Inc. - Kyonggi-do - South Korea), Dengue NS1 Ag STRIPTM (Bio-Rad), and SD BIOLINETM Dengue Duo (Standard diagnostic Inc.). The characteristics of the tests are summarized in table 2 . The PlateliaTM Dengue NS1 Ag Test and Dengue NS1 Ag STRIPTM were purchased from the local distributor while the rest were kindly donated by the manufacturers. All tests were run following the corresponding manufacturer's instructions. Dengue NS1 Ag STRIPTM was read at 15 min and 30 min. Three separate results were obtained from SD BIOLINETM Dengue Duo test based on the results of NS1 only (dengue if NS1 was positive and non-dengue if NS1 was negative, regardless of IgM/IgG results), NS1/IgM combined (dengue if one of NS1 or IgM was positive and non-dengue if both were negative, regardless of IgG results), and NS1/IgM/IgG combined (dengue if at least one of NS1, IgM or IgG was positive and non-dengue if all three were negative). Batches of samples were analyzed by all the NS1-based diagnostic tests on the same day and by the same persons who were two experienced lab scientists. Both observers were blind to the samples dengue status and each other results. Results of the ELISA-based format tests given as "equivocal" were repeated once. Persistent equivocal results were excluded from the analysis. Those results of the immunochromatography-based format tests given as "weak" were considered as positive results. Statistical analysis Data were double entered and validated using Epinfo (Centers for Disease Control and Prevention, USA, 2000). Stata 10 (Stata Corporation, 2003) was used for statistical analyses. First observer results were used to obtain sensitivity, specificity, negative (NPV) and positive (PPV) predictive values, positive and negative likelihood ratios (LR) with their corresponding 95% confidence intervals. Cochrane Q was used to compare overall performance of ELISA tests and of immunocromatographic tests. McNemar Chi squared test or the equivalent exact test was used to compare the diagnostic accuracy among each possible pair of assays. The method proposed by Roldan-Nofuentes and Del Castillo (2007) was used to identify significant statistical differences in the LR of all tests [ 34 ] and carried out in Mathematica 7 (Wolfram Research Inc., 2010). Sensitivities with their corresponding 95% confidence intervals were also calculated by stratum of duration of symptoms (≤3 and 4-7 days), primary/secondary infection (defined as absence/presence of specific IgG in acute sera based on the results of the SD BiolineTM Dengue Duo), severe and non severe infection, and viral serotype. Reproducibility of the tests (inter-observer agreement) was assessed using Kappa indexes (k). We interpreted k results as follows: values of less than 0, poor; 0 to 0.2, slight; 0.2 to 0.4, fair agreement; 0.4 to 0.6, moderate agreement; 0.6 to 0.8, substantial agreement; and values of 0.8 to 1.0 almost perfect agreement [ 35 ]. Funds allowed us to purchase a limited number of Dengue NS1 Ag STRIPTM and hence results were available for 147 samples (104 dengue and 43 non-dengue). It was not possible to assess reproducibility of this test. A P value <5% was considered as statistically significant.

Results A total of 310 samples were included in the study from which 210 were classified as dengue and 100 as non-dengue. Eight samples initially classified as non-dengue based on IgM negative results in paired serum samples were RT-PCR-positive and hence were reclassified as dengue. Therefore, for the final analysis there were 218 dengue and 92 non-dengue cases. Samples represented all age groups and had a median of 3 days of fever onset. Nine samples analyzed by PlateliaTM and 2 by Pan ETM gave equivocal results and were run twice. The second time, both Pan ETM and 2 PlateliaTM results were negative while the other 7 (1 non-dengue and 6 dengue) remained equivocal and were excluded from the final analyses (Figure 1 ). Sixty four (29.4%) dengue samples were positive for IgG in the SD BiolineTM Dengue Duo and were considered as secondary infections. Secondary infections had a median of 4 (range 2-7) days of fever onset and dengue serotype was identified in 42 of these cases: 13 DENV1, 17 DENV2, 7 DENV3, and 5 DENV4. Sensitivity and specificity of tests ranged from 51% to 80.7% and from 89.1% to 96.7%, respectively (Table 3 ). SD BIOLINETM NS1/IgM/IgG had the highest sensitivity (80.7% 95%CI 75-85.7) followed by SD BIOLINETM NS1/IgM (78.4% 95%CI 72.4-83.7) and Pan ETM (71.1% 95%CI 64.6-77). There were not statistically significant differences in the diagnostic accuracy of the three ELISA-based assays (p = 0.9). For the immunocromatographic tests, STRIPTM read at 30 min had a higher diagnostic accuracy (71.4%, 105/147) than STRIPTM read at 15 min (68.7%, 101/147) but this difference was not statistically significant (p = 0.1). The diagnostic accuracy did not differ among SD BIOLINETM NS1/IgM (82.2%, 255/310) and SD BIOLINETM NS1/IgM/IgG 83.2, 258/310 (p = 0.4). However, their diagnostic accuracy was higher than all the other immunochromatographic and ELISA tests (p<0.05 for all pair wise comparisons). In line with the relatively high specificity found, the PPVs were above 90% for all tests. In contrast, the highest NPV was 66.1% (95%CI 57.1-74.4). LR+ varied between 6.5 and 15.6 while LR-varied between 0.2 and 0.5 (Table 3 ). Statistically significant differences in LR were found between all tests pair wise comparisons except PlateliaTM Vs. PanETM, PlateliaTM Vs. STRIPTM, and ELISA SDTM Vs. STRIPTM. The sensitivity of NS1-based diagnostic tests significantly decreased in those samples taken after 3 days of fever onset, in secondary infections, viral serotypes 2 and 4, and severe dengue. Adding IgM or IgG to SD BIOLINETM NS1 increased its sensitivity in all these situations (Figure 2 ). The positive effect of adding IgM to NS1 in the sensitivity of the test was more noticeable in samples with detectable IgG regardless of the days of fever onset (Table 4 ). The inter-observer agreement was almost perfect for Pan ETM (k = 0.94 CI95% 0.88-0.99), ELISA SDTM (k = 0.89 CI95% 0.82-0.96), SD BIOLINETM NS1 (k = 0.90 CI95% 0.83-0.98), and the IgM cassette of SD BIOLINETM (k = 0.85 CI95% 0.76-0.94) while substantial agreement was observed with PlateliaTM (k = 0.75 CI95% 0.60-0.89) and the IgG cassette of SD BIOLINETM (k = 0.7 CI95% 0.56-0.84).

Discussion In the present study we compared simultaneously the performance of 5 commercially available tests for the early (within 7 days of fever onset) diagnosis of dengue. The sensitivity (51% to 80.7%) and specificity (89.1% to 96.7%) of the NS1-based tests found in the present study fell within the range described elsewhere (Table 1 ). In previous comparative studies the sensitivity of PlateliaTM (71.3%-87.4%) was consistently higher than STRIPTM (67.8%-82.4%) and, in turn, the sensitivity of STRIPTM was higher than Pan ETM (60.4%-64.9%). By contrast, we did not find differences in the diagnostic accuracy of the ELISA-format diagnostic tests (PlateliaTM, Pan ETM and ELISA SDTM). In the present study, Pan ETM sensitivity was higher than in the previous reports probably because we used Pan ETM second generation, which uses less diluted controls and samples (1:2 instead 1:10) than the previous version [ 36 ]. Despite ELISA-format tests showing comparable sensitivities, they were all below 75%. The immunocromatographic-format tests that detect only NS1 had even lower sensitivities. This means that a negative result on any of these tests does not rule out dengue. The immunocromatographic SD BiolineTM that detects simultaneously NS1 and specific IgM/IgG showed the highest sensitivity (80.7% CI95% 75-85.7) which was comparable to the 83.7% (95%CI 78.4 - 88.1) reported in Vietnam [ 6 ]. Similarly, the addition of IgM has shown to improve the sensitivity of NS1 ELISA-format tests from 63.2% to 79% on admission samples without significantly decreasing specificity [ 22 ]. Although we did not find statistically significant differences in the addition of IgM only or both antibodies to NS1, the use of IgG could have clinical significance when correlated with disease evolution and the days of fever onset. In any case, a positive result for IgM or IgG in a single sample does not confirm dengue, therefore; the impact of false positives in the routine clinical setting should be assessed. Predictive values depend on the prevalence of the disease but their trend here showed that all tests were comparable. For potential clinical use, LR measures of diagnostic tests performance are more useful than predictive values. They tell how the test results modified the pretest probability of disease independent of its prevalence. LR values above 10 and below 0.1 are considered conclusive to rule in or rule out diagnoses, respectively while values of 5 to 10 and 0.1 to 0.2 are frequently helpful to take clinical decisions [ 37 ]. In a scenario where a clinician's interest is to confirm dengue diagnosis any of the tests is likely to be useful but they should be aware that a negative test does not rule out dengue. Hence, further diagnostics such as paired IgM or IgG to assess seroconversion or titer increase would need to be done. On the contrary, if ruling out dengue is important for clinical decision then the SD BiolineTM NS1/IgM/IgG would provide more useful information than any of the other tests (LR = 0.21). However, it ought not to be considered as a screening test. Further studies with larger sample size would be required to give more precise estimates of LR. The simultaneous detection of NS1 and specific IgM/IgG appeared to overcome the limitations of using only NS1-based diagnostic tests in secondary infections, in subjects who seek treatment after 3 days of disease onset and on severe cases. Our results confirmed that sensitivity of NS1-based tests decreased in secondary infections and with time of fever onset [ 12 , 18 , 20 , 21 , 23 ]. In contrast to Chaiyaratana et al. (2009), we also found a decreased sensitivity of all assays in severe cases because, in our study, these were more likely to be secondary infections (OR = 3,01 95%CI 1.57-5.78) and presented after 4 days of fever (OR = 4.31 95%CI 2.04-9.1) than non-severe infections [ 15 ]. The addition of IgM/IgG appeared to increase the sensitivity of NS1 BiolineTM in all 4 dengue serotypes. Nevertheless, the sensitivities of all tests were consistently lower in DENV2 and DENV4 infections as was also observed in Venezuela [ 13 ] and Puerto Rico [ 19 ]. The frequency of secondary infections and time since disease onset could only partly explain these differences because, in our study, only DENV4 tended to be associated with the presence of IgG in acute sera and samples taken after 4 days of fever (data not shown). Viraemia levels were found to be significantly higher in DENV1 than DENV2 and DENV4 infected subjects in Vietnam [ 38 ]. Therefore, differences in viral loads could be an alternative explanation for the observed effect of serotype in NS1-based tests sensitivity. In spite of this, no differences in sensitivity of PlateliaTM and STRIPTM according to serotype had been reported in samples from French Guiana [ 20 , 24 ]. Therefore, other reasons such as a decreased avidity of NS1 tests for certain geographical dengue virus clades, as proposed before, could be explored [ 13 ]. One limitation of the present study was residual misclassification of non-dengue cases, which would underestimate the specificity of the tests. This is probably low because most negative (92/98) samples were analyzed by at least two gold standard methods. Confirmation of non-dengue diagnosis was not sought due to limited sample volume. Specificities were relatively high and similar to previous reports in South America [ 13 ]. Further misclassification is likely in secondary and primary infections as we did not use a gold standard method such as a quantitative HI due to limited funds and in some samples not enough available volume. The use of one of the study tests (SD BiolineTM) to define presence of both IgG and IgM would tend to overestimate the sensitivity of IgM in secondary infections. Nevertheless, the findings were consistent with previous reports of the increased test sensitivity in secondary infections provided by the addition of IgM to NS1 [ 6 ]. Yellow Fever vaccination (YFV) also is a source of potential false positives with IgG but this information was not available for the study samples. YFV is recommended in Colombia and therefore it is important to include information on YFV status of subjects in future studies to assess the degree of misclassification.

Conclusions Of the 5 tests assessed, SD BiolineTM NS1/IgM/IgG performed significantly better than the other tests. Therefore, the simultaneous detection of NS1/IgM/IgG would be potentially useful to diagnose dengue in both endemic and non endemic areas. All NS1 tests were highly reproducible. Clinicians must be aware that a negative result does not rule out dengue. To take evidence based decisions about the usefulness of this test in clinical settings, it is recommended to assess its performance in consecutive subjects with potential dengue infection under routine conditions at health centers with different levels of complexity. Further studies are required to assess the potential impact of implementing early laboratory diagnosis of dengue in terms of prognosis and cost-effectiveness. Secondary infection, viral serotype and time since fever onset should be taken into account as sources of heterogeneity in the interpretation and meta-analysis of NS1-based diagnostic tests.

Background We compared the diagnostic accuracy and reproducibility of commercially available NS1-based dengue tests and explored factors influencing their sensitivities. Methods Paired analysis of 310 samples previously characterized as positive (n = 218) and negative (n = 92) for viral isolation and/or RT-PCR and/or IgM seroconversion. Masked samples were tested by two observers with PlateliaTM Dengue NS1 Ag, second generation Pan-ETM Dengue Early ELISA, SD Dengue NS1 Ag ELISA, Dengue NS1 Ag STRIPTM, and SD BIOLINETM Dengue Duo (NS1/IgM/IgG). Results SD BIOLINETM NS1/IgM/IgG had the highest sensitivity (80.7% 95%CI 75-85.7) with likelihood ratios of 7.4 (95%CI 4.1-13.8) and 0.21 (95%CI 0.16-0.28). The ELISA-format tests showed comparable sensitivities; all below 75%. STRIPTM and SD NS1 had even lower sensitivities (<65%). The sensitivities significantly decreased in samples taken after 3 days of fever onset, in secondary infections, viral serotypes 2 and 4, and severe dengue. Adding IgM or IgG to SD NS1 increased its sensitivity in all these situations. Conclusions The simultaneous detection of NS1/IgM/IgG would be potentially useful for dengue diagnosis in both endemic and non endemic areas. A negative result does not rule out dengue. Further studies are required to assess the performance and impact of early laboratory diagnosis of dengue in the routine clinical setting.

Competing interests None of the authors have received any gifts, travel funds, have served as consultant, speaker or received previous funding from the diagnostic tests manufacturers. The present study was partially funded by Standard Diagnostics Inc. However, none of the diagnostic test manufacturers had a role, either directly or through a third party, in the gathering or preparation of data, in the writing of the manuscript or the decision to submit the manuscript for publication. There are not any other financial and non-financial competing interests. Authors' contributions LO designed the study, conducted the data analysis and wrote the manuscript; MR designed the study, entered and analyzed the data and critically reviewed the manuscript; AB contributed to acquisition of data and interpretation of the results, run the tests as a second observer and critically reviewed the manuscript; LAV contributed to acquisition of data and interpretation of the results, and critically reviewed the manuscript; BP contributed to acquisition of data and interpretation of the results, run the tests as a first observer and co-wrote the manuscript. All authors read and approved the final manuscript.

Acknowledgements We are grateful to Jaime Muñoz, Natalia Basto, Graciela Rengifo, and Olga Lucia Agudelo at Univalle who contributed with sample selection, preparation and RT-PCR and HI tests and Dr. Ruth Martinez at UIS for her contribution in data collection. The statistical function for LR comparisons was kindly provided by Dr. José Antonio Roldan-Nofuentes. We thank the manufacturers who gently provided their tests kits. This study was funded by Universidad del Valle Project ID 613 and Standard Diagnostics contract 13102009

CC BY

no

2022-01-12 15:21:44

Virol J. 2010 Dec 6; 7:361

oa_package/75/05/PMC3016282.tar.gz

PMC3016283

21156081

Background Human papillomaviruses are small circular DNA viruses that infect epithelial cells and normally replicate as nuclear plasmids. The life cycle of papillomavirus is tightly linked to epithelial differentiation [ 1 ]. Among the high-risk HPV types associated with cervical cancer, human papillomavirus type 58 (HPV58) plays a more prominent role in Asian countries. HPV58 has been found in 5.9% of cervical cancer patients in China [ 2 ], with an unusually high prevalence in cervical cancer patients in specific areas of China: 33.3% in Hong Kong [ 3 ] and 16.3% in Shanghai [ 4 ]. Despite the availability for biological study of few cell lines containing the DNA of high-risk HPVs such as 16, 18 and 31, no cell lines or animal models containing HPV58 have been established. Saccharomyces cerevisiae is a species of budding yeast. The cellular mechanism required for DNA replication in S. cerevisiae is similar to that in human cells [ 5 ]. Studies have shown yeast to be a versatile organism for the study of viruses. Many types of DNA and RNA viruses, including HPVs, can directly replicate in yeast [ 6 ]. Although HPV6, 16 and 31 can replicate stably in yeast cells as nuclear plasmids [ 7 , 8 ], whether HPV58 genome can replicate stably in yeast and whether the viral genes can be transcribed in yeast are unknown. In the present study, we first explored the replication and transcription of HPV58 genome in the yeast system and investigated the function of E2 on vrial DNA replication and transcription. We found that HPV58 genome could replicate stably as an episome, with transcription of its early and late genes in yeast. However, with mutation of the E2 gene, HPV58 genome lost its mitotic stability and the transcription levels of E6 and E7 genes were upregulated. Thus, E2 protein can facilitate the replication and maintenance of HPV58 DNA and regulate viral gene transcription in yeast cells.

Methods DNA construction HPV58-Ura The Ura gene was amplified from S. cerevisiae plasmid pRS316 with the sense (5'-GATC CACCGGTG GCAGATTGTACTGAGAGTG-3') and anti-sense (5'-CTAG CACCG GTG TAGTATACATGCATTTAC-3') primers containing the Sgr A I site (underlined). The PCR-produced Ura was subcloned into the L2 open reading frame (ORF) of pEGFPN1-HPV58 to construct a pEGFPN1-HPV58-Ura plasmid. The HPV58-Ura was released from pEGFPN1-HPV58-Ura by Bgl II digestion and recircularized by T4 DNA ligase for yeast transformation. HPV58-Ura-E2 mutant (HPV58-Ura-E2mt) A 1236 bp cassette was PCR amplified from pEGFPN1-HPV58 with the sense (5'-C CACCAGGTG TAATGATGA TTGGTAGCATC AAAGAC-3') and anti-sense (5'-CATAC CACCATGTG CAGAACCA-3') primers containing the Dra III site (underlined) and three stop codons (bold). The cassette was then substituted for the Dra III digested fragment (2924-4145 bp in HPV58 genome) in the pEGFPN1-HPV58-Ura plasmid to construct a pEGFPN1-HPV58-Ura-E2mt plasmid with three stop codons (2927-2932 bp) in the E2 ORF (2753-3829 bp). The HPV58-Ura-E2mt was released by Bgl II and recircularized for transformation. pDBLeu-E2 The E2 protein expression plasmid was constructed as follows: E2 ORF was PCR amplified with the sense primer (5'-CC AAGCTT GAAAATTGGAAATCCT-3') containing the Hind III site (underlined) and anti-sense primer (5'-CT GCTAGC TTA CAAGT CTTCTTCAGAGATCAACTTCTGTTC CAATGACATAACACCAGTACT-3') containing the Nhe I site (underlined) and a cMyc tag (bold). The E2 PCR product was subcloned into pDBLeu to construct the pDBLeu-E2. Yeast transformation S. cerevisiae strain W303-1B ( MAT α leu2-3 leu2-112 trp1-1 ura3-1 his3-11 his3-15 ade2-1 can1-100 ) were transformed with different sets of plasmids: 1). pRS316; 2). HPV58-Ura; 3). HPV58-Ura-E2mt; 4). HPV58-Ura-E2mt and pDBLeu (HPV58-Ura-E2mt/pDBLeu); 5). HPV58-Ura-E2mt and pDBLeu-E2 (HPV58-Ura-E2mt/pDBLeu-E2). The transformed yeast were spotted on selective medium plates for 3-5 days. Single colonies were selected and cultured in yeast extract/peptone/dextrose (YPD) or synthetic complete (SC) dropout media (Clontech) for further analysis. Quantitative PCR (qPCR) Yeast were cultured in 10 ml selective medium to an OD600 of 1.0 and yeast DNA was isolated as described [ 7 ]. The DNA was analysed by absolute qPCR with primers specific for E1 (sense: 5'-CTGCAATGGATGACCCTGAAG-3'; anti-sense: 5'-CCACTATCGTCTGCTGTTTCGT-3', amplicon: 136 bp, 878-1013 bp). Yeast 18S rDNA was used as internal control (sense: 5'-TTGTGCTGGCGATGGTTCA-3'; anti-sense: 5'-TGCTGCCTTCCTTGGATGTG-3', amplicon: 152 bp). A standard curve was generated by amplification of a serial dilution of pEGFPN1-HPV58-Ura. Southern blot analysis Yeast harboring HPV58-Ura and HPV58-Ura-E2mt/pDBLeu-E2 were grown in 25-ml selective medium overnight to yield an OD600 of 1.0. HPV58-Ura-E2mt transformed yeast were cultured in 25 ml selective medium for 3 days to yield an OD600 of 0.2, because of the poor growth in selective medium. Yeast DNA was isolated and digested with Xho I (no cut on HPV58 genome), Bgl II (1 cut), Hpa I (1 cut) or Dpn I (9 cuts) for 24 hr. Dpn I can digest methylated DNA isolated from bacteria only. The DNA was electrophoresed, blotted onto nylon membrane (Roche) and probed with an L1 specific mRNA probe labeled with digoxigenin by in vitro transcription according to the manufacturer's instructions (DIG RNA Labeling Kit, Roche). Western blot analysis Yeast protein was prepared from yeast harboring pDBLeu-E2, pDBLeu, and untransformed yeast as previously described [ 9 ] Protein samples were separated by SDS-PAGE electrophoresis and transferred to nitrocellulose membrane. Then the membrane was blocked with 3% BSA in PBST and immunoblotted with antibody against cMyc (Santa Cruz). Membrane was washed and incubated with HRP conjugated secondary antibody. Chemilucent ECL Detection System (Millipore) was used to detect the signals according to the manufacturer's instruction. HPV58 genome stability assay The DNA stability assay was performed as previously described [ 10 ]. Transformed yeast were first grown in selective medium to mid-log phase and diluted to an OD600 of 0.1 in new cultures containing non-selective medium. The cultures were grown for 17 hr (10 cell generations). The cultures at either 0 or 10 generations were diluted to an OD600 of 0.1. Aliquots of 5 μl were spotted to selective and non-selective media. After 3 days of growth, the percentage of colonies containing viral DNA was determined by the ratio of the number of colonies on selective medium to those on non-selective medium. The percentage of DNA loss per cell generation was calculated by subtracting the percentage DNA retained after 10 generations from that at 0 generation and divided by the total number of generations. RNA extraction, RT-PCR and quantitative RT-PCR (qRT-PCR) Yeast were cultured in 10 ml selective medium to an OD600 of 1.0 or 0.2 (HPV58-Ura-E2mt transformed yeast). Yeast DNA and RNA were isolated from the same samples. Yeast total RNA was isolated as previously described [ 11 ] and digested by DNase I (Fermentas) to remove the contaminating DNA. PCR involved use of DNase I-treated RNA as a template to ensure the complete digestion of contaminating DNA in RNA samples. The DNase I treated RNA was analysed by RT-PCR and absolute qRT-PCR with primers specific for E1 (sense: 5'-CTGCAATGGATGACCCTGAAG-3'; anti-sense: 5'-CCACTATCGTCTGCTGTTTCGT-3', amplicon: 136 bp, 878-1013 bp), E2 (sense: 5'-GACAAAGCGACGACGACT-3'; anti-sense: 5'-GTCGTTGTGTTTCCGTTGT-3', amplicon: 335 bp, 3427- 3761 bp), E6 (sense: 5'-ACTATGTTCCAGGACGCAGAG-3'; anti-sense: 5'-ACCTCAGATCGCTGCAAAG-3', amplicon: 128 bp, 107-234 bp), E7 (sense: 5'-GACGAGGATGAAATAGGCTTG-3'; anti-sense:5'-CGTCGGTTGTTGTACTGTTGA-3', amplicon: 133 bp, 670-802 bp), L1 (sense: 5'-CTTGAAATAGGTAGGGGACAG-3'; anti-sense: 5'-CAATGGAGGACAATCAGTAGC-3', amplicon: 249 bp, 5958-6206 bp) and L2 (sense: 5'-CATAGTGACATATCGCCTGCTC-3'; anti-sense: 5'-AGCCCCTATTTGCT TTCCAC-3', amplicon: 153 bp, 5051-5203 bp) respectively. Standard curves were generated by amplification of a serial dilution of pEGFPN1-HPV58-Ura. Relative qRT-PCR was ued to compare the transcription levels of E6 and E7 genes with or without E2 protein. Because HPV58 genomes have different replication efficiency in HPV58-Ura-, HPV58-Ura-E2mt- and HPV58-Ura-E2mt/pDBLeuE2-transformed yeast, we first compared the relative replication levels of the HPV58 genome in different transformants by qPCR. The E1 primers described previously and 18S rDNA primers (sense:5'-TTGTGCTGGCGATGGTTCA-3'; anti-sense: 5'-TGCTGCCTTCCTTGGATGTG -3', amplicon: 152 bp, as internal control) were used in qPCR. Then qRT-PCR was performed to analyze the relative transcription levels of E6 and E7 genes of HPV58-Ura, HPV58-Ura-E2mt and HPV58-Ura-E2mt/pDBLeu-E2 in yeast, with the 18S rRNA as internal control. The value of relative transcription levels was divided by the value of DNA relative replication levels to standardize the transcription templates.

Results Episomal replication of HPV58 in yeast Viral DNA copy number was detected by qPCR from five continuous yeast passages as shown in Table 1 . The viral DNA copy number in per microliter of yeast DNA is relatively consistent in the five continous passages. Averagely, there are 3-5 copies of viral DNA in per yeast cell. Southern blot was performed to investigate the replication form of HPV58 genome in yeast. As shown in Figure 1A and 1B , when yeast DNA was treated with Dpn I or restriction enzymes which have no recognition site in HPV58 genome, the main form of HPV58 genome was open circular (OC). The vortex process during yeast DNA isolation may disrupt the supercoiled plasmid (SC) and lead to the formation of OC plasmid. Yeast DNA digested by a single-cut restriction enzyme revealed only a single band representing the linear form of HPV58 (Figure 1A ). Therefore, HPV58 genome could replicate episomally in yeast. Furthermore, no band was detected in the Dpn I treated DNA isolated from HPV58-Ura-E2mt transformed yeast (Figure 1B , lane 1), which indicates that the E2 gene mutation induced instability and decreased the replication level of viral DNA. Transformation of pDBLeu-E2 into yeast restored the mitotic stability of HPV58-Ura-E2mt and clear bands were detected (Figure 1B , lane 2-6). We have tried to detect and compare the expression levels of E2 protein from HPV58-Ura and pDBLeu-E2 with anti-HPV16 E2 antibody, but no bands were detected. The transcription levels of E2 from HPV58-Ura and pDBLeu-E2 were compared with qRT-PCR, the mRNA level of E2 from pDBLeu-E2 is 323 folds to that from HPV58-Ura (Figure 1C ). Furthermore, we detected E2 protein in pDBLeu-E2 transformed yeast with anti-cMyc antibody as shown in Figure 1D . Therefore, E2 protein could facilitate viral genome replication and maintenance. Role of E2 protein in maintaining HPV58 genome in yeast DNA stability assay was performed to investigate the maintenance function of E2 protein on HPV58 genome. The DNA loss per cell generation for HPV58-Ura (1.9%) was comparable to that of the yeast plasmid pRS316 (2.1%) which contains centromeric elements (CEN) (Figure 2 ). Thus, the HPV58 genome was as stable as a yeast CEN containing plasmid. Yeast harboring HPV58-Ura-E2mt grew poorly in selective medium, with no colony generated on the selective plates at G0 and G10. However, with transformation of pDBLeu-E2 and re-expression of E2 protein (Figure 1C, D ), the HPV58 genome restored its mitotic stability to a DNA loss rate of 2.5% per cell generation (Figure 2 ). Therefore, E2 protein is critical to the mitotic stability of the HPV58 genome in yeast. Suppression of E6 and E7 transcription by E2 protein in yeast RT-PCR analysis revealed the transcription of both early (E1, E2, E6, E7) and late genes (L1 and L2) of HPV58-Ura in yeast (Figure 3A ). To further investigate the transcription levels of viral genes in yeast cells, qRT-PCR was performed separately with gene specific primers. The results revealed that HPV58 early and late genes have different transcription efficiency (Table 2 ). To explore the regulatory function of E2 protein on viral gene transcription, E6 and E7 genes transcription levels in the three yeast transformants were compared by qRT-PCR. We first compared the DNA relative replication levels of HPV58 genome in yeast by qPCR. With the relative replication level of HPV58-Ura set to 1.0, DNA relative replication levels were 0.03 for HPV58-Ura-E2mt and 0.54 for HPV58-Ura-E2mt/pDBLeu-E2 (Figure 3B ). Then qRT-PCR was performed to compare the transcription levels of E6 and E7 genes. Because wild type HPV58 genome and E2 mutant HPV58 genome have different replication levels, the relative transcription levels were divided by the value of relative replication levels. With E6 and E7 transcripion levels of the HPV58-Ura transformants set to 1.0, the relative transcription levels of E6 were 8.62 for HPV58-Ura-E2mt and 1.02 for HPV58-Ura-E2mt/pDBLeu-E2. The E7 relative transcription levels were 1.96 for HPV58-Ura-E2mt and 0.74 for HPV58-Ura-E2mt/pDBLeu-E2 (Figure 3C ). Therefore, E2 protein could suppress the E6 and E7 genes transcription levels when HPV58 genome persists episomally in yeast.

Discussion Previous reports have shown that HPV1, 6, 11, 16, 18 and 31 can replicate their genomes as episomal plasmids in yeast. The genomes of HPV6, 16 and 31 are mitotically stable in yeast, however HPV1, 11, and 18 are unstable in yeast [ 7 , 8 , 12 ]. In the present study, we showed that HPV58 genome, consistent with the HPV6, 16 and 31, can replicate stably as an episomal plasmid in yeast. Stable maintenance through replication of extrachromosomal DNA in yeast requires autonomous replication sequences (ARS) [ 13 , 14 ], and centromeric elements (CEN) [ 15 ]. There is a 10 of 11 nucleotide match to the consensus core ARS sequence (CAS) (5'-[A/T]TTTAT[A/G]TTT[A/T]-3') in the HPV58 genome (accession no. D90400) at nucleotides 1225 to 1216. Yeast origin recognition complex (ORC) may bind to the CAS-like element and initiate viral genome replication. It has generally been thought that replication of papillomaviruses is dependent upon the presence of E1, a DNA helicase, and E2, a transcription and maintenance factor [ 16 , 17 ]. E1 and E2 protein, perhaps in conjunction with CAS-like elements, conduct the replication of HPV58 genome in yeast. Replication of genomes in yeast is a common feature of HPVs. Different HPV types showed different genome stability in yeast cells. Two kinds of mechanisms may control the mitotic stability of the HPV58 genomes in yeast. First, HPV58 sequences may contain cis -elements that can substitute for the CEN required for the maintenance of extrochromosomal DNA in S. cerevisiae . The second mechanism may be the function of HPV58 E2 protein, a maintenance protein that can tether viral genomes to mitotic chromosomes in dividing cells [ 18 , 19 ]. Angeletti et al. (2002) reported that the HPV16 genome can replicate stably in yeast cells with the E2 gene interrupted by introducing stop codons. Kim et al (2005) also identified the CEN-like cis -elements in the HPV16 genome in yeast and mammalian cells and concluded that the maintenance of the HPV16 genome depends on the CEN-like cis -elements, perhaps in conjunction with E2 protein. HPV16 can maintain the mitotic stability of its genome in an E2-independent manner in yeast and mammalian cells [ 7 , 20 , 21 ]. In contrast to HPV16 genome, the maintenance of HPV58 genome depends on E2 protein. The wild-type HPV58 genome can replicate stably in transformed yeast cells. However, with the HPV58 E2 gene interrupted by mutation, HPV58-Ura-E2mt genome was unstable in yeast. Introduction of pDBLeu-E2 and re-expression of E2 protein restored the stability of HPV58-Ura-E2mt genome in yeast cells. Therefore, the maintenance and faithful segregation of HPV58 genome depends on E2 protein. The HPV58 genome has no CEN-like cis -elements. This specific maintenance pattern of the HPV58 genome may provide a possible way to interfere with the early stage of HPV58 infection by blocking the expression of E2 protein. Previous reports of the HPV/yeast system focused on the replication of HPV DNA. In fact, the HPV/yeast system could be a model for the study of HPV gene transcription. Although we failed to detect the mRNA of HPV58 early and late genes by Northern blot, which may be due to the low transcription levels of viral genes in yeast. RT-PCR results revealed the transcription of HPV58 early and late genes in yeast. Moreover, qRT-PCR analysis of transcription regulation function of E2 protein revealed both E6 and E7 mRNA levels were upregulated with HPV58-Ura-E2mt introduced into yeast. However, with re-expression of E2 protein, the mRNA levels were downregulated. These results confirm that E2 protein can suppress E6 and E7 genes transcription when HPV58 genome persists episomally in yeast. Bechold et al (2003) reported that E2 protein had no affect on E6/E7 expression in mammalial cells [ 22 ]. Bouvard et al (1994) reported that HPV16 and 18 E2 protein could activate their early promoter in CAT luciferase reporter plasmid. However, overexpression of E2 repressesd the early promoter[ 23 ]. These reports are different from our data that E2 protein repressed early promoter transcription in yeast. The reasons should be the different cell lines used, different expression levels of E2 protein and different conformation of viral minichromosomes in cells, especially the chromatin conformation of long control region (LCR). The early promoter of HPV locates in the LCR of the viral genome [ 24 ]. The LCR has four E2 protein binding sites (E2BS). E2BS4 is far from the TATA box. The other three E2BS (E2BS1, E2BS2 and E2BS3) are proximal to the TATA box. E2 protein binding to E2BS4, which is far from TATA box, stimulates the transcription level of E6 and E7. However, binding of E2 protein to the other three E2BS represses transcription through steric hindrance of the interaction with the transcriptional initiation factor TFIID at the proximal TATA box [ 25 , 26 ]. Binding of E2 protein to a specific E2BS is determined by the genome status. HPV58 E2 protein might interact with the E2BS proximal to the TATA box when the HPV58 genome is maintained as an episomal plasmid in yeast.

Conclusions The HPV58/yeast system we used was able to direct the stable replication of the HPV58 genome and induce the transcription of the early and late genes. As compared with maintenance of the HPV16 genome, HPV58 genome strictly depends on E2 protein. E2 protein can supress the transcription of E6 and E7 genes. With its ease of handling and similarity to mammalian system, the yeast system we described is useful for future study of HPV58 genome replication and the regulatory mechanism of viral gene transcription.

Background To establish a convenient system for the study of human papillomavirus (HPV), we inserted a Saccharomyces cerevisiae selectable marker, Ura, into HPV58 genome and transformed it into yeast. Results HPV58 genome could replicate extrachromosomally in yeast, with transcription of its early and late genes. However, with mutation of the viral E2 gene, HPV58 genome lost its mitotic stability, and the transcription levels of E6 and E7 genes were upregulated. Conclusions E2 protein could participate in viral genome maintenance, replication and transcription regulation. This yeast model could be used for the study of certain aspects of HPV life cycle.

List of abbreviations ARS: autonomously replicating sequences; CEN: centromeric elements; HPV: Human papillomavirus; LCR: long control region; qPCR: quantitative PCR; qRT-PCR: quantitative RT-PCR; RT-PCR: reverse transcription PCR; Competing interests The authors declare that they have no competing interests. Authors' contributions JL constructed the HPV58-Ura-E2mt, pDBLeu-E2, participated in Sountern blot, Northern blot, Western blot, DNA stability assay and drafted the manuscript. XW constructed HPV58-Ura, participated in Western blot, DNA stability assay and PCR. JL and HW participated in Southern blot and DNA stability assay. XLZ participated in yeast transformation and DNA stability assay. WT and YDS helped to perform the real time quantitative PCR, Southern blot and Northern blot. XW participated in primers and probe design. XPY gave advices on the design of this study. WMZ designed the study and drafted the manuscript. All authors read and approved the final manuscript.

Acknowledgements This work was funded by the National Natural Science Foundation of China (30470086 to WMZ). We thank Dr. Kong-Nan Zhao (University of Queensland Centre for Clinical Research, Australia) and Dr. Peter Angeletti (University of Nebraska-Lincoln School of Life Science) for their guidance on the experiments.

CC BY

no

2022-01-12 15:21:44

Virol J. 2010 Dec 15; 7:368

oa_package/5f/98/PMC3016283.tar.gz

PMC3016284

21143904

Background Oral tongue carcinoma is a highly curable cancer when treated with radiation therapy, especially interstitial brachytherapy [ 1 ]. Iridium-192 (Ir-192) hairpins or cesium-137(Cs-137) needles are usually used for low-dose-rate (LDR) interstitial radiotherapy in Japan. We used a high-dose-rate (HDR) remote-controlled after-loading system, using an Ir-192 microsource, the MicroSelectron-HDR (Nucletron, Veenendaal, The Netherlands) installed in 1991. Since with this system there is no risk of radiation exposure except to the patient, HDR makes it possible to treat patients in a normal ward, so that the quality of life during treatment may be better. We have already reported on the outcome of HDR brachytherapy for early oral tongue cancer which included a prospective Phase III study [ 2 - 4 ]. In addition, we reported that the efficacy of brachytherapy for T3 oral tongue cancer, especially when using HDR, was enhanced by its excellent dose distribution [ 5 ]. The number of elderly patients in Japan has been increasing steadily because of advances in both health and medical care and the leading cause of death among the elderly is cancer. The number of people aged 80 or over reached 7,130,000 in Japan in 2007, which counts for more than 5% of the population. The problems involved in treating older patients with cancer are time pressing [ 6 ]. As aging is a highly individualized process, the indication, strategy, and techniques of radiation therapy for the elderly should not be defined exclusively by chronologic landmarks [ 6 ]. We studied 21 80-year-old or older patients with oral tongue cancer treated by brachytherapy. Since to the best of our knowledge, there have been no previous reports regarding such patients, we conducted this retrospective review of the feasibility of brachytherapy for elderly patient with T1-3N0 oral tongue cancer.

Methods Patients Between 1967 and 2004, 21 patients (9 males and 12 females) with previously untreated mobile tongue cancer were treated with radiotherapy at Osaka University Hospital and Osaka National Hospital. Patients treated with radiotherapy combined with chemotherapy were excluded from the study. All tumors were histologically identified as squamous cell carcinoma. Table 1 lists patient and treatment characteristics. The patients' median age was 81, ranging from 80 to 89. There were 2 T1, 14 T2, and 5 T3 tumors (UICC TNM classification of 1987). During the study period, we also treated about 700 patients with T1-3N0 oral tongue carcinoma [ 4 ], with the elderly group accounting for 3% of all patients. The age of the 21 patients ranged from 80 to 89 years (median 81) at the start of radiation therapy, and the male-to-female ratio was 9:12. Performance status (PS) was classified as 0-1, based on the World Health Organization classification. For this study, the clinical records of consecutive these 21 patients from our database were reviewed (Table 1 ). To compare the result of treatment to younger counterpart, we reviewed population adjusted (sex, T-stage, with external radiotherapy) 226 patients treated during same time period. The background comparison was shown in Table 2 . Radiation therapy All implantation was done under local anesthesia. For patients in the LDR group, the treatment sources consisted of an Ir-192 pin for 12 patients, a Ra-226 needle for 4 and a 198 Au grain for one patient. Each needle was implanted with the Paterson-Parker system using a reference point 5 mm distant from the implant plane. The median dose and range for the LDR group treated with brachytherapy only was 70 Gy (61-84 Gy). Patients in the HDR group received a total dose of 60 Gy in ten fractions during one week at 5 mm distance from the radioactive source. Two fractions were administered per day. The time interval between fractions was more than 6 hours. Dose rates at the reference points for the LDR group were 0.30 to 0.8 Gy/h, and for the HDR group 1.0 to 3.4 Gy/min. Patients were followed up for at least 13 months or until their death, with a median follow-up time of 2.5 years (range: 1.3 - 14 years). Large T2 tumor or more including ulceration or thicker tumor received external irradiation. A total of 8 patients (T2: 3, T3: 5) underwent external radiotherapy using a Co-60 teletherapy unit or a linear accelerator. These patients received 2-3 Gy per fraction for a median dose of 30 Gy (30 - 50 Gy), and were treated with a single lateral field that involved the primary site and the upper jugular lymph nodes. Nutrition support was given by nasal tube feeding during brachytherapy. No patient required tracheostomy. The routine follow-up interval was 1 month for the first year, two months for the second year, and 3 - 6 months thereafter. We examined the outcomes in terms of local control, lymph node control, cause-specific and overall survival. Early toxicities were assessed by Common Toxicity Criteria version 3 (CTC v3). Late toxicities were counted if soft tissue (ulceration lasting 3 months or more) and/or bone (bone exposure and necrosis) reactions occurred. Statistical Analysis For a statistical analysis, a Student's t-test for normally distributed data and the Mann Whitney U-test for skewed data were used. The percentage was analyzed using a Chi-square test. Local control and survival data were estimated according to the Kaplan-Meier method, and were examined for significance with a logrank test. All analyses used the conventional p < 0.05 levels of significance.

Results Local control, regional control, cause-specific and overall survival The 2- and 5-year local control rates for the 21 elderly patients were both 91% (Figure 1 ). The 2-year (5-year) local control rates for T1, T2, T3 tumors were 100% (100%), 83%, (83%) and 80% (not available), respectively (n.s.). These figures was not inferior to that of younger counterpart (82% at 5-year, Figure 2 n.s.). Two patients showed local recurrence. An 83-year-old female (ID15) received external radiotherapy for lymph node metastasis found just after completion of brachytherapy, but local recurrence appeared and resulted in death. One more local failure occurred in an 80-year-old female with T2N0 oral tongue cancer (ID 19) treated with the Ir-192 source. During the first night of treatment in the RI ward, she tried to brush her teeth and pulled out the guide gutter of the Ir-192, so that the Ir-192 needles were replaced with Au-198, resulting in partial response and recurrence 4 months later. The 2-year and 5-year cause-specific survival (CSS) rates were both 83% (83% and 78% in control group), but the respective overall survival rates were 55% and 34% (83% and 76% in control group). Incidence of lymph-node metastasis was 21% at 2 years (34% in control group) and all four recurerce appeared at ipsilateral side. Of the 4 patients who showed nodal failure, three underwent surgery, one of whom could be salvaged. Actuarially 12 patients died because of intercurrent disease or old age. The follow-up for 5 patients had to be terminated because the patients or their family requests. Tolerance and Complications Grade 2-3 acute mucositis, pharyngitis for combined external radiotherapy and oral mucositis for solely brachytherapy, occurred but were acceptable. No grade-4 skin or mucosal acute reactions were documented. The intensity of acute reactions in the elderly patients was almost the same as that observed in younger patients. Late reactions after brachytherapy comprised one bone exposure and/or two ulcer formations lasting 3 months more (2/21 = 10%; Table 1 ). One case showed tongue deformation with ulceration scar. In previous cohort [ 4 ], 10 to 30% of delayed reaction was found according to treatment volume and addition of external radiotherapy. Aged patients showed similar ratio of delayed reaction.

Discussion The patients 80 years old or older among those who were treated by brachytherapy accounted for about 3% of our cohort. The incidence of carcinogenesis among this age group is currently unavailable, but oncologists are treating increasing numbers of elderly cancer patients, so that we should be more deeply concerned about treatment strategies for these patients. The deterioration of biological functions associated with aging leads to a diminished reserve capacity and increased vulnerability to age-related diseases and overall forces of mortality [ 6 - 8 ]. As the effects of aging depend on the individual, they manifest themselves with great variability and heterogeneity, thus making it extremely difficult if not impossible to determine a standard therapy for elderly patients based only on chronologic landmarks. When deciding on a personalized mode of treatment for older patients, it is important to assess each patient's quality of life and life expectancy. Prognostic factors related to the tumor (TNM stage, pathology, etc.), physical and/or psychological status (PS, etc.), and social support should be taken into account when estimating the outcome of treatment and life expectancy of elderly patients. However, the major part of prospective trials is carried out with patients younger than 70 so that little evidence regarding elderly patients is available. Generally, local treatment is more appropriate than systemic therapy for the elderly. Standard chemotherapy, especially combination treatment, is not encouraged because of elderly patients' physiologically impaired functions and diminished reserve capacity of important organs [ 9 - 11 ]. Unsatisfactory outcomes of combination therapy have been reported [ 8 ], although better results with less toxic antineoplastic agents or reduced doses of chemotherapeutic agents especially designed for elderly patients with non-Hodgkin's lymphoma have been reported [ 12 ]. Moreover, the rates of acute adverse effects, morbidity, and mortality remain high for the elderly, so that extended radical surgery is not encouraged for the same reasons. It is important for their quality of life and life expectancy to attain local control of symptomatic primary lesions. Carefully planned radiation therapy for the elderly is expected to become increasingly important [ 13 ]. A prospective study has also reported the usefulness of radiotherapy for esophageal cancer in elderly patients [ 14 ], and found that patients with good PS could tolerate doses that administered according to a standard radiotherapy schedule [ 9 ]. Our findings agreed with this study in that the completion rate of radiotherapy and local control rate for elderly patients were not inferior to those for younger patients. One of the limitations of this study is that its retrospective nature leads to a lack of detailed information about co-morbidity. This is important because cardiovascular and pulmonary diseases as well as diabetes and other diseases are more pronounced in elderly than younger patients. In addition, as mentioned in results, unexpected accidents will occur more frequently in elderly than younger patients. We found four cases of hypertention and a TIA records in patients' charts, however, they were able to be diagnosed as candidates for brachytherapy with local anesthesia and we noted that adverse reactions such as mucositis in HDR brachytherapy were similar for elderly patients: spotted mucositis started to appear three days after the end of brachytherapy while confluent mucositis developed and reached a peak at ten days, but disappeared by the fourth to eighth week without any major complications [ 2 ]. Fortunately, we did not encounter the aspiration pneumonia after brachytherapy in current study. Severe deterioration in QOL, such as speech disturbance, swallowing function loss, and frequent short hospital stay were also not a case enhanced than younger counterpart. Although the number of patients in this series was too small to draw definite conclusions regarding efficacy, late toxicity and tolerance, our data suggest the potential benefits of brachytherapy for elderly patients. Because radiation therapy is considered to be a minimally invasive treatment procedure, it has the advantage of preserving the shape and functions of the tongue. Brachytherapy was historically performed with Ra-226, which involved exposure of the surrounding tissue. To minimize undesirable radiation to normal tissues, an afterloading technique using Ir-192 was implemented. This LDR brachytherapy has been widely used since and become the gold standard in brachytherapy. Many institutes have reported successful results for tongue cancer treated with LDR brachytherapy [ 2 , 15 ]. Since then, HDR brachytherapy using a remote afterloading technique has been introduced in several brachytherapy centers, including ours [ 2 - 4 ]. We previously reported our phase III data and a retrospective review with good results for T1-3 N0 patients to show the comparable outcome of HDR. However, retrospective reviews including ours reported that older patients aged 65 or over showed poorer local control than their younger counterparts [ 3 , 4 ]. In a 648-patient cohort, 5-year local control rates were 87% for T1, 78% for T2, and 68% for T3 in younger patients, but 72% for T1, 67% for T2, and 54% for T3 in elderly patients aged 65 or over (p < 0.05) [ 4 ]. These findings prompted us to examine the background characteristics of older patients. We found that one possible explanation for poor local control was poor oral hygiene including dental factors in the elderly in previous study [ 12 ], which could be modified by careful intervention. In addition, in the study reported here, we found that patients aged 80 or over showed good outcome including four locally controlled HDR patients. Therefore age is not a sole factor on a local control rate by brachytherapy, other confounding factor such as tumor, oral hygiene, PS, co-morbidities have affected outcomes. Although further studies are needed to establish optimum schedules and techniques, elderly patients with good PS may tolerate brachytherapy schedules so that the advisability of definitive radiation therapy should be considered. In conclusion, patients aged 80 or over showed results comparable to those for their younger counterparts, and an aggressive approach for appropriately selected elderly patients achieved good local control.

Background To examine the role of brachytherapy for aged patients 80 or more in the trend of rapidly increasing number. Methods We examined the outcomes for elderly patients with node negative oral tongue cancer (T1-3N0M0) treated with brachytherapy. The 21 patients (2 T1, 14 T2, and 5 T3 cases) ranged in age from 80 to 89 years (median 81), and their cancer was pathologically confirmed. All patients underwent definitive radiation therapy, with low dose rate (LDR) Ra-226 brachytherapy (n = 4; median 70Gy), with Ir-192 (n = 12; 70Gy), with Au-198 (n = 1) or with high dose rate (HDR) Ir-192 brachytherapy (n = 4; 60 Gy). Eight patients also underwent external radiotherapy (median 30 Gy). The period of observation ranged from 13 months to 14 years (median 2.5 years). We selected 226 population matched younger counterpart from our medical chart. Results Definitive radiation therapy was completed for all 21 patients (100%), and acute grade 2-3 mucositis related to the therapy was tolerable. Local control (initial complete response) was attained in 19 of 21 patients (90%). The 2-year and 5-year local control rates were 91%, (100% for T1, 83% for T2 and 80% for T3 tumors after 2 years). These figures was not inferior to that of younger counterpart (82% at 5-year, n.s.). The cause-specific survival rate was 83% and the regional control rate 84% at the 2-years follow-up. However, 12 patients died because of intercurrent diseases or senility, resulting in overall survival rates of 55% at 2 years and 34% at 5 years. Conclusion Age is not a limiting factor for brachytherapy for appropriately selected elderly patients, and brachytherapy achieved good local control with acceptable morbidity.

Competing interests The authors declare that they have no competing interests. Authors' contributions HY conceived of this study and drafted manuscript. KY participated in the design of this study. TK and YY participated in the statistical analysis. MK, SF, NK, KS and TN participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.

CC BY

no

2022-01-12 15:21:44

Radiat Oncol. 2010 Dec 9; 5:116

oa_package/17/a5/PMC3016284.tar.gz

PMC3016285

21162739

Background For early stage breast cancer, whole breast irradiation (WBI) is used extensively to minimize the risk of ipsilateral breast cancer recurrence after breast conserving surgery. Over the last decade, there has been increased interest in the use of accelerated partial breast irradiation (APBI) as opposed to WBI [ 1 ]. The use of APBI offers fewer fractions and lower dose to uninvolved regions of the breast. A number of clinical trials comparing WBI with various methods of APBI treatments are ongoing [ 2 ], however mature randomized data on the efficacy and toxicity of APBI compared to standard WBI will not be available for a number of years. Publications supporting the dosimetric advantages of using APBI as an alternative to WBI have mainly focused on intra-cavitary brachytherapy, interstitial brachytherapy or intra-operative radiation therapy [ 3 - 6 ]. A common method of delivering APBI in ongoing randomized trials is linac-based, 3-dimensional conformal external beam radiation therapy (3DCRT) employing the same widely-available technology, staff, and treatment planning systems as WBI [ 7 ]. Given the potential importance of linear accelerator based delivery of APBI, the influence of dose calculation algorithms on trial eligibility and interpretation of risks to normal tissues is relevant. The impact of scatter corrections with WBI techniques comparing pencil beam convolution (PBC), the analytic anisotropic algorithm (AAA), and Monte Carlo (MC) calculations has been previously described [ 8 ], with several articles discussing the benefits of using AAA over PBC [ 9 , 10 ]. However, there are no studies that examine the accuracy of the dose to target and normal tissues for 3DCRT APBI techniques. The accuracy of the calculated dose in regions well outside the irradiated volume is particularly important when trying to ascertain the risk of secondary cancer or normal tissue toxicity [ 11 ]. Obtaining a better understanding of the potential increase, or decrease, in dose to target and normal tissues could facilitate a better understanding of the risks associated with APBI treatment strategies. This is a report of the consequences of changing dose calculation algorithms on doses to target volumes and important normal tissues during whole breast and partial breast irradiation.

Methods Treatment planning We retrospectively examined plans for 10 consecutive patients enrolled in a prospective APBI trial who met all the dosimetry constraints of the protocol when using simplified pencil beam calculations [ 12 ]. All plans were initially calculated with a pencil beam convolution (PBC) algorithm with Batho inhomogeneity corrections using the Eclipse Treatment Planning System (Version 8.617, Varian Medical Systems, Palo Alto, USA) [ 13 ]. Plans were then recomputed (keeping the monitor units, beam weights and angles fixed) within Eclipse using AAA. All calculations were performed on 2.5 mm dose grid. The WBI prescription was 42.5 Gy in 16 fractions, normalized to a point mid-plane in the breast tissue and to be delivered through a segmented MLC delivery with 6 MV photon beams. The partial breast technique employed four non-coplanar 6 MV beams that avoided direct beams into the ipsilateral lung [ 14 ]. The planning target volume (PTV) for the APBI plans was the seroma (the primary surgical site density on a planning CT scan) plus a 1 cm expansion, excluding chest wall and 0.5 cm from the skin, to form the clinical target volume (CTV) and a further 1 cm 3-dimensional expansion to form the PTV. A dose-evaluation volume (DEV) was defined as the portion of the PTV that excluded the chest wall and 0.5 cm from the skin [ 14 ]. In addition to defining these target structures, the ipsilateral breast, ipsilateral lung, heart, contra-lateral lung and contra-lateral breast were contoured (see Figure 1 ). The APBI prescription was 38.5 Gy in 10 fractions normalized to a point within the target volume. The planning guidelines for APBI patients follow those articulated in the American Society of Therapeutic Radiology and Oncology (ASTRO) consensus document [ 1 ]. Monte Carlo Verification WBI and APBI treatment plans were recomputed with the Vancouver Island Monte Carlo (VIMC) system [ 15 , 16 ]. The system provides a platform for Monte Carlo verification of the treatment plans generated by a TPS and exported in DICOM format. The main "calculation engines" within the system are BEAMnrc for modelling particle fluence and DOSXYZnrc for modelling the dose deposition within the patient [ 17 ]. The beam model for Varian 21EX treatment machine was used in this study. The model utilises a two-stage approach in calculating the dose where in the "first stage" all non-variable linac components are modelled and the particle fluence is stored in the phase space file. Then, in the "second stage" the phase space file is used in subsequent calculations as a radiation source for transporting the fluence through the patient phantom. Standard energy cut-off values were AP = PCUT = 0.01 MeV and AE = ECUT = 0.700 MeV, where AP and AE are the low energy thresholds for the production of secondary bremsstrahlung photons and knock-on electrons and PCUT and ECUT are the global cut-off energies for photon and electron transport used during electron and photon transport. In addition, "azimuthal particle redistribution" has been used to substantially reduce phase space latent variance [ 18 , 19 ]. The model has been tuned and verified (except the build-up and penumbra regions) demonstrating dose agreement with the measured open field dose profiles within 1% for the field sizes within the range of 4 × 4 to 40 × 40 cm 2 [ 11 ]. This excluded build-up and penumbra regions where the dose differences were higher, as expected, but still agreed to within 2% or within a 2 mm distance. Modelling of IMRT and RapidArc, as well as fixed-aperture fields' delivery has been performed with the dynamic multi-leaf collimators (dMLC) model by Siebers et al. and verified in our previous publications [ 11 , 20 , 21 ]. As most of the treatment fields used in the current study utilise the Varian implementation of collimator-controlled wedging, or enhanced dynamic wedges (EDWs), it is important that the dose from such fields is calculated correctly. Radiation transport through the moving jaw of EDWs is modelled in VIMC system using the method developed by Verhaegen and Liu [ 22 ]. Each particle is transported through the dynamic jaw with its position sampled from a probability density function that describes jaw motion. Then, the particle is transported through the physical jaw in its sampled position. This method naturally models the radiation transmitted through the dynamic jaw towards the patient as well as radiation backscattered from the jaw into the linac monitor chamber. The latter is essential for correct absolute dose calculation implemented in the VIMC linac model [ 23 ]. Verhaegen and Liu demonstrated excellent agreement of this EDW model with measured data. Our implementation of this model has been verified against the EDW commissioning measurements collected in our department. The measurements were done using Scanditronix Wellhofer CA24 ionization chamber array with IC-10 ionization chambers that have effective volume of 0.13 cm 3 . Examples of this verification for Monte Carlo as well as PBC and AAA calculations that include 10 × 10 and 20 × 20 cm 2 fields with 60° wedge are shown in the Results section. MC simulations of the treatment plans presented in this study were performed on 2.5 mm dose grid with less than 1% statistical uncertainty at the DEV. Statistical Analysis Volumetric and dosimetric statistics as defined in Table 1 were recorded from each of the patient's 6 plans (WBI-PBC, WBI-AAA, WBI-MC, APBI-PBC, APBI-AAA, and APBI-MC). To determine whether there is a difference to these volumes, the mean percentage differences in doses or volumes receiving a specific dose were tested using the paired Student's t-test computed in Microsoft Excel (Microsoft, Redmond WA). For a significance level of p = 0.05, the adjusted significance level with Bonferroni corrections for the 8 different tissues analyzed in this study is p = 0.006 [ 24 ].

Results Verification of MC, AAA and PBC dose calculations for EDW fields Figures 2 and 3 demonstrate agreement of the three calculation algorithms with the dose measurement in water for 10 × 10 and 20 × 20 cm 2 EDW fields at 10 cm depth. All algorithms show good overall agreement with the measurement data, however MC agrees with the measurement slightly better, especially in the out-of-field regions. Of all algorithms considered, MC has the best agreement with the 10 × 10 cm 2 measured data, and the agreement with 20 × 20 cm 2 field is excellent: the measurement points essentially overlap with MC data. Error bars on MC points demonstrate their calculated standard deviation of 1%, and most measurements fall within this range. Dose Calculation Algorithm Effects on Whole Breast Irradiation Table 2 summarizes the mean, standard deviations and ranges of the target and normal tissue statistics recorded from the three WBI plans. The volumes of the DEV and PTV receiving 95% of the prescription dose using PBC calculations were not significantly different than AAA calculations (all p > 0.127). The ipsilateral whole breast volume receiving 10% of the prescription dose in the PBC plan was 10.5% higher than the AAA dose (p = 0.004). There were no statistically significant differences between PBC and AAA, or AAA and MC calculations for target or normal tissue structures. This was also true when PBC and MC calculations were compared, with the exception that the ipsilateral breast dose was 11.8% lower than the PBC calculations with MC calculations (p = 0.004). Dose Calculation Algorithm Effects on Accelerated Partial Breast Irradiation Table 3 summarizes the mean, standard deviations and ranges of the target and normal tissue statistics recorded from the three APBI plans. The dosimetric statistics from PBC and AAA plans were not significantly different for the PTV, ipsilateral breast, heart, ipsilateral lung, and contra-lateral lung. Although not significant, the maximum dose to the contra-lateral breast was 1.9% higher for AAA compared to PBC (p = 0.030) and the average volume to the DEV receiving 95% of the prescription dose was 2.1% higher with PBC calculations compared to AAA (p = 0.012). There were no statistically significant differences between PBC and MC (p > 0.019), or AAA and MC (p = 0.100) calculations for target or normal tissue structures. Accelerated Partial Breast versus Whole Breast Irradiation Table 4 summarizes the differences in volumes and doses to the target and normal tissues when comparing WBI with APBI plans for the three different algorithms. Figure 4 illustrates the difference in dose to target and normal tissues when comparing WBI with APBI for the three different algorithms. When switching from WBI to APBI with PBC, there was significant reduction in dose to the ipsilateral breast (p = 0.002). When switching from WBI to APBI with AAA, there was significant reduction in dose to the ipsilateral lung (p = 0.001). When switching from WBI to APBI with MC, there was significant reduction in dose to the ipsilateral breast and lung and contra-lateral lung (p = 0.003, p < 0.001, p = 0.001 respectively). The magnitude of the difference in dose to these structures depends on the dose calculation algorithm used.

Discussion This study demonstrates very good agreement between the AAA and PBC algorithms when planning either WBI or ABPI. This suggests that there are no major concerns associated with target and normal tissue coverage if switching from PBC to AAA for WBI or ABPI. Given that AAA provides a significant improvement over the PBC plus Batho-heterogeneity corrections in lung tissue, our clinical practice has migrated from PBC to AAA along with dose calculations for the APBI clinical trial. For APBI plans, the dose to target and normal tissue volumes varied with the dose calculation algorithm. This result is in agreement with work that explored the impact of PBC, AAA, and MC algorithms in non-clinical scenarios [ 11 ]. The volumes of the DEV and PTV receiving 95% of the prescription dose from PBC plans were higher or equal to the plans recomputed with AAA and MC. This is predictable because the lung-tissue interface is poorly calculated with PBC. If APBI plans are switched from PBC to AAA calculations, the dose to the PTV and DEV requires re-evaluation. Based on our results, a plan generated using AAA compared to PBC calculations would deliver approximately 2% more dose within the DEV. This may not have any measurable effect on tumour control but could influence the risk of late breast fibrosis because during APBI the dose per fraction is already high. This may be a particular risk if the DEV or PTV is large. The doses (and volumes receiving a specific dose) to normal structures will also correspondingly increase. Apart from the contra-lateral breast, the treatment plan can be re-configured to ensure that normal tissue constraints are maintained. This is not difficult to achieve since the doses to normal tissues are relatively independent of the calculation algorithm, with the important exception of the contra-lateral breast. There may be a small but important difference in the contra-lateral breast dose when comparing APBI plans computed with PBC, AAA and MC algorithms. The dose to the contra-lateral breast was 2-3% higher with AAA as compared to MC. Despite the fact that dose calculation algorithms are not generally validated for dose points far away from the treatment volume and that this metric is sensitive and unstable, existing accelerated partial breast clinical trials use a maximum point dose as a constraint to the contra-lateral breast. The selection of this constraint stems from a desire to have simple planning objectives and constraints for dosimetrists. The ASTRO consensus document states that no point in the contra-lateral breast volume should receive > 3% of the prescribed dose. This work suggests that switching from the PBC to the AAA treatment planning algorithm could affect the apparent eligibility of patients for accelerated partial breast treatment. Out of ten patients in the current study, two would have failed the ASTRO contralateral breast dosimetry guideline when calculated using the PBC or MC algorithm. However, delivering an identical amount of MUs and using the same beam angles and weightings but calculated with AAA, seven patients would have not met the contra-lateral breast constraint. If reproduced across the population of patients considered for APBI, this could represent a significant reduction in eligibility. An examination of the DVH data for APBI plans suggests that relaxing the contra-lateral breast maximum dose constraint from 3% to 5% would retain eligibility for APBI without any real increase in the risk of radiation exposure or second breast cancer that is considered acceptable using existing PBC planning algorithms. A more detailed investigation on these differences was conducted to understand where these differences stem from. Figure 5 displays three dose distributions highlighting the differences between the algorithms for tissues far from the treated volume. For PBC, the isodoses are fairly parallel to the field borders, suggesting that the in-patient scatter contributes most to the peripheral dose. For AAA, this is partially true with the exception of the dose in lung tissue and the surface of the patient, far from the field borders. This suggests that the head-scatter modelling contributes the most for tissues on the surface such as the contra-lateral breast, and in-phantom scatter contributes the most for deeper tissues. With the exception of the dose in lung, the Monte Carlo isodoses agree well with PBC for isodoses higher than 5%, and with AAA for isodoses lower than 3%. The APBI technique often employs wedges to achieve tumor coverage, hence the accuracy of the dose calculation to the contra-lateral breast can be largely affected by the algorithm's ability to correctly calculate the in-field and penumbra dose for the EDW fields. The AAA algorithm uses a semi-analytic model to account for leakage radiation, jaw and multi-leaf transmission for open and wedged fields and can over-estimate the dose in penumbra by 1-2% when compared with MC [ 10 ]. In our centre, in-field open and wedged field agreement between measurement and calculations was better than 2% for AAA, and better than 1.5% for MC. This leads us to hypothesise that the dose differences in the contra-lateral breast are mostly due to head scatter and leakage modelling within AAA [ 25 ]. These contributions are modelled as extra-focal and electron contamination parameters within the treatment planning system, which are optimized in the beam fitting procedure. In the fitting procedure, these extra-focal parameters cannot be distinguished from other parameters in the beam tuning, leading to excellent agreement in the open field and penumbra, but not necessarily far from the open beam.

Conclusions There is very good agreement between PBC, AAA and MC for most tissues when treating with APBI. However, if calculation algorithms are switched from a simple pencil beam to a scatter-correction convolution/superposition algorithm, careful consideration should be given to tissues peripheral to the treated volume. In this study, it was found that a commonly used dosimetry constraint, as recommended by the ASTRO consensus document, that no point in the contra-lateral breast volume should receive >3% of the prescribed dose needs to be relaxed to >5%.

Background This paper compares the calculated dose to target and normal tissues when using pencil beam (PBC), superposition/convolution (AAA) and Monte Carlo (MC) algorithms for whole breast (WBI) and accelerated partial breast irradiation (APBI) treatment plans. Methods Plans for 10 patients who met all dosimetry constraints on a prospective APBI protocol when using PBC calculations were recomputed with AAA and MC, keeping the monitor units and beam angles fixed. Similar calculations were performed for WBI plans on the same patients. Doses to target and normal tissue volumes were tested for significance using the paired Student's t-test. Results For WBI plans the average dose to target volumes when using PBC calculations was not significantly different than AAA calculations, the average PBC dose to the ipsilateral breast was 10.5% higher than the AAA calculations and the average MC dose to the ipsilateral breast was 11.8% lower than the PBC calculations. For ABPI plans there were no differences in dose to the planning target volume, ipsilateral breast, heart, ipsilateral lung, or contra-lateral lung. Although not significant, the maximum PBC dose to the contra-lateral breast was 1.9% higher than AAA and the PBC dose to the clinical target volume was 2.1% higher than AAA. When WBI technique is switched to APBI, there was significant reduction in dose to the ipsilateral breast when using PBC, a significant reduction in dose to the ipsilateral lung when using AAA, and a significant reduction in dose to the ipsilateral breast and lung and contra-lateral lung when using MC. Conclusions There is very good agreement between PBC, AAA and MC for all target and most normal tissues when treating with APBI and WBI and most of the differences in doses to target and normal tissues are not clinically significant. However, a commonly used dosimetry constraint, as recommended by the ASTRO consensus document for APBI, that no point in the contra-lateral breast volume should receive >3% of the prescribed dose needs to be relaxed to >5%.

Competing interests The authors declare that they have no competing interests. Authors' contributions PSB calculated patient plans within the treatment planning system, performed the statistical analysis, provided the initial draft and coordinated subsequent drafts of the manuscript. SZ performed the Monte Carlo calculations and helped draft the manuscript. TB assisted in the design of the study and helped draft the manuscript. IO assisted in the design of the study and helped draft the manuscript. WB assisted in the design of the study and helped draft the manuscript. All authors read and approved the final manuscript.

Acknowledgements The authors would like to thank Michael Crane for his assistance with some of the planning of the patients in this study. The authors also greatly appreciate VIC Monte Carlo group and particularly Karl Bush for technical support of VIMC system used in this study.

CC BY

no

2022-01-12 15:21:44

Radiat Oncol. 2010 Dec 16; 5:120

oa_package/58/21/PMC3016285.tar.gz

PMC3016286

21159167

Background Malignant mesotheliomas (MMs), aggressive tumors characterized by marked local invasiveness, are poorly responsive to current therapeutic approaches. Clinical outcomes for MM are poor, resulting in average patient survival times of 7 to 12 months from initial diagnosis. We hypothesized that chemotherapeutic agents used in the treatment of MM activate survival pathways governing drug resistance [ 1 ]. For example, abnormal activation of the Raf/MEK/extracellular signal-regulated (ERK) pathway occurs in many human cancers, including MM [ 2 ], due to mutations in upstream membrane receptors, Ras and B-Raf, as well as mutations in genes regulating Raf activity that reportedly induces chemoresistance to doxorubicin (Dox) and paclitaxel in breast cancer cells [ 3 ]. Moreover, a phase II study in patients with MM shows activation of both ERK and PI3K/AKT pathways that are attributed to their resistance to erlotinib [ 4 ]. ERK activation has been identified as a potential survival pathway in several tumor types [ 5 ], and recent studies show that ERKs may also be activated in response to chemotherapeutic drugs [ 6 - 8 ] or mTOR inhibitors [ 9 ]. We focused here on whether ERK1 and 2 played critical roles in drug resistance and survival of MM, a generally incurable cancer exhibiting marked chemoresistance. To understand the mechanisms involved, we studied gene expression linked to drug resistance and metabolism, including ATP binding cassette (ABC transporters) genes. This large superfamily of membrane proteins is comprised of 48 members that are divided into 7 different families based on sequence similarities [ 10 ]. We selected doxorubicin (Dox) (Adriamycin) for our studies as this drug has been widely used as the most successful drug of choice to treat MMs in single agent studies [ 11 , 12 ] and is used currently in treatment of MMs [ 13 , 14 ]. The goal of this study was to understand how Dox-induced resistance develops, and whether it can be overcome by combination therapy. In the present study we demonstrated that Dox treatment causes activation of survival signals (ERK1/2) in MM cells. Combined treatment with a MEK1/2 inhibitor (U0126) plus Dox increased MM cell death over levels observed with Dox alone. Furthermore, using human MM lines expressing shERK constructs, we show that both ERK1 and ERK2 contribute to Dox resistance in human MMs in vitro and in vivo . Microarray and qRT-PCR analyses of these cell lines revealed that ERK1 or 2 inhibition was linked to decreases in mRNA levels of ATP binding cassette (ABC) genes. Most importantly, we demonstrate that human shERK1 and shERK2 stable MM lines (in comparison to shControl lines) have a slower growth rate after treatment with Dox in a SCID mouse xenograft model. These data suggest that combined treatment using an ERK1/2 inhibitor or RNA interference approach with Dox (or other chemotherapeutic drug) may be more beneficial than single agent therapy in treatment of MMs.

Methods Cell culture None of the human malignant mesothelioma (MM) lines described in this manuscript are commercially available. However, they have been characterized previously by cell size, doubling time, immunohistochemical analyses, electron microscopy, and chromosomal karyotyping as reported (note that the names of these lines have changed since originally reported)[ 15 ]. A sarcomatoid (MO) and epithelioid (ME-26) human pleural MM cell line were obtained from Drs. Luciano Mutti (Maugeri Foundation, Pavia, Italy) and Maurizio Bocchetta (Loyola University, Mayfield, IL), respectively. The HMESO MM line (epithelioid) was originally characterized by Reale et al [ 16 ]. PPMMill, a sarcomatoid human MM cell line, was obtained from Dr. Harvey Pass (NYU School of Medicine, New York, NY). Human mesothelial LP9/TERT-1 (LP9) cells, an hTERT-immortalized cell line phenotypically and functionally resembling normal human mesothelial cells [ 17 ], were obtained from Dr. James Rheinwald (Brigham and Women's Hospital, Harvard University, Boston, MA). Prior to initiating the studies described here, all isolates were confirmed as MM cells by immunohistochemistry using an antibody to calretinin and verified for lack of mycoplasma contamination using a polymerase chain reaction. Additionally, Hmeso tumor xenografts grown in SCID mice were resected and evaluated immunohistochemically by Dr. Michele Carbone and shown to be cytokeratin positive, indicating that they are mesothelial origin. Subsequent karyotype analysis of the Hmeso line by Dr. Joseph Testa demonstrated that the cells were human and possessed several deletions common in mesothelioma lines. These data support what was originally reported for this MM line [ 16 ]. All cells were maintained in 50:50 DMEM/F12 medium containing 10% FBS and supplemented with penicillin (50 units/ml), streptomycin (100 μg/ml), hydrocortisone (100 μg/ml), insulin (2.5 μg/ml), transferrin (2.5 μg/ml), and selenium (2.5 μg/ml), incubated at 37°C in 5% CO 2 and grown to approximately 80-90% confluency [ 18 ]. The synthetic MEK1/2 inhibitor, U0126, and its inactive analog, U0124, were obtained from Calbiochem (La Jolla, CA) and added to cells at 20 μM in medium containing ≤0.2% DMSO [ 18 ]. Control cultures received medium without compounds but with vehicle (≤0.2% DMSO) alone and were treated identically. Doxorubicin (Dox) was obtained from Sigma (St Louis, MO). Viability determination by cell counting Viability of cells after Dox treatment was studied by plating cells at 1X10 5 per well in a 12 well plate. At confluence, cells were maintained in low serum containing medium (0.5% FBS) for 24 h before treating them with different concentrations of Dox (0-100 μM) for 24 h. Cells were trypsinized and counted using a hemocytometer. MTS assay Human MM cells (transfected or untransfected) were treated with different concentrations of Dox with and without U0126 or U0124 for 24 h, and cell viability was measured in cells using the colorimetric MTS Assay, CellTiter 96 Aqueous One Solution Cell Proliferation Assay (Promega) as per the manufacturer's recommendations. Absorbance was read at 490 nm on a spectrophotometer indicating MTS bioreduction to a colored formazan product by viable cells. Western blot analysis To verify activation of ERK1/2 in MM cells after Dox exposure with and without U0126 or U0124, Western blots were performed as described previously [ 18 ] using antibodies specific to pERK1/2 (rabbit polyclonal anti-pERK1/2, 1:500, Cell Signaling Technology, Danvers, MA), total ERK1/2 (rabbit polyclonal anti-ERK1/2, 1:1000, Cell Signaling Technology, Danvers, MA), and total β-Actin 1:2000 (Abcam, Cambridge, MA). Western blots were quantitated by the Quantity One program [ 18 ] and normalized to total ERK1/2 levels. Western blotting was also performed to validate the selective inhibition of ERK1 or 2 in sh MM lines. Preparation of RNA and PCR array analyses LP9 and MM cells were grown to confluence and treated with U0126 (20 μM for 24 h). RNA was prepared and purified using a Qiagen RNeasy plus kit (Valencia, CA). After quality assessment, 1 μg of RNA was employed for cDNA synthesis using the RT 2 First Strand Kit (SABiosciences, Frederick, MD). Quantitative Real-Time PCR (qRT-PCR) was performed by the Vermont Cancer Center DNA Analysis Facility using RT 2 Real-TimeTM SYBR Green PCR Master Mix and Human drug resistance and metabolism template RT 2 ProfilerTM PCR Arrays (SABiosciences) (7900HT Sequence Detection System, Applied Biosystems). Data were analyzed using an on-line spreadsheet-based data analysis template (SABiosciences). qRT-PCR (TaqMan) was used to validate selected genes using Assay on Demand (AOD) Primers and Probes from Applied Biosystems. Creation of shERK1 and shERK2 stable MM lines HMESO cells were selected for these studies because these cells are well-characterized [ 16 ] and form MMs reproducibly after injection into SCID mice. Confluent HMESO cells were transfected with either ERK1 or ERK2, or scrambled control Sure Silencing Plasmids (4 sh sequences for each gene were used) from SA Biosciences (Frederick, MD), using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). After selection for 14 days in G418 (400 μg/ml)-containing medium, clones were screened by qRT-PCR for inhibition of ERK mRNA levels as compared to scrambled control (shControl) transfected clones. Two clones from each shERK1 and shERK2 groups were processed by limited dilution to obtain clones in which individual ERKs were inhibited by more than 70% in comparison to shControl clones. Following this procedure, shERK1 and shERK2 clones exhibiting inhibition of >80% ERK expression were obtained. Similarly, shERK1/2 lines were also created from PPMMill lines to verify observations obtained with HMESO line. The experimentally verified shRNA design algorithm (SABiosciences) assures gene-specificity and efficacy. An advanced specificity search in addition to BLAST built into the algorithm helped to reduce potential off-target effects. Flow cytometry To quantitate Dox fluorescence shControl (shCon), shERK1 and shERK2 HMESO cells were grown to confluence and then treated with Dox (0.5 or 5.0 μM) for 1 h or 5 h. Negative controls had no drug added. Cells were washed 3X with phosphate-buffered saline (PBS), trypsinized, counted, suspended in PBS, and Dox fluorescence was examined by flow cytometry using an LSRII flow cytometer (BD Biosciences, MA). A 695/40-nm-band-pass filter with a 685-nm long pass was used to measure Dox fluorescence. Fluorescence microscopy for Dox fluorescence shControl, shERK1 and shERK2 cells were grown to confluence in 4-chambered CultureSlides (BD Falcon, Bedford, MA) in medium containing 10% FBS. Media was replaced with that containing 0.5% FBS 24 h before treatment. Cells were either untreated or treated with 0.5 or 5 μM Dox for 1 h or 5 h at 37°C. Slides with attached cells were then washed in PBS and fixed in 100% methanol for 20 min at -20°C. Slides were washed in PBS and water, allowed to dry, and coverslipped with Aqua Poly/Mount (Polysciences Inc., Warrington, PA). Slides were then stored at 4°C until fluorescent images were acquired using an Olympus BX50 Light Microscope with attached mercury epi-fluorescence illumination. Affymetrix gene profiling Microarrays were performed on MM cell samples from 3 independent experiments as described previously [ 19 ]. Each of the samples was analyzed on a separate array, i.e., N = 3 arrays per MM line (3 independent biological replicates). A Human U133A 2.0 array (Affymetrix, Santa Clara, CA) was scanned twice (Hewlett-Packard GeneArray Scanner), the images overlaid, and the average intensities of each probe cell compiled. Microarray data were analyzed using GeneSifter software (VizX Labs, Seattle, WA). This program used a "t test" for pairwise comparison and a Benjamini-Hochberg test for false discovery rate (FDR 5%) to adjust for multiple comparisons. A 2-fold cut-off limit was used to assess statistical significance. Quantitative real time PCR (qRT-PCR) To validate microarray profiles and PCR Array profiles of genes, qRT-PCR (TaqMan) was performed as described previously [ 19 ]. Triplicate assays were performed with RNA samples isolated from at least 3 independent experiments. Fold changes in gene expression were calculated using the delta-delta Ct method using hypoxanthine phosphoribosyl transferase ( HPRT1 ) as the normalization control. The Assay on Demand primers and probes used were purchased from Applied Biosystems (Foster City, CA). Assessment of Dox effects on human tumors appearing after injection of stably transfected shERK1, shERK2 and shControl MM lines in a mouse xenograft model HMESO cells stably transfected with either shERK1, shERK2 or shControl were injected into 4 subcutaneous sites (5 × 10 6 cells per injection site) on the dorsa of 6 wk old Fox Chase Severe Combined Immunodeficient (SCID) mice (Jackson Laboratories, Bar Harbor, ME). All mice (N = 6/group) were weighed weekly and examined every other day for morbidity and tumor growth (measured using a digital caliper). Immediately after tumor appearance (1 wk post cell injection) each group was divided in two subgroups, each containing 3 mice. Three mice in each group were given Dox (6 μg/50 μl saline/tumor site) 3X weekly. The remaining 3 mice from each group received saline (50 μl/tumor site) 3X weekly. The Dox dose and frequency were previously determined to cause no toxicity to mice. After 6 wk of treatment, all mice were weighed and euthanized by intraperitoneal (ip) injection of sodium pentobarbital, necropsied to determine possible gross metastases, and major organs removed and stored in 4% paraformaldehyde before processing for histopathology. Tumors were characterized using previously described histochemical criteria [ 20 ] and karyotyped to prove that they were human in origin. Tumor volumes were calculated using formula (π × long axis × short axis × short axis)/6. All experiments using mice were approved by the Institutional Animal Care and Use Committee at the University of Vermont College of Medicine. Statistical analyses In all in vitro assays, at least 3 independent samples were examined at each time point per group in duplicate or triplicate experiments. Data were evaluated by ANOVA using the Student Neuman-Keul's procedure for adjustment of multiple pairwise comparisons between treatment groups or using the non-parametric Kruskal-Wallis and Mann-Whitney tests. Differences with p values ≤0.05 were considered statistically significant. The difference in tumor growth rates between different groups in in vivo studies was assessed using a hierarchical regression model to take into account the correlation between repeated measurements on the same tumor and multiple tumors in the same animal. In this analysis, the regression coefficient describing tumor growth is modeled as a function of treatment group as well as random variation due to differences between animals and tumors on the same animal.

Results Human MM lines show ERK1 and ERK2 activation in response to low (< LD 50 ) concentrations of Dox Four MM lines (MO, ME-26, HMESO, PPMMill) were treated with various concentrations (0-100 μM) of Dox for 24 h to determine LD 50 concentrations. As shown in Figure 1A , a Dox concentration of 25 μM was the approximate LD 50 concentration for MO and ME-26 lines whereas HMESO and PPMMill lines showed LD 50 concentrations of approximately 100 μM or greater, respectively (preliminary data not shown). After treatment with various concentrations of Dox, cell lysates were assessed for active (phosphorylated) and total ERK1/2 levels by Western blot analysis. The MO line showed a dose related increase in phosphorylation of both ERK1 and ERK2 that was significant (p ≤ 0.05) starting at the lowest concentrations (1 μM) of Dox used. ME-26 and HMESO lines also showed significant Dox-induced activation of ERK1 and 2 starting at 10 and 25 μM, respectively (Figure 1B ), whereas PPMMill cells showed comparable activation of ERK1 and 2 at 10-100 μM Dox. Pre-treatment of human MM cells with the MEK1/2 inhibitor U0126 (20 μM for 1 h) resulted in attenuation of Dox-induced ERK1/2 activation in all MM lines, whereas the inactive analog, U0124 (20 μM for 1 h), had no significant effects on Dox-induced ERK phosphorylation (Figure 1C ). Dox-induced ERK1/2 activation promotes survival of human MM cells To assess the role of Dox-activated ERK1/2 in cell survival, we pretreated human MM cells (MO, ME-26 and HMESO) with the MEK1/2 inhibitor (U0126, 20 μM) for 1 h before treating for 24 h with Dox at 25 (MO, ME-26) or 100 μM (HMESO), the approximate LD 50 concentration for each cell type. The MTS assay then was performed to determine cell viability. The higher concentrations of Dox (equal to LD 50 for various MM cell types) were used for viability assays as lower concentrations of Dox, (1, 5, 25 μM) had no effect on cell viability either alone or in combination with U0126 (data not shown). As shown in Figure 2A (grey bars), treatment with U0126 and Dox resulted in significantly more cell killing in all 3 MM lines evaluated as compared to Dox or U0126 alone. In HMESO and MO cells, U0126 alone also had a significant effect on reducing cell viability, suggesting the possible role of endogenous ERK1/2 activation in cell survival. The inactive analog, U0124 (20 μM for 1 h), had no toxic effects or modulation of Dox-induced cell killing in any MM line, confirming the specific effects of the U0126, MEK1/2 inhibitor. Human MM lines have high endogenous expression of many prosurvival and drug resistance related genes which are regulated by ERK1/2 A PCR Array using a 'human cancer drug resistance and metabolism' template on 2 human MM lines (MO, ME-26), compared to the nonmalignant LP9/TERT-1 human mesothelial cell line, showed that both MM lines had significantly (p ≤ 0.05) greater endogenous levels of many prosurvival and drug resistance genes (Table 1 and Additional File 1 , Table S1). Of the 10 most highly expressed genes for each line listed in Table 1 , mRNA expression of 6 genes ( BCL2, cFOS, MET, BRCA1, ESR2 and BRCA2 ) was common to both cell lines, whereas 6 genes were differentially expressed. mRNA levels of 2 common genes ( BCL2 and cFOS ) highly expressed in each MM line were also validated by qRT-PCR (*p ≤ 0.05 as compared to LP9/TERT-1 cells). In addition to the genes listed in Table 1 , many other genes were up-or down-regulated significantly in both cell types and are listed separately in Additional Table 1 . Exposure of both MM cell lines to the MEK1/2 inhibitor (U0126, 20 μM for 24 h) resulted in significantly altered levels of some of these genes ( BCL2, cFOS, BRCA1, AR, ESR2, CYP3A4, PPARγ, BRCA2, ABCC3 ) (Table 1 and Additional Table 1 ), suggesting a role of ERK1 or 2 in their regulation. Inhibition of either ERK1 or ERK2 sensitizes MM cells to Dox As the small molecule inhibitor, U0126, abrogated both ERK1 and ERK2 activation, we created stably inhibited (short hairpin, sh) ERK1 and ERK2 HMESO and PPMMill lines to determine if ERKs had similar or unique roles in Dox chemoresistance. The human HMESO and PPMMill MM lines were selected for this purpose as these lines were most insensitive (resistant) to Dox. A significant inhibition (≥70%) of ERK1 or ERK2 in respective lines was obtained as confirmed by Western blotting. In initial in vitro experiments, stable shERK1, shERK2 or shControl (shCon) MM lines were treated with Dox (100 μM, ~LD 50 dose) for 24 h, and cell viability was assessed by the MTS assay (HMESO) or by cell counting (PPMMill). As shown in Figure 2B , shERK1 and shERK2 cell lines showed significantly attenuated cell viability after Dox treatment as compared to shControl lines (Figure 2B , *p ≤ 0.05 as compared to respective untreated control; †p ≤ 0.05 as compared to Dox treated shControl). Although significantly increased Dox-induced cell killing was observed after inhibition of either ERK1 or ERK2, the shERK2 cell lines showed significantly (‡p ≤ 0.05) greater cell killing as compared to the shERK1 lines from both MMs (Figure 2B ). The shCon line, as discussed in the 'Material and Method' section, contains a vector with a scrambled sequence, which does not inhibit any gene. shCon cells are expected to behave like untransfected cells as they do in our experiments (compare HMESO in Figure 2A and 2B ). Inhibition of ERK1 or ERK2 results in greater accumulation of Dox in MM cells To show that inhibition of ERK1 or ERK2 increases Dox-induced toxicity by causing greater intracellular accumulation of Dox, we performed flow cytometry experiments on stably transfected HMESO lines treated with Dox (0.5 or 5.0 μM for 1 h or 5 h). Figure 3A shows that MM cell lines stably transfected with either shERK1 or shERK2 exhibited significant dose- and time-related increases in accumulation of intracellular Dox as compared to shControl cells treated with Dox at both time points (*p ≤ 0.05 as compared to respective shControl group). Dox at the low concentration (0.5 μM for1h) was retained marginally but significantly in the ERK1-inhibited (shERK1) HMESO line, whereas high Dox (5 μM for1 and 5 h) was retained by both ERK1 and ERK2-inhibited HMESO (sh) lines as compared to the shCon line treated with Dox. Data in Figure 3A correlate well with findings shown in Figure 2B , where Dox at the high concentration (100 μM for 24 h) shows reduced viability in the shERK2 group. Although Dox retention in both shERK1 and shERK2 groups was similar, the increased toxicity of Dox in the shERK2 group (Figure 2B , HMESO) could be attributed to additional factors. Figure 3B confirms the patterns of Dox accumulation by fluorescence microscopy in MM cells. Note the lack of Dox in the 0 (untreated groups) and the dose-related increases in intracellular fluorescence present in the shERK1 and shERK2 cells. Effect of ERK1 or ERK2 inhibition on ATP binding cassette genes (ABC transporters) in MM cells Based on data above and in Table 1 , we next hypothesized that ERKs modulated endogenous expression of ABC cassette genes that function to pump Dox and other chemotherapeutic drugs out of tumor cells, resulting in their decreased drug sensitivity. To address this hypothesis, we performed microarray analysis on shERK1, shERK2 and shControl HMESO cells (no exposure to Dox). Table 2 provides a list of 7 ABC genes (many involved in Dox transport) that had decreased mRNA levels in shERK1 and shERK2 cell lines. Validation of several changes in gene expression was performed using qRT-PCR (*Table 2 , p ≤ 0.05 as compared to shControl). We also examined endogenous levels of ABC transporter genes in HMESO MM cells as compared to nontransformed human mesothelial cells LP9/TERT-1 (p ≤ 0.05) (Table 3 ). These results showed that HMESOs showed striking decreases (> 100 fold) in mRNA levels of ABCG2 and ABCA1 as well as significant decreases in ABCA8, ABCC3, ABCB1, ABCG1 and ABCC4 expression, whereas other genes ( MDR/TAP , ABCA2 , ABCC5 and ABCA7 ) were upregulated. Tumors developing from shERK1 and shERK2 MM lines in a mouse xenograft model show decreased tumor growth rate after treatment with Dox To verify the functional effects of ERK inhibition and Dox treatment on tumor cell killing, we injected stable shERK1, shERK2 or shControl HMESO MM cells subcutaneously into SCID mice, and treated various groups with Dox or saline at the tumor site as soon as tumors appeared (1 wk post cell inoculation, 3X weekly) for a 6 wk period. As shown in Figure 4 , Dox significantly reduced the rate of tumor growth in all three animal groups compared to saline treatment, with the largest reduction occurring in the shControl group. In addition, Dox-treated animals in the shERK1 or shERK2 groups had significantly slower tumor growth than the Dox-treated animals in the shControl group. The differences between the shControl Dox-treated and shERK1 Dox-treated tumor growth rates occurred prior to 21 days post MM cell injection. All conclusions were derived by statistical analysis (described in the Method section) performed on different groups to compare alterations in tumor growth rate and not tumor volume.

Discussion Treatment of various MM lines with doses of Dox much lower than LD 50 concentrations resulted in phosphorylation of ERK1 and 2, the most abundant ERKs in mammalian cells. In addition to Dox, many other anti-cancer drugs such as paclitaxel and cisplatin induce activation of ERKs in different tumor types [ 21 - 23 ]. However, taxol inhibits ERK activation in different cell types depending upon experimental conditions [ 23 ]. In our study, Dox-induced ERK1/2 activation protected MM cells from Dox-induced cell death, as shown when MM lines were pretreated with the MEK1/2 inhibitor, U0126, prior to Dox exposure (Figure 2 ). In support of our findings, it has been reported that, in most cases, ERK activation protects cells from drug-induced cell death [ 23 , 24 ], while in some tumor cells, ERK activation contributes to cell death [ 21 , 22 ]. These different effects may be explained by differences in subcellular distribution of specific ERKs, the longevity of ERK signaling, or phosphorylation of different substrates which may dictate death or survival [ 25 ]. We studied 4 different MM lines for Dox responses after ERK1/2 manipulation either with an inhibitor (U0126) or by shRNA approaches. With the use of the ERK1/2 inhibitor (inhibits both ERK1 and 2), HMESO cells were the best responders (most susceptible) as compared to MO and ME-26 (both lines showed the same susceptibility) (Figure 2A ). A shRNA approach to inhibit either ERK1 or ERK2 was studied in 2 MM lines (HMESO and PPMMill). Of the two lines studied by this approach, HMESO again showed more sensitivity to Dox-induced killing after ERK1 or ERK2 inhibition as compared to PPMMill (Figure 2B ). In addition, in both cell lines, ERK2 inhibition was more effective than ERK1 inhibition in Dox-induced cell killing (Figure 2B ). Although regulation of apoptotic pathways has been implicated in resistance of many cancers to chemotherapy, we show that human MM lines endogenously overexpress many prosurvival genes ( BCL2, cFOS, MET , etc.) in comparison to nontransformed mesothelial cells. The increased levels of these commonly upregulated genes, as reported by our lab and others [ 18 , 26 - 29 ] may in part be responsible for drug resistance in MM cell lines. For example, BCL2 and BCL-xL antisense treatment facilitates apoptosis in mesothelioma cells, suggesting BCL2/BCL-xL bispecific antisense treatment in combination with cisplatin or gecitabine may result in a more effective therapy of MM [ 30 ]. Consistent with our findings, ERK1/2 activation has been linked to expression and activation of BCL2 in various systems [ 3 , 31 ] resulting in an anti-apoptotic or survival outcome. cFOS , a protooncogene and component of activator protein-1 (AP-1), is upregulated by crocidolite asbestos in rat pleural mesothelial cells [ 32 ], and endogenously upregulated in human mesothelioma cell lines and tumors [ 18 , 28 ]. We show for the first time that BRCA1 and BRCA2 are endogenously overexpressed in MM cells, and are pursuing their mutation and functional status in various MMs. ERK1/2 has been linked to feedback regulation of the tumor suppressor/DNA repair gene BRCA1 in irradiation induced DNA damage checkpoint activation [ 33 , 34 ]. BRCA2 was also endogenously upregulated in MM cells and ERK1/2 inhibition decreased expression of this gene (Table 1 ), consistent with already published work that ERK1/2 activation inhibits replication of prostate cells via upregulation of BRCA2 [ 35 ]. Another gene, PPARγ, which was upregulated only in ME-26 and was significantly inhibited by the U0126 MEK1/2 inhibitor is activated via an ERK1/2 dependent COX-2 pathway in macrophages [ 36 ]. Inflammatory pathways involving PPARγ or COX-2 are promising therapeutic targets in a number of cancers [ 37 ]. We also report for the first time the upregulation of a cytochrome P450 enzyme gene, CYP3A4, related to drug metabolism in the ME-26 epithelioid cell line that was decreased 3-fold after addition of U0126. The presence of the androgen receptor and its endogenous expression in sarcomatoid MM cells is also a novel finding, and both AR and ESR2 have been linked to the ERK pathway [ 38 - 40 ] as shown in Table 1 in MO cells. A recent study suggests that ER-β affects the prognosis of MM by acting as a tumor suppressor [ 41 ]. ATP-binding cassette (ABC) transporters transport various molecules, including chemotherapeutic drugs, across extra- and intracellular membranes. Increased expression of one or more of these proteins is seen in almost all resistant cancers and is considered responsible fully or in part for the observed drug resistance in most cancer cell lines. In a previous study using MM cell lines, coordinated overexpression of the multi drug resistance pump (MRP) and gamma-glutamylcysteine synthetase genes, but not MDR1, correlated with Dox resistance [ 42 ]. In the 3 MM lines we studied by PCR array or microarray analysis, different types of ABC transporter genes were endogenously overexpressed as compared to untransformed LP9/TERT-1 mesothelial cells (Table 1 , Table 3 ). The overexpression of different types of ABC genes in different MM cells further confirms the highly heterogenic nature of MM tumors that vary widely in their prognosis and response to therapy. Inhibition of ABC genes by ERK1 or 2 inhibition may be responsible for the increased accumulation of Dox observed in shERK1 and shERK2 MM cells (Figure 3 ). Among ABC genes inhibited by shERK2 in HMESO cells, ABCA8 is a relatively uncharacterized new transporter [ 43 ] whereas Dox is a known substrate for ABCC2 , ABCA2 and MDR/TAP [ 44 - 46 ]. Our data suggest that different ERKs regulate distinct ABC genes, and a detailed study is needed to understand the roles of different ERKs, including ERK5 that has been linked to chemoresistance in breast cancers [ 47 ], in ABC gene regulation. Consistent with our studies, ERK1 and 2 are linked to regulation of many ABC genes, including ABCG1 , ABCA1 , MDR1 , and MRP1 in various cancer and non-cancer cells [ 6 , 48 - 50 ].

Conclusions Our in vitro and in vivo studies here indicate that both ERK1 and ERK2 play significant roles in imparting Dox resistance to MM cells by modulating genes related to drug resistance and survival previously unidentified in MM cells. Most importantly, we demonstrate that gene expression of distinct ABC transporters is modulated by blocking ERK1 or ERK2, and show the relationship of these phenomena to Dox accumulation in human MM cells. Further, we demonstrate that blocking ERK1 and ERK2 enhances the chemotherapeutic potential of Dox in a murine xenograft model. The mechanisms of ERK1/2 action appear to involve both upregulation of prosurvival/antiapoptotic genes as well as ABC transporter genes. Based on our observations, ERK1/2 inhibitors in combination with chemotherapeutic drugs might be a better option to treat patients with MM than drugs alone.

Background Malignant mesotheliomas (MM) have a poor prognosis, largely because of their chemoresistance to anti-cancer drugs such as doxorubicin (Dox). Here we show using human MM lines that Dox activates extracellular signal-regulated kinases (ERK1 and 2), causally linked to increased expression of ABC transporter genes, decreased accumulation of Dox, and enhanced MM growth. Using the MEK1/2 inhibitor, U0126 and stably transfected shERK1 and shERK2 MM cell lines, we show that inhibition of both ERK1 and 2 sensitizes MM cells to Dox. Results U0126 significantly modulated endogenous expression of several important drug resistance ( BCL2, ABCB1, ABCC3 ), prosurvival ( BCL2 ), DNA repair ( BRCA1, BRCA2 ), hormone receptor ( AR, ESR2, PPARγ ) and drug metabolism ( CYP3A4 ) genes newly identified in MM cells. In comparison to shControl lines, MM cell lines stably transfected with shERK1 or shERK2 exhibited significant increases in intracellular accumulation of Dox and decreases in cell viability. Affymetrix microarray analysis on stable shERK1 and shERK2 MM lines showed more than 2-fold inhibition (p ≤ 0.05) of expression of ATP binding cassette genes ( ABCG1, ABCA5, ABCA2, MDR/TAP, ABCA1, ABCA8, ABCC2 ) in comparison to shControl lines. Moreover, injection of human MM lines into SCID mice showed that stable shERK1 or shERK2 lines had significantly slower tumor growth rates in comparison to shControl lines after Dox treatment. Conclusions These studies suggest that blocking ERK1 and 2, which play critical roles in multi-drug resistance and survival, may be beneficial in combination with chemotherapeutic drugs in the treatment of MMs and other tumors.

Competing interests The authors declare that they have no competing interests. Authors' contributions AS and BTM designed research; AS, JMH, MBM and SLB performed research; HIP contributed new reagents and tools; AS, BTM and PMV analyzed data; and AS and BTM wrote the paper. AS, JMH, MBM, SLB, BTM, PMV, MC, HIP and JRT critically reviewed the manuscript and given final approval of the version to be published. Supplementary Material

Acknowledgements We thank Jennifer Díaz for manuscript preparation. We acknowledge the help from the Vermont Cancer Center (VCC) DNA Analysis Facility for qRT-PCR and the INBRE/VGN Microarray Facility for microarrays. We also acknowledge the help of Colette Charland (University of Vermont) for help with Flow Cytometry. Grant Support : NCI (P01CA114047; M. Carbone, B.T. Mossman, J.R. Testa, H.I. Pass, and A. Shukla), Mesothelioma Applied Research Foundation (A. Shukla and B.T. Mossman), NIEHS T32ES077122 (B.T. Mossman and J.M. Hillegass), NIH (P30CA006927) an appropriation from the Commonwealth of Pennsylvania (J.R. Testa), and the Vermont Cancer Center/Lake Champlain Cancer Research Organization (A. Shukla). Microarray data accession number : GSE21750

CC BY

no

2022-01-12 15:21:44

Mol Cancer. 2010 Dec 15; 9:314

oa_package/2b/36/PMC3016286.tar.gz

PMC3016287

21144027

Background The infection rate of surgical cases is estimated at 2-3%, which likely represents an underestimation due to under-reporting [ 1 ]. Moreover, surgical site infections (SSIs) are becoming more frequent due to antibiotic resistant bacteria, and people living longer with more medical comorbidities [ 1 ]. Postoperative infection is a morbid complication, frequently leading to multiple operations, long-term use of antibiotics with their associated side effects, pain, and potential prolonged disability. Nationwide, SSIs may account for 10 billion dollars in health care expenditures annually [ 1 ]. The cost of a single surgical site infection can exceed 30,000 dollars [ 1 , 2 ]. Tremendous efforts to reduce hospital acquired infections are currently underway including the Surgical Care Improvement Project (SCIP) and specialty society campaigns such as the Association for Professionals in Infection Control (APIC) "Target Zero Program." Potentially preventable infections are considered avoidable by some third party payers as indicated by the policy of withholding hospital re-imbursement for some hospital acquired infections by the Centers for Medicare and Medicaid as of October 1 st 2008, and now by Blue Cross [ 3 , 4 ]. Several well-known procedures that help control the risk of surgical site infection include pre-operative antibiotics, proper hair clipping methods, and proper surgical skin preparation, as well as reduction of room activity, surgical duration, and the number of sterile field breaches [ 4 , 5 ]. Draping of the C-arm is an inefficiency that is well known to the orthopaedic trauma and spine community and experienced by the entire operative team daily. Since the introduction of intra-operative fluoroscopy in the 1950 s, no standard draping method or drape has been devised to protect the integrity of the sterile field while the lower portion of the C-arm (x-ray tube) repeatedly rotates from the unsterile zone into the sterile field. At present, un-standardized, inefficient, and wasteful techniques are employed for horizontal plane C-arm draping. These methods have been handed down as dogma to the last several generations of surgeons. As no other option was available, surgeons and nurses have done their best to protect patients with improvised draping techniques (Figure 1 ). These techniques do not adhere to AORN standards and repeatedly expose the sterile field and surgical team to contamination [ 6 ]. The sterile field is defined as the horizontal plane level with the surgical tabletop [ 6 , 7 ] (Figure 2 ). Contamination levels below the sterile field line are increased and every effort is made intra-operatively to avoid contact with this region [ 8 ]. Rotation of the C-arm into the horizontal position introduces a non-sterile object (x-ray tube) and/or contaminated drape(s) into the sterile field. This rotation is often required multiple times throughout a surgical case, which increases the potential for contamination. The Association of peri-Operative Registered Nurses; Standards and Recommended Practices states: "Nonsterile equipment (eg, Mayo stands, microscopes, C-arms) should be covered with sterile barrier material(s) before being introduced to or brought over a sterile field. Only sterile items should touch sterile surfaces. The equipment should be covered with a barrier material on the top, bottom, and all sides. Sterile barrier material also should be applied to the portion of the Mayo stand or other equipment that will be positioned immediately adjacent to the sterile field." [ 6 ] According to the accepted definition of sterile technique, the previous standard of improvised draping breeches the sterile field and this practice should be abandoned, given the availability of cost effective new technology. Presentation of the hypothesis A novel C-arm drape, which adheres to existing guidelines for sterile field maintenance will reduce risk factors linked to surgical site infection and reduce healthcare costs. The drape must provide unencumbered access to the surgical site and provide a standardized, reproducible, method of draping the x-ray tube as it enters the sterile field. The drape must also maintain its position above the sterile field line when deployed permitting maximum attention to the patient by the operative team. The drape and technique would conserve time and material by being rapidly deployable and permitting unlimited use with a single drape for each surgical case. A new standardized method using a familiar technique has been developed and accomplishes the above objectives. http://www.C-armor.com The sterile pouch concept of the drape originates from the long accepted sterile pouch present in hip and arthroscopy drapes utilized for arthroplasty and arthroscopic surgery (Figure 3 ). Using C-armor, the non-sterile portion of the C-arm is covered on all five sides in adherence with AORN standards (Figure 4 ). Additional advantages include the translucent nature of the drape, which helps the surgeon and technician properly position the beam emitter for the fluoroscopy images, thus potentially reducing hazardous radiation exposure. The sterile pouch may also be used to house instruments such as the Bovie or suction tubing during the operation. The drape often further protects the sterile field by catching dropped instruments. C-armor also protects the x-ray tube from biohazard fluids. Irrigation and body fluids can not only damage the equipment but also create a means for cross-contamination from one patient to another. Testing the hypothesis A direct comparison of draping techniques is impractical as no published C-arm draping techniques existed prior to the advent of C-armor. One study option is a study akin to Biswas et al., designed to swab the drape in order to measure the degree of contamination [ 9 ]. However, comparing the results to the wide array of existing techniques is not practical. Bacterial contamination of the sterile field is known to occur with time due to settling of aerosolized microorganisms [ 10 ]. It is not known, however, what degree of drape contamination is clinically significant. Furthermore a simple swab study will not take into account the potentially hazardous turbulent air flow created by improvised draping methods which moves air across the unsterile x-ray tube toward the wound or the additional risk factor reductions including: reduced operative time, and room activity afforded by C-armor. Unpublished industry data testing the time required to drape the C-arm using the improvised methods averages 30-45 seconds (C-armor.com). The time required to deploy C-armor is 2-5 seconds (C-armor.com). Increased room activity, increased operative time, and sterile field breeches are all known SSI risk factors [ 1 , 2 ]. Those skilled in the practice of orthopaedic trauma and spine surgery will readily understand that C-armor will reduce operative time, room activity and the number of sterile field breeches; three known SSI risk factors. Furthermore, all previous known techniques breach the sterile field according to AORN guidelines; which has potential ethical and legal implications. Whether the use of C-armor has a direct effect on the rate of surgical site infections is not yet known. The presumed reduction is SSIs is based on the existing literature of known risk factors and the logic that reducing these factors will lead to a reduction in infection rates. A study to definitively prove the drape reduces surgical site infections would require a very large sample size, and require accounting for many confounding variables such as medical co morbidities, degree of injury (open versus closed fractures), type of skin prep, antibiotic administration and timing, etc. Controlling for surgical technique would make it especially difficult to generate a large enough sample size. As interest and utilization grows, however, a multicenter trial could potentially gather this data. Until this data exists, utilization of equipment and techniques (C-armor) that adheres to published surgical safety principles would be prudent. Implications of the hypothesis Using a standardized, efficient C-arm draping method, SSI risk factors are reduced and, hence, SSIs are likely to be reduced [ 1 , 2 ]. Significant reduction in red-bag waste, and operative timesavings will also occur. http://www.C-armor.com Although the C-armor drape is a significant improvement over existing practices, several precautions must be taken to prevent fluid accumulation in the pouch and displacement of the drape below the sterile field. Blood and irrigation fluid left in the sterile pouch can be spilled onto the floor if not suctioned prior to the next horizontal plane image. When the x-ray tube is at one end of the pouch, the drape may fall below the sterile field line unless manually supported or adhered to the fastener tabs on the far end. Moreover, the drape can potentially sag below the level of the sterile field line if the existing drapes to which it is adhered are displaced in that direction intra-operatively. C-armor is an intuitive, practical, solution to a long-standing intra-operative inefficiency and offers many intuitive benefits. As experience and technology evolve, additional improvements will be incorporated to increase patient safety and to further augment existing infection control practices at healthcare facilities.

Background Intra-operative fluoroscopy for orthopaedic procedures frequently involves imaging in the horizontal plane, which requires the lower portion of the C-arm (x-ray tube) to be rotated from an unsterile zone (beneath the table) into the sterile field. To protect the integrity of the sterile field the C-arm must be draped repeatedly throughout the surgical case. The current, un-standardized, practice employs draping procedures which violate the Association of peri-Operative Registered Nurses (AORN) Standards and Recommended Practices, waste time and material, and pose an increased risk for surgical site infection. Presentation of the hypothesis Use of a novel sterile C-arm drape (C-armor) that maintains the integrity of the sterile field, will improve operating room efficiency and reduce surgical site infection risk factors. This reduction in risk factors may potentially reduce surgical site infections in orthopaedic surgical cases requiring repeated horizontal x-ray imaging. Testing the Hypothesis Savings in time and material and the reduction in surgical site infection risk factors afforded by using C-armor are intuitive to those skilled in the practice of orthopaedic surgery. Testing for a reduction in the number of microorganisms introduced to the surgical site by improved C-arm draping would be challenging due to the multiple confounding factors during a surgical operation. Determination of an absolute reduction in surgical site infections may be possible, but will require accounting for many confounding variables and a large study sample in order to achieve statistical significance. Implications of the Hypothesis Improved intraoperative workflow, healthcare savings and a reduction in surgical site infection risk factors will be achieved by utilizing a standardized and safe method of sterile field maintenance during intra-operative horizontal plane fluoroscopy.

List of Abbreviations AORN: Association of peri-Operative Nurses; APIC: Association for Professionals in Infection Control; SSI: Surgical Site Infection. Competing interests The author currently owns intellectual property including patents and trademarks related to C-armor.

Acknowledgements Contour Fabricators Inc for engineering, and design expertise.

CC BY

no

2022-01-12 15:21:44

Patient Saf Surg. 2010 Dec 13; 4:20

oa_package/80/0f/PMC3016287.tar.gz

PMC3016289

21176127

Background Despite much recent research, the archaeology of North Africa remains enigmatic, with questions of population continuity versus discontinuity taking centre-stage. Key issues concern the identity of the bearers of the Middle Palaeolithic (or Middle Stone Age) Aterian industry (the age of which has recently risen dramatically from ~40-20,000 years ago (ka) to ≥115-40 ka [ 1 , 2 ]), and whether or not there was continuity between these and the later Upper Palaeolithic populations of the Maghreb. This question has become more urgent with the discovery that the Aterian is associated in Northwest Africa with a very early appearance of evidence for behavioural modernity, such as perforated Nassarius shell beads, use of ochre and bone tools, and long-distance exchange networks - preceding those of southern Africa and making it likely that the Aterian was carried by anatomically modern (rather than archaic) humans [ 3 ]. The fate of the populations using this industry, and their possible connection with others in Africa and with the group who dispersed out of Africa ~60 ka to populate the rest of the world, has naturally become a question of great interest. Further debates have focused on the question of population replacement in the late Pleistocene and Holocene. The earliest Upper Palaeolithic industry in North Africa is the Dabban, limited to Cyrenaïca (a likely glacial refuge area [ 4 ]) and most likely dating from ~36-50 ka to ~20 ka [ 5 ]. Its similarities to Near Eastern Early Upper Palaeolithic industries have suggested an origin in the Levant [ 5 ], rather than locally in the Aterian, and similar blade industries have been found further south by 30 ka [ 1 ]. The Dabban was replaced by the Eastern Oranian (or Eastern Iberomaurusian) in Cyrenaïca possibly as early as ~18 ka [ 1 ], by which time the Upper Palaeolithic had extended eastwards for the first time along the coastal belt of the Maghreb, with the Iberomaurusian (or Oranian), the origins of which are similarly controversial, especially as it appears earliest in the Northwest. It began at the Last Glacial Maximum (LGM), ~22-20 ka, with a period of intensification in the Late Glacial ~15-13 ka, although recent work suggests a possible underlying industry as early as ~26 ka [ 1 , 6 ]. Similar debates concerning continuity versus replacement surround discussion of the widespread Epi-Palaeolithic Capsian industries, which saw expansion southwards from eastern Algeria into an increasingly humid Sahara in the early Holocene and sometimes associated with more gracile Mediterranean skeletal remains, and the subsequent emergence of the Neolithic [ 4 , 7 - 10 ]. The phylogeographic analysis of human mitochondrial DNA (mtDNA) has the potential to address questions such as these [ 11 ]. In particular, the mtDNA haplogroup U6 has a unique and highly distinctive distribution amongst human mitochondrial DNA lineages. It is found primarily in North Africa and the Canary Islands (albeit with secondary dispersals into Iberia and East Africa), with its highest frequency amongst Algerian Berbers (28%), and it has therefore been proposed to be linked to the ancestors of the indigenous Berber-speaking populations of North Africa [ 12 - 15 ]. Macaulay et al. [ 15 ] described U6 and its sister clade U5 as having evolved from a common ancestor in the Near East, approximately 50 ka; while U5 spread along the northern Mediterranean coast with the European Early Upper Palaeolithic, U6 dispersed along the southern coast, as far as Cyrenaïca, alongside the Dabban industry, ~40-50 ka, with a further expansion into Northwest Africa with the Iberomaurusian culture, ~22 ka. On this view, U6 evolved en route or within North Africa (as U5 evolved within Europe [ 11 ]), the presence of occasional derived U6 lineages in the Near East would signal more recent gene flow from North Africa [ 16 ]. These early studies were based only on information from hypervariable region I (HV-I) in the mtDNA control region and a small number of diagnostic RFLPs in the coding region, assayed in samples from several populations of North Africa and Iberia. In most of these regions, U6 frequencies are ~10% or less [ 17 ], and even appear absent from some Berber communities in Tunisia [ 18 , 19 ]. The major sub-haplogroup U6a (characterized by control-region transitions from the ancestor of haplogroup U at nucleotide positions 16172, 16219 and 16278) is highly dispersed, occurring throughout North Africa (and at low levels in the Near East and Iberia), and a further nested subclade, U6a1 (characterized by an additional transition at position 16189), follows a similar distribution. By contrast, U6b (characterized by variants at 16172-16219-16311) has a more limited range to the northwest of North Africa, the north of the Iberian Peninsula and, as a nested derivative (U6b1, characterized by a transition at position 16163), in the Canary Islands [ 13 ]. In particular, the U6b1 lineage in the Canary Islands has been considered a founder lineage for the colonization of this archipelago by the Guanches (culturally very similar to Northwest Africans), ~2-3 ka [ 13 ], a conclusion supported by studies of ancient DNA [ 20 ]. Hence its arrival suggests itself as a potentially useful calibration point for the mtDNA molecular clock, although the archaeological evidence for the colonisation time is rather insubstantial [ 21 ]. In fact, the absence of U6b1 lineages anywhere outside of the Canary Islands (the few exceptions detected in Spain and in Americas being most readily explained as recent migrants from there), and the failure to detect immediate ancestors in North Africa, seem to point to the emergence of this clade within the Canary Islands - although probably soon after their colonization, as it is observed across several islands. Maca-Meyer et al. [ 22 ] performed the first study of complete U6 mtDNA sequences (with 14 samples), defining a new U6 sub-haplogroup, U6c (characterized by the HV-I transition motif 16169-16172-16189), which was even more geographically restricted than U6b - limited to the west of North Africa and, as a derivative (U6c1, with an additional 16129 substitution), in the Canary archipelago. Using coding-region age estimates as maximum limits for radiation times, they proposed that the proto-U6 spread from the Near East to North Africa ~30 ka, alongside the Iberomaurusian industry, with U6a reflecting an African re-expansion from the Maghreb eastwards in Palaeolithic times, and U6a1 a further reverse movement from East Africa back to the Maghreb, possibly coinciding with the probable Afroasiatic linguistic expansion. The clades U6b and U6c, restricted to West Africa, had more localized expansions; they argued that U6b reached Iberia at the time of the diffusion of the Capsian culture in North Africa. However, a larger study by Olivieri et al. [ 23 ] was closer to the earlier interpretation of Macaulay et al. [ 15 ]. They confirmed the origin of U6, or at least that of its immediate ancestor, in southwest Asia, with an ancient introduction (alongside haplogroup M1, and the Dabban industry) to North Africa via the Levant, possibly during the Greenland Interstadial 12, from ~44-48 ka. They reaffirmed that the various U6 sub-groups originated in the southern Mediterranean area, dispersing subsequently to East Africa. Coalescence time estimates for U6 and it subclades have varied considerably amongst these studies. Yet these are critical for studies of prehistoric dispersals, since reliable estimates can bracket the timing of demographically significant events. For example, a regionally-specific clade may have arisen from a migration event on the edge in the tree leading to that clade, and if the diversification has then arisen in situ rather than prior to the presumed founder event, the estimate of the time to the most recent common ancestor (TMRCA) can provide a minimum bound on the age of the migration event (motivating the "founder analysis" [ 24 ]). However, success rests on a number of requirements, principally that the phylogeny can be well-estimated and the molecular clock that converts genetic differences into time depth is well-calibrated [ 25 ]. Considerable progress has recently been made on both of these fronts by more sophisticated analyses of the richer data source provided by mtDNA complete sequences. With respect to the molecular clock, there are many factors leading to uncertainty. There is the wide variation in positional mutation rates and violations of the independence of mutations at different positions. Obvious regions are the paired stems of rRNAs and tRNAs [ 26 ], which some authors remove from the analysis [ 27 ], but there are other locations in the mtDNA molecule which can also present a secondary structure related with a functional role [ 28 ]. There is the problem of multiple hits and saturation, leading to the curious observation that the total proportion of control-region polymorphisms in the African branches of the tree is lower than in the non-African ones. Selection is also an important issue, with a higher frequency of replacement substitutions in the younger branches of the human mtDNA phylogeny compared to the more internal branches [ 29 - 31 ]. Kivisild et al. [ 30 ] advocated the use of only synonymous diversity for estimation of the TMRCA, which is problematic for age estimations in young lineages, while Soares et al. [ 31 ] implemented a correction for the purifying selection effect on the mutation rate estimated for the entire molecule. The choice of calibration points is also an important issue. Traditionally, an outgroup is used, where the split time with the human lineage can be assigned in some way. For humans, the closest one is that corresponding to the human-chimpanzee split, for which the fossil evidence is controversial and which is in a time frame very distant from the TMRCA of the mtDNA of Homo sapiens , rendering the application of a strict clock problematic [ 32 ]. One recent analysis additionally used the chronometric ages of the available Neanderthal sequences as calibration points [ 33 ]. A strategy of multiple calibration points in conjunction with relaxed-clock methods, where the rate is allowed to vary among branches in the tree [ 34 ] is appealing, but this has been hard to implement in the human tree because of unavailability of secure multiple calibration points. Bandelt et al. [ 25 ] advocate that calibrated radiocarbon dates in favourable pioneer-settlement situations with a well-defined founder mtDNA scenario and a rich archaeological record could be used for calibration purposes, but consensus for both radiocarbon dates and founder mtDNA lineages are far from being achieved in most known settlement situations. Endicott and Ho [ 27 ] applied an internal calibration to the human mtDNA tree by specifying priors on the ages of three nodes in the tree associated with demographic signals: the entry into Australia and New Guinea by establishing a minimum of 40 ka for haplogroup P; and the post-Last Glacial Maximum expansion of haplogroups H1 and H3 (unfortunately suggested as 18 ka). This internal calibration was performed using a Bayesian approach with the software BEAST [ 35 ], and resulted in a substitution rate 1.4 times higher than that resulting from the human-chimpanzee calibration. BEAST also, however, allows a reconstruction of effective population size through time, by using the Bayesian skyline plot (BSP [ 36 ]), based on a coalescent model analysed by Markov Chain Monte Carlo sampling. BSPs do not require a pre-specified parametric growth model as do other methods and, although designed for use with population data, they have also been used to attempt parameter estimation from haplogroup data with some apparent success [ 37 ]. Here, we analyse an additional 39 complete U6 genomes, from the full range of the U6 geographic distribution, including the Near East, Iberia and the Canary Islands. This has enabled us to construct a phylogenetic tree including 89 U6 genomes in total, and to re-evaluate the demographic history of the haplogroup, and its role in North African prehistory, in the light of the recent developments in the calibration of mutation rates. A comparison tree was inferred for 141 U5 sequences from the literature, allowing us to test the use of four alternative internal calibration points (the settlement of the Canary Islands, the settlement of Sardinia and its internal population re-expansion, and the split between haplogroups U5 and U6 around the time of the first modern human settlement of the Near East) against the recently developed complete genome clock with a correction for purifying selection.

Methods Samples Based on information from HV-I that allowed the general classification into haplogroups, we selected 39 U6 individuals for complete sequencing: 22 from Portugal, most of them included in a published Portuguese database [ 38 ] and some new data; 5 from the Canary Islands; 1 from Morocco; 2 from a sample of 583 Ashkenazi Jews (of Polish and Russian ancestry [ 39 ]), 7 from 1143 non-Ashkenazi Jews (of Turkish, Bulgarian, Moroccan, Tunisian, and Ethiopian ancestry), and 2 from 253 Near Eastern Palestinians. Except in the case of the Canary Islands dataset, where we selected 5 U6b1 samples randomly, all the available U6 samples were sequenced from each of these datasets. The samples belonged to unrelated individuals, who gave informed consent for their biological samples to be used for mtDNA characterisation. The work complied with the Helsinki Declaration of Ethical Principles (59 th WMA General Assembly, Seoul, October 2008) and was approved by IPATIMUP Ethics Commission. Complete mtDNA sequencing and nomenclature We amplified mtDNA using 32 overlapping fragments as described elsewhere [ 40 ]. After purification, we used the forward primers for the sequencing, and in some cases, also the reverse primers (in the presence of polycytosine stretches, when polymorphisms A574C and T16189C occur). We performed sequencing on a 3100 DNA Analyzer (AB Applied Biosystems, Foster City, CA), and the resulting sequences were read with SeqScape (AB Applied Biosystems, Foster City, CA) and BioEdit version 7.0.4.1 [ 41 ], by two independent investigators. In cases of ambiguous base calls, the PCR and sequencing reactions were repeated. Furthermore, the protocols for rechecking both haplogroup-defining polymorphisms, previously well-established in the literature, as well as private mutations were followed [ 42 ]. Mutations were scored relative to the revised reference sequence, rCRS [ 43 ], and numbers 1-16569 refer to the position of the mutation in that sequence. The 39 complete mtDNA sequences have been deposited in GenBank (Accession Numbers HQ651676 - HQ651714 ). Statistical analyses For the U6 phylogeny reconstruction, besides the new 39 sequences obtained here, we used 50 complete sequences previously published [ 22 , 23 , 40 , 44 - 47 ], and some unpublished sequences deposited in GenBank from the Family Tree DNA Company (accession numbers provided in Additional File 1 ). We also inferred a U5 tree, from 141 complete sequences available in GenBank and listed in Additional File 1 . Preliminary network analyses [ 48 ] led to a suggested branching order for the trees, which we then constructed most parsimoniously by hand. The software mtDNA-GeneSyn [ 26 ] was used to convert files. For estimation of the TMRCA for specific clades in the phylogeny, we used ρ statistic frequentist, maximum likelihood and Bayesian phylogenetic analyses. We used the ρ statistic (mean sequence divergence from the inferred ancestral haplotype of the clade in question) with a mutation rate estimate for the complete mtDNA sequence of one transition in every 3,624 years [ 31 ], correcting for purifying selection by using the calculator provided with that paper. We estimated standard errors as in Saillard et al. [ 49 ]. We also obtained maximum likelihood (ML) estimates of branch lengths using PAML 3.13 [ 50 ], assuming the HKY85 mutation model with gamma-distributed rates (approximated by a discrete distribution with 32 categories). We converted mutational distance in ML to time using the same clock. Bayesian phylogenetic analyses were performed using BEAST 1.4.6 [ 35 ] with a relaxed molecular clock (lognormal in distribution across branches and uncorrelated between them) and the HKY model of mutation with gamma-distributed rates. For this analysis, we performed an independent mutation rate calibration using calibration points internal to the phylogeny, which could be associated with particular demographic events, as suggested before [ 27 , 32 ]. The four chosen calibration points were selected to be within the time scale of interest to this study. The dates of three of the calibration points were assigned a shifted exponential distribution in which the most recent age (indeed, the mode) for the given clade as well as the 95 th percentile were assigned. - For the first point, we assigned a mode age of 45 ka and a 95 th percentile of 60 ka for the split between the U5 and U6 lineages, based on the hypothesis that some of the first settlers of Europe [ 51 ] were carrying haplogroup U that would later evolve into U5 [ 11 ]. Considering this, the age of the U5-U6 split would be a minimum of 45 ka old. Reliable archaeological dates are minimum estimates of a first settlement, justifying the use of the exponential distribution. - The second calibration point with mode 2.3 ka and 95 th percentile 5 ka was the age of the U6b1 branch based on the archaeological date for the colonization of the Canary Islands. The first settlement of the Canary Islands is extremely uncertain and the archaeology of the islands is not extensive [ 21 ]. Furthermore, the fact that U6b is so uncommon in North Africa could mean that the populations that carried U6b into the Canary Islands disappeared by drift and U6b1 already existed outside the Canaries. Again an exponential distribution was used to capture this uncertainty. - The third point with mode 8 ka and 95 th percentile 14 ka was within U5b3a1, a subclade of which, U5b3a1a, is found only on Sardinia [ 52 ]. Sardinia was permanently colonized at least 8 ka [ 53 ]. The colonization time most probably took place between the radiation of the Sardinian-specific branch (U5b3a1a) and its split point with mainland Europe (U5b3a1), indicating that the latter is necessarily higher than 8 ka [ 52 ]. Again the age of the clade may be substantially higher than 8 ka and therefore an exponential distribution was used. - The fourth calibration point we used was an age estimate for the Sardinian U5b3a1a branch, assigned a normal distribution (truncated at zero) centred on 5.8 ka, with a standard deviation of 1,000 years. After the settlement of Sardinia, the population size was probably low (considering the long branch between U5b3a1 and the ancestor of U5b3a1a). It is tempting to identify the subsequent rather star-like radiation of U5b3a1a with a population expansion. The late Neolithic in Sardinia around 5-6 ka was a likely time of internal re-expansion and population increase [ 53 ] and we hypothesized that the age of U5b3a1a corresponds to this internal expansion. This interpretation gains strength from a signal elsewhere in the phylogeny: haplogroup M1 in Sardinia also presents a rather star-like clade [ 23 ] dating to 5.2 ka, based on ρ and the time-dependent clock. BEAST uses a Markov-chain Monte-Carlo (MCMC) approach to sample from the posterior distributions of model parameters (branching times in the tree and substitution rates). Specifically, we ran 150,000,000 iterations, with samples drawn every 1,000 MCMC steps, after a discarded burn-in of 15,000,000 steps. We checked for convergence to the stationary distribution and sufficient sampling by inspection of posterior samples. We also obtained Bayesian skyline plots from BEAST and visualised them with Tracer v1.3 from posterior distributions of parameters run for 50,000,000 iterations (with samples drawn every 1,000 MCMC steps, after a discarded burn-in of 5,000,000 steps). We used a generation time of 25 years to rescale the vertical axis of the BSP to years. In addition, we forced the larger sub-haplogroups (U6abd, U6a, U6bd, U6b, U6c and U6d) to be monophyletic in the analysis, as the presence of fast-evolving positions (such as 16189 and 16311) leads to the reconstruction of diverse phylogenies, which would not be comparable with the one inferred using network analysis; MCMC updates which violated this assumption were immediately rejected. To determine and visualise the geographical distribution of U6, U6a and U6bd, we constructed interpolation maps using the "Spatial Analyst Extension" of ArcView version 3.2 http://www.esri.com/software/arcview/ . We used the "Inverse Distance Weighted" (IDW) option with a power of two for the interpolation of the surface. IDW assumes that each input point has a local influence that decreases with distance. The geographic location used is the centre of the distribution area from which the individual samples of each population were collected. The data used are listed in Additional File 1 .

Results and Discussion The enlarged U6 phylogeny (Additional File 2 ) confirmed the principal sub-groups already identified [ 23 ], and did not reveal any novel ones. It highlights, however, that variation in HV-I can be misleading in regard to the branching structure, particularly when major splits in the phylogeny are supported only by mutations at hotspot positions, justifying Olivieri et al.'s caution in the naming of such clusters (their Fig. three). For example, the additional data now allows transitions at position 16189 in U6 to now be resolved into several events [ 23 ], showing that the old classification of "U6a1" based only on HV-I diversity [ 20 ], and unifying U6a1 with U6a2 and U6a3 in Olivieri et al. [ 23 ] is not reliable. The case for the postulated "U6a1" movement from East Africa back to the Maghreb advanced by Maca-Meyer et al. [ 20 ] is not favoured by our inferred phylogeny, as the 16189 transition does not identify non-monophyletic groups. Indeed, the only sub-clade which seems to have a preferred distribution in East Africa (three individuals from Ethiopia) falls within U6a2, with only coding-region diagnostic mutations at positions 6359 and 11204 (dating to 13.4 ± 4.0 ka). We display the geographic distribution of the frequencies of U6 and its major sub-clades across Europe, North Africa, the Arabian Peninsula and the Near East using HV-I data. Figure 1 confirms that U6a occurs most frequently in the west of North Africa, with two main peaks in Mauritania and Mozabites, most probably due to genetic drift, which is especially strong in the latter. U6bd (mostly likely U6b, since U6 d is very rare and cannot be distinguished from U6b on the basis of HV-I diversity) is restricted to the Canary Islands and to a few instances in North Iberia. The estimated ages for the various U6 groups (Table 1 ) obtained using the internal calibration points have a mean ratio of 0.92 (for ρ) and 0.95 (for ML) relative to dates using the Soares et al. [ 31 ] mutation rate, corrected for purifying selection. Most of the major U6 subclades coalesce at times between the Last Glacial Maximum and the early Holocene, and the age of U6 as a whole is ~37 ka, but with a large 95% confidence range encompassing roughly 25-50 ka, due to the small number of early branching events. The posterior medians and 95% intervals for the ages of the points used for calibration were: for the U5-U6 split, 46,222 years [45,000-50,293]; U5b3a1, 9,825 years [8,000-12,714]; U5b3a1a, 5,850 years [4,953-6,743]; and U6b1, 4,835 years [2,958-7,064]. The posterior distributions for the oldest points agree better with the prior dates than the youngest point of U6b1, indicating that the Bayesian estimates will readapt the priors according to the existing information in the tree. This is most obvious in the case of the age of U6b1, suggesting that either the first settlement of the Canary Archipelago was earlier than the (admittedly weak) existing archaeological evidence indicates [ 21 ] or that U6b1 had already arisen prior to colonisation and has since drifted to extinction in the northwest African source, or at least has yet to be sampled. The estimates using the rate of Soares et al. [ 31 ] are both somewhat younger, suggesting that the Bayesian calibration (with only one calibration point in this range) may be failing to correct adequately for purifying selection at this time depth. It is also worth pointing out that the age of haplogroup U reported here (the U5-U6 split at around 46 ka) is slightly lower than the age of U reported recently using ρ and ML and the complete sequence clock corrected for purifying selection [ 11 , 31 ] which tended to be > 50 ka. In fact, a ρ estimate of 44,145 years [32,460-56,254] is obtained using only the U5-U6 data with the Soares et al. rate; so the age of U in the Bayesian analysis would most probably have been higher if all the sub-branches of U had been included. The ages of U5b3a1 and U5b3a1a were 11,912 [3,456-20,777] years and 4,128 years [2,016-6,269] respectively. The comparison of the posterior mean for the ages in Bayesian with the ρ estimates for the four points indicate a mean ratio of 1.1 higher in the Bayesian analysis, in contrast to the trend obtained in the U6 ages described above. The Bayesian and ρ estimates overlap. A further check of the clock is the age of U5a2a [ 54 ], since a single HV-I sequence belonging to this clade has been obtained from an ancient skeletal sample dated by radiocarbon to 7.8 ka [ 55 ]. The date obtained for U5a2a with the internal calibration was 8,045 years [4,391-13,184], concordant with the minimum possible age of the clade. The BSPs (Figure 2 ) computed for the U6 and U5 haplogroups (based on the mutation rate obtained with the internal calibration) display quite interesting patterns. One must keep in mind that BSPs were developed to estimate (effective) population size through time from a random sample of sequences [ 56 ], and there is little doubt that haplogroups do not equate to physically separated populations. Even so, the signs of expansions displayed in the U5 and U6 phylogenies may reflect expansions of populations bearing these haplogroups (as well as others), located in Europe and North Africa, respectively (compare the approach of Atkinson et al. [ 37 ] in sub-Saharan Africa). Another model assumption, panmixia, seems inappropriate for U5 and U6 together, since they have evolved on opposite sides of the Mediterranean Sea. Thus it seems appropriate to compute the BSPs separately. We performed this using the information gleaned from the initial internal calibration on the joint U5/U6 data set (in order to exploit calibration points in both haplogroups). The BSP for U5 shows a small effective population size from the time of its origin until a strong expansion led to a ~11-fold increase within a period of ~6 ka in the late Pleistocene/early Holocene (14.3-8.3 ka). Following this, the effective size again remained more or less constant till a second strong expansion occurred, leading to an increase of ~5-fold after 4.3 ka (we took the year 1600AD as the most recent time-point [ 56 ]). For U6, the pattern is slightly more complex. The initial effective size was somewhat larger (~1.6 times) compared with U5, and the initial expansion begins earlier, ~22.2 ka, with a more gradual expansion until ~10.2 ka, leading to a 3-fold increase. As with U5, there is a further expansion in the Neolithic, after 4.8 ka with a shallower increase of ~1.5-fold.

Results and Discussion The enlarged U6 phylogeny (Additional File 2 ) confirmed the principal sub-groups already identified [ 23 ], and did not reveal any novel ones. It highlights, however, that variation in HV-I can be misleading in regard to the branching structure, particularly when major splits in the phylogeny are supported only by mutations at hotspot positions, justifying Olivieri et al.'s caution in the naming of such clusters (their Fig. three). For example, the additional data now allows transitions at position 16189 in U6 to now be resolved into several events [ 23 ], showing that the old classification of "U6a1" based only on HV-I diversity [ 20 ], and unifying U6a1 with U6a2 and U6a3 in Olivieri et al. [ 23 ] is not reliable. The case for the postulated "U6a1" movement from East Africa back to the Maghreb advanced by Maca-Meyer et al. [ 20 ] is not favoured by our inferred phylogeny, as the 16189 transition does not identify non-monophyletic groups. Indeed, the only sub-clade which seems to have a preferred distribution in East Africa (three individuals from Ethiopia) falls within U6a2, with only coding-region diagnostic mutations at positions 6359 and 11204 (dating to 13.4 ± 4.0 ka). We display the geographic distribution of the frequencies of U6 and its major sub-clades across Europe, North Africa, the Arabian Peninsula and the Near East using HV-I data. Figure 1 confirms that U6a occurs most frequently in the west of North Africa, with two main peaks in Mauritania and Mozabites, most probably due to genetic drift, which is especially strong in the latter. U6bd (mostly likely U6b, since U6 d is very rare and cannot be distinguished from U6b on the basis of HV-I diversity) is restricted to the Canary Islands and to a few instances in North Iberia. The estimated ages for the various U6 groups (Table 1 ) obtained using the internal calibration points have a mean ratio of 0.92 (for ρ) and 0.95 (for ML) relative to dates using the Soares et al. [ 31 ] mutation rate, corrected for purifying selection. Most of the major U6 subclades coalesce at times between the Last Glacial Maximum and the early Holocene, and the age of U6 as a whole is ~37 ka, but with a large 95% confidence range encompassing roughly 25-50 ka, due to the small number of early branching events. The posterior medians and 95% intervals for the ages of the points used for calibration were: for the U5-U6 split, 46,222 years [45,000-50,293]; U5b3a1, 9,825 years [8,000-12,714]; U5b3a1a, 5,850 years [4,953-6,743]; and U6b1, 4,835 years [2,958-7,064]. The posterior distributions for the oldest points agree better with the prior dates than the youngest point of U6b1, indicating that the Bayesian estimates will readapt the priors according to the existing information in the tree. This is most obvious in the case of the age of U6b1, suggesting that either the first settlement of the Canary Archipelago was earlier than the (admittedly weak) existing archaeological evidence indicates [ 21 ] or that U6b1 had already arisen prior to colonisation and has since drifted to extinction in the northwest African source, or at least has yet to be sampled. The estimates using the rate of Soares et al. [ 31 ] are both somewhat younger, suggesting that the Bayesian calibration (with only one calibration point in this range) may be failing to correct adequately for purifying selection at this time depth. It is also worth pointing out that the age of haplogroup U reported here (the U5-U6 split at around 46 ka) is slightly lower than the age of U reported recently using ρ and ML and the complete sequence clock corrected for purifying selection [ 11 , 31 ] which tended to be > 50 ka. In fact, a ρ estimate of 44,145 years [32,460-56,254] is obtained using only the U5-U6 data with the Soares et al. rate; so the age of U in the Bayesian analysis would most probably have been higher if all the sub-branches of U had been included. The ages of U5b3a1 and U5b3a1a were 11,912 [3,456-20,777] years and 4,128 years [2,016-6,269] respectively. The comparison of the posterior mean for the ages in Bayesian with the ρ estimates for the four points indicate a mean ratio of 1.1 higher in the Bayesian analysis, in contrast to the trend obtained in the U6 ages described above. The Bayesian and ρ estimates overlap. A further check of the clock is the age of U5a2a [ 54 ], since a single HV-I sequence belonging to this clade has been obtained from an ancient skeletal sample dated by radiocarbon to 7.8 ka [ 55 ]. The date obtained for U5a2a with the internal calibration was 8,045 years [4,391-13,184], concordant with the minimum possible age of the clade. The BSPs (Figure 2 ) computed for the U6 and U5 haplogroups (based on the mutation rate obtained with the internal calibration) display quite interesting patterns. One must keep in mind that BSPs were developed to estimate (effective) population size through time from a random sample of sequences [ 56 ], and there is little doubt that haplogroups do not equate to physically separated populations. Even so, the signs of expansions displayed in the U5 and U6 phylogenies may reflect expansions of populations bearing these haplogroups (as well as others), located in Europe and North Africa, respectively (compare the approach of Atkinson et al. [ 37 ] in sub-Saharan Africa). Another model assumption, panmixia, seems inappropriate for U5 and U6 together, since they have evolved on opposite sides of the Mediterranean Sea. Thus it seems appropriate to compute the BSPs separately. We performed this using the information gleaned from the initial internal calibration on the joint U5/U6 data set (in order to exploit calibration points in both haplogroups). The BSP for U5 shows a small effective population size from the time of its origin until a strong expansion led to a ~11-fold increase within a period of ~6 ka in the late Pleistocene/early Holocene (14.3-8.3 ka). Following this, the effective size again remained more or less constant till a second strong expansion occurred, leading to an increase of ~5-fold after 4.3 ka (we took the year 1600AD as the most recent time-point [ 56 ]). For U6, the pattern is slightly more complex. The initial effective size was somewhat larger (~1.6 times) compared with U5, and the initial expansion begins earlier, ~22.2 ka, with a more gradual expansion until ~10.2 ka, leading to a 3-fold increase. As with U5, there is a further expansion in the Neolithic, after 4.8 ka with a shallower increase of ~1.5-fold.

Conclusions The Bayesian evolutionary analyses conducted in this work, as well as the comparison between recently updated mutation rates and the improvement of the phylogenetic resolution for the mitochondrial haplogroup U6, enable us to revise our interpretation of the evolutionary history of haplogroup U6 and sketch a demographic prehistory for North Africa, which we can then compare with that of Europe. In particular, they allow us to address many of the archaeological questions raised in the introduction. The recently revised archaeological dates for the Aterian industry of North Africa emphasize that the makers of this industry do not appear to have left any imprint in the maternal lineages of present-day North Africans. The oldest arrivals amongst extant mtDNAs appear to be the U6 and M1 lineages, which date to 36.6 (24.9; 48.8) and 25.4 (17.9; 33.1) ka respectively [ 31 ]. As with U5 in Europe [ 11 ], the arrival time could be older in each case, since the haplogroups appear likely to have arisen within the southern Mediterranean region from haplogroup U and M ancestors, making dating the arrival time very imprecise. Nevertheless, the estimates seem to match best the appearance of the Upper Palaeolithic Dabban industry in Cyrenaïca, as suggested before [ 15 , 23 ]. There is an intriguing further signal in the U6 data, witnessed by the Bayesian skyline plot. For the European haplogroup U5, which is one of the most ancient in Europe [ 11 ], we identified a strong expansion (an ~11-fold increase in effective population size) occurring in the Lateglacial period between the LGM and the beginning of the Holocene, followed by another large population expansion (~5-fold) after 5 ka, evidently associated with late Neolithic/early Bronze Age (rather than, for example, the early Neolithic expansion in Europe, which began ~8.5 ka). For U6, by contrast, the corresponding increases in effective sizes were less marked (~3-fold and ~1.5-fold, respectively), and the signal indicates that the expansion began earlier, ~22 ka. This coincides closely with the beginning of the Iberomaurusian industry in the Maghreb. These results therefore suggest that the Iberomaurusian was initiated by an expansion of modern humans of ultimately Near Eastern, carrying mtDNA haplogroup U6, who had spread into Cyrenaïca ~35-45 ka and produced the Dabban industry. The link back to the Near East and the European Early Upper Palaeolithic (which likely has the same source) may explain the suggested skeletal similarities between the robust Iberomaurusian "Mechta-Afalou" burials and European Cro-Magnon remains, as well as the case for continuity of the bearers of the Iberomaurusian industry from Morocco with later northwest African populations suggested by the dental evidence [ 57 ]. We can compare the U5 and U6 BSPs with the ones for geographic regions published in Atkinson et al. [ 56 ], inferred from a more random sample of mtDNA sequences observed in those regions - for which the model behind the BSP is, on the surface, a better match. Similar to the situation reported for sub-Saharan Africa [ 37 ], the picture that emerges from U5 and U6 appears to represent well the general demographic patterns observed in, respectively, Europe and North Africa (although the latter was combined the Near East). This observation suggests that by investigating in depth U6 and U5, the oldest lineages present in North Africa and Europe, respectively, we are indeed receiving signals from the demographic pre-history of modern humans in these regions. Aside from U6, North Africa was also the recipient of European, Near Eastern and sub-Saharan African lineages most of which most likely arrived in the Holocene. Haplogroups H1, H3 and V expanded in Iberia in the Lateglacial/postglacial [ 11 , 58 - 61 ], and evidently spread into North Africa from Iberia across the Gibraltar Straits, most likely in the early Holocene [ 62 - 65 ]. Although the postglacial Capsian industry appears to have originated in eastern Algeria, it is tempting to hypothesize a connection with the arrival of these new populations from southwest Europe. Intriguingly, although U5b1, which also expanded from southwest Europe in the Lateglacial, has not been seen in Moroccan Berbers, it has been identified amongst Algerian Berbers and Fulbe from Senegal, as well as Iberia, Italy and northern Eurasian Saami and Yakut [ 44 ]. Most sub-Saharan lineages observed in North Africa are presently difficult to date and probably arrived at various times, but the age of the sub-Saharan subclade L3e5 indicates its arrival in North Africa from the south ~7 ka, following its expansion in the immediate postglacial humid phase ~11.5 ka [ 66 ]. Other L3 lineages seem to have been introduced even in more recent times, during the slave trade initiated by the Arab conquest of North Africa [ 67 ]. The Near Eastern haplogroups J and T (and probably K) appear to be concentrated more towards the east [ 68 ], mirroring the higher densities of U6, H and V in the west [ 64 ]. These may reflect the spread of the Neolithic into North Africa from the Levant, but their phylogeography awaits detailed analysis.

Background The archaeology of North Africa remains enigmatic, with questions of population continuity versus discontinuity taking centre-stage. Debates have focused on population transitions between the bearers of the Middle Palaeolithic Aterian industry and the later Upper Palaeolithic populations of the Maghreb, as well as between the late Pleistocene and Holocene. Results Improved resolution of the mitochondrial DNA (mtDNA) haplogroup U6 phylogeny, by the screening of 39 new complete sequences, has enabled us to infer a signal of moderate population expansion using Bayesian coalescent methods. To ascertain the time for this expansion, we applied both a mutation rate accounting for purifying selection and one with an internal calibration based on four approximate archaeological dates: the settlement of the Canary Islands, the settlement of Sardinia and its internal population re-expansion, and the split between haplogroups U5 and U6 around the time of the first modern human settlement of the Near East. Conclusions A Bayesian skyline plot placed the main expansion in the time frame of the Late Pleistocene, around 20 ka, and spatial smoothing techniques suggested that the most probable geographic region for this demographic event was to the west of North Africa. A comparison with U6's European sister clade, U5, revealed a stronger population expansion at around this time in Europe. Also in contrast with U5, a weak signal of a recent population expansion in the last 5,000 years was observed in North Africa, pointing to a moderate impact of the late Neolithic on the local population size of the southern Mediterranean coast.

Authors' contributions LP conceived the study, coordinated its design and drafted the manuscript. RFD, VF, JBP and MDC carried out the molecular genetic studies, participated in the sequence alignment and in the phylogeny reconstruction. MBR and HM revised the manuscript, in particular focusing on the archaeology of North Africa and interpretation of the genetic results against the archaeological background. NMS and PS performed the statistical analyses and interpreted results with LP, DMB, MBR and VM. All authors read and approved the final manuscript. Supplementary Material

Acknowledgements FCT, the Portuguese Foundation for Science and Technology, supported this work through the research project PTDC/ANT/66275/2006 and the personal grants to VF (SFRH/BD/61342/2009), JBP (SFRH/BD/45657/2008), MDC (SFRH/BD/48372/2008), HM (SFRH/BD/44089/2008) and PS (SFRH/BPD/64233/2009). IPATIMUP is an Associate Laboratory of the Portuguese Ministry of Science, Technology and Higher Education and is partially supported by FCT. PS was also supported by a Marie Curie Early Stage Training Grant. We thank Sturla Ellingvåg (of Explico) for the Canarian samples.

CC BY

no

2022-01-12 15:21:44

BMC Evol Biol. 2010 Dec 21; 10:390

oa_package/2d/15/PMC3016289.tar.gz

PMC3016290

21122143

Background Adolescent Medicine is a pediatric subspecialty that addresses clinical issues that are often unique and different from other aspects of training in pediatrics. Pediatric postgraduate training programs in North America are required to provide at least a one-month block in Adolescent Medicine [ 1 , 2 ] for trainees. During this training, residents may be exposed to ethically challenging or controversial issues and to patients who are stereotypically labeled as being 'difficult'. Unlike with younger children, trainees most often communicate directly with the adolescent patient for both assessment and management issues. Some studies in the literature have examined pediatric residents' and other trainees' clinical skills and/or knowledge of adolescent health care [ 3 - 8 ]. Others have assessed pediatric residency training programs for the adolescent medicine training provided to their residents [ 9 ], and some have even looked at physicians' personal adolescent experiences or values and their subsequent effects on the delivery of health care to adolescent patients [ 10 , 11 ]. However, no studies to date have explored the experiences and perceptions of residents themselves during their postgraduate training in Adolescent Medicine and how such clinical experiences and exposure to different patient populations and patient-provider interactions may differ from the rest of their training in pediatrics. The aim of this study was to explore the general experiences and possible challenges faced by pediatric residents during their rotation in Adolescent Medicine. Such information may be used to a) better address the needs of residents during this rotation; b) create an environment that supports learning; and c) develop curricula that foster the development of clinical competence in the provision of care to adolescents. By addressing these issues, one hopes that the ultimate goal of providing quality patient care and improved patient outcomes is achieved.

Methods Because of the nature of the question and the lack of literature on the topic, an exploratory, qualitative design, as described by Strauss and Corbin [ 12 ], was adopted for this study. Setting The study was conducted between August 2006 and July 2007 at an academic tertiary care pediatric hospital in Canada, where the Division of Adolescent Medicine provides outpatient and inpatient services for a range of adolescent issues. Approval for the project was obtained from the institution's Research Ethics Board. Participants As part of their core training, pediatric residents have a four-week rotation in Adolescent Medicine during their first year of residency (PGY1). Using convenience sampling [ 13 ], all PGY1 pediatric residents who completed their rotation were asked to participate in the study. All other trainees were excluded. Eligible participants were contacted by a research assistant external to the Division. Each resident was contacted at the end of his/her rotation, after his/her end-of-rotation evaluation had been completed and submitted. Written consent was obtained from all interested participants. Recruitment continued until theoretical saturation [ 12 , 13 ] was achieved. Data collection Data was collected using open-ended, semi-structured, individual interviews, which were approximately 30-60 minutes in duration. The interview questions were prepared and tested beforehand and modified accordingly. Interviews were audio-recorded and conducted in a private setting by the same research assistant who recruited the participants. The interviewer also took field notes and collected demographic information from each participant. Audiotapes were transcribed verbatim with the exclusion of all identifying data. Two of the interviews were not audio-recorded because of technical problems; analyses of those two interviews were based on the information gathered from the field notes. Because only 1-2 residents rotated through Adolescent Medicine each month, transcripts were randomly assigned non-consecutive numbers for identification to further ensure participants' anonymity. Data analysis Data analysis occurred alongside data collection, a method referred to as the constant comparative method [ 13 ]. Three transcripts were analyzed at a time to ensure participants' confidentiality and anonymity. Transcripts were independently analyzed for emergent themes by all three investigators, and a manual coding structure was developed through group negotiation [ 12 , 13 ]. Analysis occurred in a hierarchical manner with the identification of codes, concepts, and themes. When new themes failed to emerge, theoretical saturation [ 12 ] was considered to have been achieved, and no additional participants were recruited. Findings from the initial analysis were shared with a representative number of the study participants on an individual basis. These member checks [ 13 ] provided additional validation of the data.

Results Of all eligible trainees approached to participate, one declined and two others agreed but did not confirm interview appointments, even after several attempts were made to schedule them. Theoretical saturation was achieved by conducting interviews with 13 participants, 7 males and 6 females. Their mean age was 28.8 years, with a range of 25-42 years. Five participants received their medical education in countries other than the USA and Canada. Three key themes emerged in the data (Table 1 ): 1) gaining exposure , 2) taking on a professional role , and 3) achieving self-awareness . These three themes were further subcategorized and are reviewed below. Gaining Exposure Knowledge Participants spoke of their exposure to adolescents and described gaining knowledge about various adolescent issues, including the development and behavior of adolescents. Residents expressed enhanced insight into the complexity and reality of adolescents' lives and decisions: " The lived experience of having actually worked with the conditions that I've read about and putting that into practice certainly made my feelings about the conditions and the adolescent's suffering from them more real" (Participant 3). Insight The insight gained through their experiences with adolescents with chronic illness was described in the following quote: Like, they (adolescents with chronic illness) have an active home life and lifestyle, but at the same time the medical aspects impact on those other areas in some way or another. It's hard to describe in words, but a sense I guess of wonderment too that these guys have gone through so much hardship through their childhood, were learning now how to engage with other peers who probably had not had that kind of experience, and try to be as normal as possible, where they may not appear normal to their peers... (Participant 2). Comprehensiveness The biopsychosocial approach to adolescent health care was regularly brought up by participants. This holistic and comprehensive approach was typical of the care provided to adolescents during this rotation, as opposed to strictly focusing on the main concern/problem as residents had experienced in other areas of pediatric medicine. As one participant said: "...an advantage of the rotation is that we don't really have much exposure to adolescents in other areas of the hospital, and when we do, it's very focused on their medical issue and not looking at like everything else..." (Participant 1). Another resident referred to Adolescent Medicine as a "crossroad of medical issues and psychosocial issues" (Participant 2), reflecting the comprehensive nature of the care provided. Taking on a Professional Role The second theme that emerged in this study was the professional role that participants acquired. Two subcategories emerged within this theme: the professional relationship with the adolescent patient and the role of the resident within the health care team. Relationship with patient Residents felt that as a result of their experiences in their rotations in Adolescent Medicine, they were better able to establish rapport and engage adolescent patients. Residents discussed directly communicating with the adolescents, rather than through their parents, and approaches for communicating with adolescents were individually tailored to meet the needs, personalities, and developmental stages of the adolescent patients. One of the study participants commented: I feel more comfortable working with them. ...[a]nd now my out-patient (who was previously) in the unit is saying 'can you be my doctor? Can you be my doctor?' Because you know, ... getting along with them, talking with them, know how do they feel, how do they think at this age ... (Participant 5). Residents' desire to help this patient population was evident: "...and you felt that you should be there all the time helping, not just for medical issues and besides emotional. Then just helping to figure out what's going on and find out what is the best way for them" (Participant 5). Participants also identified an advocacy role with their adolescent patients: "... [a]nd they (adolescents) have rights, especially rights to decide for themselves" (Participant 5). The participants described a variety of feelings toward their adolescent patients. A participant shared how he felt after an interaction with a teen mom: Frustration. A little bit of shock at some of the presentations, and sadness. But also significant and profound moments of connection and happiness that headway was being made or that understanding seemed to be created and a therapeutic bond developed (Participant 3). Other residents also reported feeling frustrated at times. These feelings of frustration were usually related to patients not adhering to treatment recommendations and appointment scheduling, as adolescent patients were often either late to or did not attend their appointments. Frustration was also expressed toward patients with eating disorders; this seemed to be related to participants' lack of understanding of the underlying pathology: "[a]nd I was frustrated because she was choosing, I felt, to take on a sick role. And it was her choice. And I didn't know. I thought it was all behavior and not organic in origin" (Participant 13). Participants also shared the positive feelings and satisfaction they experienced in working with adolescents patients: " I looked forward to coming into work to work with them (the patients) " (Participant 9). Role within health care team The role of the resident within the health care team was the second subcategory that emerged within this theme. Involvement with an interprofessional team not only aided in the comprehensive appreciation of the issues from various viewpoints, but also provided a supportive environment for the trainees: "I think the team's been really good in a confidential way, just in discussing (issues) among health care professionals in the team of sharing emotions and kind of venting that way, which I think is a professional way to do things" (Participant 11). Participants regularly communicated with other members of the team, whether nurses, social workers, dietitians, or other physicians. Participants viewed the rest of the team as providing a supportive, safe and healthy learning environment and recommended that future trainees discuss challenging situations with members of the health care team. One of the residents made the following statement: "...there is a profound amount of wisdom to be learned from all the people you're working with" (Participant 3). Self-Awareness The final theme that emerged was self-awareness. All of the participants were reflective of their experiences in Adolescent Medicine and were conscious of their personal values and beliefs. They expressed an awareness of their own experiences as adolescents and were aware of shifts in their attitudes toward adolescents and/or Adolescent Medicine. Personal values All participants identified some of their personal values and biases; for some, these were consistent with those of their adolescent patients, while for others, there was tension between their own personal values and those of their patients. For the latter group, there were some residents who were aware of this conflict and believed that it was necessary to be nonjudgmental and offer adolescents all possible options, even if these options conflicted with their own personal values and beliefs. These residents also felt that it was important to try not to influence adolescents with their own personal beliefs/values. For others who had conflicting values, this conflict posed a significant challenge, and a resident's main coping mechanism was to avoid or not take part in the situation: "I think I avoided situations that would have been the most difficult, or situations that I would not have been able to handle. So no, I think I was just consciously aware of situations and didn't want to be part of (them)" (Participant 11). Other forms of coping strategies identified by residents included discussing situations with other members of the health care team, with a member of their own family, or utilizing faith-based support. The specific clinical scenarios that were avoided by a few of the residents, because of conflicting personal values and beliefs, were situations involving discussions about contraception and/or counseling a pregnant teenager. Past personal experiences Participants were also reflective of their own adolescent experiences: "I don't think that anyone could say that their adolescence didn't influence it (the experience in Adolescent Medicine) in some way. Like everyone's experience affects every subsequent experience" (Participant 1). Many of the participants compared their adolescent lives and experiences to those of their patients, with some being consciously aware of their internal biases: "... (I was) a bit judgmental on how teens act today compared to when I was a teenager. Like the things that they dare said, or like the drugs they took, the amount of people they slept with..." (Participant 13). Others spoke about trying to maintain a more neutral standpoint: "(I) kept an open mind and didn't assume anything about any of the patients that I met" (Participant 2). Attitudinal shift Participants were also aware of their attitudinal shifts. Several had preconceived ideas about adolescents and/or Adolescent Medicine, which shifted during the rotation. These attitudinal shifts were generally positive: "... I was kind of a bit scared...but then when I got to know them (street involved youth) better, they're actually more friendly than I thought they are. They're just teenagers and they're very nice essentially" (Participant 12). This general sense of self-awareness prompted several participants to recommend that future trainees begin the rotation with an open mind and a conscious awareness of their own beliefs. As one participant advised, "keep as broad and as open a mind as possible..." (Participant 3), and another resident stated, "...you really have to keep an open mind and see how to best work with these kids, learn from these kids, and how you can alter your own approach and your own practice in working with these adolescents to the best of your ability for the best possible outcome" (Participant 7).

Discussion This study contributes to our understanding of the experiences of pediatric residents during their postgraduate training in Adolescent Medicine in several ways. It is not surprising that residents gained knowledge through this training, as it is expected that people will learn as the result of any given experience, and other researchers have similarly reported the improvement in knowledge and clinical skills in adolescent health after participating in such a rotation [ 8 ] . The experience during this rotation, however, went beyond a simple gain in knowledge; residents gained insight and a comprehensive understanding of adolescents' lives and issues. They acted on the wealth of information that they had acquired and the skills that they had learned by engaging with and advocating for their adolescent patients. They communicated and collaborated with other members of the interprofessional health care team, and finally, they reflected upon their whole experience. In other words, this rotation not only assisted residents in becoming the "medical expert" but also assisted residents in developing some of the "non-expert" physician competencies that postgraduate medical education programs aim to achieve. CanMEDS physician competencies [ 14 ] exemplify such a framework, and like other similar frameworks, include competencies such as communicator, collaborator, health advocate, and professional. These competencies have been developed with the ultimate goal of improving patient care. They were not specifically assessed or evaluated in this study, but interestingly, the residents' experiences during their Adolescent Medicine rotation reflected these areas that are essential components of postgraduate and other medical education training programs. The biopsychosocial and comprehensive approach to adolescent patients was repeatedly mentioned by the residents. They compared this experience to other areas of training in pediatrics in which such an approach was not consistently modeled. Trainees were generally familiar with a problem-focused approach, where a patient's chief complaint was addressed and other aspects of a patient's life not routinely explored. With the complexity of adolescents' issues, the residents recognized that focusing solely on the chief complaint would be inadequate and/or misleading, as one might not get the opportunity to really 'get to know' his/her patient in so doing. Patient care, in general, is known to be complex and requires that multidisciplinary professionals work together in an effective manner to deliver quality care [ 15 ]. Interprofessional education (IPE) has been suggested as a way of improving interprofessional collaboration and patient care, yet the results of health outcomes are mixed [ 15 ]. The training in Adolescent Medicine was not of an IPE nature, meaning that students were not of different professional backgrounds, yet the learning occurred in an interprofessional team environment. Residents' roles within the interprofessional health care team were established. The value of communicating and collaborating with the team in managing challenging situations, seeking resources, or discussing particular clinical experiences was reported. The interprofessional nature of the team allowed for a variety of perspectives on issues and complemented the biopsychosocial approach to patient care. Trainees learned from the other health professionals and their respective roles in the care of their patients. The team was also considered a source of support for trainees when they encountered a clinically or ethically challenging situation. The interprofessional team environment could serve as a model for other postgraduate medical training programs and may be used to assess the role of the interprofessional team environment in the learning of "non-expert" physician competencies. Engaging with their adolescent patients promoted feelings of empathy, and the trainees' roles as patient advocates became increasingly evident to them. Participants who previously acknowledged being indifferent toward adolescents and their behaviors expressed a shift in these attitudes and described an enhanced awareness and understanding of adolescent behavior. The mixed feelings that emerged reflect the countertransference that is known to be a part of the doctor-patient relationship [ 16 , 17 ] . Not only was there a shift in attitude toward adolescents, but there was also a shift in attitudes toward Adolescent Medicine, with a few participants reporting that they were positively influenced by their rotation and were now considering Adolescent Medicine as a future career path. One might wonder if trainees had not previously considered Adolescent Medicine as a future specialty because of a lack of exposure to this area or because of negative preconceived ideas about working with adolescents. The exposure of participants to potentially controversial issues during their rotation, as well as the nature of the interviews that were conducted for this study, might have resulted in the heightened self-awareness that all participants described. Most of the participants were in their mid to late 20s; the age difference between them and the adolescent patient population was not considerable, which could have contributed to participants reflecting on their own adolescent experiences. The fact that some participants recommended that future trainees commence their rotation in Adolescent Medicine with an open mind and a self-awareness of their own values and biases is consistent with published reports about clarification of values and how this may assist with decision making [ 11 , 18 ]. Cultural stereotypes about teenagers, as well as personal and ideological beliefs and values, are known to shape a clinician's approach to adolescent patients [ 11 , 19 ]. Furthermore, the personal backgrounds, values, and own adolescent experiences of residents have been associated with their attitudes about adolescents and their approach to adolescent health care [ 10 , 11 ]. Being conscious of one's opinion can assist a trainee in his/her approach to patient management as well as facilitate discussion with members of the health care team. Awareness of physicians' feelings and responses toward their patients is known to impact the quality and character of the doctor-patient relationship [ 10 , 18 ]. In essence, pediatric residents' experiences during their clinical rotation in Adolescent Medicine may have reflected a learning process . Residents gained knowledge and insight through clinical exposure, took on a professional role as a member of a health care team, and reflected through self-awareness. One may wonder if these elements were related and if so, if they occurred in series, in a unidirectional or bidirectional manner. Figure 1 illustrates a potential model of such a learning process. Though unclear if and how each of these elements led to the next, residents, through their experiences, built their "non-expert" physician competencies in addition to their "medical expert" physician competencies in an interprofessional team environment. This particular gain in the "non-expert" physician competencies during the rotation may support the need for pediatric residents and/or other specialty residents to take their Adolescent Medicine rotation during PGY1, as this exposure and learning may enrich subsequent clinical experiences with adolescent patients seen during other rotations where the biopsychosocial approach to adolescent health care may not be modeled. This study's findings reflect the experience at one particular institute with diverse patient and trainee populations. Therefore, these findings may not be generalizable to all Adolescent Medicine training programs. Other centers, nationally or internationally, may have a different trainee and/or adolescent patient population that could influence trainee experiences. The methodology utilized, however, could be used for further studies examining and comparing resident and other trainee experiences in clinical settings.

Conclusions Pediatric residents' experiences of their clinical rotations in Adolescent Medicine reflected their learning, notably gains in the "non-expert' as well as "medical expert" physician competencies. Future studies should explore the proposed potential model of the learning process, whether this learning is similar in other Adolescent Medicine training programs and in other specialty/subspecialty postgraduate medical training programs, and whether the interprofessional nature of an Adolescent Medicine team and the patient populations themselves contribute to this learning. Further studies could also explore the effect of values clarification prior to engaging in an Adolescent Medicine rotation, as a way of 'priming' the learning experience and might inform the development of orientation materials for future trainees.

Background Although Adolescent Medicine is a pediatric subspecialty, it addresses many issues that differ from other aspects of pediatrics clinical training. The aim of this study was to explore the general experiences of pediatric residents during their rotations in Adolescent Medicine. Methods Qualitative methods were applied. Semi-structured individual interviews were conducted with pediatric residents who had completed a rotation in Adolescent Medicine. Emergent themes were identified. Results Three key themes emerged: gaining exposure, taking on a professional role, and achieving self-awareness. Subcategories were also identified. There was particular emphasis on the multidisciplinary team and the biopsychosocial approach to adolescent health care. Conclusions The experiences in Adolescent Medicine reflected residents' learning, notably gains in the "non-expert" as well as "medical expert" physician competencies. Future studies should explore how the interprofessional nature of an Adolescent Medicine team and the patient populations themselves contribute to this learning.

Competing interests The authors declare that they have no competing interests. Authors' contributions FA conceived of the study, participated in its design and analysis, drafted the manuscript, and obtained the funding. KL and EG participated in the design and analysis of the study, as well as in the critical review of the manuscript. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1472-6920/10/88/prepub

Acknowledgements We thank our research assistant, Ms. Gillian Watson, and all of the participants that made this study possible. We would also like to thank Dr. Juliet Corbin for her review and constructive feedback of the manuscript. This study was funded by the Trainee's Start-Up Fund, Research Training Center at The Hospital for Sick Children and University of Toronto.

CC BY

no

2022-01-12 15:21:44

BMC Med Educ. 2010 Dec 1; 10:88

oa_package/29/04/PMC3016290.tar.gz

PMC3016291

21126358

Background Breast cancer is the most common female malignancy in Germany and worldwide, and its prevalence is predicted to increase [ 1 ]. Growing numbers of clinical trials are being conducted to collect the information needed to provide patients with tailored treatments, and the guidelines for certified breast centres in Germany require that 20% of all treated patients be enrolled in clinical trials [ 2 ]. However, trial recruitment is difficult, expensive and time-consuming [ 3 ]. In the everyday clinical setting, physicians recruit fewer than half of the potentially eligible patients into clinical trials [ 4 ]. Major factors preventing patients from participation have been shown to include lack of interest [ 3 , 5 ], low acceptance of the study treatment, concerns about work-related problems, travel and long treatment periods [ 6 , 7 ]. Physicians fail to recruit patients due to reservations about clinical trials and negative referral policies at certain centres, and their patients' comorbidities, mental state and disabilities [ 8 , 9 ]. In addition, the inclusion and exclusion criteria for trials can vary to such an extent that physicians may not have sufficient knowledge to identify eligible patients [ 10 ]. On the other hand, 40-50% of cancer patients already resort to the Internet for healthcare information, and this number is likely to increase dramatically over the next few years [ 4 ]. As a result of "patient empowerment", patients have come to expect access to reliable, high quality, patient-centred information on their medical condition and all available treatment options. To meet these demands, the German Society of Senology (DGS) and the Baden-Württemberg Institute for Women's Health (IFG) in 2000 jointly launched the internet portal http://www.brustkrebs-studien.de (BKS) as a not-for-profit service to breast cancer patients and physicians alike [ 11 , 12 ], offering patients extensive information on the disease, its treatment and study participation, and providing investigators with a registry database for their clinical trials. We here describe how BKS was developed and present the results of the implementation phase as a proof of principle.

Methods Development of the BKS website Design of the BKS platform Existing breast cancer trial registries such as ClincialTrials.gov (maintained by the National Cancer Institute (NCI)), BreastCancerTrials.org, OncoLink.org and a German clinical trials registry maintained by the German Cancer Society, Studien.de, were reviewed and analysed [ 13 - 16 ]. All features were compared, evaluated and prioritized. These were reviewed by an advisory board of breast cancer experts from the German Society of Senology (DGS) and patient support and advocacy organizations to assist with the development process. The advisory board regularly reviewed drafts of BKS to ensure early detection of usability issues as work progressed. Software implementation HTML/PHP (Hypertext Markup Language/Hypertext Preprocessor) and MySQL (Structured Query Language) were chosen because they are well-established standards and supported by virtually all web browsers. They also offer the benefits of Open Source Software for maximum scalability and extensibility [ 17 , 18 ]. Security measures implemented included the encryption of data transmitted to and from users via HTTPS/SSL (HyperText Transport Protocol Secure/Secure Sockets Layer) and digital certificates [ 19 ]. User friendliness was given the highest priority during the design and development phase, in accordance with the findings of Nielsen [ 20 ]. Trial registry A registry of all relevant therapeutic breast cancer trials, including protocols, inclusion and exclusion criteria and recruitment status was designed and implemented as one of the three core BKS resources. The advisory board determined the studies to be included in the registry. The principal investigators were assigned the responsibility for maintaining and updating the information on their trials. All trial entries underwent regular quality assurance reviews by the DGS support team. Breast cancer guidebook To strengthen the portal as a knowledge hub on breast cancer, the DGS published its patient-centred guidebook on BKS. Expert advice A service was implemented via which patients could contact an expert panel for second opinions or advice on treatment and participation in clinical trials. Evaluation of the BKS website Access statistics User and hit statistics were obtained by log file analysis using the Webalizer 2.1 [ 21 ]. User interests, portal acceptance and satisfaction A two-week online survey was conducted during the 2nd and 3rd week of March 2008. To this end, a questionnaire designed to assess user acceptance and satisfaction was announced and made accessible on the BKS homepage. The questionnaire comprised 29 questions with Likert-type scale answers. The questions addressed topics such as the reasons for visiting the website, information quality, quantity and structure of content, awareness of, and interest in, clinical trial participation, and usability aspects of the portal, e.g. page loading time. Statistical analysis Descriptive statistics were performed using SPSS software (version 9.0 for Windows; SPSS Inc, Chicago IL, USA) to analyse the information submitted via the online questionnaire. Ethics This study was noninterventional, and no patient-identifiable data were used in the analyses. Therefore the study did not require ethics committee approval or informed patient consent according to the relevant German laws and regulations.

Results The BKS website In 2000, the German Society of Senology (DGS) launched the BKS portal http://www.brustkrebs-studien.de as an information resource and a registry-based matching service for breast cancer patients to facilitate their enrolment in clinical trials. Figure 1 is a schematic representation of the website's core features that can be accessed directly from the home page. Figure 2 shows exemplary screenshots of the page on the BKS portal where patients can match themselves to relevant clinical trials, pages with search results and study information as well as part of the input mask principal investigators can access to register clinical trials. The three main components of the BKS portal and their features and functionalities are described in the following. Trial registry A consistent layout and step-by-step wizards were implemented to help users familiarize themselves rapidly with the system. Investigators receive immediate feedback on information validity when entering protocol information, an approach also used in the NCI's ClinicalTrials.gov database [ 22 ]. Investigators can securely publish classified trial documents for physicians via DocCheck ® authentication [ 23 ]. An automated notification system reminds investigators to update trial information and alerts registered patients to new studies as soon as they are entered into the registry database. Potential study centres can easily contact principal investigators via the system to enquire about inclusion of their site in the database. Patients can search the trial registry for suitable clinical trials via a simplified wizard-based matching service. Three basic screening questions establish the patient's disease and treatment profile (Figure 2 , panel A). The search for suitable studies is then based upon individual parameters and refined by specific criteria, such as treatment modalities, anti-cancer drugs, trial status and randomization type. Following the basic search, patients can securely forward their details anonymously to a particular principal investigator for further eligibility testing. Breast cancer guidebook Written by prominent members of the DGS, this core resource of the BKS portal provides detailed information on the pathogenesis, types, prevalence, early detection and diagnosis of breast cancer, treatment options, follow-up care and support resources, while also including interactive modules, such as an online audio book ("From diagnosis to follow-up care") and 3D video animations. The guidebook is based on the high quality criteria for health related websites [ 24 ]. Another section, added to facilitate the provision of basic informed consent, offers information on clinical trial participation, different trial designs and the procedures involved such as randomization and blinding, the quality of trials and participants' rights and obligations. Expert advice An expert advisory board of leading breast cancer specialists can be contacted via the platform for second opinions and personal treatment recommendations with regard to clinical trials. Evaluation of the BKS website Access statistics Between 1 October 2005 and 30 June 2010, 702,655 visitors logged onto the portal, generating a total of 15,507,454 page views (quarterly average: 46,183 visitors [range 23,583-55,738]). Figure 3 shows access details from the log file analysis. Mean website visit duration was 3.2 minutes. Patient enquiries for trial participation and second opinions In 2009, 143 patient enquiries about trial participation were securely forwarded to the principal investigators for eligibility screening. In the same period, 119 patients sought a second opinion or individual treatment advice online via the expert panel. User evaluation of BKS The portal attracted 5,702 visitors during the two-week online survey period. In total, 568 questionnaires were submitted, of which 507 (89%) were evaluable. Visitors were predominantly patients with breast cancer (67%) or friends and relatives (25%). Most visitors stated that they used the Internet as a source for information about breast cancer (81%), and some regularly forwarded information on the disease to others or exchanged such information (33%). Most participants found the information on BKS well organized (79%), useful (79%), adequate (63%) and easy to navigate through (87%). The most frequently accessed content in the guidebook was information on the pathogenesis and diagnosis of breast cancer (42%) and treatment options (34%). Most users were confident that the information offered was correct (80%). Pages loaded quickly enough (83%). More than three-quarters (83%) of respondents called for a discussion forum as an additional feature to the portal. About two-thirds (68%) stated they would be happy to store their medical details online in an anonymous breast cancer record and about three-quarters (74%) were interested in communicating individually with an expert (e.g. for a second opinion and treatment options). Most visitors indicated that they intended to use the portal again in the future (76%) and would recommend it to others (81%). Patient attitudes towards trial participation Most patients considered it valuable to be able to search for clinical trials themselves (79%), and a majority had already considered trial participation (62%) and previously sought information on participation in clinical trials (54%). The information sources on trial participation that respondents considered useful were the Internet (29%), the patients' physicians (27%) and other patients or support groups (15%). Trial registry By 30 June 2010, a community of 189 investigators and physicians was contributing to the portal, including all major breast cancer trials groups in Germany, who regularly updated the information on their trials. At that time, the registry covered 163 trials, of which 89 (55%) investigated recurrent disease and 51 (31%) evaluated adjuvant therapy, 15 (9%) neoadjuvant therapy, 4 radiotherapy and 4 surgical treatment (2% each). No trials of preventive interventions had been registered. The 89 (55%) of studies in patients with recurrent disease comprised 27 (30%) with chemotherapy, 8 (9%) with biologicals, 6 (7%) with endocrine treatment, 2 (2%) with radiotherapy and 6 (7%) with other treatment approaches. Of the 51 (31%) adjuvant trials 6 (12%) used chemotherapy, 4 (8%) biologicals, 2 (4%) bisphosphonates, 3 (6%) endocrine treatment, and 1 (2%) other treatments. Table 1 lists the most frequently visited protocols during the 5-month period from 1 August to 12 December 2008. Study types in the BKS registry compared with ClinicalTrials.gov (NCI) An analysis conducted in March 2008 showed that 52% of the BKS trials were in patients with recurrent disease, as opposed to only 2.3% of the NCI trials. The majority (70%) of the NCI trials investigated neoadjuvant therapies (26%), radiotherapy (20%) or surgical treatment (24%).

Discussion Over the past few decades, the concept of tailored treatment for breast cancer has led to an ever-increasing need for clinical studies, but patient enrolment has generally been too slow to meet the demands of valid guidelines. As a result of patient empowerment, the demands for reliable, patient-centred information have also grown. The integration of the patient as an active participant in the decision-making process and a partner requiring comprehensive, up-to-date, correct, comprehensible information reflects the changing patient-physician relationship. The Internet has become a key source for patients who wish to be well informed [ 25 ], and it has been identified as a powerful vector in increasing trial recruitment rates [ 4 ]. It has also been shown that the knowledge- and data-intensive processes of determining patient eligibility can be facilitated by using computerized systems [ 26 ]. Here, we describe how the German Society of Senology (DGS) and the Baden-Württemberg Institute for Women's Health (IFG), in collaboration with patient support and advocacy groups, developed and established a trusted online information and communication platform featuring up-to-date trial protocols and resources for the growing physician and patient community. Similar to the NCI's ClinicalTrials.gov, the BKS portal is an official resource backed by a national specialist medical society and aims to serve patients, doctors and researchers alike. It is unrealistic to assume that physicians can screen every patient for trial eligibility in view of the vast number of clinical trials available today. A large number of potential participants are therefore lost. It has been suggested that only 2-4% of patients with newly diagnosed cancer participate in clinical trials [ 27 ]. Although online trial registries seem to be a preferred solution to this dilemma, trend analysis has indicated that it is unlikely that a single trial database would function worldwide, enabling multiple domain-, funder- or country-specific registers to be created [ 28 ]. Over the past few years, we have established the BKS website as a highly frequented breast cancer portal with a total of over 700,000 visitors by the middle of 2010. On average, 46,183 visitors accessed the BKS portal per quarter during the period from 1 October 2005 to 30 June 2010. The true value of the BKS became apparent in 2009, during which year 143 patients contacted investigators via the website to enquire about trial enrolment and 119 requests for second opinions were submitted via the secure e-messaging service. The benefits of e-messaging have been widely discussed in the literature and include decreased numbers of telephone calls [ 29 ], unnecessary clinic visits [ 29 - 31 ], improved patient care [ 32 ], greater efficiency in the exchange of information between physicians and patients [ 29 ] and cost savings [ 33 ]. The respondents to our online survey showed a high degree of satisfaction with both the design of the portal and the information provided, and also expressed great trust in the information. The great majority of respondents stated that they would recommend the portal to others and continue using it in the future. Patients have been shown to respond positively to portals backed by a national specialist medical society because they are perceived as more trustworthy [ 34 ], and this is a likely reason for the success of our portal, which is officially supported by the German Society of Senology (DGS). By the middle of 2010, the BKS covered a total of 163 current breast cancer trials, which the advisory board had selected as the most important trials. The active scientific BKS community at the time consisted of 189 physicians and investigators. The majority of trials (55%) focused on the use of chemotherapy in recurrent disease. However, the most visited trials were those investigating adjuvant chemotherapy regimens. This is especially relevant since the majority of trials registered with the NCI during our study period were neoadjuvant studies, possibly reflecting the future focus of research in breast cancer trials. The success of trial registries depends on a number of factors. One of the main issues in the development of the BKS portal was the need for up-to-date information on trial status, which meant that principal investigators had to take responsibility for updating the information on their trials and removing closed trials from the database. To achieve this, the system was equipped with an automated notification and feedback service to encourage commitment on the part of both the support team and the investigators. Studies suggest that female patients are generally in favour of clinical trials [ 35 , 36 ], and this is supported by our finding that almost two-thirds of the BKS visitors considered participating in a trial. However, complex trial descriptions and eligibility criteria, the extensive use of medical terminology and expecting patients to determine preferences early on in the registration process act as strong barriers to trial participation [ 8 ]. Patients with previous clinical trial experience have fewer concerns than those facing this option for the first time, and recommendations by clinicians play a significant role in their decision [ 10 ]. For these reasons, and based on the suggestions of Gillen et al. [ 22 ], the emphasis in the development of the BKS portal was placed on patient-friendly versions of trial protocols embedded into relevant patient information and supported by leading breast cancer experts. Our user analysis showed that most patients appreciated being able to search for suitable trials themselves, and that patients relied on the Internet, and not their doctor, for information on trials. To concentrate on patient-friendliness and provide high-quality content was therefore the right approach. To ensure that patients' needs were addressed appropriately, all development work was, and continues to be, conducted in close collaboration with patient support and advocacy organizations, notably "Frauenselbsthilfe nach Krebs" (Women's Self-Help after Cancer) and "Mammazone". A further trend observed in our analysis is that patients are interested in online personal health records, which is consistent with earlier reports [ 10 ]. Although this trend has been known for some years, no approach to patient disease management has been successfully adopted at a national level so far. The BreastCancerTrials.org service [ 13 ], however, offers a promising approach. It functions as an online patient breast cancer record combined with a matching service to suitable trials. OnkoLink is a similar portal but is not breast cancer specific [ 37 ]. A matching service for trial eligibility based on the US National Cancer Institute's Physician Data Query (PDQ) database was introduced by Ohno-Machado et al. [ 38 ]. Not only has it been shown that computer-based, automated screening is superior to physician-based individual screening [ 39 ], but also that an online system gives patients more information and more time to decide, and also increases their confidence in their decision-making [ 40 ]. Before the BKS portal was initiated, no such approach had been pursued in Germany for breast cancer - a shortcoming that has now been addressed. One of the limitations of online services for study recruitment in this population is that female breast cancer most frequently occurs after the sixth decade of life, at which age women may be less likely to use online services. Moreover, minorities tend to be underserved with regard to Internet access, especially in rural areas [ 41 ]. Clinicians should therefore bear in mind that clinical trial recruitment via the Internet may introduce bias [ 42 ]. Thus, potential selection bias represented another limitation to our user analysis. Further limitations included the relatively small sample size, the short period during which the survey was conducted, and that it is unclear whether the participants were representative of the general population, e.g. with regard to age distribution. Despite these limitations, we are confident that the largely very positive answers from the respondents confirmed our approach to launch the BKS portal and that they provided useful information for further enhancements to the service. Web-based clinical trial portals have great potential as a tool for physicians to manage an increasing number of clinical studies and for patients to access accurate and up-to-date information. Developed by a team physicians and investigators, patient support and advocacy groups and the German Society of Senology, the BKS website already effectively meets the needs of all types of users with regard to data security, privacy and ease of use. A conceivable future refinement of our system will include the creation of pools of eligible patients who have expressed interest in clinical trial participation, undergone preliminary screening and given basic consent. Work is in progress to expand the service by introducing a tool for cross-matching patient demographic and clinical data with trial inclusion and exclusion criteria. Integration of such services into hospital information systems would greatly facilitate preliminary screening and mean that a web-based tool of this kind could be integrated into everyday clinical practice. Most importantly, outcome research is now needed to determine whether such tools actually result in improved recruitment to clinical studies.

Conclusions We conclude that http://www.brustkrebs-studien.de is an established and trusted interactive platform providing information on breast cancer for patients and physicians alike. With the aim of encouraging participation in clinical trials, it offers the growing community of patients and physicians seeking information on the internet a range of up-to-date resources including expert-written content on the disease, current treatment options and clinical trial protocols. Further studies are needed and being undertaken to assess potential increases in trial enrolment by eligibility matching services.

Background The internet portal http://www.brustkrebs-studien.de (BKS) was launched in 2000 by the German Society of Senology (DGS) and the Baden-Württemberg Institute for Women's Health (IFG) to provide expert-written information on breast cancer online and to encourage and facilitate the participation of breast cancer patients in clinical trials. We describe the development of BKS and its applications, and report on website statistics and user acceptance. Methods Existing registries, including ClinicalTrials.gov, were analysed before we designed BKS, which combines a trial registry, a knowledge portal, and an online second opinion service. An advisory board guided the process. Log files and patient enquiries for trial participation and second opinions were analysed. A two-week user satisfaction survey was conducted online. Results During 10/2005-06/2010, the portal attracted 702,655 visitors, generating 15,507,454 page views. By 06/2010, the website's active scientific community consisted of 189 investigators and physicians, and the registry covered 163 clinical trial protocols. In 2009, 143 patients requested trial enrolment and 119 sought second opinions or individual treatment advice from the expert panel. During the two-week survey in 2008, 5,702 BKS visitors submitted 507 evaluable questionnaires. Portal acceptance was high. Respondents trusted information correctness (80%), welcomed self-matching to clinical trials (79%) and planned to use the portal in the future (76%) and recommend it to others (81%). Conclusions BKS is an established and trusted breast cancer information platform offering up-to-date resources and protocols to the growing physician and patient community to encourage participation in clinical trials. Further studies are needed to assess potential increases in trial enrolment by eligibility matching services.

Abbreviations BKS : http://www.brustkrebs-studien.de ; DGS : German Society of Senology (Deutsche Gesellschaft für Senologie); HTML : Hypertext Markup Language; HTTPS/SSL : HyperText Transport Protocol Secure/Secure Sockets Layer; IFG : Baden-Württemberg Institute for Women's Health (Institut für Frauengesundheit Baden-Württemberg); NCI : National Cancer Institute; PHP : Hypertext Preprocessor Competing interests The authors declare that they have no competing interests. Authors' contributions MW participated in the development of the website described, conceived of the present study and its design, participated in data analysis and drafted and finalised the manuscript. CWW participated in the development of the website described, contributed to data collection and analysis and reviewed the draft manuscript. SYB, AH, TNF and JKK participated in the development of the website described, contributed to data collection and analysis and reviewed the draft manuscript. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/663/prepub

Acknowledgements This study was supported by a grant from the "Landesakademie Baden-Württemberg" via the Baden-Württemberg Institute for Women's Health. We thank the patient support and advocacy groups "Frauenselbsthilfe nach Krebs" and "Mammazone" for their ongoing support.

CC BY

no

2022-01-12 15:21:44

BMC Cancer. 2010 Dec 2; 10:663

oa_package/d8/ca/PMC3016291.tar.gz

PMC3016292

21129184

Background Renal cell carcinoma (RCC) accounts for 3% of all adult cancers [ 1 ]. Approximately 30% of patients are diagnosed with metastases and an additional 20-40% of patients develop metastases after radical nephrectomy with curative intent [ 2 , 3 ]. The outcome of patients with metastatic RCC is poor, with a median survival time of 10 to 21 months [ 4 , 5 ] Classical cytokine therapies have been the only systematic treatments available for advanced RCC for a long time [ 6 - 9 ]. The oncogenic mechanism of RCC has been elucidated and agents that target relevant biological pathways have been investigated. Multiple tyrosine kinase inhibitors (multiple TKIs) targeting vascular endothelial growth factor receptor (VEGFR) such as sunitinib and sorafenib have revolutionized the treatment of RCC [ 10 , 11 ]. Although mammalian target of rapamycin (mTOR) inhibitor was not available in Japan at the time of this study, the efficacies of mTOR inhibitors have been reported [ 12 , 13 ]. These developments have made it necessary to predict the prognosis of individual patients with advanced RCC and to select optimal management. Many clinical risk factors have been proposed, and classifications of patients using these risk factors have been established. The most common classification was proposed by the Memorial Sloan-Kettering Cancer Center group for cytokine-based therapies (MSKCC classification)[ 14 ], and modified criteria adapted for the new era of molecular targeting was reported recently and recommended in the National Comprehensive Cancer Network guideline (NCCN classification)[ 12 , 15 ]. However, these classifications are not enough to determine the best treatment selection for an individual patient. Novel biomarkers to predict the prognosis of individual patients are therefore desired. During the last decade, 18-fluoro-2-deoxy-D-glucose positron emission tomography ( 18 F-FDG-PET) emerged as a useful non-invasive tool to evaluate the metabolic status of tumors. Numerous recent studies of various types of malignancies have reported an association between the 18 F-FDG accumulation rate evaluated by PET and patient prognosis. The standardized uptake value (SUV) is a semiquantitative simplified measurement of the tissue FDG accumulation rate, and studies of the head-and-neck, lung, and cervical cancer have explored the prognostic significance of the maximum standardized uptake value (SUVmax) [ 16 - 19 ]. However, the role of the SUVmax as a prognostic factor for patients with advanced RCC has not yet been evaluated. In the present study, we evaluated prospectively the impact of SUVmax on the survival of patients with advanced RCC.

Methods Patients This was a prospective study to clinically follow enrolled patients planning to undergo systematic therapies for advanced RCC. In principle, the pathologies of enrolled cases were confirmed by prior nephrectomy or biopsy, but only one case was diagnosed clinically by conventional imaging because the patient wished to be treated immediately and did not consent to biopsy. The patients were initially assessed by conventional imaging techniques (computed tomography [CT], magnet resonance imaging [MRI], or bone scintigraphy) and diagnosed as stage IV or metastatic RCC. Patients with uncontrolled diabetes mellitus, with other known malignancies and treated with therapeutics the last 2 weeks before the scan were excluded. The study protocol was approved by the Yokohama City University Institutional Review Board. Written informed consent was obtained from all patients. The patients underwent various therapeutic interventions decided before the evaluation by PET/CT at Yokohama City University Hospital and Kanagawa Cancer Center. Imaging Patients fasted for at least 6 hours prior to intravenous injection of [ 18 F] FDG. PET/CT images were obtained using a PET/CT system (Aquiduo 16; Toshiba Medical Systems, Tokyo, Japan). PET/CT images were acquired from the top of the head to the mid thigh at 60 min after intravenous injection of 2.5 MBq/kg of [ 18 F] FDG. A low-dose non-contrasted CT scan was acquired first and used for attenuation correction. Emission images were acquired in 3-dimensional mode for 2 min per bed position. After PET acquisition, CECT was performed with a 2-mm slice thickness, 120 kV, 400 mA, 0.5 s/tube rotation, from the top of the head to the mid thigh, with breath holding. A total of 100 ml contrast medium (iopamidol) was administered intravenously at a rate of 1.0 ml/s. The scan delay was set at 120 s after starting the injection of contrast material. Images were reconstructed by attenuation-weighted ordered-subset expectation maximization (OSEM) (four iterations, fourteen subsets, 128' 128 matrix, with 5-mm Gaussian smoothing). The highest SUV in all RCC tumors of each patient was defined as SUVmax. To obtain the SUVmax, the SUV values of all lesions in tumors diagnosed as RCC by CT imaging were analyzed. Statistical analysis Survival time was calculated from the date of evaluation by 18 F-FDG PET/CT to the date of death. Cox proportional hazards model was used to assess the effects of SUVmax on survival. The cancer-specific survival curve was estimated by the Kaplan-Meier methods, and the resulting curves were compared using the log-rank test. All statistical analyses were carried out with SPSS software (SPSS, Inc, Chicago, IL).

Results Patient characteristics and intervention A total of 26 patients (21 males and 5 females) were enrolled in this study between 2008 Jun and 2009 October (Table 1 ). The median age was 61 years (range of 32-82). There were 17 patients with recurrent diseases and 9 with stage IV disease. Pathological examination showed 18 cases of clear cell carcinoma, 5 of papillary, and 2 of clear/sarcomatoid; in one case, the pathological type was unknown. As for prior surgeries, 19 patients had undergone nephrectomy and 4 metastatectomy. Thirteen patients had not undergone previous systematic therapies. Of the other 13 cases with previous systematic therapies, 9 patients had undergone interferon-alpha (IFN-α therapies, one sorafenib, one S1, one combined therapy with IFN-α and sorafenib, and one combined therapies with IFN-α and UFT. After the evaluation by PET/CT, 20 patients were treated with multiple TKIs (9 sorafenib, 9 sunitinib, 2 sequential therapy with sorafenib and sunitinib, and 1 sequential therapy with sorafenib and IFN-α), and 6 patients underwent cytokine therapies. At the follow-up end (January 2010), there were 9 cases with cancer death, and we confirmed the other 17 patients alive. There were no cases with death due to other causes and no cases dropped out during follow-up. The median follow-up period was 262 days (range, 43 to 531 days). Accumulation of FDG in the lesions diagnosed as RCC tumor by CT imaging We first, examined the FDG accumulation in all 368 tumor lesions in 26 patients who were diagnosed as stage IV or metastatic RCC by CT imaging. FDG uptake was detected in 230 of 243 lesions (94.7%) excluding lung or liver metastasis with diameters less than 1 cm. On the other hand, among 125 lung or liver lesions with diameters between 5 mm and 9 mm, FDG accumulations were detected in only 21 lesions (16.8%). The SUV in RCC lesions demonstrated various values from undetectable levels to 16.6. In 6 of 7 patients without prior nephrectomy, the primary tumor demonstrated the highest SUV in all RCC tumor lesions (Figure 1 ), and lung metastasis showed the highest SUV in another. In 19 cases with metastases or recurrence after nephrectomy, bone metastasis demonstrated the highest SUV in 5 cases, lung metastasis in 4 cases, lymph node metastases in 3 cases, and local recurrence in 2 cases (Figure 2 ). The uterus, pancreas, Inferior Vena Cava thrombus, muscle metastasis, and contra-lateral kidney metastasis demonstrated the highest SUV in one case each. The impact of SUVmax on patient survival time We next analyzed the association between SUVmax and patient survival time. The SUVmax of all patients ranged between 1.4 and 16.6 (mean 8.8 ± 4.0). The patients with RCC tumors showing high SUVmax tended to demonstrate poor prognosis, as shown in Figure 1 , 2 , 3 (26 patients were lined up in order of SUVmax in Figure 3 ). When the patient population was subdivided using the mean SUVmax (8.8), only 2 (13%) of 15 patients with RCC tumors having an SUVmax less than 8.8 were dead due to cancer and the median survival time of the 15 patients was not calculated because the number of dead patients was less than half, whereas 7 (64%) of 11 patients RCC tumors having SUVmax equal to 8.8 or more and the median survival time of the 11 patients was 156 day (95% CI 33-279). The survival for these patient subgroups were significantly different (Figure 4 ) ( P = 0.0012). When SUVmax was analyzed as a continuous variable, it was correlated with survival time ( P = 0.005 hazard ratio 1.326 95% CI 1.089-1.614).

Discussion In the present study, we demonstrated that the SUVmax evaluated by 18 F-FDG-PET/CT is a useful predictive "imaging biomarker" for survival of patients with advanced RCC. PET has not been generally used for the screening of RCC due to the urinary excretion of the radiotracer, which can mask the presence of primary lesions [ 20 , 21 ]. However the large RCCs often presenting in stage IV could be evaluated without the influence of urinary excretion of the radiotracer by PET/CT providing combined morphological and functional information (Figure 1 ). In this study, 7 primary RCC lesions, with diameters ranging from 8.5 cm to 14.7 cm, were examined by 18 F-FDG-PET/CT, and abnormal FDG accumulations sufficient to evaluate SUV were detected in all lesions. Pathological diagnosis was confirmed in 6 cases. Distant metastases of RCC could also be detected without interference of excretory radiotracers. We did not confirm the pathologies of the individual metastatic lesions, but the previous report by Majhail et al . warranted the accuracy of metastasis diagnosis by 18 F-FDG-PET. They performed biopsy or surgical resection of 36 distant metastatic lesions in 24 patients that were diagnosed by 18 F-FDG-PET, and pathological findings revealed metastatic RCC in 33 lesions (89%) [ 22 ]. In this study, FDG accumulation was evaluated in 94.9% of all RCC lesions diagnosed by CT scan except for lung or liver metastases less than 1 cm. These results were consistent with a previous report [ 23 ] and indicated that the information gained by 18 F-FDG-PET/CT was sufficient to characterize advanced RCCs. In this era of molecular targeting therapy when various systematic treatments can be selected, prognostic biomarkers are required for the purpose of risk-directed therapy selection. We revealed that the SUVmax has the potency as a novel biomarker to predict the survival time of patients with advanced RCC, by multivariate analyses with standard risk factors or risk classifications. FDG accumulation is thought to be indicative of the metabolic activity of a targeted lesion and it has been found to be a useful index in a variety of cancers. It is reasonable that a tumor with high metabolism would show rapid progression and a poor prognosis. It has been reported recently that 18 F-FDG PET/CT is useful for evaluating the response to sorafenib and sunitinib treatment of RCC [ 24 , 25 ]. The results showing that these therapeutics decrease the FDG accumulation of RCC lesions encourage the hypothesis that the FDG accumulation is indicative of the biological activity of RCC. Additionally, it has been reported that intratumoral neutrophils were detected in RCCs showing poor prognosis [ 26 ]. SUV may reflect not only the biological activity of cancer cells but also the presence of migrating neutrophils. To our knowledge, this is the first report to evaluate the impact of SUVmax on survival of patients with advanced RCC. However, the number of patients and the follow-up period were limited. Enrollment for this study continues now, and the impact of SUVmax on the survival of patients with advanced RCC will be more apparent from results from an expanded number of patients and follow-up period.

Conclusions These preliminary data indicate that the SUVmax evaluated by 18 F-FDG-PET/CT has an impact on survival in patients with advanced RCC. Additional study with an expanded number of patients and period of follow-up is necessary.

Background In this era of molecular targeting therapy when various systematic treatments can be selected, prognostic biomarkers are required for the purpose of risk-directed therapy selection. Numerous reports of various malignancies have revealed that 18-Fluoro-2-deoxy-D-glucose ( 18 F-FDG) accumulation, as evaluated by positron emission tomography, can be used to predict the prognosis of patients. The purpose of this study was to evaluate the impact of the maximum standardized uptake value (SUVmax) from 18-fluoro-2-deoxy-D-glucose positron emission tomography/computed tomography ( 18 F-FDG PET/CT) on survival for patients with advanced renal cell carcinoma (RCC). Methods A total of 26 patients with advanced or metastatic RCC were enrolled in this study. The FDG uptake of all RCC lesions diagnosed by conventional CT was evaluated by 18 F-FDG PET/CT. The impact of SUVmax on patient survival was analyzed prospectively. Results FDG uptake was detected in 230 of 243 lesions (94.7%) excluding lung or liver metastases with diameters of less than 1 cm. The SUVmax of 26 patients ranged between 1.4 and 16.6 (mean 8.8 ± 4.0). The patients with RCC tumors showing high SUVmax demonstrated poor prognosis ( P = 0.005 hazard ratio 1.326, 95% CI 1.089-1.614). The survival between patients with SUVmax equal to the mean of SUVmax, 8.8 or more and patients with SUVmax less than 8.8 were statistically different ( P = 0.0012). This is the first report to evaluate the impact of SUVmax on advanced RCC patient survival. However, the number of patients and the follow-up period were still not extensive enough to settle this important question conclusively. Conclusions The survival of patients with advanced RCC can be predicted by evaluating their SUVmax using 18 F-FDG-PET/CT. 18 F-FDG-PET/CT has potency as an "imaging biomarker" to provide helpful information for the clinical decision-making.

Conflicts of interests The authors declare that they have no competing interests. Authors' contributions Noboru Nakaigawa had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript. Study concept and design: KN, MY, TI, YK, NN Acquisition of data: KN, RM, KM, NH, TM, FS, UT, KK, SN, HI, TK, TM Analysis and interpretation of data: KN, MY, NN Administrative, technical, or material support: TI, YK, NN Drafting of the manuscript: KN Critical revision of the manuscript for important intellectual content: MY Obtaining funding and supervision: NN Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/667/prepub

Acknowledgements This works was supported by Grant for Research and Development Project II (No. 17) of Yokohama City University, Japan and " Establishment of research center for clinical proteomics of post-translational modifications" as part of the Special Coordination Fund for Promoting Science and Technology "Creation and Innovation Centers for Advanced Interdisciplinary ", and Grants-in-Aid for Scientific Research (No. 19591864 and 22591775) from the Ministry of Education, Science, Sports and Culture of Japan.

CC BY

no

2022-01-12 15:21:44

BMC Cancer. 2010 Dec 3; 10:667

oa_package/1f/06/PMC3016292.tar.gz

PMC3016293

21134270

Background The use of high-dose chemotherapy (HDC) combined with autologous stem cell transplantation (ASCT) has improved the outcome of haematological malignancies such as multiple myeloma (MM) and (non-)Hodgkin's lymphoma (NHL/HL). It has become standard of care in these diseases in first line and relapse, respectively. However, this treatment has long term negative side effects. Symptoms like fatigue [ 1 - 7 ] and dyspnoea [ 1 , 4 , 5 , 8 ] are highly prevalent among ASCT survivors. In addition, survivors have a reduced global quality of life, role and physical function when compared to population norms [ 9 ]. For instance, up to 60% of the patients 3 years post-transplant had a compromised ability to engage in activities as carrying a heavy bag and taking a long walk [ 10 ]. 23-56% of the patients were not able to return to work during the course of the first year after ASCT [ 6 , 11 - 13 ]. The persistent fatigue and deficits in health related quality of life (HRQoL) might reflect a self-perpetuating condition [ 14 - 18 ]. Cancer, its treatment and the associated bed rest can lead to poor physical fitness (a.o. impaired cardiorespiratory function and reduced muscle strength). As a result, greater effort is required to fulfil the activities of daily living, and performance of these activities can induce an abnormally high level of fatigue. In order to minimize fatigue, patients will limit physical activities, which will eventually lead to a greater decline in physical fitness. An exercise intervention might break this downward sequence [ 14 - 18 ]. Previous studies have shown that exercise intervention programs can improve physical fitness, fatigue level and quality of life among haematological cancer patients [ 14 , 15 , 17 , 19 - 25 ]. However, based on systematic reviews, Liu et al. (2009) and Wiskemann & Huber (2008) conclude that more high quality research is necessary [ 26 , 27 ]. In the review of Liu et al. (2009) [ 26 ] only three of the ten included studies were randomized controlled trials. The overall quality of many studies reviewed was limited, with shortcomings related to trial design, sample size, choice of comparison groups, outcome measures and duration of follow up [ 26 ]. Wiskemann & Huber (2008) reached similar conclusions [ 27 ]. Both reviews show that there is a need for well designed, randomized controlled trials that verify the findings of the previous trials and can lead to evidence-based interventions. In addition to the limited methodological quality, the trials performed to date were heterogeneous in terms of the type of exercise interventions studied. Most of the studies focussed on isolated aerobic exercise during or after the stem cell transplantation. Resistance exercise programs and combined training strategies have been evaluated more rarely [ 27 ]. This is somewhat surprising since muscle atrophy is a common problem in cancer patients [ 28 - 30 ]. The muscle athrophy is likely to be even more pronounced in patients undergoing HDC and ASCT because of the nature of drugs being used (a.o. high dose glucocorticoids) [ 16 ], and because of the morbidity associated with the neutropenic phase after ASCT, which often leads to prolonged bed rest. As considerable evidence suggests that the ability to perform physical tasks in daily life is determined by a certain threshold level of muscular strength [ 31 ], it seems important that exercise interventions not only aim to improve aerobic capacity but also aim to minimize muscle atrophy or even stimulate muscle hypertrophy. To our knowledge, there are currently no data on the cost-effectiveness of exercise intervention programs in cancer patients. The common inability to return to work during the course of the first year after ASCT [ 6 , 11 - 13 ], the frequent use of health care resources [ 32 , 33 ] and the reported financial problems by patients [ 8 , 34 ] show the importance of determining the cost-effectiveness of exercise intervention programs. The current study will evaluate an individualized high intensity strength and interval training program developed and pilot-tested by De Backer et al. (2007, 2008) at the Maxima Medical Center (MMC) in Veldhoven, the Netherlands [ 35 , 36 ]. This program has shown promising results with respect to rehabilitation of cancer patients after chemotherapy, but needs to be further explored and tested in ASCT survivors. The aims of the current study are (1) to determine the effectiveness of a state-of-the-art individualized high intensity strength and interval training program with respect to physiological and psychological status in patients with MM, NHL or HL who have recently undergone HDC followed by ASCT; and (2) to evaluate the cost-effectiveness of this exercise program. We hypothesize that this exercise program will lead to (1) improved physical fitness; (2) lower levels of fatigue; (3) less mood disturbance; (4) higher levels of daily activities; (5) improved HRQoL; (6) a higher partial and full return to work rate; and that the program (7) will be cost-effective when compared to standard care only.

Methods The EXIST (EXercise Intervention after Stem cell Transplantation) study is one of four randomised controlled trials included in the Alpe d'HuZes Cancer Rehabilitation (A-CaRe) program. This study consists of a pilot study which is followed by a multicenter, prospective, single blind randomized controlled trial (RCT). The protocol of the pilot study will be similar to the RCT protocol described in this manuscript, with the exception of the long term follow up. The aims of this pilot study are to evaluate the feasibility of the intervention and to test the study logistics. For these aims, the patients will also be interviewed during and after completion of the program about: (1) the perceived efficacy of and satisfaction with the intervention program and (2) the need for changes to the program. If necessary fine tuning of the intervention will take place. The aim of the RCT is to compare the exercise intervention and usual care with usual care only (Figure 1 ). Eligible patients will be randomly assigned to the intervention or control group. In addition to usual care, patients in the intervention group will take part in an 18-weeks individualized supervised high-intensity exercise program. This program will start 7-14 weeks after ASCT. Patients in the control group are treated according to usual care. The study protocol was approved by the Medical Ethics Committee of the Academic Medical Center (METC AMC 10/106). Study sample Eligibility criteria are: (1) diagnosed with MM in first line or with HL/NHL in first relapse and treated with HDC and ASCT 6 to 12 weeks ago; (2) sufficiently recovered from the ASCT: Hb > 6.5 mmol/L, WBC > 3.0 × 10 9 /L, platelets > 100 × 10 9 /L; (3) aged between 18 and 65 years; (4) able to cycle on a bicycle ergometer with a load of at least 25 Watt; (5) able to walk at least 100 meters independently without crutches/cane(s) or walking frame; (6) give written informed consent. Exclusion criteria are (1) treatment with autologous-allogeneic SCT; presence of (2) severe cognitive impairment; (3) severe emotional instability; (4) extensive osteolytic lesions with risk of fracture; (5) serious cardiorespiratory and/or cardiovascular conditions; (6) severe infections; (7) progression/relapse of the disease; (8) other disabling comorbidity interfering with the intervention program or influencing outcome parameters (a.o. having a pacemaker, epileptic seizures and/or poorly regulated diabetes mellitus); (9) insufficient mastery of the Dutch language. Recruitment and randomization Our goal is to recruit 120 patients within an inclusion period of two years. Recruitment takes place at the Academic Medical Center (AMC; Amsterdam), Antoni van Leeuwenhoek Hospital (Amsterdam), Haga Hospital (Den Haag), Meander Medical Center (Amersfoort), Onze Lieve Vrouwe Gasthuis (Amsterdam), St. Antonius Hospital (Nieuwegein) and University Medical Center Utrecht. All potentially eligible patients are asked to complete a short screening questionnaire and are informed about EXIST by their treating haematologist. This screening questionnaire assesses co-morbidity, pre-illness lifestyle, current attitudes toward and beliefs about exercise in general and exercising after ASCT. These questions are adapted from measures developed by Courneya and colleagues [ 37 , 38 ] for use in evaluating exercise in cancer survivors, and are based on established health behaviour theories, in particular the Theory of Planned Behaviour [ 38 ]. Patients who are not willing to participate in EXIST are asked for the reason for non-participation. Patients who are willing to participate are asked to provide written informed consent. After the baseline measurements a randomization will take place conform a stratified block randomization method with a block size of 4 and by using a validated software package [ 39 ]. Stratification factors include transplant center and diagnosis. A member of the research team will inform the patient of the randomization outcome. Study outcomes will be assessed by blinded professionals and patients will be instructed not to reveal their group allocation. Intervention Besides the usual care, patients in the intervention group follow an 18-week exercise program similar to the program developed by De Backer et al. (2007, 2008) [ 35 , 36 ]. This program consists of high-intensity resistance and interval training. Before the start of the intervention (T = 0) a sports physician screens the patients and, adapts the program in case of physical limitations. Training takes place in local physiotherapy practices supervised by physiotherapists. Patients will train on specialized resistance training equipment and bicycle ergometers. Furthermore, the physiotherapist will motivate the patient to maintain an active lifestyle. A detailed training manual will be available. Table 1 presents the structure of the intervention program. Resistance exercises The six resistance exercises target the large muscle groups: (1) vertical row (longissimus, biceps brachii, rhomboideus); (2) leg press (quadriceps, glutei, gastrocnemius); (3) bench press (pectoralis major, triceps); (4) pull over (pectoralis, triceps brachii, deltoideus, trapezius); (5) abdominal crunch (rectus abdominis); (6) lunge (quadriceps, glutei, hamstrings). Indirect one repetition maximum (1-RM) measurements will be performed every four weeks for all six exercises. In the first 12 weeks, resistance exercise consists of two sets of 10 repetitions at 65 to 80% of the 1-RM. From week 12 onwards it comprises more repetitions (20 repetitions per set) at a lower resistance (35-40%). Interval training Before and after the resistance exercises patients cycle two times eight minutes. To determine the right resistance a steep ramp test [ 40 - 42 ] is performed every four weeks. With this test the subject is instructed to cycle at a speed between 70 and 80 revolutions per minute (RPM), starting at a work load of 25W for 30 seconds. Hereafter the load is increased by 25W every 10 seconds until exhaustion. Maximal short exercise capacity (the maximal workload; MSEC) is recorded. In the first eight weeks, blocks of 30 seconds at 65% of the MSEC will be alternated with blocks of 60 seconds at 30%. From week nine onwards, the duration of the latter block is reduced to 30 seconds. Counselling A behavioural motivation component is included to improve compliance and to stimulate physical activity outside the exercise program in addition to and after completion of the exercise program. From week 12 onwards patients are encouraged to meet the recommendations made by the American College of Sports Medicine and the American Heart Association [ 43 ]. Specific program elements include the provision of general and motivational information, both verbally and via folders, about physical activity and the desired intensity of activity based on the Borg Scale rating perceived exertion [ 44 ]. The physiotherapist uses basic counselling techniques and instruction sheets. Standard care Both the patients in the intervention group and the patients in the control group receive 'usual care'. In the Netherlands usual care varies according to doctors' and patients' preferences. Patients in the control group are allowed to participate in sports or existing rehabilitation programs. Information on physical activity and exercise of all patients will be obtained from the physical activity questionnaire, recordings of the accelerometer and cost-diaries (see outcome measures). Study outcomes All studies within A-CaRe Clinical Research use similar methodologies and a comparable set of outcome measures. Figure 2 shows the timing of the assessments. The physical tests are performed centrally at the AMC according to a detailed and standardized protocol. The questionnaires will be filled out at home. Primary outcomes measures Cardiorespiratory fitness At rest respiratory function will be assessed by measuring the forced air expiratory volume in 1 second (FEV-1), the forced vital capacity (FVC), inspiratory capacity (IC) and by estimating the maximal voluntary ventilation (MVV) [ 45 , 46 ]. A maximal exercise test will be performed to assess cardiorespiratory fitness and potential cardiovascular and cardiorespiratory limitations. This technique has been shown to be feasible in cancer survivors [ 18 ]. We will follow international guidelines concerning standardization and interpretation strategies [ 47 ]. The maximal exercise test will be performed on a cycle ergometer (Lode Corival, Groningen, the Netherlands) under supervision of a sports physician. A ramp test design will be applied: after four minutes of unloaded cycling, the load will be gradually increased until exhaustion or inability to maintain pedal frequency above 60 RPM. Throughout the test the ECG, saturation and blood pressure will be monitored and heart rate (HR) and gas exchange variables (MasterScreen CPX, CareFusion, Hoechberg, Germany) will be recorded continuously and averaged at 30 seconds intervals. We will register peak oxygen uptake (peakVO 2 ), maximal heart rate (peak HR) and mean power output (peak W). Ventilatory threshold (VT) will be determined using the ventilatory equivalents method [ 47 ]. The respiratory exchange ratio (RER) and HR will be used to review peak exercise. After the start of recovery, the patient will be asked about his/her exercise-related perceptions using the Borg Scale of perceived exertion [ 44 ]. Muscle strength Upper extremity muscle strength is assessed using a grip strength dynamometer (Hydraulic Hand Dynamometer, North Coast Medical Inc., Morgan Hill, USA), and lower extremity muscle strength with the 30 seconds chair stand test. Maximal handgrip strength is measured following the standard procedures recommended by the American Society of Hand Therapists (ASHT) [ 48 ]. The mean score attained for each side will be recorded. The 30 seconds chair stand test has been demonstrated to be a valid and reliable measure of proximal lower limb strength in older adults [ 49 ]. The subject is asked to stand upright from a chair with arms crossed at the wrist and held against the chest, then return to a fully seated position and repeat the action at his/her fastest pace over a 30 seconds period. The number of full stands within this time period is recorded. Fatigue Two self-report questionnaires will be used to assess fatigue: the Multidimensional Fatigue Inventory (MFI) and the Fatigue Quality List (FQL). The MFI [ 50 ] is a questionnaire consisting of 20 statements for which the person has to indicate on a 0-5 scale to what extent the particular statement applies to him or her. This self-report instrument consists of five subscales based on different dimensions: general fatigue, physical fatigue, reduced activity, reduced motivation and mental fatigue. The patients' perception and appraisal of experienced fatigue will be assessed with the FQL [ 51 ]. The FQL consists of 25 adjectives describing the fatigue experience, organized into four subscales: frustrating, exhausting, pleasant, and frightening. Secondary outcomes measures and moderating variables Secondary outcomes are body composition, bone mineral density, HRQoL, neuropathy, objective and self reported physical activity level, mood disturbance, functioning in daily life, return to work and cost from a social perspective. In addition, clinical data, disease status and treatment, sociodemographic characteristics, adverse events and potential predictors of compliance and satisfaction with the exercise program will be recorded. A complete overview of assessments and instruments are presented in Table 2 . A small selection of these measures is described in detail here. Objectively assessed level of physical activity The Actitrainer accelerometer (Actigraph, Fort Walton Beach, USA) will be used to measure physical activity. The Actitrainer is able to measure accelerations from 0.05 to 2.00 G. These accelerations are scored in "counts" per minute. The Actitrainer will be set at 60 seconds and the measurement period will include five days including at least one weekend day. Costs from a societal perspective Costs will be measured from a societal perspective. The following are being considered in this study: (1) Health care costs: the costs of oncological care, general practice care and physiotherapy; additional visits to other health care providers, prescriptions of medication, professional home care and hospitalization. (2) Patient and family costs: out-of-pocket expenses (e.g. travel expenses), costs for sports and sports equipment, and costs of paid and unpaid help. (3) Costs due to loss of production (absenteeism for patients with paid jobs and hours of inactivity for patients without a paid job). These data will be collected though retrospective cost questionnaires administered on a monthly basis during the period between T = 0 and T = 1. After T = 1 the cost questionnaires will be administered once every three months till T = 2. Health care utilization will be valued using Dutch cost prices [ 52 ]. Return to work The following indicators of return to work will be measured: (1) Time to partial and to full return to work (meaning number of calendar days between end of treatment and first day at work), (2) time to full return to work corrected for partial return to work, (3) partial and full return of work rate at T = 1 and T = 2. (4) Details on hours worked per week, nature of work, and return to a different job will also be recorded. Adverse Events All adverse events noticed by treating physicians and/or physiotherapists or mentioned by the participant will be recorded and monitored. The grading of adverse events will be done using the most recent version of the NCI Common Terminology Criteria for Adverse Events, CTCAE version 4. A complete document may be downloaded from [ 53 ]. Power calculations The sample size calculations were estimated for our primary study outcomes using a two-sided α = 0.05 and a power of 80%. Based on the results of De Backer et al. (2007, 2008) [ 35 , 36 ] and Hayes et al. (2004) [ 17 ], we expect a between group difference of 7.5 ± 7 ml/kg/min in the peak oxygen uptake (peakVO 2 ); 0.2 ± 0.1 kg in handgrip strength and 3.5 ± 4 points in fatigue (MFI). We need between 25 and 42 subjects per group to detect these differences between the intervention and control group. We expect a drop-out rate of 30%; 15% of the patients who have undergone autologous SCT for MM, HL or NHL have an early relapse within six months and may not be able to complete the study assessments, and 15% may drop out because of other reasons. Consequently, we need to enrol 60 subjects per group. Statistical analyses Differences in baseline characteristics between intervention and control groups will be tested using independent t-tests, Mann Whitney U tests and chi-square tests. Data will be analyzed on an intention-to-treat basis. Additionally, a per-protocol analysis will be performed. Scores on the self-report measures of fatigue, mood state and health-related quality of life will be calculated according to published scoring algorithms. Longitudinal regression analysis will be used to assess differences in each outcome measure between the intervention and control group. The two follow-up measurements will be defined as dependent variable and multi-level analysis with three levels will be used, (1) transplant center, (2) time of follow-up measurement (values corresponding with performance at T = 1 and T = 2), (3) individual. Regression coefficients indicate differences between intervention and control group. Regression models will be adjusted for baseline values, age and gender. Test results are considered significant for p -values < 0.05. All analyses will be performed using the statistics program SPSS. Cost-effectiveness analysis Both cost-effectiveness and cost-utility analyses will be performed. The cost-effectiveness ratios will be calculated by dividing the difference between the mean costs of the two treatment groups by the difference in the mean effects of the two treatment groups. This ratio will include the primary outcome measures of the trial. The cost-utility ratio will express the additional costs of the intervention per quality adjusted life year (QALY). Utilities will be measured using the EuroQol ([ 54 ] EQ5D) at baseline, at the end of the treatment and at twelve months.

Discussion The EXIST study will assess the effectiveness of a high intensity strength and interval training program on physiological and psychological variables in patients with MM, NHL or HL who have recently undergone HDC followed by ASCT. In addition the cost-effectiveness of the program will be determined. This study has several strengths. Firstly, the trial design is solid; we will conduct a randomised controlled trial among well defined patient groups, using standardized outcome measures, and including a long term follow up to assess the therapeutic sustainability of the program. Secondly, the exercise program is developed based on existing evidence and consists of both strength and endurance exercise. Therefore, the exercise program is considered to have a higher potential to restore muscle mass compared to an endurance exercise program alone, and consequently, we expect an improvement in the physical as well as the psychological health status of ASCT patients. Furthermore, counselling sessions are included to increase motivation and compliance to physical exercise both during and after completion of the intervention. Thirdly, a cost-effectiveness evaluation will give insight in the costs of the program with respect tot the outcomes.

Background The use of high-dose chemotherapy combined with autologous stem cell transplantation has improved the outcome of hematologic malignancies. Nevertheless, this treatment can cause persistent fatigue and a reduced global quality of life, role and physical function. Physical exercise interventions may be beneficial for physical fitness, fatigue and quality of life. However, the trials conducted so far to test the effects of physical exercise interventions in this group of patients were of poor to moderate methodological quality and economic evaluations are lacking. Hence there is need for a rigorous, appropriately controlled assessment of the effectiveness of exercise programs in these patients. The aims of the present study are (1) to determine the effectiveness of an individualized high intensity strength and interval training program with respect to physiological and psychological health status in patients with multiple myeloma or (non-)Hodgkin's lymphoma who have recently undergone high dose chemotherapy followed by autologous stem cell transplantation; and (2) to evaluate the cost-effectiveness of this program. Methods A multicenter, prospective, single blind randomized controlled trial will be performed. We aim to recruit 120 patients within an inclusion period of 2 years at 7 hospitals in the Netherlands. The patients will be randomly assigned to one of two groups: (1) intervention plus usual care; or (2) usual care. The intervention consists of an 18-week individualized supervised high-intensity exercise program and counselling. The primary outcomes (cardiorespiratory fitness, muscle strength and fatigue) and secondary outcomes are assessed at baseline, at completion of the intervention and at 12 months follow-up. Discussion The strengths of this study include the solid trial design with clearly defined research groups and standardized outcome measures, the inclusion of an economic evaluation and the inclusion of both resistance and endurance exercise in the intervention program. Trial registration This study is registered at the Netherlands Trial Register (NTR2341)

Competing interests The authors declare that they have no competing interests. Authors' contributions SP contributed to the study design, is responsible for data collection, analysis and interpretation and wrote the manuscript. MJK, MC, FN and LB contributed to the conception and the design of the study and revised the manuscript. HB has made contributions to the design of the intervention and contacted the participating physiotherapists. GS developed the intervention program at the MMC. JB revised this manuscript critically. All authors have read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/671/prepub

Acknowledgements The research is supported by the Alpe d'HuZes/KWF Fund. The research grant is provided by the Dutch Cancer Society. LB is further supported by a fellowship granted by the EMGO Institute for Health and Care Research. The authors would like to acknowledge Marjolein Spiering for her contribution to the study design and Marieke van Wier for her contribution to the cost-effectiveness analysis. The authors further acknowledge the A-CaRe Clinical Research group.

CC BY

no

2022-01-12 15:21:44

BMC Cancer. 2010 Dec 6; 10:671

oa_package/f0/54/PMC3016293.tar.gz

PMC3016294

21143897

Background Androgen-deprivation remains the cornerstone of treatment for patients with hormone-sensitive advanced prostate cancer [ 1 ]. The combination of hormonal suppression and the radiotherapy is able to significantly decrease the recurrences and the mortality of patients affected by locally advanced prostate cancer. Substantial toxicities were also identified when hormone therapy was used.[ 2 - 11 ]. A recently conducted meta-analysis has demonstrated an overall absolute benefit of 7.5-10% in favor of the addition of the hormone therapy to in terms of biochemical failure and clinical progression free survival [ 12 ]. The optimal duration of androgen ablation therapy is still controversial despite a growing number of prognostic factors that have been shown to predict disease outcome [ 13 ]. In all prospective randomized trials, long term hormonal therapy (more than 2 years) improves survival in patients with locally high risk (Gleason score ≥ 8, median PSA ≥ 20, stage ≥ T3) advanced prostate cancer. This strategy, however, is also associated with an increased risk of morbidities, such as osteoporosis, diabetes, depression, dislipidemia and abdominal obesity [ 14 , 15 ]. The value of short-term hormonal treatment in patients with intermediate risk disease (Gleason score ≥ 7, median PSA ≥ 10, stage ≥ T2c) has also been detected [ 16 , 17 ], although its role in high-risk disease is still controversial. Nevertheless, short courses of hormonal suppression have been shown to be very effective in downstaging localized disease. Surgical series have provided some histopathological confirmation, with significant decreases in margin-positive rates in prostatectomy after neoadjuvant hormonal therapy [ 18 , 19 ]. In order to determine if difference exists between a shorter and a longer hormone therapy (HT) in combination with radiotherapy, a literature based meta-analysis was conducted.

Methods The analysis was conducted following 4 steps: definition of the outcomes (definition of the question the analysis was designed to answer), definition of the trial selection criteria, definition of the search strategy, and a detailed description of the statistical methods used [ 20 , 21 ]. Outcome definition The combination of a short-term HT (ST) and RT was considered as the experimental arm and a long-term HT (LT) and RT as the standard comparator. Analysis was conducted in order to find significant differences in primary and secondary outcomes, according to a previously published meta-analysis and the reported sequence and definitions in the selected trials. Primary outcomes for the magnitude of the benefit analysis were both the Biochemical Failure (BF, time between randomization and PSA increase) and the CSS (time between randomization and death for prostate cancer). Secondary end-points were: OS (time between randomization and death for any cause), local failure rate (LF) and distant metastases rate (DM). Search strategy Deadline for trial publication and/or presentation was September 30 th , 2010. Updates of Randomized Clinical Trials (RCTs) were gathered through Medline (PubMed: http://www.ncbi.nlm.nih.gov/PubMed ), ASCO (American Society of Clinical Oncology, http://www.asco.org ), ESMO (European Society for Medical Oncology, http://www.esmo.org ), FECS (Federation of European Cancer Societies, http://www.fecs.be ), and ASTRO (American Society for Therapeutic Radiology and Oncology, http://www.astro.org ) website searches. Key-words used for searching were: adjuvant hormone therapy, prostate cancer, radiotherapy, duration, longer, shorter, review, metanalysis, meta-analysis, pooled analysis, randomized, phase III, comprehensive review, systematic review. In addition to computer browsing, review and original papers were also scanned in the reference section to look for missing trials. Furthermore, lectures at major meetings (ASCO, ESMO, ECCO, and ASTRO) having 'hormone treatment and radiotherapy for prostate cancer' as the topic were checked. No language restrictions were applied. Trial identification criteria All prospective phase III RCTs published in peer-reviewed journals or presented at the ASCO, ECCO, ESMO and ASTRO meetings until September 2010, in which previously untreated patients with locally advanced prostate cancer were prospectively randomized to short or long-HT plus radiotherapy were gathered. Any trial exploring a shorter versus a longer HT (regardless of their absolute values) in combination with RT for the treatment of LAPC was considered eligible. Data extraction The number of events for primary and secondary end-points were extracted; the last trial's available update was considered as the original source. All data were reviewed and separately computed by five investigators (F.Cu., E.B., D.G., I.S., and P.C.). Data synthesis The log of relative risk ratio (RR) was estimated for each considered endpoint [ 22 ], and 95% Confidence Intervals (CI) were derived [ 23 ]. A random-effect model according to the DerSimonian method was preferred to the fixed, given the known clinical heterogeneity of trials [ 24 - 26 ]; a Q-statistic heterogeneity test was used [ 27 - 29 ]. Absolute benefits for each outcome were calculated (i.e. absolute benefit=exp {RR × log[control survival]} - control survival [ 30 ]; modified by Parmar and Machin [ 31 ]). The number of patients needed to be treated for one single beneficial patient was determined (NNT: 1/[(Absolute Benefit)/100]) [ 32 ]. Results were depicted in all figures as conventional meta-analysis forest plots; a RR <1.0 indicates fewer events in the experimental arm. In order to find possible correlations between outcome effect and negative prognostic factors (selected among trials' reported factors: the median PSA, the Gleason score of 7-10 and the T3-4), a meta-regression approach was adopted (i.e. regression of the selected predictor on the Log RR of the corresponding outcome); the time delay in months for each considered trial between the duration of the short and the long treatment arm was considered in the meta-regression analysis as treatment predictor as well. Calculations were accomplished using the Comprehensive Meta-Analysis Software, version v. 2.0 (CMA, Biostat, Englewood, NJ, USA). Correlation Potential correlations to test surrogacy between primary and secondary end-points, were explored according to regression between the calculated RRs and their logs for each outcome.

Results Selected trials Twelve trials (7,811 patients) were identified (Figure 1 ) [ 2 - 11 , 33 - 37 ]. Seven RCTs were excluded because HT was administered in one arm only (4,387 patients) [ 2 - 11 , 33 ], question already answered in a previous meta-analysis [ 38 ]. Five RCTs were included in the meta-analysis, all evaluable for BF (3,424 patients) [ 7 , 34 - 37 ]. According to the trial' selection, the shorter approach ranged from 3 to 6 months, while the longer from 8 to 36 months; trials characteristics are listed in Table 1 . Four were evaluable for CSS, OS, and three for LR and DM [ 34 - 37 ]. Toxicity analysis was not performed because data about toxicity were available only in two studies. Median follow-up ranged from 3.7 to 10 years. Three predictors were identified: median PSA (range 9.5-20.35), Gleason score 7-10 (range 27-55% of patients/trial) and T3-4 (range 13-77% patients/trial). Combined Analysis With regard to the primary outcomes, the longer HT significantly decreased biochemical failure over the shorter HT with an absolute benefit of 10.1% (RR 1.32, 95% CI 1.09, 1.60, p = 0.004), corresponding to 9-10 patients to be treated for one to benefit, although with significant heterogeneity (p = 0.003) (Table 2 ). A non significant trend in the prostate-cancer specific survival was found (RR 1.21, 95% CI 0.82, 1.79, p = 0.32) when adopting a longer HT, without significant heterogeneity. Concerning the secondary outcomes, the longer HT significantly reduced both the risk of local recurrences (RR 1.87, 95% CI 1.22, 2.86, p = 0.004) and the risk of distant metastases (RR 1.77, 95% CI 1.16, 2.69, p = 0.007) by 11.7% and 11.5%, without significant heterogeneity, which translate into 9 patients to be treated for one to benefit (Table 2 ). No significant differences in overall survival were observed by comparing the two arms (Table 2 ). According to the performed meta-regression analysis, none of the considered negative predictors significantly affected outcome. Conversely, the treatment predictor (the time delay) significantly correlates with all explored outcomes (Table 3 ). Correlation Analysis The correlation analysis was performed in the 4 RCTs, in which BF could be considered as a potential surrogate endpoint [ 34 - 36 ]. The regression between the RR of BF and the Log of CSS (p = 0.005) was statistically significant. The regression between the RR of OS and the Log of BF and CSS was not significant (p = 0.052 and p = 0.16, respectively). The regression of the RR of DM and the Log of BF and CSS was statistically significant (p = 0.029 and (p = 0.041, respectively). The regression of the RR of LR and the Log of BF was statistically significant (p = 0.049).

Discussion Although the number of studies is small to produce reliable estimates, the data presented herein strengthen the role of the hormone-therapy duration in patients candidates to receive androgen suppression in association with radiotherapy for locally advanced prostate cancer (Table 2 Figures 2 , 3 ). Hormonal suppression in patients affected by locally advanced prostate cancer candidates to receive exclusive radiotherapy is able to improve outcome, as a series of trials recently demonstrated; according to the results of our previous meta-analysis, hormonal suppression plus radiotherapy significantly decreases recurrence and mortality of patients with localized prostate cancer. The magnitude of this survival benefit ranges from 4.9% to 5.5% for OS and CSS, respectively, which translates into 20 and 18 NNT. Local and distant relapse were significantly decreased by hormonal treatment, by 36% and 28%, without significantly affect toxicity [ 38 ]. Although the use of adjuvant hormonal therapy has been shown in randomized trials to improve outcomes, two issues still need to be clarified: 1) the optimal hormone therapy duration and 2) the impact of the radiation dose on outcome and local control. Within this context, the current meta-analysis was aimed to assess if difference exist between a shorter and longer androgen suppression, according to the trialist' definition. The pooled results of the considered trials show that a longer treatment approach improves all outcomes when compared to a shorter one (Table 2 Figure 2 , 3 ). In particular, BF is significantly reduced when hormone-therapy is administered in a longer fashion, with an overall absolute difference of 10.1%, which corresponds to only 9-10 patients needed to be treated for one to benefit (Table 2 ). Although the meta-regression analysis could be not valid given the small number of gathered trials, the benefits present across all outcomes are provided in favour of a longer strategy, regardless of any of the selected negative predictors, suggesting an overall effect independent by any clinico-pathological feature. Conversely to the these predictors, the lenght of treatment seems to significantly affect all outcomes, suggesting the even more independent value of a longer approach (Table 3 ). Although this interpretation should be softened, given the limitations related to a meta-regression analysis, a further individual patient data meta-analysis could eventually clarify if a particular subset of patients would better benefit of such approach. Moreover, these conclusion should be driven only when an homogenous risk classes classification was likely to be prospectively adopted in all analyzed trials. Indeed, the significant heterogeneity in the selected primary outcome can easily reflect that the magnitude of the reported benefit is different across trials, and this can likely be a clear effect of the different patient' selection. The adopted meta-analytic method did not allow to extract patients' subgroups data from those trials where either low, intermediate and high risk patients were accrued together. For example, no advantage for longer duration was found in those trials enrolling low risk patients [ 7 , 39 ] (Figure 2 , 3 ). Conversely, the positive effect of long term hormonal duration can be partially driven by the inclusion of those 2 trials including patients with node-positive disease [ 34 , 36 ]. So, we are not able to actually exclude a 'carry-over' effect due to the inclusion of such trials with these characteristics. The mixture of such confounders (such as different cut-offs in the Gleason score) in trial selection does not allow to derive clear and definitive conclusions. Nevertheless, even taking into account all the drawbacks in correlation and meta-regression estimations of cumulating few trials, the absolute difference in favour of the longer HT (and the lower NNT) strongly justifies a prolonged treatment strategy. Moreover, the meta-regression analysis did also show a significant correlation between the length of the longer strategy and the effect in the reduction of BF, LR and DM and the improvement of CSS (Table 3 Figure 2 ). The latest conclusion allows to carefully speculate upon a 'treat-until-progression' strategy, at least in the context of a randomized trial. Recent reports suggest that the risk of cardio metabolic problems with long-term castration deprivation therapy could counteract the benefits of hormone therapy, although this has also been questioned. Long term androgen suppression can reduce the quality of life and increase the risk of fatal myocardial infarction, fractures, and the development of a metabolic syndrome. In the context of neoadjuvant approaches, it has been recently reported that hormone-therapy is associated with a significant increase of all-cause mortality in those patients with a cardiac comorbidity, but not in those who did not show any history of cardiovascular disease [ 40 ]. This represent a crucial step-forward in the selection of those patient who could be considered as appropriate candidates for a longer hormonal suppression. The radiotherapy component of the combined treatment is crucial: cancer-specific and overall mortality rates at 10 years are significantly lower with the combined treatment in respect to androgen suppression alone. Moreover, the survival benefit seem to depend on the duration of homone-therapy, and was also reported by the Early Prostate Cancer Programme using adjuvant antiandrogen treatment as well. In studies with short or intermediate androgen deprivation of, survival prolongation has only been reported in subgroups. At the start of these trials, the standard radiation dose to the prostate was 70 Gy or less. With the advent of intensity-modulated and image-guided radiotherapy techniques, radiation doses of 78 Gy or higher are now possible, and randomised studies have shown that biochemical relapse-free survival improves with high radiation doses. As a consequence, the overall survival benefit might further increase further with higher safely radiation doses [ 41 - 44 ]. Clearly, open issues remain whether the duration of the hormonal suppression could eventually affect the outcome when radiation therapy is administered with higher dosages, and what should be irradiated (prostate or the whole pelvis). The general shared consensus is that high risk patients would better benefit from two to three years of adjuvant androgen deprivation, while those with intermediate' risk prostate cancer from six months of adjuvant androgen deprivation plus radiotherapy delivered with a dose lower or equal to 70Gy (for doses higher than 70 Gy the appropriate duration of adjuvant androgen deprivation is still unknown) [ 16 , 17 ]. According to the trial recently published by Bolla et al, the effect of short-term and long-term androgen suppression upon five-year mortality is likely to be maintained at ten years [ 39 , 45 ], whereas the short-term benefit may actually be no longer effective. The intriguing factor that can help in the daily decision between one strategy over the other is the relative weight of adverse events and the adverse effects on quality of life between the two groups of patients; indeed, although no significant difference in the overall quality of life did emerge between the two groups in the Bolla et al trial (p = 0.37), a clinically meaningful difference was present for hot flushes, sexual interest, and sexual activity [ 34 ]. Unfortunately, our analysis could investigate neither the pooled toxicities nor the quality of life data into a cumulative fashion, given the lack of a complete reporting in the original trials. Although the treatment duration seems to significantly impact upon outcome, it should be acknowledged that other factors may have a critical role, or may be more important: indeed, in the recent update of the Crook et al study, the more relevant weight for benefit seems related more to the early biochemical response to hormonal treatment before radiotherapy, rather than its overall duration [ 46 ]. For these reasons, a critical view of all data as shown in our analysis (taking into account all the limitations related to the adopted approach and methodology) is recommended.

Conclusions Taking into account the current data available in literature, and the results reported herein, the significant differences in outcome in favour of a longer hormonal treatment duration do support the adoption of such a strategy for patients affected by locally advanced prostate cancer, although several limitations should be considered given the small number of trials included. Nevertheless, a careful patient selection upon the presence/absence of cardiovascular and/or metabolic comorbidities should be adopted, in order to avoid significan late-term toxicities. On the contrary, a wide adoption of a longer administration regardless of the risk stratification (notwithstanding the results of the meta-regression analysis) seems in some way not enough mature, in absence of more reliable data coming from an individual patient data meta-analysis.

Background Hormone therapy plus radiotherapy significantly decreases recurrences and mortality of patients affected by locally advanced prostate cancer. In order to determine if difference exists according to the hormonal treatment duration, a literature-based meta-analysis was performed. Methods Relative risks (RR) were derived through a random-effect model. Differences in primary (biochemical failure, BF; cancer-specific survival, CSS), and secondary outcomes (overall survival, OS; local or distant recurrence, LR/DM) were explored. Absolute differences (AD) and the number needed to treat (NNT) were calculated. Heterogeneity, a meta-regression for clinic-pathological predictors and a correlation test for surrogates were conducted. Results Five trials (3,424 patients) were included. Patient population ranged from 267 to 1,521 patients. The longer hormonal treatment significantly improves BF (with significant heterogeneity) with an absolute benefit of 10.1%, and a non significant trend in CSS. With regard to secondary end-points, the longer hormonal treatment significantly decrease both the LR and the DM with an absolute difference of 11.7% and 11.5%. Any significant difference in OS was observed. None of the three identified clinico-pathological predictors (median PSA, range 9.5-20.35, Gleason score 7-10, 27-55% patients/trial, and T3-4, 13-77% patients/trial), did significantly affect outcomes. At the meta-regression analysis a significant correlation between the overall treatment benefit in BF, CSS, OS, LR and DM, and the length of the treatment was found (p≤0.03). Conclusions Although with significant heterogeneity (reflecting different patient' risk stratifications), a longer hormonal treatment duration significantly decreases biochemical, local and distant recurrences, with a trend for longer cancer specific survival.

Competing interests None of the authors have competing interests or potential conflicts with this work. In particular: • In the past five years have you received reimbursements, fees, funding, or salary from an organization that may in any way gain or lose financially from the publication of this manuscript, either now or in the future? NO • Do you hold any stocks or shares in an organization that may in any way gain or lose financially from the publication of this manuscript, either now or in the future? NO • Do you hold or are you currently applying for any patents relating to the content of the manuscript? Have you received reimbursements, fees, funding, or salary from an organization that holds or has applied for patents relating to the content of the manuscript? NO • Do you have any other financial competing interests? NO Authors' contributions section FCu, EB, DG, and PC conceived the analysis, and supervised the calculations; FCu, EB, DG, IS, and PC performed the calculations in a blinded fashion; VV, MM, EMR, IS, and PC participated in the trials recruitment and selection process; FCu, EB, SB and PC drafted and revised the manuscript; DG, CN, PP, GL, PM, FC and PC did coordinate the overall study process and did provide the funding. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/675/prepub

Acknowledgements Supported by a grant of the National Ministry of Health and the Italian Association for Cancer Research (AIRC). Previous Presentation: Presented at the 45 th ASCO (American Society of Medical Oncology) annual meeting, Orlando, Florida (US), May 29 th - June 2 nd , 2009.

CC BY

no

2022-01-12 15:21:44

BMC Cancer. 2010 Dec 9; 10:675

oa_package/64/82/PMC3016294.tar.gz

PMC3016295

21143927

Background Medulloblastoma (MB) is a malignant pediatric brain tumor that is thought to arise in many cases from transformation of granule neuron precursors (GNPs) within the external granular layer (EGL) of the developing cerebellum reviewed in [ 1 ]. Medulloblastoma is classified into the following subtypes based on their phenotypic [ 2 ] and genetic characteristics [ 3 ]: classic, large cell, anaplastic, desmoplastic and MB with extensive nodularity. Analyses of human desmoplastic MB samples revealed altered activity of, and/or mutations in, molecules of the sonic hedgehog (HH) signaling pathway including patched-1 ( ptch1 ) and glioma-associated oncogene homolog 1 ( gli-1 ) [ 1 , 3 - 5 ]. HH signaling is pivotal in the regulation of CNS patterning and development [ 6 - 8 ] stem cell self-renewal [ 9 ] and MB pathogenesis [ 4 , 5 , 10 ], and has also been implicated in a variety of additional human cancers including lung, pancreatic, and prostate reviewed in [ 11 ]. Mutations in ptch1 (the gene that codes for the Patched-1 12 pass transmembrane protein, a negative regulator of the HH signaling cascade), are observed with Gorlin's (Basal Cell Nevus) Syndrome - an autosomal hereditary disease characterized by the development of MB, basal cell carcinomas, and rhabdomyosarcoma [ 12 , 13 ]. Ptch1 -/- mice die in utero while ptch1 +/- mice typically survive but develop cerebellar tumors closely resembling MB in about 15% of the cases [ 10 ]. Our present understanding of the HH signaling cascade suggests that binding of the hedgehog ligands to PTCH-1 de-represses smoothened (SMO), allowing for the cascade of events leading to the activation of cubitus interruptus (Ci) in flies and the Gli transcription factors in vertebrates reviewed in [ 14 , 17 ]. The Gli transcription factors bind to the promoters of several genes considered to be HH targets (including gli1 and ptch1 ) [ 16 - 19 ]. Presently, molecules that act directly on components of the canonical HH pathway, including several that target SMO, are in various stages of clinical trials [ 20 ]. On the other hand, the protein kinase A (PKA) signalling pathway is known to antagonize hedgehog (HH) signaling in both invertebrates [ 21 ] and vertebrates [ 22 , 23 ], suggesting an alternative approach to blocking overactive HH activity. Nonetheless, little is know about the significance of the PKA/HH interaction in the pathogenesis of MB and other tumors. PACAP is a 38-amino-acid peptide originally identified on the basis of its ability to induce production of cAMP in pituitary cells [ 24 ]. PACAP regulates a variety of biological functions including neuroblast proliferation [ 25 , 26 ], and neuroblast survival [ 27 ]. In the developing brain PACAP binds to, and signals through, the G-protein coupled receptor PAC1, increasing cAMP production and PKA activity [ 28 ]. PAC1 is also coupled to other signaling cascades in some cells, including phospholipase C (PLC) [ 29 ], phosphatidylinositol 3-Kinase (PI3-K) [ 30 ], src [ 31 ], and the MAP kinase pathways [ 32 - 34 ]. Gene transcripts for the HH targets genes ptch1 and gli1 are expressed within the EGL in the developing rat and mouse cerebella [ 26 , 35 ], indicating that the HH pathway is active in these cells. PAC1 gene transcripts are also present in the EGL [ 26 ], suggesting that the PACAP and HH pathways might interact within the proliferating cells in this layer. Moreover, the robust proliferative action of SHH on cultured GNPs was completely blocked by either treatment with PACAP or pharmacological induction of cAMP/PKA [ 26 ]. The potential significance of a PACAP/HH interaction in MB was recently demonstrated. First, PAC1 gene transcripts were found to be abundantly expressed in MB preneoplastic lesions in ptch1 mutant mice [ 36 ]. Second, deletion of a single copy of PACAP in ptch1 +/- mutant mice was found to increase the incidence of MB approximately 2.5 fold [ 36 ]. Taken together, these data suggest that PACAP may regulate HH signaling in MB pathogenesis. In this work we investigate the interaction between HH and PACAP/PKA signaling in murine primary tumorspheres derived from MB tumors arising in ptch1 +/- / p53 +/- double mutant mice. Our data suggests that regulation of HH signaling by PACAP/PKA may provide a novel alternative to SMO inhibition for the treatment of medulloblastoma (MB).

Methods Animal Generation, Care, and Assessment of Medulloblastoma These studies utilized mice with single copy mutations in both the ptch1 and the p53 genes. The creation of these mice has been reported previously [ 10 , 37 ]. Generation of ptch1 +/- /p53 +/- mice were achieved by crossing p53 -/- and ptch1 +/- mice as previously described by Wetmore, Eberhart and Curran, 2001 [ 38 ]. Mice were housed according to UCLA Institutional Guidelines, including standard light-dark cycles with food and water ad libitum. All mice were monitored daily for evidence of tumor presence or illness. Drugs and Stock Concentrations PACAP-38, SANT-1 (a HH pathway antagonist), purmorphamine (a HH pathway agonist), and H-89 (a PKA antagonist) were obtained from Calbiochem. DMSO (vehicle) and forskolin (a cAMP/PKA inducer) were obtained from Sigma. Treatment concentrations for the drugs, unless otherwise stated, were as follows: 10 nM PACAP, 5 μM forskolin, 3 μM purmorphamine, 1 μM SANT-1 and 30 μM H89. When DMSO was used as drug solvent its concentration was adjusted to 0.25% (v/v) for all samples, except for experiments where H89 was used, in which case the final DMSO concentration was adjusted to 0.55%. In all experiments using H89, cells were pre-treated with this agent 30 minutes prior to addition of other drugs. Isolation of Primary Tumorspheres, Maintenance, and Treatment GNP-like tumor cells were isolated in a manner similar to that of Zhao, H et al., 2008 [ 39 ]. A total of thirteen individual tumor isolations and preparations were performed. Briefly, tumors were minced in 1 mL of trypsin/DNAse, placed into 1.5 mL eppendorf tubes and centrifuged at 750 × g for 5 min at 4°C. Pellets were resuspended in 800 μL of cold DNAse and immediately triturated with a 1000 μL filtered pipet, followed by trituration with a 200 μL pipet, and lastly with a fire polished Pasteur pipet. After another centrifugation as above, 500 μL of DNAse was added to the pellet with a trituration. This sample was resuspended in 4.5 ml of DMEM-F12 and applied to a 40 μm strainer (BD Falcon). The cell suspension was carefully layered on top of a 32%/60% percoll gradient and centrifuged at 2985 × g for 20 min at 4°C. Cells collected from the layer between the 32% and 60% percoll were washed once with cold DMEM-F12 by centrifuging at 2985 × g for 10 min. The supernatant was aspirated and the resulting pellet was resuspended in serum free medium comprising of DMEM/F12 medium (Invitrogen), Heparin [2 μg/mL](Sigma), Penicillin/Streptomycin/Glutamine [1%] (Invitrogen), bFGF [0.2 μg/mL] (Peprotech), EGF [0.02 μg/mL](Sigma), and B27 supplement [1:50] (Invitrogen). The cells were then counted and plated into 6-well plates containing 2 mL of medium such that the final cell concentrations were either 1 × 10 6 cells/mL or 0.5 × 10 6 cells/mL. Cells were then cultured for 6 or 7 days. During this period the cells were supplemented with additional bFGF (0.05 μg) and EGF (0.04 μg) on days 3 and 5. On day 6 or 7, serum-free media was added when necessary to equalize all volumes at 2 mL, and then drug treatments were added to the cells in the same wells as indicated in the figure legends. RNA Extraction and RT-PCR Tumorspheres were pelleted in 1.5 mL eppendorf tube by microcentrifugation (750 × g for 5 minutes at 4°C). RNA was extracted from the samples using the TRIzol reagent (Invitrogen) according to manufacturer's instructions. cDNA was generated using the iScript cDNA synthesis kit (BioRad). Semi-quantitative RT-PCR for pac1 and SYBR-Green quantitative real-time RT-PCR for gapdh and gli1 were performed as described previously [ 26 , 36 ]. Individual sample differences were calculated based on the ratio of standard quantities of gli1 and gapdh transcripts determined by the standard curve method. Western Blotting Whole-cell lysates, prepared in NP-40 lysis buffer along with 1 mM PMSF, 1 μg/ml each of aprotinin, leupeptin, and pepstatin, 1 mM Na 3 VO 4 , 1 mM NaF (Sigma), were analyzed by immunoblotting. Equal amounts of proteins from whole-cell extracts were resolved on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to PVDF membranes. The membranes were subjected to Western blot analysis with rabbit anti-phospho-PKA substrate antibody (1:2000; Cell Signaling Technology) or mouse anti-β-actin antibody (1:5000; Cell Signaling Technology). The signals from the primary antibody were amplified by horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse immunoglobulin G (Santa Cruz Biotechnology) and detected with enhanced chemiluminescence (GE Healthcare). Proliferation Studies Thymidine incorporation in tumorsphere cultures was performed using the trichloroacetic acid (TCA) precipitation method. Dissociated cells isolated from tumors by percoll gradient as described above, were seeded in 12-well tissue culture plates at the initial density of 1×10 6 cells/well and cultured in serum free media as indicated above. After 24 hours, the tumorspheres were treated for either 24 or 48 hours with indicated drugs. [ 3 H]-thymidine (NEN/PerkinElmer), 2 μCi/well, was added to the wells during the last 18 hours of treatment. Tumorspheres were harvested and centrifuged at 800 × g for 10 min at 4°C. After removal of the supernatant, the pellets were washed twice with 1 mL of PBS then lysed in 500 μl of NaOH (0.5 M) for 10 minutes at room temperature. Cell extracts were transferred into fresh tubes and DNA was precipitated with 500 μl of cold 20% TCA on ice for 20 min. The samples were centrifuged at 20,000 × g for 20 min at 4°C. Supernatants were aspirated and the DNA-containing pellets washed with 500 μl of 10% TCA. After a 10 min centrifugation, pellets were finally reconstituted in 500 μl of 0.1 M NaOH and transferred into scintillation vials. Incorporated radioactivity was counted on a scintillation counter (Beckman) using AquasafeTM scintillation cocktail (Wallac). Statistics Statistical analyses were performed using the GraphPad Prism 4.0 software. All P values were calculated using Student's t-test for comparison between two samples only, and ANOVA followed by Bonferroni's post hoc test for multiple comparisons. A value of P < 0.05 was considered significant. All data were normalized to vehicle controls.

Results Primary Mouse Medulloblastoma Cells Grow as Tumorspheres and Exhibit Constitutive Hedgehog Pathway Activity Cell lines derived from human or murine MB tumors have been used to understand the biology of MB, but recent evidence indicate that propagated MB cell lines lose much or all of their responsiveness to HH pathway antagonism [ 40 ]. We therefore harvested fresh primary MB tumor cells for each experiment. We utilized ptch1 +/- /p53 +/- double mutant mice for our studies because mutations in the p53 tumor suppressor gene in the context of ptch1 haploinsufficiency results in a higher incidence of MB than that in mice with the ptch1 mutation alone [ 38 ]. Tumor cells were harvested and purified by percoll gradient, and single cell suspensions were plated in serum-free media for 1 week using a modification of methods reported by others to preserve HH pathway-dependent proliferation [ 39 ]. After several days of culture the cells formed aggregates of increasing size, referred to as "tumorspheres" (Figure 1A ). To determine if the HH pathway was active in these cells, we treated these cultures at the end of the 1-week period with the HH pathway antagonist SANT-1 (a known inhibitor of smoothened (SMO), 1 μM [ 41 ]) and measured the expression of the HH target gene gli1 by real time RT-PCR. SANT-1 significantly decreased gli1 expression (Figure 1B ) relative to vehicle. These data implied that the HH pathway is constitutively active in tumorsphere cultures generated using our experimental paradigm. To determine if the HH pathway could be activated beyond basal levels in these tumorsphere cultures, we treated them with the SMO agonist purmorphamine at a concentration (3 μM) shown to induce maximum HH signalling without cellular toxicity [ 42 - 44 ]. Purmorphamine failed to induce gli1 expression above control levels, suggesting that the HH pathway was already maximally activated in the tumorspheres under these experimental conditions (Figure 1C ). Pituitary adenylyl cyclase activating polypeptide (PACAP) and Forskolin Inhibit Hedgehog Pathway Activity in Primary Tumorspheres Because the HH pathway exhibited a significant level of constitutive activity in primary tumorsphere cultures, we investigated if this activity could be inhibited by either PACAP (10 nM) or forskolin (5 μM), an activator of adenylyl cyclase. We first performed RT-PCR analysis for the PACAP receptor PAC1 and confirmed that its gene expression is preserved in tumorspheres cultures (Figure 2B ). Treatment with 10 nM PACAP and 5 μM forskolin resulted in significant down-regulation of gli1 in these cultures compared to vehicle treated controls (Figure 2A ). Attenuation of Gli1 Expression in Primary Tumorspheres by PACAP was Associated with PKA Activation and was Reversed by the PKA Inhibitor H89 As both PACAP and forskolin typically exert their effects in a PKA dependent manner, we investigated if their effects could be reversed by the PKA inhibitor H89. To confirm their effects on PKA activity, we performed Western blot analyses using an antibody that detects proteins containing a phospho-serine/threonine residue with arginine at the -3 and -2 positions, which are typically found on proteins phosphorylated by PKA. We observed that treatment of primary tumorspheres with either 10 nM PACAP or 5 μM forskolin induced similar increases in PKA activity (Figure 3A ) compared to vehicle treated controls. Treatment with H89 (30 μM) reversed the inductions of PKA activity by these treatments (Figure 3A ). Moreover, treatment with H89 reversed the reduction in gli1 expression by 10 nM PACAP (Figure 3B ). These data indicate that PACAP attenuates HH signaling in primary tumorspheres, most likely via a PKA dependent mechanism. PACAP and Forskolin Inhibit Cellular Proliferation in Primary Tumorspheres The above studies suggest that PACAP treatment or stimulation of PKA activity by other methods might provide strategies to inhibit the growth of MB tumors. We thus investigated the effects of PACAP (10 nM) and forskolin (5 μM) on cellular proliferation using [ 3 H]-thymidine to measure DNA synthesis. As predicted, proliferation was reduced by 24 or 48 hour treatment with PACAP in a dose dependent manner. Forskolin (5 μM) and SANT-1 (1 μM), also decreased proliferation at these time periods (Figures. 4A and 4B ). These data suggest that PACAP and PKA activation might provide therapeutic alternatives to SMO antagonism in the treatment of MB.

Discussion A large and diverse array of cancer types show evidence of overactive HH signalling [ 11 ]. This has triggered an intensive search for agents that selectively inhibit this pathway [ 20 ]. As a result, several molecules have been discovered or designed that directly interact and inhibit the activity of SMO, some of which are in clinical trials [ 20 ]. Despite the promise of this approach, two potential problems are that 1) some tumors acquire mutations in SMO that render them relatively resistant to SMO antagonists [ 45 ], and 2) SMO is critically involved in the growth of bones and other tissues [ 11 , 46 ], and appears to regulate stem cell biology [ 9 ] and [ 47 ]. Thus, there is a need for molecules that act downstream of SMO in the HH signalling pathway and that inhibit HH signalling specifically in tumor cells. PKA is thought to act in the HH pathway downstream of SMO, phosphorylating and triggering proteosome-mediated degradation of Gli2 and Gli3 [ 48 , 50 ]. Thus PKA activators might be expected to block HH signalling even in the context of drug-resistant SMO mutations. However, given that PKA is used in so many biological processes, it is unlikely that general activation of PKA will be useful as a treatment strategy. An alternative strategy would be to activate PKA specifically in tumor cells. This could potentially be achieved by targeting appropriate heterotrimeric G protein-coupled receptors expressed on the cells. The data reported here suggest that a potentially-effective strategy to treat MB tumors exhibiting overactive HH signaling would be to activate PACAP receptors. We investigated this possibility by using genetically engineered mice that spontaneously develop MB, and a tissue culture system that preserves a significant level of constitutive HH activity in tumor cells. We found that PACAP inhibited proliferation and the expression of the HH target gene gli1 in these cells. Moreover, the reduction of gli1 gene expression by PACAP was reversed by PKA inhibition, and was mimicked by stimulation of the cAMP/PKA pathway by forskolin. Overall, the data suggest that PACAP, via activation of PKA, inhibits HH signalling and HH-driven proliferation in MB cells. However, our findings do not rule out alternative mechanisms of PACAP/PKA action in these assays. For example, despite its action on gli1 in MB tumorspheres, the ability of PACAP to inhibit proliferation might be unrelated to its effects in HH pathway activity. In this respect, PACAP (and PKA activation in general) is known on inhibit the proliferation of many types of cells [ 51 , 53 ], some of which might not exhibit HH pathway activity. In any case, the findings reported here suggest that PACAP or PAC1 agonists could be useful as alternatives to, or in conjunction with, SMO antagonists, for the treatment of MB and other HH dependent tumors.

Conclusions Primary murine medulloblastoma derived tumorspheres from ptch1 +/- /p53 +/- mice, cultured in a serum-free environment, express the PACAP receptor PAC1, and exhibit constitutive HH pathway activity as evidenced by sensitivity to HH pathway antagonist SANT-1. Treatment of tumorspheres with PACAP or the cAMP/PKA activator forskolin significantly reduces gli1 expression in a manner that is reversed by the PKA inhibitor H89. PACAP and forskolin also block tumorsphere proliferation. These finding suggest that PACAP or PKA activation may be useful for therapeutic intervention in the treatment of MB.

Background Hedgehog (HH) signaling is critical for the expansion of granule neuron precursors (GNPs) within the external granular layer (EGL) during cerebellar development. Aberrant HH signaling within GNPs is thought to give rise to medulloblastoma (MB) - the most commonly-observed form of malignant pediatric brain tumor. Evidence in both invertebrates and vertebrates indicates that cyclic AMP-dependent protein kinase A (PKA) antagonizes HH signalling. Receptors specific for the neuropeptide pituitary adenylyl cyclase activating polypeptide (PACAP, gene name ADCYAP1) are expressed in GNPs. PACAP has been shown to protect GNPs from apoptosis in vitro , and to interact with HH signaling to regulate GNP proliferation. PACAP/ ptch1 double mutant mice exhibit an increased incidence of MB compared to ptch1 mice, indicating that PACAP may regulate HH pathway-mediated MB pathogenesis. Methods Primary MB tumorsphere cultures were prepared from thirteen ptch1 +/- /p53 +/- double mutant mice and treated with the smoothened (SMO) agonist purmorphamine, the SMO antagonist SANT-1, the neuropeptide PACAP, the PKA activator forskolin, and the PKA inhibitor H89. Gene expression of gli1 and [ 3 H]-thymidine incorporation were assessed to determine drug effects on HH pathway activity and proliferation, respectively. PKA activity was determined in cell extracts by Western blotting using a phospho-PKA substrate antibody. Results Primary tumor cells cultured for 1-week under serum-free conditions grew as tumorspheres and were found to express PAC1 receptor transcripts. Gli1 gene expression was significantly reduced by SANT-1, PACAP and forskolin, but was unaffected by purmorphamine. The attenuation of gli1 gene expression by PACAP was reversed by the PKA inhibitor H89, which also blocked PKA activation. Treatment of tumorsphere cultures with PACAP, forskolin, and SANT-1 for 24 or 48 hours reduced proliferation. Conclusions Primary tumorspheres derived from ptch1 +/- /p53 +/- mice exhibit constitutive HH pathway activity. PACAP antagonizes HH signalling in these cells in a manner blocked by the PKA antagonist H89. PACAP and pharmacological activation of PKA also inhibited proliferation. Our data suggests that regulation of HH signaling by PACAP/PKA signaling may provide an alternative to SMO inhibition for the treatment of MB.

Competing interests The authors declare that they have no competing interests. Authors' contributions JRC: designed and performed the experiments, analyzed and interpreted the data, and wrote the manuscript; DZR: helped in performing experiments; PN: helped with experimental design, technical expertise and data interpretation; HD: performed all animal care, husbandry, and genotyping; LML: helped with experimental design and interpretation of the data; JAW: coordinated the design and execution of experiments, supervised and participated in data analysis, and interpretation, and editing of the manuscript. All authors have read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/676/prepub

Acknowledgements We would like to thank the members of Dr. Waschek's laboratory for their valuable insight and support throughout this work. We would also like to thank Drs. Matthew P. Scott and Owen N. Witte for generously providing the ptch1 and p53 mutant mice, respectively. This present work was supported by NIH grant CA110384 to JW. JRC was supported by the Molecular and Cellular Neurobiology and Intellectual and Developmental Disabilities Research Center training grants (5T32DH007032-30 and 5T32MH19384-13, respectively).

CC BY

no

2022-01-12 15:21:44

BMC Cancer. 2010 Dec 9; 10:676

oa_package/1f/02/PMC3016295.tar.gz

PMC3016296

21176237

Background The emergence of personalized medicine as a central healthcare theme has led to an explosion of research dedicated to finding molecular markers for prediction of disease outcome and treatment benefit. However, achieving clinical validation for molecular tests has proven to be challenging given the dual needs for selection of high quality molecular markers and the provision of an analytically sound test process for their measurement. Indeed, the clinical utility of molecular biomarkers may be limited unless they are provided in the context of a robust, well-characterized diagnostic test with documented accuracy, precision, and reproducibility. The Onco type DX Breast Cancer Assay has met this standard as an analytically and clinically validated high complexity, multi-analyte, RT-PCR test to predict the likelihood of recurrence and response to chemotherapy in estrogen receptor-positive, node-negative and node-positive breast cancer patients [ 1 , 2 ]. With over 160,000 tests performed since its launch in 2004, and its inclusion in the National Comprehensive Cancer Network (NCCN) and American Society Clinical Oncology (ASCO) clinical practice guidelines, the Onco type DX Breast Cancer Assay has become standard of care for individualized treatment decision-making in early stage breast cancer [ 3 - 5 ]. For patients diagnosed with stage II colon cancer, the decision of whether or not to receive adjuvant chemotherapy has been a significant challenge, in part due to limitations in existing methods for assessing recurrence risk. Although pathologic T4 stage and mismatch repair deficiency (MMR-D) can identify patients at significantly higher or lower recurrence risk, respectively [ 6 , 7 ], these two markers account for only 25-30% of stage II patients. For the majority of stage II colon cancer patients, who would be characterized as having standard risk, there has been a major need for better prediction of individual recurrence risk [ 6 ]. Based on the results of the clinical validation study of the 12 gene Onco type DX Colon Cancer Assay [ 7 ], stage II colon cancer patients (particularly those with T3 disease and mismatch repair proficient (MMR-P) tumors) are now empowered to make more informed treatment decisions with quantitative, individualized recurrence risk information based on the molecular profile of their tumor. The seven cancer-related genes in the Onco type DX Colon assay were selected from a panel of 761 genes for their consistent association with colon cancer recurrence in 4 large and independent development studies [ 6 ]. Three of these genes are associated with activated stroma ( BGN, INHBA and FAP ), three represent a cell cycle pathway ( MK167, MYBL2 and MYC ) and one can be characterized as part of an early-response or genotoxic stress pathway ( GADD45B ). For each patient, expression values from the co-expressed stromal genes are aggregated to form a stromal gene score (SGS) and the expression values of the three cell cycle genes are aggregated to form a cell cycle gene score (CCGS). The SGS, CCGS and the expression of GADD45B are combined to generate the final RS which provides an individualized risk estimate for colon cancer recurrence (see Figure 1 ). The expression levels of these genes may vary both within and between FPE blocks from the same patient's tumor. However, all sources of variability are inherently incorporated into the clinical validation of the Onco type DX Colon Cancer Assay, and therefore embedded within the estimated risk of disease recurrence obtained for each RS value. However, one source of variability which can be controlled in a molecular diagnostic test is that derived from the assay process (analytical variability). In fact, it is this source of variability which has to be controlled if a consistent and reliable test result is to be provided to patients on a daily basis. Prior to performing the clinical validation study of the 12 gene RS assay we established pre-determined assay performance criteria and performed analytical validation studies, designed to demonstrate that the process was well-controlled and capable of reproducibly translating tumor biology into clinically actionable information. We report here the design and results of the suite of analytical validation studies that characterize the performance of the assay.

Methods Gene selection The seven cancer-related genes ( BGN, MYC, FAP, GADD45B, INHBA, MK167 and MYBL2 ) and five reference normalization genes ( ATP5E, GPX1, PGK1, VDAC2 and UBB ) were selected from 761 genes, based on the results of four development studies which included over 1800 patients [ 6 ]. Tumor blocks and samples FPE colon cancer blocks (Stage II adenocarcinoma and mucinous carcinoma) from recent surgeries (within 3-6 months of testing) were obtained under an Institutional Review Board approved protocol. The selected blocks were representative of the tumor block type expected to be received in the clinical reference laboratory (with respect to tumor type, fixative and age). Samples with no tumor or very little tumor (< 5% tumor present) were excluded. In studies where pooled RNA samples were required, samples were selected to assure acceptable ranges of expression levels for the seven cancer-related genes. RNA extraction RNA extraction was performed using a semi-automated method, as previously described [ 8 ]. RNA quantification RNA was quantified using the RiboGreen fluorescence method (Invitrogen, Carlsbad, CA.), as previously described [ 9 ]. The minimum concentration required for accurate quantification was 5 ng/μl. Any sample with less than this amount was excluded from the study. Genomic DNA detection RNA was tested for residual genomic DNA (gDNA) by quantitative PCR (qPCR) using an ATCB TaqMan ® assay (see section "TaqMan Primer Probe Design and Gene Expression Analysis" for details). Six wells containing 2 ng/well RNA were assayed per sample. RNA samples were failed if more than 2 wells gave a Cycle Threshold (C T ) value of <37. This value represents less than 5 pg per well (less than 3 copies). Reverse transcription and qPCR Reverse transcription was performed as a single, master reaction using the Omniscript kit (Qiagen, Valencia, CA) with a gene specific primer (one reverse primer) for each assay gene. Each primer in the RT reaction was at a concentration of 50 nmol/L. For the standard assay, a Tecan workstation was used to dispense RNA at a concentration of 2 ng/μl into the RT reaction (2 ng/well complementary DNA (cDNA) for quantitative PCR (qPCR)). The dilution series was manually pipetted using RNA input ranging from 0 to 32 ng/μl in the RT reaction (0 to 32 ng/well of cDNA for quantitative PCR). RT reactions were performed either in a single tube (dilution series) or 96-well plate (for standard reactions). The resulting cDNA was distributed to 384-well plates and PCR forward and reverse primers and TaqMan probe were added using a Tecan workstation. TaqMan ® primer probe design and gene expression analysis Reference sequences were obtained from NCBI Entrez, and TaqMan assays designed using a proprietary primer-design module. Where possible, TaqMan assays were designed to span introns ( ATP5E, BGN, MK167, MYBL2, PGK1 and VDAC2 ). Oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA). Dual labeled TaqMan probes have a Fluorescein ( 6-FAMTM) 5' reporter and a Black Hole Quencher-2 (BHQ-2 ® ) 3' quencher. Largest amplicon size was 84 base pairs and smallest size was 66 base pairs. See Additional file 1 Table S1 for the oligonucleotide sequences and Additional file 2 Table S2 for amplicon sequences. TaqMan RT-PCR was performed using Applied Biosystems (ABI) Prism 7900HT instruments (7900) according to manufacturer's instructions. Reactions were 10 μl volume with cDNA equivalent to 2 ng total RNA (with the exception of the linearity study, where cDNA input ranged from 2 -10 to 2 5 ng/well). Final primer and probe concentrations were 0.9 μmol/L (primers) and 0.2 μmol/L (probe). PCR cycling conditions were 95°C for 10 minutes for one cycle, 95°C for 20 seconds, and 60°C for 45 seconds for 40 cycles. Triplicate wells were run for each assay and each well was classified as "valid" or "invalid" using pre-defined amplification curve metrics. The mean C T of the valid wells was used in subsequent statistical analyses. Reference gene normalization Where required, normalization was performed using the average expression of the five reference genes ( ATP5E , GPX1 , PGK1, UBB and VDAC2 ). These genes were selected due to low expression variability between patients in the preliminary clinical development studies [ 6 ] (and data on file, Genomic Health Inc). The mean (valid) C T for each gene was subtracted from the mean (valid) C T for the 5 reference genes, then 10 was added to give a normalized range of expression from 0-15, where each unit reflects a 2-fold change in expression. Gene group and Recurrence Score calculation Genes in the stromal gene group and cell cycle gene group were selected on the basis of co-expression and meaningful biological pathway associations, as previously described [ 6 ]. The CCGS was derived from the aggregated, normalized mean C T values for MYC, MKI67 and MYBL2 . The SGS was derived from the aggregated, normalized mean C T values for BGN, FAP and INHBA . The CCGS, SGS and normalized mean C T for GADD45B are then entered into an algorithm to generate the RS (see Figure 1 ). Amplification efficiency Using a pool of FPE RNA samples, two independent 17-point serial dilutions (representing 2 -10 to 2 5 ng RNA equivalent/well) were manually prepared and taken through RT and then qPCR, with triplicate wells per gene assay. The mean C T for the two dilution series, at each RNA input, was averaged to obtain a final non-normalized C T value. The estimates of amplification efficiencies were calculated for each of the 12 genes using the formula where slope was estimated from the regression of C T measurements versus RNA concentration. Linearity Using the aggregated, non-normalized C T values from the 17-point serial dilution, the linearity of C T value as a function of RNA concentration was evaluated for each gene assay. The polynomial method proposed by Krouwer et al., 1993 [ 10 ] and recommended by the National Committee of Clinical Laboratory Standards (NCCLS) [ 11 ], was used to test for linearity over the range of input RNA concentrations. The polynomial method evaluates nonlinearity in two parts. The first part examines whether a second or third order polynomial model fits better than a linear model. If none of the nonlinear terms in either the second or third order polynomial models is significant, then linearity is assumed. If significant nonlinearity is detected, then comparisons are made between the best-fitting polynomial model and linear polynomial to determine if the degree of bias is within a predefined allowable limit. For each gene assay, orthogonal polynomial regression was used to obtain coefficients and associated tests of significance for the first (linear), second (quadratic), and third (cubic) order polynomials. Since precision is known to vary significantly across different RNA concentrations, heteroscedasticity in error variance was modeled through a log-linear variance model. The degree of nonlinearity in signal response was assessed by examining the standard error of the regression and selecting the higher-order (nonlinear) polynomial model with the best fit. This statistic constituted the average difference between the data and the model, and so the model with the lowest value provided the best fit. At each (known) input RNA concentration, the deviation from linearity (DL) was calculated as follows: where the values x range from x 1... x 15 , and p ( x i ) was the value of the best-fitting polynomial at point x i . Consequently, DL i is a measure of the difference between the nonlinear model and the best-fit straight line at each of the RNA concentrations. Analytical sensitivity The analytical sensitivity of each of the 12 genes was evaluated separately. Specifically, for each gene assay, a nonlinear mixed effects model with log-linear variance function was used to model the heteroscedasticity in intra-assay response as a function of RNA concentration. This information was used to estimate the Limit of Detection (LOD) and Limit of Quantitation (LOQ) of the assay. For each of the 12 genes, a lower one-sided 95% confidence interval on the C T at zero RNA concentration, y min , was calculated. The LOD was estimated by a lower 95% confidence bound on the mean expression at zero concentration (natural scale). Similarly, the LOQ was estimated by the inverse prediction of y min with the upper one-sided 95% confidence interval of the fitted calibration model. Analyses were performed using an iterative estimation scheme involving the PROC MIXED procedure in SAS version 9.1. Assay precision and reproducibility Two FPE RNA pools were created to provide homogeneous templates for measuring precision and reproducibility; one pool had a RS in the high recurrence risk group (RS≥41) and the other had a RS in the low recurrence risk group (RS < 30). Using the standard 2 ng/well RNA equivalent, precision was assessed by estimating between-day, between-lot, between-7900 instrument (7900HT Fast Real-Time PCR System, Applied Biosystems), between-plate (within day), and within-plate variability components and total variability. Reproducibility was assessed by estimating differences in mean C T s between Tecan workstations used to assemble RT and qPCR plates. Analyses were performed using the PROC MIXED procedure in SAS version 9.1. For assessing precision, a mixed effects analysis of variance (ANOVA) model was used to decompose the total variability in C T measurements into components of variance due to day, 7900 instrument, oligonucleotide lot and plate (run) within day, treating Tecan workstations as a fixed effect. An efficient (algorithmic) experimental design with G-efficiency > 50% was followed. Two Tecan workstations were used to assemble the RT and qPCR plates on five different days, using three different lots of oligonucleotides and data generated on five different 7900 instruments. For each gene, and each FPE RNA pool, Restricted Maximum Likelihood (REML) estimates of the precision of the assay were obtained along with approximate 95% confidence intervals expressed as relative standard deviations (RSD) of total variance. For assessing the reproducibility of the assay, (least square) mean C T and mean scores between different combinations of RT and qPCR Tecan workstations were contrasted. Assay controls Prior to clinical validation, a standard RNA template was run across several 7900 instruments, using various primer-probe lots to monitor control of the process. During clinical validation [ 7 ], the same RNA was included on every RT plate at the same concentration as test samples, and was used as an RT positive control. Nuclease-free water was used as the RT negative control. For the quantitative PCR controls, an RNaseP TaqMan primer-probe set (Applied Biosystems, Foster City, CA) was distributed in twelve wells across the 384-well plate; genomic DNA was added to the qPCR positive control wells and nuclease-free water to the qPCR negative control wells.

Results When an assay result depends on multiple single-value measurements, it is imperative that the analytical characteristics of each analyte are assessed independently. Here we describe the analytical performance of the individual components of the Onco type DX Colon Cancer Assay, which allowed us to confidently proceed to clinical validation. Amplification efficiency Amplification efficiencies close to 100% are characteristic of qPCR assays that are robust, reproducible and specific. The 17-point serial dilution demonstrated that each gene assay achieved this. Specifically, amplification efficiencies (summarized in Table 1 ) were excellent and ranged from 96% to 107%. Pipette error, the presence of PCR inhibitors or non-specific amplification (primer-dimer) may account for efficiencies greater than 100%. In addition, all genes showed similar efficiencies. This is an important characteristic of tests which utilize expression of reference genes to normalize expression of the test genes [ 12 ]. Linearity The range of expression across patients for each of the 12 genes is large, and therefore it is important to demonstrate linearity of each gene assay over the potential range, to ensure accuracy of the result irrespective of the level of expression. Using the same dataset generated to assess amplification efficiency, the linearity of signal response between C T value and RNA concentration was evaluated for each of the 12 genes. Linear range estimates and estimated maximal deviation from linearity at the extremes are provided in Table 2 . All 12 genes met the pre-specified acceptance criteria (a maximum deviation from linearity of 1 C T ) over at least an 11 log 2 concentration range (2 -6 to 2 5 ) with a median deviation at the extremes of 0.4 C T . Analytical sensitivity Since background noise could impact signal, it is important to determine the LOD for each gene. Similarly, it is important to establish the LOQ for each gene assay to achieve confidence in the level of expression being reported. Table 3 summarizes the estimated Limit of Detection (LOD) and Limit of Quantitation (LOQ) for each of the 12 genes. The LOD for all 12 genes was 40 C T , thus meeting the pre-specified acceptance criterion of ≥37 C T . The LOQ for all 12 genes was greater than 36 C T , also meeting the pre-specified acceptance criterion of ≥35 C T . These values are similar to those obtained for the 21 genes in the Onco type DX Breast Cancer Assay [ 13 ]. Analytical precision Between-7900 instrument, between-primer-probe lot, and within-PCR plate components of variance accounted for greater than 80% of total variance in C T measurements. Between-instrument and between-primer-probe lot constituted the largest components of variance. Table 4 lists the estimates of total variance for non-normalized C T values for each of the two FPE RNA pools investigated. All genes met the pre-specified acceptance criterion of 10% on the total % RSD. Total means and total SD for the SGS, CCGS, GADD45B and RS are summarized in Table 5 . In both the low RS and high RS FPE RNA pools, the highest SD was 0.04 (normalized/aggregate) C T units for the SGS, 0.05 (normalized/aggregate) C T units for the CCGS, 0.13 (normalized/aggregate) C T units for GADD45B and 1.38 RS units for RS. These results demonstrate that the analytical variability would be very unlikely to produce a clinically meaningful change in the risk of recurrence reported for each patient. Reproducibility Table 6 summarizes the largest pair-wise differences in (least square) mean non-normalized C T between Tecan workstations by gene for the two FPE RNA pools. Overall, the differences in mean C T between Tecan workstations across all 12 genes and the two FPE RNA pools were small, all ≤ 0.28 C T . All the genes in the 12-gene panel met the pre-specified acceptance criteria for reproducibility. Assay controls A control sample was run on twenty-one RT plates prior to clinical validation of the Onco type DX Colon Cancer Assay, and on thirty one RT plates during clinical validation. Figure 2 is a variability chart (Box Plots) of gene expression for the 12 genes for the RT positive control run on these fifty two RT plates. It illustrates a tightly controlled process with standard deviations in aggregate C T measurements between samples ranging from 0.19 to 0.33 C T units. Figure 3 provides a histogram of the RNase P (Applied Biosystems. Foster City, CA) PCR positive controls from all PCR plates generated prior to and during clinical validation, where a pre-specified acceptance criterion of 1% coefficient of variance (CV) was applied. During the clinical validation study, seven PCR plates containing twenty eight samples (< 2%) failed this 1% CV specification and were therefore repeated.

Discussion A molecular diagnostic test used for treatment planning in cancer patients must be analytically robust and reproducible so as to enable consistent and accurate translation of each patient's tumor biology into clinically actionable information. The Onco type DX Colon Cancer Assay generates a RS for individual patients from the reference normalized tumor expression level of 7 cancer-related genes. These genes were chosen from a refined list of 48 genes (from an original set of 761 candidate genes) which were consistently associated with colon cancer recurrence in four independent clinical development studies comprising more than 1800 patients [ 6 ]. Among these 48 genes, two major pathways are represented; cell cycle and stroma-associated. The heart of the co-expressed stromal gene group is a collection of extra-cellular matrix (ECM) proteins including BGN, COL1A1, SPARC and CTHRC1 . Cluster analysis from the development studies show these genes shared the lowest average distance between clusters (1-Pearson's R distance as low as 0.15) [ 6 ]. Other subgroups observed in the stroma-associated gene cluster included the TGFβ signaling pathway ( TGFBI, TGFB3, INHBA ), early response genes ( EGR1, GADD45B ), the WNT pathway ( SFRP2, SFRP4 ), and invasion related genes ( PAI1, OPN, TIMP2, TIMP3 ) [ 6 ]. Genes within the cell cycle group included checkpoint control genes and cell cycle regulation genes (such as CDC20, MCM2, MYBL2, CSE1L, MYC and MK167 ). After identifying these biological pathways and genes as important markers of clinical outcome in stage II colon cancer, they were used to develop a test to enable translation of tumor biology into clinically useful information. Since biological heterogeneity was inherently embedded within the clinical validation of the Onco type DX Colon Cancer Assay, any such variability is already incorporated into the prediction of disease recurrence provided by each individual patient's RS. However, the most important source of variability from the perspective of daily testing is assay process, or analytical variability. The 12 RT-qPCR assays and associated processes were therefore analytically validated using pre-defined performance criteria prior to clinical validation of the RS. Analytical validation for the Onco type DX Colon Cancer Assay was patterned after the approach used for the widely accepted Onco type DX Breast Cancer Assay [ 3 - 5 ]. At the time the Onco type DX Breast Cancer Assay was analytically validated there was no widely accepted standard for multi-analyte RT-PCR tests. Therefore, methods commonly used to validate single-analyte laboratory tests were adapted for the purpose [ 13 ]. Here, all 12 individual genes were shown to be linear over a 2,000 fold range, and 5 genes ( ATP5E, MYC, GPX1, UBB and VDAC2 ) were linear over a 32,000 fold range. For all genes, the limit of detection was at a C T of 40, and the limit of quantitation at a C T of 36 or greater, providing uniformly high analytical sensitivity for all genes being reported. The sensitivity and accuracy of the Onco type DX Colon Cancer Assay ensures robust reporting irrespective of the level of RNA expression, an attribute which may not be achieved with DNA microarrays given the lesser dynamic range of this platform [ 14 , 15 ]. This may in part account for why assessments and validations of microarray systems have focused on precision or reproducibility, rather than accuracy [ 16 - 20 ]. The Onco type DX Colon Cancer Assay is performed in a high-throughput process using multiple Tecan robotic workstations. In addition, multiple reagent lots and 7900 qPCR instruments provide potential sources of process variability. Therefore, all measureable analytical sources of variability were assessed to determine total system variability. Using two FPE RNA pools, an analysis of variance (ANOVA) model was applied to estimate the total analytical variability in C T measurements for each separate component of variance. The greatest source of analytical variability in C T (and therefore gene scores and RS) came from between-qPCR instruments and between-primer-probe lot components. However, even with these components the relative standard deviations (RSD) associated with each gene was still very small, and well within the pre-defined acceptance criterion of 10%. In fact, the upper bounds of the 95% confidence intervals on the RSD for all the genes in the 12-gene panel were within 10%. The high precision of the individual genes translates into a similarly high level of precision for the stromal gene group score (SD≤0.04), the cell cycle gene group scores (SD≤0.05) and the RS (SD≤1.38). Plots from a control sample run prior to and during the clinical validation study demonstrate consistent stability for each gene assay. During the clinical validation study, less than 2% of the qPCR plates had to be repeated because of failure to meet both qPCR positive and negative control specifications. Such a well-controlled assay process is an important element of any prognostic or predictive molecular test performed in a clinical reference laboratory used for treatment planning. The two principal co-expressed gene groups in the Onco type DX Colon Cancer Assay have long been known as important in cancer progression. The expression levels of an ECM protein ( BGN ), a fibroblast specific integral membrane serine protease ( FAP ), and a TGFβ family member ( INHBA ) are aggregated in the colon RS algorithm to form the SGS, where higher expression is associated with a higher risk of recurrence. Dvorak first described cancers as wounds that do not heal [ 21 ], and it is now generally accepted that activated stroma represents a "wound healing response" that can promote tumor growth, cell migration, invasion and angiogenesis [ 22 - 27 ]. In the same way that tissue regeneration during wound healing involves a complex relationship between the host and the microenvironment, tumorgenesis is also dependant on extra-cellular interactions and signals from the stroma. High amounts of stroma have been associated with poor clinical outcome in patients with colon cancer [ 28 , 29 ], but as demonstrated by clinical validation of the Onco type DX Colon Cancer Assay, the level of activation and associated gene expression within the stroma is strongly associated with risk of recurrence. GADD45B is entered into the algorithm as an individual gene, although it tends to associate with the larger stromal group on cluster analysis [ 6 ]. Interestingly, GADD45B is believed to stimulate BGN expression [ 30 ]. The second co-expressed gene group that proved to be clinically informative as a component of the colon cancer RS was the cell cycle gene group. In contrast to the Onco type DX Breast Cancer Assay, where higher expression of cell cycle genes ( STK15, MYBL2, MK167 and CCNB1 ) is associated with increased risk of recurrence [ 1 , 2 ], higher expression of the colon cell cycle genes (such as CDC20, MCM2, MYBL2, CSE1L, MYC and MK167 ), was found to correlate with a lower risk of recurrence [ 6 ]. This is consistent with other reported evidence that cell cycle gene expression correlates with a good prognosis in colon cancer [ 31 - 35 ]. Garrity et al. reported only a weak correlation in colon cancer between MK167 levels and S-phase (the standard measure of proliferation) [ 32 ], indicating that expression of this gene may not signify rapidly dividing tumors. Instead, increased expression of these cell cycle checkpoint and control genes may represent tightened control of various stages of the cell cycle in response to DNA damage or misalignment of chromosomes during mitosis. APC is mutated in 80% of sporadic colon carcinomas [ 36 ] by either allelic loss or mutations in the multi cluster region (MCR), and there appears to be an interdependence of the two hits [ 37 ]. Homozygous deletions of APC are very rare, and residual APC activity is associated with difference biological characteristics depending on the type of mutation and its associated truncated protein (N-APC) [ 38 - 41 ]. For example, different APC mutations have been shown to result in various levels of β-catenin activation [ 38 , 39 , 42 , 43 ] and thus different growth advantages [ 40 ]. APC is also involved in chromosomal segregation, whereby it localizes to the ends of microtubules within the kinetochore and forms a complex with checkpoint proteins [ 35 , 44 ]. Some N-APC mutants have been shown to impair spindle checkpoint and contribute to mis-segregation of chromosomes [ 41 , 45 - 47 ]. With tighter cell cycle control and the ability (albeit a modest one) to undergo apoptosis or mitotic catastrophe in response to such mitotic errors, a tumor could reduce its abundance of aneuploidy and chromosomal instability (CIN) [ 48 ]. Given that CIN is associated with poor outcome [ 49 - 52 ] it could explain how high expression of cell cycle and checkpoint genes (such as CDC20, MCM2, MYBL2, CSE1L, MYC and MK167 ) were correlated with a lower risk of recurrence [ 6 ]. Since APC mutations are much rarer in breast cancer, it is likely that cell cycle genes (and specifically cell cycle control genes) do not harbor the same prognostic information as they do for colon cancer. This possible connection in colon cancer between different types of APC mutations, cell cycle control gene expression, CIN and prognosis warrants further investigation.

Conclusions In summary, after molecular markers and biological pathways had been identified in four independent clinical studies, the Onco type DX Colon Cancer Assay was developed to translate individual tumor biology into treatment planning. By requiring the individual components of RS to meet stringent analytical performance criteria, it was possible to clinically validate the Onco type DX Colon Cancer Assay and have confidence that every test performed in the clinical reference laboratory is done so using a sound and well controlled process.

Background The Onco type DX ® Colon Cancer Assay is a new diagnostic test for determining the likelihood of recurrence in stage II colon cancer patients after surgical resection using fixed paraffin embedded (FPE) primary colon tumor tissue. Like the Onco type DX Breast Cancer Assay, this is a high complexity, multi-analyte, reverse transcription (RT) polymerase chain reaction (PCR) assay that measures the expression levels of specific cancer-related genes. By capturing the biology underlying each patient's tumor, the Onco type DX Colon Cancer Assay provides a Recurrence Score (RS) that reflects an individualized risk of disease recurrence. Here we describe its analytical performance using pre-determined performance criteria, which is a critical component of molecular diagnostic test validation. Results All analytical measurements met pre-specified performance criteria. PCR amplification efficiency for all 12 assays was high, ranging from 96% to 107%, while linearity was demonstrated over an 11 log 2 concentration range for all assays. Based on estimated components of variance for FPE RNA pools, analytical reproducibility and precision demonstrated low SDs for individual genes (0.16 to 0.32 C T s), gene groups (≤0.05 normalized/aggregate C T s) and RS (≤1.38 RS units). Conclusions Analytical performance characteristics shown here for both individual genes and gene groups in the Onco type DX Colon Cancer Assay demonstrate consistent translation of specific biology of individual tumors into clinically useful diagnostic information. The results of these studies illustrate how the analytical capability of the Onco type DX Colon Cancer Assay has enabled clinical validation of a test to determine individualized recurrence risk after colon cancer surgery.

Competing interests All authors are, or have been (in the last 5 years), employees of Genomic Health Inc. and hold shares and/or stock. Patents relating to the content of this manuscript have been applied for by Genomic Health Inc. Genomic Health Inc. funded this project. Authors' contributions KMCL led the study and drafted the manuscript. CS and DW were responsible for study design and statistical analysis. JK coordinated and performed the analytical procedures. All authors read and approved the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/691/prepub Supplementary Material

Acknowledgements The authors gratefully acknowledge Maureen Cronin, Audrey Goddard, Mark Lee, Joffre Baker and Steve Shak for their review of the manuscript, Muriel Osugi, Daniel Paragas and Jinhui Yan for their assistance with the analytical procedures and Jackie Brooks for data management. This study was funded by Genomic Health Inc, Redwood City, CA.

CC BY

no

2022-01-12 15:21:44

BMC Cancer. 2010 Dec 23; 10:691

oa_package/32/b7/PMC3016296.tar.gz

PMC3016298

21129168

Background Up to now, different methodologies to select sperm have been described in the hope of selecting a viable sperm without - or with a low level of - DNA damage. Jakab et al. [ 1 ] was the first group that reported the use of a hyaluronic acid (HA) assay as a method to select a healthy sperm for use with ICSI. HA is a linear polysaccharide present in the extracellular matrix of cumulus oophorus around the oocyte that seems to play an important role in natural human fertilization. The use of this polysaccharide is based on the theory that hyaluronan is a major constituent of the cumulus oophorous matrix and may play a critical role in the selection of mature, functionally competent spermatozoa during in vivo fertilization. The principles of this assay are: (1) the expression of the protein HspA2, which indicates sperm maturation; (2) cytoplasmic membrane remodeling, which is responsible for the formation of sperm binding sites for the zona pellucida of oocytes and for HA binding sites. The Jakab group [ 1 ] suggested that immature spermatozoa present low HspA2 levels, fail to undergo cytoplasmic membrane remodeling and consequently are unable to bind to HA. Previous studies on sperm surface markers have demonstrated that HA-bound spermatozoa are mature and devoid of cytoplasmic retention, persistent histones, apoptotic markers and DNA fragmentation [ 1 - 4 ]. In addition, a normal frequency of chromosomal aneuplodies [ 1 ], normal Tygerberg strict [ 5 - 8 ] and normal nucleus morphology criteria [ 9 ] have been correlated positively with HA-bound spermatozoa. It was shown that binding to hyaluronic acid seems to be related to one or more conventional and one or more functional sperm tests, indicating that spermatozoa from patients with abnormal conventional semen parameters have a higher likelihood for multiple functional abnormalities [ 10 ]. In addition, freezing and thawing seems not alter the HA-binding properties of the spermatozoa [ 11 ]. On the other hand, another method to select healthy sperm proposed by Bartoov et al. [ 12 ] consists of using an inverted microscope equipped with high power Nomarski optics enhanced by digital imaging to achieve a magnification ≥6300×, sufficiently high to select spermatozoa according to their nuclear fine morphological integrity, and much higher than the magnification used habitually by embryologists in sperm selection for the ICSI procedure (200×-400×) or even the one employed in the routine semen exam (1000×). The use of high magnification motile sperm organellar morphology examination (MSOME) has revealed that the selection of a morphologically normal sperm nucleus before ICSI is an important factor in improving ICSI fertilization, embryo development, pregnancy rates [ 13 - 19 ] and the chance of having a healthy normal child [ 20 ]. Given this context, the present study aimed to evaluate the efficacy of the HA binding assay as a method to improve selection of motile spermatozoa with normal morphology at high magnification.

Methods Patients and sperm preparation Semen samples were obtained by masturbation after 2-5 days abstinence from 56 selected men with mean age 37.9 ± 5.9 year who attended the infertility investigation and treatment. This study received internal Institutional Review Board approval and all patients signed an informed consent. Inclusion criteria: samples with ≥20 × 10 6 spermatozoa/ml, ≥50% progressive motility and ≤1.0 × 10 6 leukocytes according to the World Health Organization criteria [ 21 ].The mean percentage of sperm normal form by MSOME analysis was 1.8 ± 2.5%, as in our previous experience [ 22 , 23 ]. The liquefied fresh semen samples were prepared by using swim up method which consisted of leaving the ejaculated sperm sample to migrate up to the human tubal fluid (HTF) medium with 10% human serum albumin (10% HSA) deposited on the top of the fresh semen sample, at 2:1 proportion, for 30 minutes at the temperature of 37°C. The supernatant, with motile spermatozoa was removed, mixed with 1 ml of HTF 10%HSA and centrifuged at 1200 g for 10 minutes. The final pellet was adjusted to a concentration of 2 × 10 6 motile sperm/ml and used for the sperm-HA binding assay. Hyaluronic acid binding assay For the determination and selection of the HA-bound spermatozoa, a PICSI dish for sperm selection (MidAtlantic Diagnosis - USA) was used. The three hyaluronan microdots existent in the PICSI dish were first hydrated by placing a 5 μl microdrop of mHTF 10% HSA on each microdot covered with mineral oil for 15 minutes. After this, 2 μl of prepared sperm was added into each microdrop covered by oil and incubated for at least 10 minutes to allow the spermatozoa to bind to the hyaluronan microdots. Following this period, two groups of selected spermatozoa were evaluated: Group I (HA-bound spermatozoa): spermatozoa that presented vigorous beating, i.e. those with an increased tail cross-beat frequency and were attached by the head to the hyaluronan microdots were collected and transferred to a 3 μl microdroplet containing 7% polyvinylpyrrolidone solution (PVP - medium Irvine Scientific-USA) presented in a sterile glass dish (FluoroDishTM-Word Precision Instrument, USA), under paraffin oil (Ovoil-100, Vitrolife, Goteborg, Sweden), using an ICSI micropipette (Humagen-USA) for morphological classification by MSOME (Figure 1 ) Group II (HA-unbound spermatozoa): motile spermatozoa with no head attached to the hyaluronan microdots were collected and transferred to a 3 μl microdrop of 7% PVP presented in another glass dish for morphological classification by MSOME. The entire HA procedure was performed at room temperature according to PICSI dish manufactures' guidelines (MidAtlantic diagnostic, Mount Laurel, NJ, USA), since the binding of sperm to the hyaluronan microdot is reduced at temperature above 30°C. The same embryologist, who is checked by the lab procedures in an internal lab control quality, carried out the entire HA analysis Classification of sperm morphology by MSOME Both HA-bound and HA-unbound spermatozoa were analyzed at high magnification. The microdroplets were placed on a microscope with an Uplan Apo 100× oil/1.35 objective lens previously covered by a droplet of immersion oil. In this manner, suspended motile spermatozoa in the observation droplet could be examined at high magnification using an inverted microscope (Eclipse TE 2000 U Nikon, Japan) equipped with high-power differential interference contrast optics (DIC/Nomarski). The total calculated magnification was 8400× (total magnification: objective magnification = 100X magnification selector = 1.0X video coupler magnification = 1.0X calculated video magnification = 84.00). Other technician blinded to HA classification performed all sperm morphology. A total of at least 250 spermatozoa/patient were evaluated and the percentages of the following sperm forms were determined: Normal spermatozoa A spermatozoon was classified as morphologically normal (Figure 2A ) when it exhibited a normal nucleus as well as acrosome, post-acrosomal lamina, neck, tail, besides not presenting a cytoplasmic droplet or cytoplasm around the head [ 12 ]. For the nucleus, the morphological state was defined by the form and content of the chromatin. The criterion for normality of nuclear form was a smooth, symmetric and oval configuration. Normal means for length and width were estimated as 4.75 ± 2.8 and 3.28 ± 0.20 μm [ 12 ], respectively, whereas the form classified as abnormal presented a minimum variation of 2SD in at least one of the axes (length: ≥5.31 or ≤4.19 μm, width: >3.7 or <2.9 mm). For rapid evaluation of nuclear form, a fixed, transparent, celluloid form of sperm nucleus fitting the criteria was superimposed on the examined cell (chablon construction based on ASTM E 1951-2 [ 24 ]). In the same manner, the form of the nucleus was considered normal if no extrusion or invagination of the nuclear chromatin mass had been detected (regional abnormality of nuclear form). Chromatin content was considered normal if vacuoles occupied no more than 4% of the nuclear area. A nucleus was considered normal if both nuclear form and chromatin content were normal. Abnormalities of nuclear form a-Spermatozoa with small or large oval nuclear forms (Figure 2B ) - Sperm cells exhibiting an abnormal but oval nuclear shape and a morphologically normal nucleus [ 16 ], content length ≤4.19 μm or≥5.31 μm. b-Spermatozoa with wide or narrow nuclear forms (Figure 2C ) - Sperm cells with non-oval, abnormal nuclear shapes, but with normal nuclear content [ 16 ]. Width: >3.7 or <2.9 μm. c-Spermatozoa with regional shape abnormality of nuclear form (Figure 2D ) - sperm cells with an extrusion or invagination of the nuclear mass [ 16 ]. Abnormalities of nuclear chromatin content a-Spermatozoa with vacuoles occupying >4-50% of the nuclear area (Figure 2E ) b-Spermatozoa with large nuclear vacuoles (Figure 2F ) - sperm cells with vacuoles occupying >50% of the nuclear area Sperm cells with a severe abnormality (such as: pin, amorphous, tapered, round or multinucleated head, double tail) easily identified at low magnification (200×-400×) were not assessed in this study. The abnormalities observed at high magnification, in both form and nuclear content, also presented normal acrosome, post-acrosomal lamina, neck, tail, and did not show a cytoplasmic droplet or cytoplasm around the head. Spermatozoids that presented more than one alteration were classified as having the most severe alteration [ 15 , 16 ] (small/large < wide/narrow < regional shape abnormality < with vacuoles occupying >4% to 50% of the nuclear area < with vacuoles occupying >50% of the nuclear area) Statistical analysis Data reported as means SD were analyzed using Instat version 3.0 (GraphPad Software, San Diego, CA, USA) on a Macintosh computer (Apple Computer In, Cupertino, CA, USA). Based on our previous experience in sperm morphology classification using high magnification [ 22 , 23 , 25 ], at least 200 motile spermatozoa per patient were selected. The percentages of sperm forms by MSOME and the HA-bound and HA-unbound spermatozoa were evaluated using Chi-square test. The significance level was set at P < 0.05.

Results The general characteristics of the men in the studied population are summarized in Table 1 . From a total of 16.592 of 56 patients (± 300 spermatozoa/patient), out of a total of 5579 HA-bound spermatozoa evaluated, 2.7% presented normal morphology, 1.6% had large/small nuclear form, 3.1% showed wide/narrow nuclear form, 4.7% presented regional disorder, 72.2% presented vacuoles on four to 50% of the nuclear area and 15.7% had large vacuoles. These results did not differ statistically ( P > 0.05) from the group of HA unbound spermatozoa (n = 11013) of which 2.5% presented normal morphology, 1.6% had large/small nuclear form, 2.7% displayed wide/narrow nuclear form, 4.4% possessed a regional disorder, 72.5% displayed vacuoles on four to 50% of the nuclear area and 16.3% had large vacuoles (Table 2 ).

Discussion In the last decade, both MSOME and the HA method have been demonstrated efficacious in selecting spermatozoa with high DNA integrity and normal morphology. Bertovitz et al. [ 16 ] have suggested that the sperm nuclear morphology, under high magnification, is one of the major parameters for sperm quality. They argued the hypothesis that vacuolization of the sperm nucleus reflects some DNA defect. In concordance with their hypothesis, Franco et al. [ 26 ] demonstrated an association between large vacuoles and the presence of DNA damage (fragmentation and denaturation) in the spermatozoa. The authors have demonstrated that spermatozoa with large vacuoles, selected at a high magnification (8400×) and directly assayed for DNA fragmentation by the Tunnel methodology, presented twice the chance of having DNA fragmentation compared to those spermatozoa with normal nuclei, i.e., without large vacuoles. These data have been confirmed by Garolla et al. [ 27 ], who have shown that the presence of nuclear vacuoles affects mitochondrial function, chromatin status, and aneuploidy rates. With respect to the HA assay, in a consecutive series of studies on sperm surface makers, Huszar's group was the first to argue that its assay permits the selection of mature spermatozoa with no DNA damage [ 1 - 3 , 28 , 29 ]. In contrast to this hypothesis, Petersen et al. [ 30 ] found no correlation between the HA-binding assay (PICSI) and a low degree of DNA damage. The HA-bound spermatozoa did not differ from HA-unbound ones as to DNA fragmentation (19.6% versus 21.4%, respectively). More recently, Parmegiani et al. [ 9 ] using a medium with HA (Sperm Slow-Medicult) to select HA-bound sperm has demonstrated an optimized ICSI outcome by favoring the selection of spermatozoa without DNA fragmentation. They studied 20 patients, and after analyzing 4000 HA-bound spermatozoa by the sperm chromatin dispersion (SCD) method, showed a reduction in rate of DNA fragmentation (5.3%) that was slightly less than one half the percentage obtained by collecting 4000 spermatozoa directly from the PVP (11%), as in a conventional ICSI procedure. In addition, Tarozzi et al. [ 8 ] have demonstrated a very low DNA fragmentation level by the Tunnel assay in both HA-bound sperm and sperm prepared by density gradient separation (1.17% and 1.59%, respectively). However, despite such low rates, their results underscore the ability of HA to select spermatozoa with higher DNA integrity. The discrepancies among these three studies [ 8 , 9 , 30 ] may be due to the different HA-binding methods used: PICSI dish, sperm slow medium and HA coated chamber. In addition, the low DNA fragmentation rate demonstrated in the Tarozzi et al. [ 8 ] study differs from the literature data [ 31 , 32 ], thus hindering the ability to draw any conclusion. On the other hand, a study evaluating the variations in the structural character and stability of the nuclear chromatin in morphologically normal human spermatozoa has demonstrated that even a normally shaped human sperm nucleus can be abnormal at the molecular or ultrastructural level [ 33 ]. Recent studies reveal a positive correlation between sperm morphology, assessed by the Tygerberg strict criteria, and HA binding [ 5 - 8 ]. The criteria by which the morphological normality of spermatozoa can be assessed depend on the resolution power of the optical magnification system utilized. Spermatozoa appearing as morphologically normal at 1000× magnification, as in Kruger morphology, may in fact carry various structural abnormalities that can only be observed at higher optical magnification (> 6000×). Oliveira et al. [ 25 ] found a significantly higher ( P < 0.0001) incidence, almost three-fold greater, of normal spermatozoa by the Tygerberg criteria (9.4%) than by MSOME (3.3%), and concluded that MSOME is a superior criterion for sperm morphology classification since it identifies vacuoles and chromatin abnormalities that are not found with the same precision by the Tygerberg criteria. More specifically, aim of the present study was to evaluate the capacity of the HA binding assay (PICSI dish procedure) to select motile spermatozoa with normal morphology. In this study we found no difference in the sperm morphology between HA-bound and HA-unbound spermatozoa, when investigated by MSOME. A total of 151 out of 5579 HA-bound spermatozoa, i.e., 2.7%, demonstrated a normal morphology compared to 2.5% of the 11013 HA-unbound sperm analyzed. Up to now, only one study has attempted to analyze HA-bound spermatozoa using MSOME as the morphological sperm classification method [ 9 ]. The authors assessed 1500 HA-bound spermatozoa and found a significantly higher percentage with nuclear normality (14.5%) than among spermatozoa collected from PVP (11%). In their study, a medium containing HA (SpermSlow-Medicult) was deposited into a glass dish and after the 15-minute sperm incubation period, HA-bound sperm morphology was evaluated at high magnification. The study's difference may be due to the different HA binding methods used, the PICSI dish, sperm slow medium and also the sperm morphology criteria: normal nucleus morphology [ 9 ] and normal spermatozoa morphology [ 30 ]. Moreover, in the present study we also evaluated for the first time, specific malformation of the nucleus shape (such as: small/long; wide/narrow; extrusion/invagination) and chromatin (medium and large vacuoles), and found that HA-bound and HA-unbound spermatozoa did not differ in any of these morphologies. On the other hand, since it was demonstrated significant positive correlation between HA binding and morphology [ 10 ], the normozoospermic samples [ 21 ] included here could have biased the outcome. In the study of Parmegiani et al. [ 9 ] 40% of their patients presented oligozoospermic semen samples, an additional fact that might have contributed for the different results. Other important point to emphasize is the possible influence of the sperm preparation on the outcome of HA binding. It is well defined in literature that semen sample preparations improve motility and morphology. Besides, the kind of semen preparation could impact the final sample quality. Different from the others papers in which no prepared sperm sample were carried out before hyaluronic binding assay, in this study we used only prepared semen samples. In addition, difference in sperm preparation between our work (swim up) and the study of Parmegiani et al. [ 9 ] (density gradient) could also contributed for contradictory results. The experimental design raises other questions. HA-bound is well characterized, however the HA-unbound sperm is a mixed population, as lack of binding may occur for several different reasons. The latter fraction may also contain sperm that are in the process of binding, but yet to be bound. The distribution 2:1 (200 HA-unbound spermatozoa: 100 AH-bound), by increasing the sample of HA-unbound spermatozoid, reduces this problem but without completely eliminate it. Still, it is unclear how the study results would be influenced by the lack normal spermatozoa in same sperm samples. In addition, it is not clear how sperm morphology is relevant in hyaluronic binding assay [ 10 ]. It is possible that some spermatozoa can be mature and contain normal DNA but morphologically abnormal [ 26 ]. In conclusion, the HA binding PICSI dish assay is not efficacious at improving the selection of motile spermatozoa with normal morphology at high magnification. However the same cannot be concluded for others HA assay methods.

Background The present study aimed to evaluate the efficacy of the hyaluronic acid (HA) binding assay in the selection of motile spermatozoa with normal morphology at high magnification (8400x). Methods A total of 16592 prepared spermatozoa were selected and classified into two groups: Group I, spermatozoa which presented their head attached to an HA substance (HA-bound sperm), and Group II, those spermatozoa that did not attach to the HA substance (HA-unbound sperm). HA-bound and HA-unbound spermatozoa were evaluated according to the following sperm forms: 1-Normal morphology: normal nucleus (smooth, symmetric and oval configuration, length: 4.75+/-2.8 μm and width: 3.28+/-0.20 μm, no extrusion or invagination and no vacuoles occupied more than 4% of the nuclear area) as well as acrosome, post-acrosomal lamina, neck, tail, besides not presenting a cytoplasmic droplet or cytoplasm around the head; 2-Abnormalities of nuclear form (a-Large/small; b-Wide/narrow; c-Regional disorder); 3-Abnormalities of nuclear chromatin content (a-Vacuoles: occupy >4% to 50% of the nuclear area and b-Large vacuoles: occupy >50% of the nuclear area) using a high magnification (8400x) microscopy system. Results No significant differences were obtained with respect to sperm morphological forms and the groups HA-bound and HA-unbound. 1-Normal morphology: HA-bound 2.7% and HA-unbound 2.5% (P = 0.56). 2-Abnormalities of nuclear form: a-Large/small: HA-bound 1.6% vs. HA-unbound 1.6% (P = 0.63); b-Wide/narrow: HA-bound 3.1% vs. HA-unbound 2.7% (P = 0.13); c-Regional disorders: HA-bound 4.7% vs. HA-unbound 4.4% (P = 0.34). 3. Abnormalities of nuclear chromatin content: a-Vacuoles >4% to 50%: HA-bound 72.2% vs. HA-unbound 72.5% (P = 0.74); b-Large vacuoles: HA-bound 15.7% vs. HA-unbound 16.3% (P = 0.36). Conclusions The findings suggest that HA binding assay has limited efficacy in selecting motile spermatozoa with normal morphology at high magnification.

Competing interests The authors declare that they have no competing interests. Authors' contributions CGP was responsible for designing and coordinating the study. All authors were responsible for data collection, data analysis, and data interpretation in the manuscript. CGP, JBAO and JF were responsible for the statistical work and for writing the manuscript. JF was responsible for reviewing the manuscript. All authors read and approved the final manuscript.

Acknowledgements The authors wish to thank the Research Support Group - UNESP for revising the English text.

CC BY

no

2022-01-12 15:21:44

Reprod Biol Endocrinol. 2010 Dec 3; 8:149

oa_package/65/54/PMC3016298.tar.gz

PMC3016299

21176181

Background Invasion of aerobic and anaerobic bacteria into the uterus is common in dairy cattle after parturition. The uterine bacterial colonisation is non-specific and includes a variety of pathogenic bacteria, e.g. Escherichia coli, Arcanobacterium pyogenes, Fusobacterium necrophorum and Prevotella species . Within 2-4 weeks after parturition most of the bacteria are eliminated spontaneously [ 1 ]. Depending on the pathogenicity of bacteria, bacterial load, responsiveness of the immune system and other risk factors, infection may result in acute metritis or, if persistent, in clinical endometritis. Clinical endometritis is defined as an inflammation of the endometrium identified by the presence of purulent or mucopurulent discharge detectable in the vagina later than 21 days postpartum (pp), and is not accompanied by systemic signs [ 2 ]. The first response of the innate immune system is the invasion of neutrophils to the site of bacterial infection, i.e. the endometrium. This increase of polymorphonuclear neutrophils (PMN) is used to define and diagnose subclinical endometritis [ 3 , 4 ]. Subclinical endometritis is described as an inflammation of the uterine wall in clinically healthy cows, i.e. in cows with no vulvar discharge. To detect subclinical endometritis, cytological samples can be obtained from the uterus by low-volume uterine lavage or by using the cytobrush technique [ 3 , 4 ]. Clinical as well as subclinical endometritis impair reproductive performance resulting in decreased conception rates and prolonged days open [ 3 - 5 ], leading to economic losses to the dairy farmers. However, the treatment of subclinical and clinical endometritis is still under discussion. Additional information about the underlying mechanism of inflammation is needed to understand the pathology of these diseases and to develop new therapeutic approaches in the future. The invasion of PMN determined by cytology describes only the cellular reaction to an inflammatory impact. Therefore, pro-inflammatory enzymes and cytokines have been suggested to be synthesized as mediators for inflammatory processes. A variety of factors are secreted to chemoattract other immune cells and to control the inflammatory response. Furthermore, the epithelium is also involved in the innate immune response against bacteria. Bovine endometrial epithelial cells express Toll-like-receptors that recognise specific bacterial components [ 6 ]. It is apparent that a complex and well-balanced network of cytokines is present in the uterus [ 1 , 7 ]. Bacteria that invade through the open cervix after parturition and persist in the uterus can disturb this fine-tuned system. Inflammatory responses may lead to perturbation of hypothalamic and pituitary function in turn which may contribute to impaired fertility of cows affected with clinical and subclinical endometritis [ 8 ]. Potential candidates involved in physiological and pathological events in the bovine endometrium are the chemokine CXC ligand 5 (CXCL5), interleukins 1β, 6 and 8 (IL1B, IL6, and IL8), tumour necrosis factor alpha (TNF), prostaglandin-endoperoxide synthase 2 (PTGS2), and haptoglobin (HP) [ 7 ]. Real-time RT-PCR has revealed that CXCL5 , IL1B , IL8 and TNF mRNA are significantly higher expressed in the endometrium of cows with subclinical or clinical endometritis than in healthy cows [ 7 ]. Therefore, the aim of the present study was to elucidate processes taking place in the bovine endometrium during the postpartum period on a molecular basis. Focus of the investigation was the mRNA expression of CXCL5 , PTGS2 , HP , IL1B , IL6 , IL8 and TNF . Specific objectives were to analyse the selected mRNA expression patterns in endometrial samples from primiparous cows: 1) in a time-course related manner starting on day 10 pp until day 45 pp and 2) in correlation to the proportion of PMN in the cytological samples and to the health status (healthy endometrium versus inflamed endometrium).

Methods Collection of endometrial samples during the postpartum period Holstein Friesian heifers (n = 5) were housed at the Clinic for Animal Reproduction, Freie Universität Berlin. All heifers calved without assistance. Fetal membranes were expelled within 12 h after calving. Rectal temperature was measured daily. None of the cows showed signs of acute metritis. All cows were examined by palpation of the uterus per rectum and vaginoscopy starting on day 10 pp. The first endometrial samples from each cow were collected on that day. Further examination of cows and sampling was performed on days 17, 24, 31, 38 and 45 pp, respectively. Endometrial samples for cytological assessment and mRNA analysis were collected from the uterus using the cytobrush technique [ 9 , 10 ]. Briefly, a cytobrush (Cytobrush plus GT, Medscand, Malmö, Sweden), screwed onto a 70-cm long rod, protected by a metallic catheter, was inserted into the uterine body via the cervix. This approach required only one passage through the cervix. Three different samples per cow were taken by introducing a new cytobrush through the protection catheter for each collection. Cells were collected by rotating the cytobrush clockwise while in contact with the uterine wall. Cell samples from two cytobrushs per animal and day were transferred into two separate reaction tubes containing RNAlater (Sigma, Deisenhofen, Germany). One cytobrush served for cytological analysis. The brush was rolled onto a sterile glass microscope slide, fixed and stained with the Hemacolor staining set (Merck, Darmstadt, Germany) following the manufacturer's protocol. A total of 300 cells were counted under a microscope (x 400 magnification) to determine the proportion of PMN. These data served to classify the health status of the uterus. Two criteria were applied to define an inflammation of the endometrium: presence of purulent or mucopurulent vaginal discharge detected by palpation of the uterus per rectum and vaginoscopy (criteria in the literature for the definition of clinical endometritis [ 2 ]) and/or percentage of more than 5% of PMN in the cytological sample (criteria in the literature for the definition of subclinical endometritis [ 3 , 10 ]). The uteri without signs for inflammation (no mucopurulent/purulent discharge and <5% PMN in the cytological sample) were defined as healthy. Blood samples were drawn on the days of cell collection. The serum concentrations of progesterone and 17β-oestradiol were determined using a chemiluminescence detection system (Immulite, Diagnostic Products, Los Angeles, CA, USA) on an automated analyser (Immulite 2000, DPC Biermann, Bad Nauheim, Germany) at the Institute for Zoo and Wildlife Research (Berlin, Germany). In addition, ovaries and uterus were examined by transrectal ultrasonography with a 5 MHz linear array transducer (Pie Medical, Maastricht, Netherlands), and results were documented. Total RNA extraction and reverse transcription Harvested endometrial cell samples were subjected to RNA extraction using Invisorb Spin Cell RNA Mini Kit (Invitek, Berlin, Germany) as previously described in detail [ 10 ]. Isolated total RNA was quantified photometrically at a wavelength of 260 nm. Quality of RNA was verified by loading 1 μL of total RNA onto a RNA 6000 Nano Chip using the Agilent 2100 Bioanalyzer (Agilent, Waldbronn, Germany) following the manufacturer's instructions. Single stranded cDNA was generated out of 0.5 μg total RNA using 200 U Moloney Murine Leukemia Virus reverse transcriptase (M-MLV RT) (Fermentas, St. Leon-Rot, Germany) and 3.75 μM random hexamers (Amersham Bioscienes, Freiburg, Germany) in a total volume of 40 μL [ 11 ]. The cDNA obtained was stored at -20°C until further analysis. Before reverse transcription, a genomic DNA removal by DNase digestion was performed. To monitor the absence of genomic DNA or other contaminations, reactions containing RNA samples without M-MLV RT or template (H 2 O) were carried out at the same time as negative controls. Real-time PCR For mRNA quantification of the selected factors, real-time PCR was performed as previously reported using the primer pairs listed in Table 1 [ 11 ]. In short, 1 μL cDNA was subjected to real-time PCR in a total volume of 10 μL containing 0.4 μM of each primer (forward and reverse) and 1x SensiMix (dt) (Quantace, Berlin, Germany). Amplification was done using the Rotor-Gene 3000 (Corbett Research, Mortlake, Australia) and the following parameters: a denaturation step at 95°C for 10 minutes, a three-step cycling amplification including denaturation at 95°C for 15 sec, annealing at the indicated temperature in Table 1 for 20 sec and extension at 72°C for 30 sec, a melting curve program (50-99°C) with continuous fluorescence measurement and a final cooling step to 40°C. Data acquisition was carried out at the end of each extension step at 72°C. A total of 45 cycles were performed (exception 18S rRNA: 25 cycles). A dilution series of the according PCR products with known concentrations was amplified simultaneously with the samples as a standard. Quantities of specific mRNA were calculated in comparison with the standard curves using Rotor-Gene 6.0 software. The obtained melting points of the amplified PCR products confirmed specific amplification. Negative controls containing no template (H 2 O) or non reverse-transcribed RNA did not show a signal verifying that obtained amplicons were not derived from contaminations or genomic DNA. Statistical analysis Gene expression of the investigated factors obtained by real-time PCR was normalised. To this end, absolute quantities of each gene were divided through the quantities of 18S rRNA in each sample. Means were obtained from the independently processed duplicates for each animal and collection time. Normalised values were analysed by the Kruskal-Wallis-H-test. When this test indicated significant differences, the Mann-Whitney-U-test was used to compare the samples from day 17 pp with day 31 pp. These points in time were chosen because of the results obtained from cytological analysis (day 17 pp: high PMN content; day 31 pp: first point in time of low PMN content). Furthermore, expression patterns of the selected factors were analysed by comparing samples from an inflamed endometrium with samples obtained from cows with a healthy endometrium using the Mann-Whitney-U-test. The Pearson's test was used to calculate the correlation between each of the factors as well as between the percentage of PMN and pro-inflammatory factors. All statistical calculations were performed by using SPSS for windows version 18.0 (SPSS, Chicago, IL, USA). Values of P < 0.05 were considered to be significant.

Results PMN content in cytobrush samples and clinical observations The assessment of the cytological samples revealed that the PMN content changed in a time-dependent manner. In detail, the PMN content, which was > 5% during the postpartum period (on day 10 pp or day 17 pp), declined to percentages < 5% on day 31 pp (Figure 1 ). During the first 2-3 weeks of the puerperium, individual differences were wide, especially on day 10 pp. Some animals had a content of less than 5% PMN in their cytological sample, while others revealed values higher than 30%. In general, however, a PMN content higher than 5% was associated with vaginal discharge (except for two samples) (Table 2 ). In the later postpartum phase (day 31-45 pp), most animals showed neither vaginal discharge nor an increased proportion of PMN. One cow (animal 5) showed mucopurulent vaginal discharge from day 10 pp until day 38 pp (diagnosed from day 31 pp on as clinical endometritis) which resulted in an elevated PMN level on day 45 pp and a persisting subclinical endometritis. Out of 29 samples obtained from the cows over the sampling period, 16 samples were classified as "healthy", and 13 samples were considered obtained from an "inflamed endometrium". The observations from ultrasonography, adspection and palpation as well as the serum concentrations of progesterone (P 4 ) and 17β-oestradiol (E 2 ) were used to allocate the cows in the postpartum period to distinct oestrous cycle phases. Only cow 2 was in the luteal phase on day 45 pp (corpus luteum detected and >1.5 ng P 4 /ml serum), and none of the cows were in oestrus at sample collection (signs of oestrus and >2 pg E 2 /ml serum). Extracted RNA from all samples harvested by the cytobrush method was of a good quality, evaluated by standard methods (data not shown) as previously demonstrated [ 10 ]. The two independently processed samples of each cow revealed a similar mRNA expression pattern for each investigated factor (median of variation 30%). Expression of CXCL5 mRNA The expression pattern of CXCL5 mRNA in bovine endometrial samples varied during the postpartum period (Figure 2A ). In general, CXCL5 mRNA expression was higher in samples obtained from an inflamed endometrium compared with the healthy endometrium (P < 0.05; mean of normalised expression ± S.E.M. × 10 -6 : 74 ± 23 and 25 ± 8, respectively). The mean expression of CXCL5 mRNA was significantly higher on day 17 pp than on day 31 pp. The endometrial sample of one animal (cow 5) showed an increase of CXCL5 mRNA expression on day 45 pp, which was accompanied by an elevated PMN level (Figure 1 ). The range of CXCL5 mRNA expression in endometrial cell samples during the postpartum period was 64-280 fg/μg total RNA. Expression of IL1B , IL6 and IL8 mRNA The expression patterns in the luminal endometrium of IL1B and IL8 were almost identical during the postpartum period, in contrast to IL6 (Figure 2B-D ). However, the mRNA expression of these selected interleukins was significantly higher (P < 0.05) in samples from an inflamed endometrium than in samples from a healthy endometrium (mean of normalised expression ± S.E.M. × 10 -6 ; IL1B : 1594 ± 776 and 178 ± 168; IL6 : 3.7 ± 1.4 and 2.3 ± 1.7; IL8 : 693 ± 322 and 57 ± 45, respectively). In detail, an increase of IL1B mRNA expression was observed from day 10 pp to the highest levels on day 17 pp (Figure 2B ). Thereafter, a significant decrease of IL1B mRNA content was noted up to day 31 pp. After this point, IL1B mRNA expression was very low or not detectable. In contrast, mRNA expression of IL6 was inconsistent (Figure 2C ). Average IL6 mRNA contents were significantly higher on day 17 pp compared with day 31 pp (P < 0.05). The IL6 mRNA content declined after day 17 pp. However, in endometrial samples of two cows an increased IL6 mRNA expression was detected on day 45 pp. The PMN content in these cows was <5%. Pro-inflammatory IL8 mRNA was detected in the early phase of the postpartum period with a distinct expression pattern (Figure 2D ). As observed for IL1B , expression of IL8 mRNA was significantly higher in samples obtained on day 17 pp compared with samples collected on day 31 pp. The expression level of IL8 was 20-100-fold lower from day 31 pp onwards than on day 17 pp. One cow (animal 3) showed an approximately fourfold higher expression of all investigated interleukins on day 17 pp compared with the other cows. The absolute range of mRNA expression for the selected interleukins differed: the highest range was noted for IL8 (400-2,000 fg/μg total RNA) followed by IL1B (80-400 fg/μg total RNA). IL6 expression showed the lowest range, with 4-80 fg/μg total RNA. Expression of TNF mRNA Transcripts for TNF were detected in all samples with a time-dependent variation (Figure 3A ). The higher expression observed on day 17 pp was significant compared with day 31 pp (P < 0.05). In the late postpartum period, a slight increase of TNF mRNA expression was noted in each cow on day 45 pp compared with day 31 pp or 38 pp. In addition, TNF mRNA expression was dependent on the health status. A significantly higher TNF mRNA content was observed in samples from cows with an inflamed endometrium compared with samples from healthy cows (P < 0.05; mean of normalised expression ± S.E.M. × 10 -6 : 34 ± 6.7 and 24 ± 14, respectively). The range of TNF mRNA expression in luminal endometrium samples during the postpartum period was 40-160 fg/μg total RNA. Expression of PTGS2 and HP mRNA Endometrial mRNA expression of PTGS2 was found in all samples from day 10-45 pp (Figure 3B ). The mean PTGS2 mRNA expression declined from day 10 pp to lower levels on day 31 pp and remained on such low expression levels in the late puerperial phase. As observed for the cytokines, this enzyme was also higher expressed on day 17 pp compared with day 31 pp (P < 0.05). One animal (cow 4) showed a dramatic increase of PTGS2 mRNA expression on day 45 pp. This was not correlated with an increase of the PMN content but accompanied by an increased IL6 mRNA expression. However, the health status correlated with the PTGS2 expression, showing higher PTGS2 transcript contents in samples from cows with an inflamed endometrium compared with samples from a healthy endometrium (P < 0.05; median of normalised expression × 10 -6 : 6.5 and 0.5, respectively). The range of PTGS2 mRNA expression in endometrial cell samples during the postpartum period was 160-800 fg/μg total RNA. In contrast, the smallest and lowest range of the mRNA expression in the bovine endometrium for all investigated factors in the postpartum period was noted for HP (Figure 3C ). The absolute amount of HP mRNA expression in luminal endometrium samples was in the range of 4-16 fg/μg total RNA. An increase was observed from day 10 pp to day 17 pp. Furthermore, the HP mRNA content in the endometrium declined significantly thereafter to about fourfold lower levels on day 31 pp (P < 0.05). There was a tendency towards a slight increase of HP mRNA expression on day 45 pp. In contrast to PTGS2 , no significant difference of the HP transcript amount was detected between samples from cows with an inflamed endometrium versus healthy cows. Correlation analysis Significant correlations between the mRNA expression of several factors in endometrial cell samples harvested during the puerperium are listed in Table 3 . Especially IL8 mRNA contents showed a high correlation with the mRNA expression of all factors except PTGS2 . The highest correlation was noted between IL1B and IL8 mRNA expression. HP as an acute phase protein showed a high correlation in its local mRNA expression to the mRNA expression of the PMN chemoattractant factors IL1B and IL8 . PTGS2 transcript content was only significantly correlated with IL6 mRNA expression. In addition, mRNA expression of CXCL5 , IL1B , IL8 and HP correlated significantly with the proportion of PMN (Table 3 ). In contrast, a correlation was not observed between the number of PMN and the mRNA expression of TNF , IL6 and PTGS2 .

Discussion The present study demonstrates a time-dependent presence of PMN in endometrial samples collected until week 8 postpartum. A peak in the proportion of PMN was detected between day 10-24 pp. Furthermore, the percentage of PMN in the cytological samples correlated with an increased mRNA expression of several pro-inflammatory factors. These findings are supported by the literature. The prevalence of subclinical endometritis, defined by an increased proportion of PMN, in a time-dependent manner has been previously reported by other authors. Gilbert et al . [ 3 ] diagnosed 90-100% of the cows examined two and four weeks after parturition by uterine lavage and subsequent cytological assessment as affected with subclinical endometritis. A prevalence of subclinical endometritis of 41% was still noted after two months [ 3 ]. In a study using both the cytobrush and lavage techniques for the diagnosis of subclinical endometritis, the percentage of PMN decreased from day 20-33 pp to day 34-47 pp [ 9 ]. Our study, although conducted with only five cows, is the first that describes an initial increase of PMN from day 10 pp to day 24 pp, followed by a decrease to percentages about 0% on day 31 pp. This time course of PMN content is in accordance with findings describing a recovery from endometritis or infection within four weeks postpartum [ 5 , 12 , 13 ]. In the present study, all cows had an increased number of PMN in the uterus in the postpartum period. Thus, it can be hypothesised that a PMN influx is a physiological response to a bacterial infection of the uterus postpartum [ 14 ]. Our findings for the presence of PMN and the mRNA expression analysis support previous reports that the diagnosis of clinical or subclinical endometritis should be performed after day 24 pp [ 15 , 16 ]. There was a significant correlation between the presence of PMN and mRNA expression of IL1B , IL8 and CXCL5 . These data support evidence that PMN are chemoattracted by substances secreted by epithelial cells and the latter substances are suggested to act together directing PMN to the site of inflammation [ 17 ]. In the present study, mRNA expression analysis revealed a significant expression peak on day 17 pp compared with contents on day 31 pp for all investigated factors. This is in agreement with higher expression of CXCL5, IL1B, IL8 and TNF described for several bacterial infections [ 18 - 20 ], indicating that the immune system is activated to eliminate bacteria. Recently, it was suggested that CXCL5 , IL1B , IL8 and TNF may be suitable markers for the detection of subclinical endometritis because mRNA levels were elevated compared with samples from healthy cows [ 7 ]. Interestingly, significant differences in the expression of the selected factors did not exist between samples derived from cows with subclinical and clinical endometritis [ 7 ]. This may be explained by the fact that both diseases are an inflammation of the endometrium, only with a different clinical appearance. The similar expression pattern for IL1B , IL8 and CXCL5 is in accordance with data from the literature stating that IL1B stimulates the production of CXCL5 and IL8 [ 17 , 21 ]. However, it has to be pointed out that the expression of IL1B must occur earlier to stimulate the expression of IL8 and CXCL5, but a weekly sampling can not provide such data. In addition, IL8 and CXCL5 are responsible for the recruitment of PMN [ 22 , 23 ]. In an in vivo bovine model, it has been demonstrated that peak numbers of uterine PMN were attracted 6 h after intrauterine application of IL8 [ 24 ]. Furthermore, an infiltration of neutrophils into the bovine uterus around ovulation has been reported [ 25 , 26 ]. This was supported by finding an increased mRNA expression of the PMN chemoattractant factors IL1B , IL8 and CXCL5 in the endometrium around ovulation [ 7 ]. Therefore, an up-regulated expression of these factors can lead to a subsequent increased influx of PMN. The mRNA expression pattern of all investigated factors in the endometrial samples of this study was not affected by the presence of steroid hormones because concentrations of P4 and E2 were not elevated before day 45 pp. The cytokine TNF was also expressed in a time-related pattern during the postpartum period. The highest levels on day 17 pp and day 24 pp and the following decline may be explained by the pro-inflammatory response that has been described in the literature [ 20 ]. TNF may act as a potent regulator of PGE 2 as well as of PGF 2α secretion [ 27 ]. TLR stimulation via bacterial lipopolysaccharids leads to a switch from PGF to the PGE series [ 8 ]. This has been suggested as a pathogenic mechanism after uterine bacterial infection which could result in an extended luteal phase and thus contribute to an impaired reproductive performance. In this context, PTGS2 , one of the key-enzymes of the PG synthesis, was up-regulated in the present study in its expression in parallel to TNF . Interestingly, HP as an acute phase protein showed a similar local expression pattern as the pro-inflammatory interleukins. In a former study, HP was found unspecific for the detection of subclinical endometritis [ 7 ]. A recently published study describes the expression of several pro-inflammatory cytokines including IL1A , IL1B and IL6 in the first week postpartum in endometrial biopsies [ 28 ]. A higher expression of IL1 correlated with decreased fertility parameters. That study compared healthy cows (n = 4) without uterine disorders during the postpartum period and conceiving at the time of first insemination (day 59-74 pp) with cows (n = 4) showing uterine diseases and failing to conceive within 200 days pp. These findings are complementary to the results of the present study and enhance the understanding of events occurring during the postpartum period. The heifers in our study were not selected specifically, and the number of animals did not allow a comparison of reproductive performance data.

Conclusions In conclusion, the results of the present study showed that inflammatory cytokines and acute phase proteins are expressed in the bovine endometrium in a time-related manner during the postpartum period with a peak of expression around day 17 pp. Therefore, the evaluation of the expression patterns of such candidate genes reveals more information than only determining the percentage of PMN to judge the severity of inflammation. This knowledge may give further information for new therapeutic strategies, which may be monitored by a local expression analysis of these selected biomarkers. However, this study is only a descriptive and preliminary one. Further investigations with a greater number of cows are required to follow the time-dependent expression profile of such candidate genes in healthy cows, cows developing a subclinical endometritis and in cows showing a clinical endometritis.

Background Inflammatory processes and infections of the uterine wall must be accepted as a physiological event in dairy cows after calving. This might result in clinical or subclinical endometritis which is assumed to impair reproductive performance in the current lactation. Several cytokines and acute phase proteins have been discussed as local and systemic mediators of these inflammatory processes. The aim of the present study was to investigate the endometrial mRNA expression of the chemokine CXC ligand 5 ( CXCL5 ), interleukin 1β ( IL1B ), IL6 , IL8 , tumour necrosis factor alpha ( TNF ), prostaglandin-endoperoxide synthase 2 ( PTGS2 ) and haptoglobin ( HP ) in the postpartum period. Methods Endometrial samples were obtained from primiparous cows (n = 5) on days 10, 17, 24, 31, 38 and 45 postpartum (pp) using the cytobrush technique. Cytological smears were prepared from cytobrush samples to determine the proportion of polymorphonuclear neutrophils (PMN). Total RNA was extracted from endometrial samples, and real-time RT-PCR was performed. Results A time-dependent mRNA expression of the investigated factors was found for the course of the postpartum period. In detail, a significantly higher expression of these factors was observed on day 17 pp compared to day 31 pp. Furthermore, the proportion of PMN peaked between days 10-24 pp and decreased thereafter to low percentages (< 5%) on day 31 pp and thereafter. In addition, CXCL5 , IL1B , IL8 and HP mRNA expression correlated significantly with the proportion of PMN (P < 0.05). A significantly higher CXCL5 , IL1B , IL6 , IL8 , PTGS2 and TNF mRNA content was observed in samples from cows with an inflamed endometrium compared with samples from cows with a healthy endometrium (P < 0.05). Conclusions These results show that inflammatory cytokines and acute phase proteins are expressed in the bovine endometrium in a time-related manner during the postpartum period, with a significant expression peak on day 17 pp as a possible mucosal immune response in the uterus. The evaluation of the expression patterns of such candidate genes may reveal more information than only determining the percentage of PMN to judge the severity of an inflammation.

Competing interests The authors declare that they have no competing interests. Authors' contributions CG, MD, RE and WH planned and devised the experiments. CF conducted the animal study, collected the samples and performed cytological analysis. CF and CG performed the molecular biology analysis. CG, MD, RE and WH wrote the manuscript. All authors read and approved the manuscript.

Acknowledgements We are grateful to Christoph Holder, Annekathrin Lincke and Doris Forderung for technical assistance, and to Rose Schmitz for helpful suggestions in performing the statistical analysis. This work was partly supported by the DFG (Ei 296/10-3) and by the BMBF (FUGATO plus - REMEDY).

CC BY

no

2022-01-12 15:21:44

Reprod Biol Endocrinol. 2010 Dec 22; 8:152

oa_package/c7/2e/PMC3016299.tar.gz

PMC3016300

21176150

Introduction The purpose of the present review is to give a short overview of how engineered nanoparticles (ENPs) can translocate from the respiratory tract to the circulation, pass the blood-brain-barrier (BBB), affect the brain, and to discuss possible adverse health effects and associated risks. We also suggest that there is a need for focused research to support risk assessment. This research should use standardized and proper methods and experimental designs including the selection of the right in vitro and/or in vivo models, controls, ENP characteristics, doses, etc. Nanoparticles (NPs) can be generated through both natural (e.g., combustion by-products, volcano eruption etc.) and synthetic processes. In the present article, we focus on engineered nanoparticles and their unintended exposure of the CNS. In principal, researchers have agreed to use the term nanoparticle if the material size is smaller than 100 nm in three dimensions and are singular particles; although different terms are still used in the literature, like nanosized materials, ultrafine particles (UFP), engineered nanomaterials, manmade nanoparticles [ 1 ]. This shows that the expression "nanomaterial" is related to the size dimension only, but not to the material itself which can contain any kind of substance. This is relevant from different perspectives, e.g. in political discussions and decisions but also for dosimetry aspects. For the latter, it is important to characterize the kind of the nanomaterial, to define concentration(s), establish dose response relationships etc. Dosimetry is furthermore necessary for risk estimation and for the establishment of thresholds and/or limit values. The general use of the term nanomaterial does not say much about the chemical conditions. Therefore, the physico-chemical properties have to be known for exposure calculations, including size, shape and composition of the material.

Conclusion The aim of the present study is to assess if there is a risk to especially the CNS after unintended exposure to inhaled ENPs. A possible risk has two components, viz. exposure and hazard. Regarding exposure, there are at present very few if any data on exposure of the general public to either acute high dose exposure or on chronic exposure to low dose levels of air-borne ENPs. It is furthermore unlikely, with the exception of possibly a few occupational situations that acute high dose exposures would happen. The risks from such exposures for damaging CNS effects is thus probably very low, irrespective of any biological hazards that ENPs could constitute. The situation is more complicated regarding chronic exposures, at low doses. The long term accumulation of ENPs can not be excluded. However, we do not have access to exposure data for the general public regarding ENPs. We also know that translocation to the brain via respiratory organs and the circulation is very low, even in cases where ENPs have such surface modifications as to be able pass the BBB. At higher concentrations, ENP can possibly enter the olfactory bulb via the olfactory nerve, and then possibly distribute to other areas of the brain. It is also shown in both in vivo and in vitro studies that several types of ENP have various types of biological effects. The relevance of these data is unclear. However, a possibility remains that chronic exposures, and/or biopersistent ENPs, can influence processes within the brain that are triggering or aggravating pathological processes. In general, the present state of knowledge is unsatisfactory for a proper risk assessment in this area. Improvements of the study qualities as well as increased number of relevant studies are strongly recommended.

There are certain concerns regarding the safety for the environment and human health from the use of engineered nanoparticles (ENPs) which leads to unintended exposures, as opposed to the use of ENPs for medical purposes. This review focuses on the unintended human exposure of ENPs. In particular, possible effects in the brain are discussed and an attempt to assess risks is performed. Animal experiments have shown that investigated ENPs (metallic nanoparticles, quantum dots, carbon nanotubes) can translocate to the brain from different entry points (skin, blood, respiratory pathways). After inhalation or instillation into parts of the respiratory tract a very small fraction of the inhaled or instilled ENPs reaches the blood and subsequently secondary organs, including the CNS, at a low translocation rate. Experimental in vivo and in vitro studies have shown that several types of ENPs can have various biological effects in the nervous system. Some of these effects could also imply that ENPs can cause hazards, both acutely and in the long term. The relevance of these data for risk assessment is far from clear. There are at present very few data on exposure of the general public to either acute high dose exposure or on chronic exposure to low levels of air-borne ENPs. It is furthermore unlikely that acute high dose exposures would occur. The risk from such exposures for damaging CNS effects is thus probably very low, irrespective of any biological hazard associated with ENPs. The situation is more complicated regarding chronic exposures, at low doses. The long term accumulation of ENPs can not be excluded. However, we do not have exposure data for the general public regarding ENPs. Although translocation to the brain via respiratory organs and the circulation appears to be very low, there remains a possibility that chronic exposures, and/or biopersistent ENPs, can influence processes within the brain that are triggering or aggravating pathological processes. In general, the present state of knowledge is unsatisfactory for a proper risk assessment in this area. Crucial deficits include lack of exposure data, the absence of a proper dose concept, and that studies often fail in adequate description of the investigated ENPs.

ENPs and dose For the calculation of the biological or chemical reactivity of the material, knowledge about the physico-chemical properties, the number of molecules on the surface of the nanosized material is needed, as well as the number of particles per cell. The number per cell is important to determine the effective dose , since nanoparticles have larger surface area than the corresponding bulk material including a higher number of molecules on the surface which can interact with the biological material, and their larger number per mass allows their dispersion into more cells. Information about the physico-chemical properties including size and shape are important in order to estimate ENP specific effective dose as well. In in vitro studies it is difficult to estimate this dose because NPs diffuse, settle, and agglomerate in cell culture media depending on different factors like media density and viscosity, particle size, shape, charge and density. Teeguarden et al. [ 2 ] developed a particokinetic model to estimate cellular dose in vitro considering different factors like the dynamic precipitation rate in cell culture media which depends on particle size and follows more the Brownian motion than gravitation. Another important dose measure might be the relative biological effectiveness (RBE) . If the physico-chemical properties and the number of molecules on the surface of the nanosized material are known, as well as the number of particles per cell, weighting factors might be introduced as in dose calculation for ionizing radiation, where RBE is calculated as a function of the quality of the radiation. Thus, for the same absorbed dose, alpha radiation is 20 times more biologically potent than x-rays or gamma radiation. Accordingly, RBEs can then be calculated for specific nanomaterials. The RBEs would then be dependent of the material itself and the number of internalized/taken up nanoparticles per cell. The so called biological effective dose (BED) concept describes oxygen radical generation, as an indirect measure (or marker) for BED [ 3 ] considering the physico-chemical properties of the material. This concept is very useful. However, specific cell type dependent redox potential capacities have also to be considered. This is reviewed by Valco et al. [ 4 ] where pH dependent effects are shown to be due to the specific redox capacity of the cell type in question, with cell type dependent effects on e.g. cell cycle and developmental events. In addition, the work by Sohaebuddin et al. [ 5 ] shows such cell type dependent effects of various ENPs. Furthermore, for dose calculations relevant for both chronic and for acute exposure, the dose rate has to be known, which includes the time factor. To determine the retention time (how long an ENP is present in a cell or a body) of a certain ENP, knowledge about the physico-chemical properties, but also about the biological deposition time in each site (deposition and retention time are depending on deposition site) is needed. In other words, knowledge about site dependent retention time and bioavailability is needed to calculate the time factor for dose rate. Other factors that will affect the dose rate are that certain ENPs are biodegradable with a relatively short half-life whereas others will not be metabolized within the body. In addition, some ENPs may be excreted, whereas others may accumulate over time. The present knowledge regarding the different dose concepts relevant for nanomaterials is however very limited, with possible exception of data from a few in vitro studies. Relevant data from in vivo situations is mainly absent. Drug delivery systems and the blood-brain-barrier ENPs have the potential to revolutionize medicine because of their ability to reach and to affect target organs and tissues, even "as distant" as tumours in the brain, at the molecular and cellular levels. Medical and pharmacological research is focused on applications of nanosized materials, whereas side effects associated with their use are generally not taken into consideration. In fact, the knowledge about potential toxicity of ENPs is far from comprehensive [ 6 , 7 ]. Drug delivery systems or nanocarriers should and may overcome solubility or stability issues for the drug, and minimize drug induced side effects. However, the nanomaterials themselves can also induce significant toxic effects (for reviews see [ 8 , 9 ]). Besides the chemical properties, this can be due to their electric, optical, and magnetic properties that are related to physical dimensions, but also the surface of the material can be involved in catalytic and oxidative reactions which themselves can induce cytotoxicity. This toxicity can be greater than that of a bulk material because the surface area-to-volume ratio for nanomaterials is much greater. Moreover, some nanomaterials contain metals or compounds with known toxicity, and thus the breakdown of these materials could elicit similar toxic responses. A number of questions pertaining to the safety of nanomaterials in this context are thus obvious. What is the ultimate fate of the drug delivery systems/nanocarriers, and their components within the body? What happens with those which are not bio-degradable and those which are functionalized, like carbon nanotubes, or coated with different agents? Further on, what are the consequences after long term exposure? The blood-brain barrier (BBB) protects the central nervous system from potentially harmful xenobiotics and endogenous molecules (for review see [ 9 ]). The BBB, formed by brain capillary endothelial cells linked together by tight junctions, together with adjacent processes from astrocytes, restricts the transfer of most substances from the bloodstream into the brain. Therefore, substances may gain access to the central nervous system by (lipid-mediated) free diffusion or potentially by receptor-mediated endocytosis. Since tight junctions in the BBB have a gap of only 4-6 nm, it has been suggested that nanoparticles pass through the endothelial cell membrane rather than via inter-endothelial junctions [ 10 ]. It has been shown that nanoparticles from the blood circulation may influence endothelial cell membrane integrity and/or disrupt the BBB [ 11 ], and may induce vesicular transport to gain access into the CNS (see below). Moreover, it seems to be accepted that nanoparticles can induce oxidative stress leading to the generation of free radicals that could disrupt the BBB and cause certain dysfunctions. It is also known that nanoparticles without a surfactant coating are mainly internalized by phagocytes and are thus unable to reach the brain in desirable quantities, therefore almost no pharmaceutical can reach the brain tissues by administering it with uncoated nanoparticles [ 12 ]. (See Figure 1 for a description of how ENPs can enter the cell and exert different actions) However, surface modifications of nanoparticles are presently intensely studied for nanomedicinal applications like diagnosis and therapy aiming to influence the target-oriented pharmacokinetic behaviour of nanocarriers. Nanocarriers require surface modifications or other forms of functional modifications for receptor-mediated transport through the brain capillary endothelium to deliver drugs to the central nervous system. Different approaches to obtain suitable modifications are discussed [ 13 ]. Kreuter et al. [ 14 ] demonstrated that polysorbate 80-coated polybutylcyanoacrylate (P80-PBCA) nanoparticles can deliver the peptide "dalargin" into CNS to induce its analgesic effects. Coating with alternative surfactants did not produce the expected effects [ 15 ]. P80-PBCA nanoparticles can thus deliver drugs to the brain, however these nanoparticles seem to have limitations due to their potential toxicity [ 16 ]. Thus, Calvo et al. [ 17 ] showed a drastic increase in sucrose permeability (as a sign of BBB permeabilization) of the BBB in rats following intravenous administration of P80-PBCA. Other functional modifications of ENPs include for example the conjugation of cell surface ligands or of antibodies. In order to facilitate the crossing of the BBB, apolipoprotein-E coating of nanoparticles for LDL receptor mediated endocytosis in brain capillaries has also been discussed [ 15 , 18 ]. Apolipoprotein, especially ApoE combined with nanoparticles, behaves like low-density lipoprotein (LDL) in the sense that LDL receptor-mediated transcytosis enhances the drug delivery together with nanoparticles across the BBB and is very effective in drug delivery [ 19 , 20 ]. Another important factor in penetration of the BBB by nanoparticles is their electrostatic charge. Cationic charged molecules occupy anionic areas at the BBB endothelium [ 21 ] and increase the endothelial cell permeability [ 22 ]. In in vitro studies, the cationized nanoparticles translocate more readily to the brain compared with anionic or neutral nanoparticles [ 23 ]. Thus, both the size and the charge of colloidal drug carriers are important factors in determining drug or nanoparticle delivery across the BBB or in brain parenchyma [ 24 ]. However, there are little in vivo data regarding brain permeability of cationized nanoparticles. One exception is the work by Lockman et al. [ 25 , 26 ] who investigated the effect of neutral, anionic and cationic charged ENPs on the blood brain barrier (BBB) integrity and permeability in situ with rat brain perfusion. Neutral ENPs and low concentrations of anionic ENPs were found to have no effect on BBB integrity, whereas high concentrations of anionic ENPs and cationic ENPs disrupted the BBB structure. Especially cationic NPs displayed an immediate toxic effect. It has been suggested that the BBB disruption by cationic nanoparticles may be due to opening of inter-endothelial routes, i.e., widening of tight junctions [ 21 ]. The structural changes in tight junction may occur without any decrease in the high electrical resistance of the cerebral endothelium. Since cationic nanoparticles are possibly more neurotoxic than other forms, it seems that neutral and low concentrations of anionic nanoparticles are better suited as colloidal drug carriers for enhanced delivery to the brain [ 26 ]. Kim et al. [ 27 ] synthesized silica-coated magnetic nanoparticles containing rhodamine B isothiocyanate within a silica shell of controllable thickness (50 nm). After intraperitoneal administration into mice (10 mg/kg), ENPs were detected in the mouse brain, indicating a blood-brain barrier penetration without disturbing its function or producing apparent toxicity. In another study they demonstrated that poly(methoxypolyethyleneglycol cyanoacrylate-co-hexadecylcyanoacrylate) (PEG-PHDCA) nanoparticles have the capacity to diffuse through the blood-brain barrier after intravenous administration. The authors suggested that the LDL receptor-mediated pathway was involved in the endocytosis mechanism [ 28 ]. It has to be pointed out, that nanoparticles administered intravenously are rapidly cleared from the blood stream by the mononuclear phagocyte system and mainly accumulate in liver and spleen [ 29 ]. Specially prepared ENPs with surface modifications using either PLA, PLGA or PEG, or combinations of all, as well as functionalized pegylated PLA/PLGA nanoparticles seem to offer possibilities for drug delivery to the brain. It seems that these special ENPs are more biocompatible and are having a better safety profile, and can furthermore pass the BBB without inducing substantial toxicity even at very high doses (440 mg/kg in mice, [ 13 , 30 ]. This suggests that the possibility for ENP uptake in the CNS is very complex and therefore it is not likely that inhaled or ingested ENPs are reaching the CNS in significant amounts. Furthermore, many NPs are agglomerates or covered by a protein corona, undergoing a fast metabolism and/or excretion. It is known that certain ENPs need surface coating to be internalized by phagocytes which in turn induces oxidative stress by the generation of free radicals. The question arises how long this oxidative stress is present within the CNS and to if and what kind of dysfunctions it is leading? It is very likely that the effects (chronic or acute) are dose dependent, therefore a dose definition is strongly needed for the purpose of risk assessment. Translocation of nanoparticles from the respiratory tract to the CNS Since inhalation is one of the main portals of ENP entry into the body and the majority of knowledge is available on that field we focus on uptake of ENPs in the lungs by inhalation or instillation followed by retention and distribution to secondary organs. Epidemiological and toxicological studies have shown that inhalation and subsequent deposition of ambient ultrafine particles (UFP, < 100 nm) into the lungs have adverse health effects, especially respiratory and cardiovascular effects (e.g. [ 31 , 32 ]). These particles also seem to have effects on CNS properties and functions, as shown by both experimental (e.g. [ 33 - 35 ]) and epidemiological studies [ 36 ]. UFPs dominate outdoor particle number concentration and particle surface area and are therefore also capable of carrying large concentrations of adsorbed or condensed toxic air pollutants [ 37 ]. The existing information regarding health effects of UFP can possibly be used for predicting some hazardous effects of ENPs. It is known that inhaled particles are size dependently deposited in three different regions, namely the nasopharyngeal, tracheobronchial and in the alveolar region of the respiratory tract. Different studies have shown that 90% of the smaller particles (1 nm) are deposited in the nasopharyngeal and the rest in the tracheobronchial region (for review see [ 38 ]). Particles in the range of 1-5 nm deposit in nasopharyngeal, tracheobronchial and in the alveolar region, whereas 20 nm ENPs deposit to around 50% in the alveolar region [ 39 ]. Larger particles (0.5-10 μm) are remaining on the epithelial surface in airways and alveoli [ 40 ]. The retention time seem to depend on the deposition site. For microparticles (0.5-10 μm) the retention time is 24-48 h in rodent airways [ 41 ] and it is likely that this is increasing in humans because of the airway length. Kreyling and colleagues have shown [ 42 ] that 75-80% of ENPs (< 100 nm) were long-term retained in the alveolar region where particles are interfering with or within cells, like epithelial cells and macrophages, but also with the serous lining fluid (mucus). The alveolar region of the lungs is the most permeable since gas exchange between blood and air is taking part here. The air-blood barrier in this region is approximately 2 μm thick [ 43 ]. If particles are deposited in a certain area they will be either dissolved and/or metabolized, undergoing clearance mechanisms, or insoluble particles will be enriched in particular areas or even in individual cells of the lungs causing biological or toxicological effects [ 40 , 43 ]. ENPs can pass through the interstitium and can be taken up by epithelial cells. However, Geiser and Kreyling [ 40 ] summarized that the main pathway for particle clearance in airways, for any kind of particles, is towards the larynx via mucociliary clearance. The authors pointed out that even particles that were relocated into the underlying interstitium re-appear again on the lung surface to be cleared this way. Alveolar and airway macrophages are on the inner surface and within the lining layer of the lungs, and are constantly exposed to inhaled particles. Phagocytic uptake is the main mechanism to remove insoluble inhaled microsized particles. Monocytes/macrophages are also circulating within the body to take part in the main pathway for monocytes/macrophage-associated particles clearance, which is the mucociliary transport. It cannot be excluded that circulating particle-containing macrophages may re-enter the interstitium and/or lymph nodes and thus the lymphatic system. On the other hand, it is suggested that inhaled and deposited nanoparticles are not efficiently taken up by surface macrophages. Therefore passive mechanisms like diffusion, adhesive interaction, and also pinocytic uptake are currently discussed for translocation [ 40 , 43 ]. Particles which penetrate cells may enrich, interact with organelles, cause oxidative stress, induce different cellular signalling pathways and leading to cellular effects like the release of inflammatory intermediates such as cytokines and free radicals (see also Figure 1 ). Mühlfeld et al. [ 44 ] have shown that inhaled aerosols of 20 nm TiO 2 in rats were distributed after one hour to all lung compartments in proportion to the compartment volume, and some particles were detected in erythrocytes within the pulmonary capillaries. Peters et al. [ 45 ] hypothesized that the way how particles translocate to secondary organs is by the blood circulation. Nemmar et al. [ 46 ] documented by using technetium-labeled carbon NPs by an inhalation study in humans, that a certain amount of NPs diffuse rapidly into the systemic circulation. However, certain published studies report that translocation rates for NPs into the blood circulation are very low [ 47 ]. Clearance mechanisms in airways and alveoli are reducing the retention time of NPs in the lungs, therefore only relatively few nanosized particles can translocate to secondary organs. Chen et al. [ 48 ] have shown that intratracheal-instilled polystyrene particles with an average diameter of 56.4 or 202 nm, are passing into the blood circulation, but this translocation is between 1-2.5% independently of the particle size. Liu et al. [ 49 ] investigated the overall toxicity of nasally instilled nanoscale copper particles (23.5 nm) in comparison with micro-sized copper particles (17 μm) in mice and found only in the high-dose group (40 mg/kg, three times per week) significant pathological changes. It has to be pointed out, that this is an enormously high dose without any physiological significance. These kinds of experiments are useful only for toxicity tests or for hazard identification but not for risk assessment. There are several studies performed using different nanomaterials and sizes and concentrations showing translocation to secondary organs, basically to the liver, spleen and kidney (detailed below). Kreyling et al. [ 50 ] performed an iridium (2-4 nm) and/or carbon (5-10 nm) ENP inhalation study with rats to learn about the translocation rate from lungs to blood circulation and secondary organs. The authors detected from 0.1 to 1% Ir-NPs of the retained fraction in liver, spleen, kidneys, heart, and brain, and 1-5% in the remaining carcass (soft tissue and bone). The mixed fraction of Ir with the carbon ENP retained in secondary organs at significantly lower levels than pure Ir-NP. Furthermore, 80 nm aggregates translocated and accumulated significantly less than the 20 nm ones. In a recent review, Geiser and Kreyling [ 40 ] summarized the evidence for translocation of certain ENPs like gold, silver, TiO 2 , polystyrene and carbon nanoparticles in the size range of 5 - 100 nm across the air-blood barrier from animal studies. In summary, the translocation fraction out of the lung seems not to exceed 5% for any of the investigated ENPs. Mills et al. [ 51 ] investigated the extent to which inhaled radioactive labelled carbon nanoparticles (Technegas, 99m Tc, 4-20 nm, aggregates ca. 100 nm) were able to access the systemic circulation on human volunteers. The authors detected more than 95% of Technegas retention in the lungs, with no accumulation in liver or spleen and concluded that the majority of carbon nanoparticles remain within the lung up to 6 h after inhalation and do not pass directly from the lungs into the systemic circulation. Nemmar et al. [ 46 ] showed that 20% of initial lung radioactive carbon nanoparticles were detected in the liver, meaning that 80% remained in the lung. The translocation rate from the respiratory tract to the central nervous system has been shown to be very low. It is questionable if the amount of nanomaterials which reaches the brain can cause hazardous effects. However, Chen et al. [ 48 ] reported, that pulmonary inflammation induced by instillation plays the major role in enhancing the extrapulmonary translocation of particles (using LPS coated particles). This fact is indicating that at least LPS coated nanomaterials can induce inflammatory effects which themselves are changing the microenviroment leading to higher translocation rates to secondary organs [ 35 ]. A systemic inflammation can contribute to local inflammation in the brain which in turn can lead to enhancement of ongoing inflammatory reactions in the brain [ 52 ]. If ENPs are injected or translocated to the blood circulation, proteins are associating with the nanoparticles, which in turn can lead to an in vivo response [ 53 ]. It has been shown that a so called "corona" on the surface of the ENPs is the result of the adsorption of different serum/plasma proteins on ENPs. Cedervall and colleagues [ 53 ] reported that proteins compete for the nanoparticle "surface," and the resulting "corona" largely defines the biological identity of the particle. Lundqvist et al. [ 54 ] have shown that the nature of the corona is determined by the local chemical property of the nanomaterial including size and surface properties. Therefore the kinetics of the ENPs are also depending on the local corona-structure which is different in each microenviroment. In drug delivery studies using polysorbate 80-coated nanoparticles, it was shown that the ENP adsorbs apolipoproteins from the blood after injection. These particles mimic lipoprotein particles which could be taken up by the brain capillary endothelial cells via LDL receptors [ 15 ]. In conclusion, the translocation rate of deposited ENPs from the lung to the blood circulation and then to secondary organs seems not to exceed 5%. Furthermore, the translocation from the blood to the CNS is lower than 1% according to available studies (see [ 50 , 55 ] and Table 1 ). Corona formation can change the translocation rate and possibly increase the hazardous effects. Axonal transport of ENPs to the brain An important mechanism of particle endocytosis involves the uptake by sensory nerve endings embedded in airway epithelia. In the nasal region it is the olfactory and trigeminus nerve system, and in the tracheobronchial region it is the extensive sensory nerve network. Translocation to ganglia and the CNS can then be accomplished by axonal transport. The olfactory nerve pathway may be a critical portal of ENP entry to the central nervous system of humans, especially under high environmental or occupational ENP exposures but also under chronic exposure. Using colloidal gold particles (50 nm) that were intranasally instilled in monkeys it was shown that particles translocate in the axons of the olfactory nerves to the olfactory bulbs, where nanoparticles were seen in the mitochondria but not in the cytoplasm (as cited by [ 56 ]). A study by Hunter and Dey [ 57 ] in rats demonstrated the translocation of intranasally instilled rhodamine-labeled microspheres (20-200 nm) to the trigeminal ganglion inside the cranium via uptake into the ophthalmic and maxillary branches of the trigeminus nerve. In another study, Hunter and Undem [ 58 ] instilled similar particles intratracheally into guinea pigs and reported a neuronal translocation to the ganglion nodosum in the neck area, which is integrated into the vagal system. More recent studies indicated that neuronal translocation pathways are also operational for other inhaled ENPs. Inhalation of elemental 13 C ENPs (36 nm, 160 μg/m 3 ) resulted in a significant accumulation of these particles in the olfactory bulb of rats on the first day, which constantly increased further throughout day seven after the initial 6 h exposure [ 38 ]. Results from another inhalation study with solid nanosized manganese oxide particles (30 nm, 500 μg/m 3 ) in rats also demonstrated an increase of particles in the olfactory bulb. Inhalation exposures were for 6 h/day, 5 days/week for up to 12 days. After 12 days of exposure with both nostrils patent, Mn concentrations in the olfactory bulb increased 3.5-fold (from 0.5 to 1.75 ng Mn/mg tissue), whereas lung Mn concentrations doubled; there were also increases in striatum, frontal cortex, and cerebellum. When one nostril was occluded during a 6-h exposure, the accumulation of Mn was seen only in the olfactory bulb of the open nostril [ 56 , 59 ]. These observations suggest that nanoparticles in the air can enter into the CNS via the olfactory nerve during accidental or prolonged environmental or occupational exposure to humans. Another study showed that inhaled 20 nm nanogold particles (2 × 10 6 particles/cm 3 ) can accumulate in the olfactory bulb of rats [ 60 ]. The exposure for 5 days resulted in a significant increase of gold ENPs in the olfactory bulb (8 ng Au/g body weight). After 15 days of exposure, significant accumulations of gold particles were detected in the septum and entorhinal cortex. Both brain structures receive direct neuronal projections from the olfactory bulb, and are important in attention and new memory formation. After nasal instillation (500 μg of TiO 2 nanoparticle suspension every other day for 30 days), the micro-distributions of TiO 2 NPs (80 nm) and fine TiO 2 particles (155 nm) in the olfactory bulb of mice were investigated by [ 61 ]. It could be demonstrated that both types of investigated TiO 2 particles were taken up by the olfactory bulb via the primary olfactory neurons and then accumulated in the olfactory nerve layer, olfactory ventricle, and granular cell layer of the olfactory bulb. The TiO 2 content was increased in all investigated brain regions (olfactory bulb, cerebral cortex, hippocampus and cerebellum), with the most significant increases seen in the hippocampus. The presence of TiO 2 in hippocampus was furthermore accompanied by changes in neuron morphology and increased amount of GFAP-positive cells in the CA4 region. Signs of oxidative stress were documented in all regions of the brain. Interestingly, in general anatase TiO 2 gave rise to stronger effects than the rutile form. Taken together, it seems that nanoparticles can translocate to the nervous system through sensory nerves. Translocation of 20 nm particles is 2-10 times higher in the human olfactory bulb than in rats [ 6 ]. Thus, the translocated nanoparticles in humans can enter into the deeper brain structures in short exposure time. Based on the limited data available, it is presently difficult to assess to what extent accumulation in the brain via axonal transport is a realistic possibility (see also Figure 2 ). Neurobiological effects of ENPs In vivo studies Of the two principal cell types in the nervous system (neurons and glia cells), the neurons have characteristics that make them especially sensitive to various types of stressors. The neurons have an especially vulnerable anatomy due to their extensive and very thin and fragile extensions (dendrites and especially axons). In addition, these cells are metabolically very sensitive since they rely solely on aerobic metabolism of glucose. The neurons are extremely sensitive to oxidative stress, which in many cases also is a contributing factor to a number of neurodegenerative diseases. In addition, with very few exceptions, neurons are not renewable in mammals, making the nervous system functions very sensitive to agents that cause cell death. Studies performed on intact animals can specifically address both exposure and dose requirements for potential effects, as well as the specific effects on processes in the brain (Table 2 ). That nanosized MnO 2 NP can translocate into the brain after long-term inhalation/instillation has been shown by Elder et al. [ 56 ] and Sarközi et al. [ 62 ]. In the former study, rats were exposed to MnO 2 ENP by inhalation for 12 days. Mn levels were seen to increase in the olfactory bulb (3.5 times the background levels), striatum, frontal cortex, and cerebellum. No functional endpoints were investigated in this study, but molecular signs of inflammatory changes were possible to detect. In the work by Sarközi and co-workers [ 62 ], male Wistar rats were subjected to MnO 2 (23 nm) instillation for 3, 6, or 9 weeks at 2.63 or 5.26 mg/kg bodyweight. After the end of exposures, Mn was detected in the brain by x-ray spectroscopy. The rats' spontaneous motility was negatively affected. In addition, electrophysiological changes in cortical activity and the conduction velocity of the tail nerve were documented. Chen et al. [ 11 ] infused rats intravenously with Al 2 O 3 NP (8-12 nm; 29 mg/kg). The rats were sacrificed 20 h after the infusion and the brains were subsequently investigated with immunohistochemistry for certain tight junction proteins normally present in the endothelium of the BBB. The data indicate that the proteins claudin-5 and occludin are down-regulated in the vessels of treated animals, suggesting impairment of the BBB. However, the data are only qualitatively expressed, no proper quantification of the protein levels were made. It is also unclear how many animals were investigated. In a recent paper by Viswaprakash et al. [ 63 ] rat olfactory epithelia were exposed to 1-2 nm zinc particles and the responses to odorants were measured by electroolfactogram and whole-cell patch clamp. The addition of the Zn particles to the odorant suspension enhanced the response to the odorant. Interestingly, this response was seen to be specific for Zn particles, whereas neither Zn 2+ ions nor other metal particles (Cu, Ag, Au) elicited similar responses. The increased production and presence of nanosized TiO 2 particles in consumer products and in processes has generated an interest into the possible effects of these particles on human health. Regarding in vivo studies of nervous system function, a few recent articles have rendered relevant data. Thus, Wang et al. [ 61 ] subjected female mice to nasal instillation with TiO 2 NP (80 nm, rutile and 155 nm anatase; 500 μg every 2 nd day for 30 days) at an extremely high dose. Titanium particles were mainly accumulated in the cerebral cortex, thalamus, olfactory bulb and hippocampus, (especially in the CA1 and CA3 regions). There was an obviously dispersed arrangement of neurons in the hippocampal CA1 region after TiO 2 exposure. Furthermore, the investigation of cell numbers in the stratum pyramidale of the CA1 region indicated a drastic neuronal loss. There was 30% and 25% cell loss in the 80 and 155 nm TiO 2 -exposed groups, respectively. Apparent morphological changes of hippocampal neurons and increased GFAP-positive astrocytes in the CA4 region were also found, which were in good agreements with the high TiO 2 contents in this hippocampus region. GFAP is best viewed as a biomarker of early pathological effects, indicated by the activation of astrocytes. Oxidative stress such as lipid peroxidation, protein oxidation and increased activities of catalase, as well as excessive release of glutamic acid and nitric oxide occurred in the whole brain of exposed mice [ 61 ]. In a follow- up study, mice were again intranasally instilled every second day with the two types of TiO 2 particles (80 nm, rutile or 155 nm, anatase; purity > 99%, about 500 μg per mouse, respectively). This time, brain tissues were collected at post-instillation time points of 2, 10, 20 and 30 days and evaluated for accumulation of TiO 2 , histopathology, oxidative stress, and inflammatory markers. It is shown in this study, that instilled TiO 2 nanoparticles entered the brain directly through the olfactory bulb during the whole exposure period. In all brain parts and at all post-exposure time periods, the measured concentrations of both types of TiO 2 particles were higher than in any of the controls. The anatase form of the TiO2 exhibited stronger effects on some of the investigated endpoints in both these studies. In the olfactory bulb, TiO 2 contents increased gradually with time. TiO 2 particles were mainly deposited in the hippocampus, where TiO 2 contents were significantly increased after exposure for 2 days, then stayed constant for 10 and 20 days, before reaching the highest values after 30 days of exposure [ 64 ]. After 30 days of exposure, pathological changes were observed in olfactory bulb and hippocampus. Irregular arrangements of neurons in the olfactory nerve layers and dispersed arrangement and loss of neurons in the CA1 region of hippocampus were demonstrated. Hippocampal nerve cells were degenerated, together with changes in the nuclear membrane, mitochondria, rough endoplasmic reticulum, chromatin condensation, and elevated amounts of free ribosomes [ 64 ]. Shimizu et al [ 65 ] studied effects of anatase TiO 2 (the particle size is not given in the article) that they injected subcutaneously (s.c.) into pregnant mice. Male embryos and pups were then investigated for certain gene expression patterns. The expression of genes associated with brain development, motor activity, oxidative stress, and apoptosis was changed compared to control animals during various periods of investigation (embryonic day 16 to 21 days post partum). Also Takeda et al [ 66 ] injected TiO 2 (anatase, 25-70 nm) s.c. into pregnant mice. The nanoparticles were found in the brains (cortex, olfactory bulb) of the offspring. In addition, cells expressing the apoptosis marker Caspase-3 increased in the olfactory bulb of these animals. Abdominal injection of high dose anatase TiO 2 (5 nm; 5-150 mg/g) to mice were performed daily for 14 days in a recent study [ 67 ]. The TiO 2 content of the brains increased with increasing injection "doses". Also changes in neuronal morphology, transmitter levels and signs of oxidative stress were seen to follow a dose-response relationship. The effects of various types of quantum dots (QDs) in the hippocampus of rats were investigated by Tang et al [ 68 ]. They found that both unmodified (CdSe) and modified (streptavidin-CdSe/ZnS) QDs can negatively affect synaptic transmission and plasticity in the rat hippocampus. The QDs were directly applied into the hippocampus and the effects on the electrophysiological properties of the neurons in the area were recorded after 20 min. Pair-pulse relation and long-term potentiation were significantly decreased after treatments. The effects were seen at two investigated concentrations of QD, 0.5 and 10 nM. The authors also reported that signs of oxidative stress were seen immediately after completion of the electrophysiological measurements. The results showed that SOD activity, GSH content and MDA levels all increased in the animals treated with QDs. These responses were stronger in the unmodified (CdSe) QDs, and more pronounced at the higher investigated concentration. This finding is possibly due to the toxic effects of Cd, which can be expected to be released from the unmodified QD. In a study by Maysinger et al [ 69 ] intracortical injection (μM concentrations) of various types of PEGylated QDs, non-PEGylated CdTe QDs, and CeO 2 all caused activation of the glial cell marker GFAP to various degrees in mice. The effects were strongest in animals injected with CdTe QDs, and weakest after CeO 2 treatment. Furthermore, the study showed that one of the PEGylated QDs, QD705, primarily accumulated in glia, whereas a small fraction (0.5%) could be found in neurons. Both these studies indicated that especially non-PEGylated QDs can cause inflammation and possibly gliosis in the brain. It is difficult to evaluate if the used concentrations are relevant for any real exposure situation. In conclusion, the referenced studies point to that the investigated metallic nanoparticles all can translocate from the point of application (respiratory tract, skin, circulatory system) to the brains of the animals. However, it is unclear from these studies under which specific conditions this can be accomplished since the studies have not investigated dose-response relationships, properties of the ENPs in question etc. Certain of the observations are furthermore made in experiments where unrealistically high doses have been applied. A single study also indicates that a high dose of TiO 2 can pass the placenta and be taken up into the brains of embryos. Regarding the physiological effects of these exposures, it is unclear to what extent, and at what exposure levels, nervous system functions can be affected by ENPs. However, the available data are suggestive of effects on neurotransmission, and possibly behaviour. Several signs of changes in oxygen radical homeostasis were also seen. The consequence of this could be that long-term exposures cause permanent inflammatory states, which can be a contributing factor in certain neurodegenerative diseases. In vitro studies Synaptic transmission between neurons involves a number of structures and processes both in the pre- and the postsynaptic neuron. The presynaptic neuron needs to have the necessary machinery for synthesis of the neuron-specific transmitter. Furthermore, it is necessary to have structures for transmitter release and transmitter re-uptake or enzymatic degradation of transmitter. The post-synaptic neuron needs to express receptors for transmitters and together with its synaptic partner it has to express the structures that make the physical contacts in the synapse. To some extent, the development and function of neurotransmission in ENP exposed neurons have been investigated, along with studies on the toxicity of ENP on nervous system components. Several studies have dealt with toxicity and oxidative stress due to ENP exposures. Thus, Pisanic et al. [ 70 ] showed that iron oxide ENP (5-12 nm) have cytotoxic effects on PC12 cells (a neuroendocrine cell line derived from rat pheochromocytoma). At 1.5 and 15 mM iron concentrations (but not at 0.15 mM) the ENP caused decreased cell viability. The response to NGF (nerve growth factor, inducing differentiation) in the PC12 cells was also negatively affected, seen as diminished neurite extension and number of neurites per cell. Also, the level of the GAP43 protein, a marker for neuronal differentiation, was decreased. Maysinger et al. [ 69 ] showed that certain QDs were taken up into differentiated PC12 cells, whereas other (non-PEGylated) QDs caused cell death. Increased oxidative stress, measured as H 2 O 2 production, was seen in immortalized microglia cells exposed to Degussa P25 ENP that formed aggregates. The doses (2.5-120 ppm) of the P25 aggregates were non-cytotoxic [ 71 ]. Alekseenko and coworkers [ 72 ] used rat brain synaptosomes to test ferritin molecules that contain Fe 3+ iron particles (7 nm). The treatment caused ROS formation at high doses (800 μg/ml), whereas the effects of 80 and 8 μg/ml were not significantly different from controls. The higher concentration could not induce glutamate release, but inhibited uptake of glutamate. Consequently, at 800 μg/ml, iron-based nanoparticles can cause conditions that can lead to neurodegeneration. Schubert et al. [ 73 ] showed that both CeO ENPs (6 and 12 nm) and YO ENPs (12 nm) are neuroprotective in cultured hippocampal neurons (HT22 cell line). The cells were treated with glutamate to generate ROS at levels that were cytotoxic, which was counteracted by addition of the mentioned ENPs. Conflicting results are available regarding the effects on cytotoxicity and oxidative stress from fullerenes (C 60 ). Sayes et al. [ 74 ] used a water-soluble fullerene species, nano-C 60 that was cytotoxic to several human cell types, including astrocytes. The cytotoxicity was mediated by lipid peroxidation according to the authors. On the other hand, polyhydroxylated C 60 fullerenols at μM concentrations acted as antagonists to glutamate receptors in a study by Jin et al. [ 75 ]. The C 60 particles blocked primarily the AMPA-type glutamate receptor in neuronal cultures from rat brain, and also to some extent NMDA and KA receptors. The antagonistic behaviour on glutamate receptors were not seen in GABA or taurine receptors. In the absence of C 60 , higher concentrations of glutamate were needed to elicit similar effect. The fullerenes could also act as antioxidants, inhibiting effects of added H 2 O 2 and Fe 2+ . An earlier study by Dugan et al. [ 76 ] also showed neuroprotective effects of C 60 fullerenes on cortical cell cultures exposed to NMDA or AMPA at concentrations that caused excitotoxicity. The effects of Mn ENP on transmitter levels in PC12 cells were seen by Hussain et al. [ 77 ]. The Mn nanoparticles specifically caused depletion of dopamine stores in the PC12 cells. This occurred in a dose-dependent fashion (concentrations ranging from 1-100 μg/ml) after cells were exposed for 24 h to 40 nm particles. The effect was similar to effects of added Mn 2+ ions. However, the levels of ROS were much higher after Mn ENP addition compared to Mn 2+ , or compared to Ag ENP (15 nm). The latter ENP also caused dopamine depletion, although to a lesser extent than Mn ENP. In two studies Wang et al. have documented that ENP inhibit the acetylcholine degrading enzymes acetylcholine esterase [ 78 ] and butyrylcholine esterase [ 79 ] in solution. Several different types of ENP (MWCNT, SWCNT, Cu, TiO 2 ) could adsorb and thus inhibit enzyme activities in a dose-dependent fashion. The authors suggested that the inhibitory effects were caused by ion dissolution from the ENPs. The communication between neurons relies on transmitter release which in turn is dependent on changes in ion concentrations on the in and outside of the neuronal membrane. Since such changes lead to displacement of charged entities, it is possible to measure these events by analyzing electric potentials that are present over the membrane. Several studies have investigated whether currents that pass through ion specific membrane channels are affected by ENP. Tang et al. [ 68 ] studied the effects of CdSe QDs (2.38 nm) on primary cultures of rat hippocampal neurons. At 10 nM or higher concentrations, these particles caused cell death, due to sustained increases in intracellular Ca 2+ levels. The particles also had effects on voltage gated Na-channels, where patch-clamp analyses revealed enhanced activation and inactivation of the sodium current, and also a prolonged activation time and increased recovery time for the Na 2+ current. Thus, fewer Na-dependent potentials would occur in these cells, interfering with normal synaptic transmission. Another study revealed that silver particles (244.4 nm; 12.5 m 2 /g) could inhibit Na + currents in rat hippocampal slices [ 80 ]. This occurred at 10 μg/ml, but not at lower concentrations. Zhao et al [ 81 ] could show that ZnO ENP (20-80 nm; 2-3 crystal forms; 100 μg/ml, but not at lower concentrations), increased amplitudes of both Na + and K + currents, by increasing the number of open Na + channels, delaying rectifier K + channels and thus enhancing the excitability of neurons. Xu et al. [ 82 ] have shown that CuO ENP (60.6 nm; 15.7 m 2 /g; 50 μg/ml) could inhibit the rectifier K + current in rat CA1 pyramidal hippocampus neurons. Jakubek et al. [ 83 ] demonstrated that carbon nanotubes could inhibit the function of Ca 2+ channels expressed in human embryonic kidney tsA201 cells. This effect was probably due to the release of Ni + and Y + ions from the carbon nanotubes, and that these ions displaced Ca 2+ from the channel pore. Taken together, these findings give some support for the concept that several types of ENPs under specific in vitro conditions can influence the electrophysiological properties of neurons. Exposure to silver is likely to increase due to the increased use of silver nanoparticles. A recent study [ 84 ] asked the question if silver ions (AgNO 3 ) can have effects on the developing nervous system. The reason why the authors investigated silver ions was that silver nanoparticles are releasing ions according to the authors. The experimental model was the mouse PC12 neuroblastoma cell line which can be induced to differentiate into neurons in the presence of NGF (nerve growth factor). The cells were exposed for 1 h to Ag + at 1 or 10 μM, or to control substances (chlorpyrifos, which is a known developmental neurotoxicant; NaNO 3 to investigate if effects were due to NO 3 - ions or to Ag + ). In undifferentiated cells, both concentrations of Ag + inhibited DNA and protein synthesis. The higher concentration furthermore caused cell death and oxidative stress, to an extent which was larger than the positive control chlorpyrifos. Furthermore, it was clear from the experiments that it was the Ag + ion and not the NO 3 - that was responsible for the effects. Continuous exposure to Ag + in cells that were induced to differentiate caused DNA synthesis inhibition and oxidative stress, and also inhibition of the differentiated phenotype (dopaminergic neurons), whereas cholinergic neuron differentiation was favoured. This study suggests that Ag + can exert a developmental neurotoxic effect at higher concentrations that are even stronger than a known neurotoxicant. Also at the lower Ag + level, effects were present, although less pronounced. However, one has to keep in mind that this study was not dealing with silver nanoparticles but instead was designed on the assumption that silver ENPs would act as a depot for release of silver ions. In the mouse neural stem cell line C17.2, TiO 2 ENP (50-250 μg/ml; 80-100 nm; rutile form) could lower the proliferation rate and induce neuronal differentiation [ 85 ]. The authors also performed protein expression profiling and found that the induction of differentiation by TiO 2 was accompanied by changes in the levels of nine of the investigated proteins. These data suggest that TiO 2 effects include modulation of the PKC-epsilon pathway. A common feature for several neurodegenerative diseases is the formation of extra- or intracellular protein complexes or aggregates. In e.g. Alzheimer's disease, the so-called amyloid hypothesis states that aggregates of the beta-amyloid peptide are neurotoxic and cause local inflammations that are detrimental for neurons. The reasons for formation of these aggregates are manifold, including both genetic and environmental factors. Since so many patients are diagnosed with diseases like Alzheimer's every year, there is a constant interest into potential aggravating factors. Thus, the question is if ENPs can trigger or promote formation of beta-amyloid aggregates. Wu et al. [ 86 ] have seen that TiO 2 ENP (20 nm; 80:20 anatase: rutile) in a concentration dependent manner (4-20 μM) accelerates the fibrillation, and thus aggregate formation, of the beta-amyloid peptide in solution. The proposed mechanism is that the nucleation process, which is rate limiting for fibril formation, is shortened. Also other ENP were previously seen to stimulate protein fibril formation. Thus, several types of nanoparticles (copolymer particles, cerium oxide particles, QDs, carbon nanotubes) stimulated faster formation of fibrils of the beta 2 -microglobulin protein [ 87 ]. Both these studies thus suggest that formation of potentially neurotoxic protein fibrils can be enhanced by ENPs. However, these studies have to be treated with caution, since the experiments were performed in solution, and not in any living system. Whether this is relevant for any in vivo situation is unclear (Table 2 ). Risk assessment and research needs A health risk assessment has to consider data from various lines of evidence (e.g. human epidemiological and clinical studies, experimental animal and in vitro studies, in silico studies) and integrate these into a cohesive evaluation. It is furthermore essential to have relevant information on exposure. A risk can then be deduced from exposure data together with the hazard assessment. Needless to say, the assessment becomes more reliable when more relevant information is available. Here we try to make an assessment of the risks for neurological effects in humans that are subjected to unintended air-borne exposures (i.e. non-clinical) of ENP (see also Figure 3 ). Data on assessment of human exposure to ENPs is very sparse. However, there is at present very little reason to expect that the general public is exposed to any significant amounts of air-borne ENPs, although ENPs are present in certain consumer products. It is more likely that occupational exposures can be a factor in at least some settings. Besides the few data on exposure that makes risk assessment difficult, the absence of a relevant dose concept for quantification of hazards is an obstacle. We consider this deficiency to be one of the biggest problems for risk assessment of ENPs today. There are different models available to study toxicological effects of nanomaterials in the human body, like physiologically-based pharmacotoxic and pharmacokinetic models, but in addition the experience from radiobiology in generating dose concepts could be very valuable. Thus, knowledge about the retention time of the nanomaterial in the body, and also half-life, (cf. radionuclides that have dual effects; the effects of the element itself plus the effect of the irradiation that the nuclide generates) is essential to get an idea of both dose and unintended reactions in vivo. Furthermore, the dose rate (the kinetics of the uptake of ENPs per unit of time (acute high dose exposure vs. chronic low dose exposure)) is an important aspect of exposure, as well as the ENPs physicochemical structure. Even during clinical situations, where ENPs are created to act as drug delivery systems, translocation to CNS is difficult to assess. It has been shown that special coverings and functional modifications of the surface of ENPs are necessary for them to reach the target organ, in this case the CNS. Also experimental studies on animals suggest that translocation even after instillation or inhalation of substantial amounts is very low, but can occur (see also Figure 2 for an overview of translocation routes). The knowledge regarding the specific physico-chemical characteristics that are important for translocation is sparse. It is feasible that also in humans, translocation to at least some degree can occur as a consequence of environmental and/or occupational exposure. Importantly, there are no long term data available which could demonstrate chronic exposure conditions. It has to be pointed out that chronic exposure is relevant for non-biodegradable and non-excreted ENPs, which can accumulate over time within the brain leading to long term (toxic) effects. In addition, long term and low "dose" exposure to biodegradable ENPs can induce chronic inflammation-like conditions by oxidative stress. Such a condition can lead to pathological processes in the CNS. Chronic exposure to ENPs within the CNS could possibly also aggravate ongoing pathological processes. Regrettably, this is presently only speculation since knowledge about the effects of chronic and long term/low dose exposure is entirely missing. If ENPs are reaching the CNS through the olfactory nerve after inhalation, the numbers of particles (dose) can be higher (acute exposure) then by translocation through the lungs. This circumstance can be relevant for occupational exposure. Also chronic exposure in occupational settings can lead to a brain exposure, both through the lung and/or the olfactory nerve. However, if a high CNS-exposure would occur, other parts of the body would experience even higher exposures and thus stronger toxic effects. Table 2 summarizes in vivo and vitro data on effects caused by ENPs on properties and functions of the CNS. The noted effects suggest that several types of ENPs can have various types of biological effects. Some of these effects could also imply that ENPs can cause hazards, both in an acute fashion and in the long term. However, the relevance of these data for risk assessment is far from clear. At issue is especially if these effects would occur at levels that are relevant for environmental or occupational exposure. Since investigations into the possible harmful effects of ENP have been performed only for a few years, it is not surprising that many studies suffer from shortcomings. It is nevertheless the view of these authors that it is possible to improve the quality of the studies with a few means, as outlined below. It is essential • that exposure assessments are performed so that experimental studies can investigate effects of specific ENPs at relevant "doses". Some of the presently available studies are using enormously high doses of ENPs without any relevance for risk assessment. However, it should also be mentioned that also exposures to high doses can be informative, especially in the identification of possible hazards • to perform more dose-response studies • to use appropriate controls and studies should be performed in a blinded manner. Very often the informative value of a study would vastly improve if, for toxicology studies, a relevant positive control was applied. Positive controls are furthermore essential for validation of the methodology used. Admittedly, positive controls are sometimes difficult to identify, but are basically physical or chemical agents with known mechanisms of action. • to correctly describe the physico-chemical properties of the ENP, which is sometimes missing. Essential information includes data on size, shape and composition (which includes surface charge and adsorbed species) and also redox-reactivity. • to have knowledge about possible surface modifications, whether the ENP aggregates and their dissolution or degradation is needed (see e.g. [ 82 ] for details). • for risk assessment to report negative findings. For adequate risk assessment of chronic exposure, information about metabolism of ENPs within the CNS, accumulation, dose definition etc is needed. Obviously, at the present state of knowledge, the risk assessment needs to be performed on a case by case basis. Competing interests The authors declare that they have no competing interests. Authors' contributions MS conceived of the study and participated in data collection and screening, data analysis, drawing of conclusions. MOM participated in data collection and screening, data analysis, drawing of conclusions. Both authors drafted the manuscript, read and approved the final manuscript.

Acknowledgements This work is supported by the Federal Ministry for Transport, Innovation and Technology, Austria.

CC BY

no

2022-01-12 15:21:44

Part Fibre Toxicol. 2010 Dec 21; 7:42

oa_package/d4/c1/PMC3016300.tar.gz

PMC3016301

21143886

Introduction Stroke is the most common cause of disability. Prevention of stroke by carotid endarterectomy (CEA) or carotid artery stenting (CAS) is widely accepted, and they are basically equivalent treatment modalities [ 1 , 2 ]. The endovascular treatment of internal carotid stenoses is an appropriate treatment method not just for patients at high surgical risk. It is not unexpected that procedural safety and complication rates of CAS are closely related to the operator's skill and the institutional experience with the technique. This can be expressed in terms of caseload or patient enrollment numbers into clinical trials [ 3 ]. With the widespread and unregulated use of CAS, however, complication rates could increase and their management is sometimes a challenge for vascular specialists. Apart from the inherent risks of stenting procedures (for example, stent thrombosis, distal emboli, hyperperfusion, hemorrhage and so on), a variety of technical failures have also been observed. They include, among others, sizing issues with overdilation or underdilatation, distal wire injury, and disconnection of protection filters, with their respective clinical sequelae. Here, we report the case of a patient in whom a proximal internal carotid artery (ICA) stenosis was stented in another hospital. At presentation to our institution, the apparent misdeployment of the two stents in a partly overlapping position was found ('hugging' stents). Well co-ordinated endovascular and surgical management saved our patient from further impairment.

Discussion With experienced staff at dedicated centers, CEA and CAS are considered equally safe and efficient methods for the treatment of proximal carotid artery stenoses. However, carotid artery angiography and stenting requires proper training [ 4 ]. In institutions with a sufficiently large caseload, low complication rates of CAS can be achieved [ 5 ]. However, the technical risks and the clinical sequelae of CAS procedures performed by inexperienced operators have also been noted [ 6 ]. Our case report deserves some discussion. Whether an asymptomatic mid-grade ICA stenoses with contralateral ICA occlusion should be treated is the subject of an ongoing debate [ 7 , 8 ]. The deployment of a self-expanding stent without balloon angioplasty has been proposed by others [ 9 ], but would certainly not be our preferred technique. The deployment of two stents is hard to justify, and in this particular case is likely just a technical mistake. Unfortunately, the reason why two stents were inserted in this way remains unexplained by the operator at the other hospital. The non-coaxial deployment, which apparently remained unrecognized by the operator, made the situation even worse. The significant residual stenosis (Figure 1B ) can be interpreted as a failure to improve the cerebral perfusion and should have prompted immediate action. The omission of the examination of the posterior circulation vessels, which would have shown the complexity of the cerebral blood supply, is also an area for criticism. Our subsequent efforts applied generally accepted endovascular and surgical methods, and started with the deployment of a short drug-eluting stent into the proximal vertebral artery stenoses [ 10 ]. For treatment of the basilar artery stenosis, a combination of undersized balloon dilatation followed by the deployment of a moderately oversized self-expanding stent was used [ 11 ]. The following surgical stentectomy and CEA was not carried out in a standard fashion. At present, surgical experience with complications after stenting is limited [ 12 ]. In the case of our patient, with a symptomatic occlusion of the left ICA and a high-grade mechanical obstruction of the right proximal ICA, we 'prepared' him for the operative stentectomy with temporary clamping of the right ICA by improving the collateral supply using endovascular means. With regard to the anesthesiological and surgical methods, local anesthesia during CEA with the eversion technique offers the possibility of continuous neurological assessment; an inherent advantage over general anesthesia. The eversion endarterectomy enabled the simultaneous removal of the two stents and the underlying carotid plaque. As the proximal part of the two 'hugging' stents could be removed easily, we decided not to use the standard endarterectomy with patch plasty. Thus, a biological carotid reconstruction without extraneous tissue was possible in this re-do procedure. To the best of our knowledge this is the first description of a stentectomy by carotid endarterectomy with the eversion technique [ 13 ]. A single suture for reanastomosis of ICA and CCA after plaque eversion reduces the risk of bleeding under high antiplatelet medication with acetylsalicylic acid and clopidogrel in comparison to the standard patch plasty. The strategy of removal of the two 'hugging' stents with prior hemodynamic 'preparation' led to a good clinical outcome for our patient, without any further neurological deficit during follow-up of 12 months.

Conclusion Post-procedural complication management in vascular medicine is a continuous task requiring interdisciplinary co-operation. The increasing numbers of stent procedures will increase the related surgical expertise. Recommendations for re-do procedures required by local complications of carotid stenting are: (1) the earlier the better, (2) biological reconstruction with the eversion technique is beneficial, and (3) institutions which offer the full range of therapeutic options on site are advantageous. Further evaluation in larger studies is highly recommended.

Introduction With the widespread use of carotid artery stenting, previously unknown technical mistakes of this treatment modality are now being encountered. There are multiple strategies for the treatment of in-stent restenosis. With regard to surgical management, endarterectomy and patch plasty are favored. To the best of our knowledge, this report is the first description of a complete stent removal by the eversion technique. Case presentation We report the case of a 63-year-old Caucasian man with misdeployment of two stents into his stenotic proximal internal carotid artery, resulting in a high-grade mechanical obstruction of the internal carotid artery lumen. With the contralateral internal carotid artery already occluded and associated stenoses of both proximal and distal vertebral arteries, an interdisciplinary therapeutic concept was applied. Bilateral balloon angioplasty and stenting of the proximal and distal stenotic vertebral arteries were carried out to provide sufficient posterior collateral blood flow, followed by successful surgical stentectomy and carotid endarterectomy using the eversion technique. Duplex scanning and neurological assessments were normal over a 12-month follow-up period. Conclusions Interdisciplinary treatment is a recommended option to protect patients from further impairment. Further evaluation in larger studies is highly recommended.

Case presentation A 63-year-old Caucasian man initially presented with an asymptomatic 55% stenosis of the right proximal ICA (Figure 1A ) to another institution. The left ICA was found to be occluded; the vessels of the posterior circulation were not examined. Our patient underwent an endovascular procedure at the other institution, which included the deployment of two 7 mm/40 mm Wallstents (Wallstent, Boston Scientific Corporation, Natick, MA, USA) (Figure 1B ). The reason why two stents were inserted without balloon angioplasty remains unexplained. Then, three months later, our patient was referred to our hospital with clinical signs and symptoms of a transient left hemispheric ischemia, including global aphasia, right hemiparesis, and paresthesia of the right upper extremity. His cardiovascular risk factors included arterial hypertension, hyperlipidemia and non-insulin dependent diabetes mellitus. Our patient was a non-smoker. Our patient's history included severe coronary heart disease, cardiac insufficiency (New York Heart Association stage II) and recurrent atrial fibrillation, which altogether resulted in an American Society of Anesthesiologists (ASA) physical status of IV. Magnetic resonance imaging and angiography including diffusion-weighted imaging did not reveal any ischemic lesions of the left hemisphere. Digital subtracted angiography (DSA) of the supra-aortic and intra-cranial vessels confirmed the occlusion of the left ICA and revealed a highly effective mechanical obstruction of the right ICA caused by two stents deployed in an overlapping side-by-side position (Figure 2A ). Further atherosclerotic lesions included significant stenoses of both proximal vertebral arteries (V1), a stenosis of the entire left V4 segment and a focal stenosis at the junction of the right V4 segment with the basilar artery. The anterior communicating artery and the right posterior communicating artery were widely patent and the left external carotid artery contributed to the supply of the left hemisphere via the ophthalmic artery. When giving his informed consent, our patient was informed of a peri-procedural and post-procedural mortality and morbidity rate of between 3% and 5%, and an increased risk of peri-operative bleeding due to the antiplatelet medication. Our patient accepted this, and the off-label use of both the Coroflex (Coroflex Please, B Braun Melsungen AG, Melsungen, Germany) and Enterprise (Codman Enterprise, Raynham, MA, USA) stents. The first step of the treatment strategy focused on the posterior circulation stenoses in order to improve the potential collateral supply during subsequent stentectomy and CEA. Dual platelet antiaggregation with acetylsalicylic acid and clopidogrel was initiated. Under general anesthesia the stenoses of both vertebral artery origins were treated with short drug-eluting stents (Coroflex), followed by the stent percutaneous transluminal angioplasty of the left and right V4 stenosis using a combination of moderately undersized balloon dilatation and deployment of oversized self-expanding stents (Enterprise). The surgical stent removal from the right ICA combined with CEA completed the treatment. Stentectomy and CEA were carried out under regional anesthesia. As routinely, we operated using ultrasound-guided regional anesthesia (MicroMaxx, Sonosite GmbH, Erlangen, Germany) of the cervical plexus using 20 cc of lidocaine 1% (Xylocain, Braun Melsungen AG, Melsungen, Germany) and 50 cc of ropivacaine 0.375% (Naropin, Astra Zeneca GmbH, Wedel, Germany). The dispensation of analgosedation allowed our patient to be awake throughout the operation, while neurological function was monitored by assessing the level of consciousness and our patient's motor function on the left side. At first the proximal ICA was dissected circumferentially beyond the level of the carotid bifurcation after uneventful clamping. The removal of the proximal stent from the common carotid artery was then possible without any neurological deficit, and we decided to proceed with the eversion technique. The simultaneous removal of the two 'hugging' stents together with the atherosclerotic plaque in the proximal ICA was possible. The ICA and common carotid artery (CCA) were reanastomosed using a 6/0 polypropylene suture in a continuous fashion. Intra-operative angiographic assessment was performed to ensure patency and in order to control the distal end of the plaque removal. Our patient made an uneventful recovery with no additional neurological deficit. Histological examination revealed a thickened layer of arterial neointima. Duplex scanning was within normal limits after five days and three months, and was confirmed by DSA after six months (Figure 2B ). Follow-up duplex scanning surveillance and neurological assessments were unremarkable after 12 months. Consent Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Competing interests The authors declare that they have no competing interests. Authors' contributions DJ, HvL and TH performed the surgical procedure and drafted the case report. TG and HH performed the endovascular procedure. HM participated in the diagnostic and therapeutic decisions and was responsible for follow-up examinations. TH and HH made major contributions to writing the manuscript. All authors read and approved the final manuscript.

CC BY

no

2022-01-12 15:21:44

J Med Case Reports. 2010 Dec 9; 4:397

oa_package/7f/41/PMC3016301.tar.gz

PMC3016302

21143956

Introduction The standard operation for pancreatic cancer within the head of the pancreas or uncinate process is a partial pancreaticoduodenectomy ('Whipple procedure'). The standard Whipple procedure involves the removal of the head of the pancreas, the distal part of the stomach, the duodenum, the first part of the jejunum, the common bile duct and the gallbladder. Pancreaticoduodenectomy was popularized by Dr Allen Whipple after his success in his initial three cases in 1935 [ 1 ]. The technique has undergone several technical modifications and revisions, so that morbidity and mortality rates have dramatically decreased over the past several decades [ 2 ]. In 1978, Traverso and Longmire described the pylorus-preserving pancreaticoduodenectomy (PPPD), a procedure already mentioned by Watson in 1944 [ 3 ]. The expected advantages of this procedure were less dumping, improved gastrointestinal function and reduced jejunal ulceration. Occasionally, unusual pathology was present in the surgical specimens which were not initially suspected. We present three unusual histological findings.

Discussion The most common indications for pancreatic resection are ductal adenocarcinoma, a neuroendocrine tumor or chronic pancreatitis. The three cases we have described are rare. We would like to present an overview of these unusual conditions: Benign multicystic mesothelioma was first described by Mennemeyer and Smith in 1979 [ 4 ]. It is a very rare tumor originating from the peritoneum but it can also be found on serosal surfaces (for example, intestine, liver and kidney). To our knowledge, this is the first description in the literature of benign multicystic mesothelioma of the pancreas. Although it is a tumor with a benign clinical behaviour, it frequently recurs after surgical resection. A report of 15 cases revealed a recurrence rate of 26.7% [ 5 , 6 ]. Time between recurrences may range from four months to 12 years [ 7 ]. As it has a tendency to recur, a neoplastic etiology is supported. In our patient, 10 months postoperatively there had been no evidence of recurrence. Benign multicystic mesothelioma is found mainly in women of reproductive age with a past history of abdominal surgery or endometriosis [ 8 ]. Only a few reports exist of this gynecologic diagnosis in elderly women [ 9 ]. Our patient had not undergone any previous abdominal surgery. The clinical findings of benign multicystic mesotheliomas are non-specific and include nausea, vomiting and abdominal pain. There is often a palpable abdominal mass, although this was not present in our case [ 10 ]. Preoperative diagnosis is often difficult and several differential diagnoses must be considered. As this tumor has a high risk of recurrence, a close follow-up should be performed in every patient. Adenomyomas are common in the stomach, gallbladder and jejunum but are very unusual in the common bile duct or the ampullary region [ 11 ]. Adenomyomas of the extrahepatic bile duct are defined as non-neoplastic, tumor-like localized lesions characterized by glandular and myomatous hyperplasia without cellular atypia [ 12 ]. In the absence of gallstones, extrahepatic bile duct adenomyomas often stay dormant for long periods and may eventually cause vague symptoms due to local pressure [ 13 ]. They also cause obstructive jaundice, as in this case, which can easily be misinterpreted as a malignant neoplasm. It is difficult to differentiate between biliary adenomyomas and other malignant lesions by radiology (CT). Since malignancy is often suspected preoperatively, surgery with the intent to cure should be performed. The etiology of adenomyomas of the extrahepatic bile duct and the ampulla of Vater is unknown. As different terms are used in the literature to describe the same histological lesion, the true incidence of adenomyomas of the ampullary region is unclear. In the literature, 30 cases of adenomyomas of the ampulla of Vater and 13 cases of the extrahepatic bile duct have been reported [ 11 , 13 ]. Adenomyomas are usually diagnosed by histopathologic examination after surgery. As in our case, it often turns out to be a surprising diagnosis. Undifferentiated carcinoma, sarcomatoid variant, is a rare variant of pancreatic tumors. The malignant tumor arises de novo or occurs in association with mucinous cysts or other pancreatic neoplasms. Men are affected more often than women with a ratio of 3:1 and a mean age at diagnosis of 63 years. The most common symptoms are fatigue, nausea and vomiting, weight loss and abdominal pain. As in our case, radiologic imaging often shows an aggressive pancreatic mass which is not specific for undifferentiated carcinoma. Undifferentiated carcinomas are usually very large, aggressive neoplasms (average size 9-10 cm size). The prognosis is often limited to several months after resection [ 14 ]. Our patient remains in good physical condition seven months postoperatively, although he did develop two new liver metastases under chemotherapy which were resected. In contrast to our case, metastases are usually present at the time of diagnosis, the most common sites being lymph nodes, liver and lung. Surgical resection is the treatment of choice, although most of these neoplasms have already metastasized in the majority of patients by the time of diagnosis [ 15 ]. Histologically, the neoplasm may resemble a sarcoma due to the spindle cells. In contrast to a carcinomasarcoma, which consists of a mixture of glandular and spindle cell differentiation, an undifferentiated carcinoma, sarcomatoid variant, has only a prominent spindle cell differentiation. Cytokeratin is expressed in over 80%. Sarcomatoid carcinomas can be differentiated from carcinosarcomas which immunohistochemically show cytokeratin and vimentin reactivity and are, therefore, not considered to be true carcinomas [ 16 ]. The spindle cells of sarcomatoid carcinomas often label for actin, although desmin is not frequently expressed. In our case desmin was expressed and actin was negative.

Conclusion In light of these factors, complete resection offers the only potentially curative treatment option in malignant tumors of the head of the pancreas. A preoperative diagnosis could not be obtained in any of our cases without a biopsy. Although adenocarcinoma of the pancreas was suspected preoperatively in each case, the final pathology demonstrated unexpected and unusual diagnoses. Due to the suspicion of a carcinoma, partial pancreaticoduodenectomy was justified in all three patients. Although the Whipple procedure is still considered to be a major surgical intervention with high morbidity and low long-term survival, partial pancreaticoduodenectomy is indicated for all suspicious tumor-like lesions of the head of the pancreas. These unexpected pathologic diagnoses underscore the fact that confirmatory evidence of suspicious diagnoses by surgery is still the gold standard.

Introduction The standard operation for carcinoma of the pancreatic head is a partial pancreaticoduodenectomy. Unusual histological findings may occasionally occur in the surgical specimen. We present three unusual histologic diagnoses after pancreaticoduodenectomy. Case presentations In the first case, an 86-year-old Caucasian woman was admitted with abdominal pain and nausea. Preoperative evaluation showed a 3 cm cystic lesion in the head of the pancreas. Pathology revealed a benign multicystic mesothelioma. In the second case, a 45-year-old Caucasian man complained of nausea, vomiting and general malaise for several months. Endoscopic retrograde cholangiopancreatographic examination and a computed tomography scan showed a stenosis of the distal bile duct secondary to a mass in the head of the pancreas and duodenum. Histology showed an adenomyoma of the ampulla. In the third case, a 59-year-old Caucasian man presented with chronic alcoholic pancreatitis. A computed tomography scan revealed a 3.5 cm lesion in the head of the pancreas with cystic and solid components. Pathology showed an undifferentiated carcinoma, sarcomatoid variant. Conclusion Partial pancreaticoduodenectomy is usually performed for ductal adenocarcinomas, neuroendocrine tumors or chronic pancreatitis. Compared to the majority of the above diagnoses, the three cases in our study are very rare. Benign multicystic mesothelioma is a very rare tumor that originates from the peritoneum. Although it demonstrates a benign clinical behaviour, it frequently recurs after resection. Adenomyoma of the bile duct or ampullary region is a very unusual, benign, localized lesion characterized by adenomyomatous hyperplasia. Undifferentiated carcinoma, sarcomatoid variant, is an aggressive tumor and is characterized by spindle cells. As the lesions were suspicious for carcinoma, partial pancreaticoduodenectomy was justified in all three patients. The histologic diagnosis after partial pancreaticoduodenectomy may differ from the preoperative and intraoperative findings. These cases demonstrate that a definitive diagnosis may only be obtained by a pathologic examination of the surgical specimen.

Case presentations Case 1 An 86-year-old Caucasian woman was admitted with nausea and abdominal pain. Abdominal and endoscopic ultrasound, as well as a computed tomography (CT) scan, showed a 3 cm cystic lesion in the head of the pancreas with a dilated pancreatic duct and regional lymphadenopathy (Figure 1 ). Ultrasound-guided fine needle aspiration showed no malignant cells. Preoperative CA19-9 level was 54.1 U/mL (reference value < 27 U/mL). Nevertheless, based upon the morphological and radiological findings, a malignant cystic pancreatic tumor (for example, cystadenocarcinoma) was suspected. Intraoperatively, a firm, cystic lesion was found in the head of the pancreas with suspicious infiltration of the superior mesenteric vein (SMV). We were unable to exclude malignancy on frozen section, so we performed a modification on Whipple procedure with a partial resection and end-to-end anastomosis of the SMV. The postoperative course was uneventful. Ten months later, the patient had no signs of recurrence. A gross 2.5 cm tumor of the head of the pancreas was described. Histopathology showed multiple small cysts covered by cubic mesothelial cells with uniform, small nuclei (Figure 2A and 2B ). Immunohistochemistry revealed a strong immunoreactivity for cytokeratin, vimentin and CK5. CD31 and CD34 immunoreactivity was not expressed by the cystic epithelium. There was no invasive growth or malignancy. Based on the histopathological and immunohistochemical findings, the final diagnosis was benign multicystic mesothelioma. Case 2 A 45-year-old Caucasian man complained of nausea, vomiting and general malaise for several months. CA19-9 was within normal range (20.5 U/mL). A CT scan of the abdomen and endoscopic retrograde cholangiopancreatographic (ERCP) examination showed a mass in the head of the pancreas and duodenum with a stenosis of the duodenum and distal bile duct (Figure 3 ). Due to the duodenal stenosis and the recent onset of chronic pancreatitis we performed a duodenum-preserving head of the pancreas resection. There was severe chronic inflammation of the head of the pancreas and body intraoperatively. There was no evidence of peritoneal dissemination or hepatic metastases. An intraoperative frozen section showed atypical cells in the resection margin which could not be clearly identified as benign. Thus, the operation was extended to a pancreaticoduodenectomy. The patient recovered and was discharged on postoperative day (POD) 13. A macroscopic examination revealed a 5.5 cm lesion in the ampulla of Vater. Microscopically, multiple hyperplastic glands and cysts were found. They were covered by a single-layer epithelium, consisting of cuboidal, columnar and, in part, mucinous cells. Hyperplastic glandular lobes were surrounded by hyperplastic mesenchymal tissue which consisted of muscle fibers, fibroblasts and myofibroblasts (Figure 4A and 4B ). There was no invasive component. The lesion was described as an adenomyoma of the ampulla. Case 3 A 59-year-old Caucasian man presented with chronic alcoholic pancreatitis. An ultrasound and a CT scan of the abdomen showed a 3.5 cm lesion of the head of the pancreas with cystic and solid parts (Figure 5 ). The tumor marker CA19-9 was strongly elevated (74.3 U/mL). Due to our suspicion of pancreatic carcinoma on a background of chronic pancreatitis, we performed a partial pancreaticoduodenectomy with radical lymphadenectomy. The postoperative course was uneventful and the patient was discharged on POD 13. The patient was given chemotherapy with gemcitabine according to the protocol for adenocarcinoma following surgery. On a follow-up CT scan eight months after surgery two new liver metastases were discovered. As the patient was in good health, we resected the two liver metastases by the left hemihepatectomy. The patient tolerated the second operation well. Macroscopically, a 3.7 cm cystic malignant tumor of the head of the pancreas was found. Microscopically, the malignant tumor contained undifferentiated spindle cells (Figure 6A and 6B ). All lymph nodes and resection margins were tumor free. Immunohistochemistry stains were strongly positive for vimentin. Some cells expressed cytokeratin, CK5, desmin and myogenine (Figure 6C ). The proliferation index was 30%-40%. Tumor cells were negative for CD34, c-kit, sm-actin and S100. In conclusion, the diagnosis of an undifferentiated carcinoma, sarcomatoid variant was established. The Union for International Cancer Control classification was pT2, pN0 (0/12), pM0, G4 and R0. Abbreviations CT: Computed Tomography; ERCP: Endoscopic Retrograde Cholangiopancreatography; CK: cytokeratin; CD: Cluster of differentiation; CA19-9: carbohydrate antigen 19-9; POD: postoperative day; PPPD: pylorus-preserving pancreaticoduodenectomy. Competing interests The authors declare that they have no competing interests. Consent Written informed consent was obtained from all patients for publication of this case report and accompanying images. A copy of the written consents is available for review by the Editor-in-Chief of this journal. Authors' contributions NL was involved in the conception and design and wrote the first draft of the manuscript. KC was involved in the preparation and review of the manuscript, the literature review and made a substantial intellectual contribution. SEB performed the histopathological work-up and contributed to the pathology part of the manuscript. PK performed the radiological work-up and contributed to the radiology part of the manuscript. JSaE and CFE performed surgery and critically revised the manuscript. WTK was involved in the initiation of the report and made a substantial intellectual contribution to the conception, acquisition and interpretation of the data. All authors read and approved the final manuscript.

CC BY

no

2022-01-12 15:21:44

J Med Case Reports. 2010 Dec 10; 4:402

oa_package/59/5b/PMC3016302.tar.gz

PMC3016303

21176139

Introduction Dry beriberi or 'acute nutritional polyneuropathy' is considered to be rare in the western world. Rapid deterioration can occur, typically with weakness, paraesthesiae and neuropathic pain. Striking motor nerve involvement can occur, mimicking Guillain-Barré syndrome (GBS). Indeed, a GBS-like deterioration has previously been reported [ 1 , 2 ]. In the context of increasing alcohol abuse in the western world, it is possible that alcoholic neuropathy associated with abrupt deterioration due to concomitant nutritional hypovitaminosis B1 may be seen increasingly often. Case presentation Initial presentation A 49-year-old Caucasian British man with a long history of type 2 diabetes mellitus (DM) and excessive alcohol intake presented with progressive weakness of both lower limbs of approximately seven months' duration. He was morbidly obese with a body mass index (BMI) of 45. Over the same period he had noticed thinning of his muscles, particularly noticeable in his thighs. As well as lethargy, he had recently gained weight. Previously he had consumed very high levels of alcohol, but he did not admit to excessive alcohol intake in the five years prior to this presentation. He remained ambulant. Abnormalities on clinical examination included grade 4 power throughout both his lower limbs, absent ankle jerks and glove and stocking distribution reduction in pinprick sensation to his mid forearms and proximal thighs. Initial electromyogram (EMG)/nerve conduction studies (NCS) demonstrated a mild sensory peripheral polyneuropathy (consistent with long standing type 2 DM and chronic alcohol abuse). There were also patchy non-specific EMG abnormalities suggesting a non-specific myopathic process. Clinical course Rapid deterioration occurred over the course of approximately three months, with relatively rapid evolution of tetraparesis. Urgent admission was arranged, by which stage he was unable to walk. He had also developed tingling in his legs and arms. On examination, his cranial nerves remained intact but he had marked weakness of upper limbs (grade 3) and lower limbs (grade 2), with absent lower limb and reduced upper limb deep tendon reflexes bilaterally. He also had prominent sensory impairment in his upper and lower limbs, including loss of pain sensation to his waist. At one stage this was suggestive of a possible spinal sensory level. Repeat EMG/NCS four months after the first study showed evidence of a widespread polyneuropathy, with sensory nerve action potentials absent or reduced and evidence of extensive acute and sub-acute denervation in his upper and lower limbs. Although an EMG showed extensive motor denervation suggestive of anterior horn cell disease, this could be ruled out due to the extensive degree of sensory impairment. Other investigations His c erebrospinal fluid was acellular with normal protein and glucose concentrations. Oligoclonal bands were absent. A muscle biopsy of his left quadriceps showed mild non-specific morphological abnormalities with mainly selective type 2-fibre atrophy, possibly associated with an element of neurogenic atrophy. This was thought to correlate with underlying alcohol-related or endocrine myopathy. There were no features of metabolic, mitochondrial, inflammatory or necrotising myopathies. A whole body computed tomography (CT)- 18 fluorodeoxyglucose (FDG) positron emission tomography (PET) scan identified no abnormality other than lobular increased thyroid intake, a lesion for which he had previously undergone a negative fine needle aspiration biopsy. Despite the possible spinal sensory level, magnetic resonance imaging of his spine was not undertaken given the absence of any other myelopathic features coupled with the confirmation of a severe neuropathy. Serum immunoglobulins revealed raised immunoglobulin M (IgM) at 3.65 g/L (normal range, 0.5-2.0 g/L). Electrophoresis identified no monoclonal band. Gamma-glutamyl transferase was elevated at 385 u/L (normal range, 1-71 u/L). No other cause was identified other than excessive alcohol intake prior to admission. A full blood count showed a haemoglobin level of 12.3 g/dL (normal range, 13-18 g/dL) and a mean cell volume of 104.3fL (normal range 82-98fL). Vitamin B 12 deficiency (191 ng/L, normal range 200-900 ng/L) had been identified when our patient initially presented with symptoms, and was treated with B 12 replacement (instituted by his general practitioner). Other relevant investigation results which were negative or did not show significant abnormalities included renal function, the remainder of liver function tests and full blood count, bone profile, coagulation screen, creatine kinase, thyroid function tests, HIV-1, HIV-2, Borrelia burgdorferi serology, and autoantibody testing for anti-nuclear, extractable nuclear antigen and anti-neutrophil cytoplasmic antibodies. Anti-glycolipid antibody testing revealed an elevated titre of anti-GM1 IgM antibody at 1700 units (normal range, 0-500 units). Glycated hemoglobin (HbA1C) was 5.9%. His vitamin B1 level was measured prior to thiamine supplementation and hypovitaminosis B1 was confirmed (45 nmol/L, reference range 66-200 nmol/L), in keeping with his rapid clinical deterioration being attributable to dry beriberi. Treatment High potency intravenous (iv) multivitamins (Pabrinex) and then oral thiamine (200 mg twice a day) were commenced shortly after admission. On the basis of his anti-GM1 antibody positivity and the remaining possibility of a primary immune-mediated polyneuropathy, he received a trial of treatment with intravenous immunoglobulin. Anti-GM1 positivity was subsequently regarded to be a probable non-pathogenic epiphenomenon. Our patient also received his usual medications for established medical co-morbidities (oral hypoglycaemics and anti-hypertensives). He received gabapentin for neuropathic symptoms. Progress Within two weeks of treatment with intravenous Pabrinex and subsequent oral thiamine, our patient started to improve significantly, with dramatic improvement in limb power and reduction in sensory loss. He was discharged six weeks after admission. When subsequently followed up as an out-patient, 15-16 months from initial onset of symptoms and approximately six months after the onset of his marked tetraparesis, he was able to stand independently and was gradually gaining confidence in walking pending a period of in-patient neurorehabilitation. On examination he had grade 5 power throughout his upper limbs, and grade 4+ throughout his lower limbs. Pinprick sensation was only reduced in his legs in stocking distribution. Reflexes remained absent in his lower limbs, and were depressed in his upper limbs, although triceps jerks were easily elicited bilaterally.

Discussion The rates of chronic and acute alcohol abuse are increasing, with 4.6% of all ill-health and premature deaths worldwide due to alcohol [ 3 ]. For individuals who chronically consume high levels of alcohol, there is often associated liver disease which impairs storage and phosphorylation of thiamine in the liver. This inadequate liver function and low thiamine level is exaggerated by poor diet and the increased requirement for thiamine due to the continued alcohol consumption. The lack of thiamine impairs aerobic glucose metabolism which in turn causes nervous system dysfunction, given its need for glucose as an energy source [ 4 ]. The neuropathy seen in most alcoholics is usually chronic, with insidious onset, is length-dependent, and is predominantly sensory. However, the physician needs to remember that in alcoholics with dry beriberi, neuropathy can present as an acute non-length-dependent process mimicking GBS. The acute denervation changes on EMG can be profuse and widespread, mimicking motor neurone disease. The thoraco-lumbar and cranial segments are often involved and this may be deceptive to the unsuspecting mind. Anti-GM1 antibodies have previously been found in the sera of patients with polyneuropathy and either insulin-dependent or non-insulin-dependent diabetes [ 5 ]. There are a number of possible explanations, including the independent occurrence of polyneuropathy and raised anti-GM1 antibody titre in diabetic patients, anti-GM1 antibody as an epiphenomenon associated with motor nerve damage (without direct pathogenecity), and anti-GM1 antibody mediated motor nerve destruction in some patients. Taking into account the temporal profile of our patient's clinical deterioration and the overall clinical context, it seems unlikely that anti-GM1 antibody contributed substantially to his rapid deterioration. Whilst a trial of treatment with intravenous immunoglobulin was given prior to the result of the vitamin B1 assay being available, we do not consider intravenous immunoglobulin treatment to have modified his clinical course. It is likely that this man's background history of DM, alcohol abuse and associated nutritional deficiencies all contributed to a multifactorial initial presentation, with dry beriberi as the cause of his subsequent rapid clinical deterioration. A recent case of dry beriberi mimicking GBS [ 2 ] has been reported, although hypovitaminosis B1 had not been confirmed in that case. Although thought to be rare, we feel dry beriberi may represent an under-recognized and increasingly common cause of acute polyneuropathy, particularly against a backdrop of increasing alcohol abuse. An increasingly high index of suspicion is likely to be needed especially given that this is a treatable condition. The clinical spectrum of beriberi encompasses dry beriberi with predominant neurological involvement and wet beriberi which mainly affects the cardiovascular system, with presentations including unexplained congestive cardiac failure [ 6 ] and lactic acidosis [ 7 ]. Bariatric surgery patients represent another group at risk of micronutrient deficiency, including B1 deficiency [ 8 , 9 ]. Thiamine deficiency can also occur in re-feeding syndrome [ 10 ], and in patients on parenteral nutritional support [ 11 ].

Conclusion A greater index of suspicion is required for nutritional hypovitaminosis B1 as a cause of rapidly worsening polyneuropathy, particularly in view of rising alcohol abuse in the western world. Confirmation of hypovitaminosis B1 by requesting the assay prior to vitamin replacement ensures accurate diagnosis and appropriate ongoing treatment.

Introduction We describe a case of rapidly progressive and severely debilitating polyneuropathy in a patient with confirmed hypovitaminosis B1, consistent with dry beriberi. Crucially, this is a treatable condition, although sometimes with incomplete recovery, but it is probably under-recognized yet increasingly common given increasing levels of alcohol abuse in the western world. Case presentation A 49-year-old Caucasian British man presented with progressive weakness of both lower limbs of approximately seven months' duration. He noted difficulty climbing stairs. He also complained of lethargy, and loss of muscle bulk, including his thighs. He had a history of type 2 diabetes mellitus and admitted prior alcohol abuse but denied excessive alcohol intake in the five years prior to presentation. Initial clinical and neurophysiological examinations were consistent with a mild peripheral neuropathy and probable proximal myopathy. However, over the subsequent four months he evolved a marked tetraparesis, with profound sensory disturbance of all limbs. Repeat neurophysiology revealed a widespread polyneuropathy with extensive acute and sub-acute denervation changes in all four limbs, and reduced or absent sensory nerve action potentials. Hypovitaminosis B1 was confirmed (45 nmol/L, reference range 66-200 nmol/L). His rapid clinical deterioration was in keeping with dry beriberi. He was treated with thiamine. Subsequent follow-up revealed slow but significant improvement, such that by 15-16 months from the initial onset of symptoms, and approximately six months after the onset of his marked tetraparesis, he was able to stand independently and was gradually gaining confidence in walking pending a period of in-patient neurorehabilitation. Conclusion A potentially wide differential diagnosis exists for this type of presentation. Confirming hypovitaminosis B1 by requesting the assay prior to vitamin replacement ensures accurate diagnosis and appropriate ongoing treatment. An increasingly high index of suspicion is likely to be required in the context of increasing levels of alcohol abuse in the western world and the possible increasing prevalence of dry beriberi.

Consent Written informed consent was obtained from the patient for publication of this case report. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Competing interests The authors declare that they have no competing interests. Authors' contributions HE and AH jointly drafted the manuscript, with OK and GL contributing. GL also undertook neurophysiological examination and contributed to the discussion of differential diagnosis. All authors read and approved the manuscript.

Acknowledgements Drs Ranjith Ramdass and Caroline Ross also undertook neurophysiological examinations on this patient and contributed to the discussion of differential diagnosis.

CC BY

no

2022-01-12 15:21:44

J Med Case Reports. 2010 Dec 21; 4:409

oa_package/0d/4b/PMC3016303.tar.gz

PMC3016304

21176182

Introduction Every year approximately 130,000 lower-limb amputations are performed in the United States, and approximately 500 amputations are carried out in the province of Manitoba (total population of 1.1 million people) [ 1 , 2 ]. Rehabilitation of a patient who has undergone amputation is an intricate process, as several factors determine successful ambulation with a limb prosthesis. These factors include pre-existing pulmonary disease, cardiovascular disease, peripheral vascular disease, diabetes, hypertension, hyperlipidemia, the status of the other limb and functional level prior to amputation [ 3 ]. The level of the amputation is also a key determinant for successful ambulation. Peripheral vascular disease accounts for over 90% of all amputations, and more than half occur in people diagnosed with diabetes [ 4 , 5 ]. Increases in the amputation rate can be expected as both the number of patients with diabetes and the number of elderly in the general population is rising, with estimated five-year survival of 30-40% after amputation [ 1 ]. The age range of the patients who undergo dysvascular amputations in North America is between 55 and 85 years. It has been estimated that the five-year survival after amputation is 30-40% [ 1 ]. The two-year survival after lower-extremity amputations is encouraging and averages at 50-60%, with most deaths attributed to the cardiovascular complications [ 6 ]. However, there is a 20-50% risk of losing the contralateral leg to the peripheral vascular disease during the four years after amputation [ 7 ]. The more proximal levels of amputation are associated with decreased survival rates [ 8 ]. Compared to normal biped ambulation, the energy costs for prosthetic ambulation are much higher. Pinzur et al . [ 9 ] reported increased energy expenditure of walking with limb prostheses over normal ambulation as follows: unilateral trans-tibial amputation 40-60%, unilateral trans-femoral amputation 90 to 120%, bilateral trans-tibial amputation 60-100% and bilateral trans-femoral amputation > 200%. Patients with severe chronic obstructive pulmonary disease (COPD) are rarely offered limb prostheses as many patients are limited by their ventilatory status and unlikely to achieve the high energy expenditure required for successful prosthetic ambulation.

Discussion This case highlights a satisfying functional outcome for a patient with trans-tibial amputation with severe COPD among other comorbidities, who is currently living in her own house and is participating in housework. Her successful outcome was secondary to oxygen therapy and optimization of underlying severe COPD. Patients with severe COPD are unlikely to achieve prerequisite high oxygen consumption levels for prosthetic ambulation because of ventilatory and gas exchange limitations. However, with supplemental oxygen, our patient was rehabilitated successfully. Only a few cases of successful rehabilitation of a patient with severe COPD have been reported in the literature [ 11 , 12 ]. Our case highlights the point that age and other co-morbidities should not be considered a barrier to rehabilitation and prosthetic fitting in patients with limb amputations. The energy required for ambulation in trans-tibial amputation is about 40-60% above normal [ 9 ]. This energy demand becomes even higher when patients have COPD and additional significant co-morbidities. Thus rehabilitation in this population is challenging, and these patients require optimization of the underlying medical condition and close medical monitoring to avoid cardiovascular complications. Generally, patients with trans-tibial amputation, whether unilateral or bilateral, cope better than those who undergo above-knee amputation. This is particularly important in patients with COPD because preserving the knee joint helps decrease the energy demands on an already taxed cardiovascular and pulmonary system. With optimization of airflow obstruction and supplemental oxygen, it is possible to achieve the high energy consumption required of prosthetic gait ambulation and successful rehabilitation. Sioson et al . [ 11 ] previously reported three cases in which they demonstrated successful rehabilitation of older adults with COPD. McAnelly et al . [ 12 ] described a case of hip disarticulation and successful rehabilitation of an individual with COPD. None of these papers reported the use of supplemental oxygen specifically for the purpose of prosthetic ambulation. Oxygen consumption during exercise can be measured by formal cardiopulmonary exercise testing. Since these patients are incapable of performing objective testing on a bicycle ergometer or treadmill, the use of an arm ergometer is suggested. We recommend formal testing utilizing an arm ergometer and oxygen supplementation to assess whether a patient can meet the required oxygen consumption criteria during such testing. Our case highlights the importance of prospective investigations to document the benefits of supplemental oxygen during rehabilitation of patients who have undergone limb amputation and have severe underlying COPD.

Conclusions Patients with lower-limb amputations with severe or advanced COPD are generally not considered candidates for prosthetic fitting and rehabilitation. However, we describe a case of a 67-year-old woman with severe COPD who was fitted with a lower-limb prosthesis and successfully rehabilitated. In our opinion, patients with severe COPD should be carefully assessed, regardless of their age and preexisting respiratory disorders, and a trial period of rehabilitation should be considered to explore the possibility of prosthetic fitting. We also suggest that the use of supplemental oxygen during rehabilitation and prosthetic gait ambulation may be of additional benefit to these individuals.

Introduction Dysvascular amputations are increasingly performed in patients with underlying cardiac and pulmonary disorders. A limb prosthesis is rarely offered to patients with severe chronic obstructive pulmonary disease because of their inability to achieve the high energy expenditure required for prosthetic ambulation. We describe a case of successful prosthetic fitting and rehabilitation of a patient with severe chronic obstructive pulmonary disease with the aid of oxygen supplementation. Case presentation A 67-year-old aboriginal woman with severe chronic obstructive pulmonary disease and hypercapnic respiratory failure underwent right trans-tibial (below the knee) amputation for severe foot gangrene. An aggressive rehabilitation program of conditioning exercises and gait training utilizing oxygen therapy was initiated. She was custom-fitted with a right trans-tibial prosthesis. A rehabilitation program improved her strength, endurance and stump contracture, and she was able to walk for short distances with the prosthesis. The motion analysis studies showed a cadence of 73.5 steps per minute, a velocity of 0.29 meters per second and no difference in right and left step time and step length. Conclusion This case report illustrates that patients with significant severe chronic obstructive pulmonary disease can be successfully fitted with limb prostheses and undergo rehabilitation using supplemental oxygen along with optimization of their underlying comorbidities. Despite the paucity of published information in this area, prosthesis fitting and rehabilitation should be considered in patients who have undergone amputation and have severe chronic obstructive disease.

Case Presentation We report a case of a 67-year-old aboriginal woman who was admitted to the rehabilitation ward. In March 2007, the patient underwent right femoral popliteal artery bypass surgery for occlusive peripheral vascular disease. Four months later, intermittent claudication recurred; she also complained of right leg pain at rest and developed ulceration of the right toes. The patient underwent right trans-tibial amputation in August 2007 because of ischemia and gangrene of the foot. This was followed by left superficial femoral artery stent placement in November 2007. Her ankle brachial pressure index was markedly reduced at 0.11 (normal, 0.95-1.2). Despite previous surgical treatment, her peripheral vascular disease progressed to gangrene of the right foot, thus necessitating the right trans-tibial amputation. The stump healing was initially delayed because of the wound infection, but eventually healed well. Her past medical history included a 61 pack-year smoking history, severe COPD, type 2 diabetes mellitus, hypertension, ischemic heart disease and a myocardial infarction three years ago treated with percutaneous coronary intervention and stent placement. The patient had a supportive husband, lived in a wheelchair-accessible bi-level home and was using a wheelchair for ambulating long distances and was mobilized with a walker for short distances. Her physical examination revealed a well-oriented individual with normal vital signs and oxygen saturation at 88% on room air. Her neurological and cardiac examinations were normal. The respiratory examination showed hyperinflation of the thorax, decreased air entry to the lung bases bilaterally and occasional expiratory wheezing. Her residual limb length was 5 cm from the tibial tuberosity and had a 15-degree flexion contracture. The incision line was well healed with an adherent scar. Her left lower extremity showed some atrophic changes: loss of hair with absent dorsalis pedis and posterior tibial pulses were noted. The popliteal pulse was palpable but weak. General strength was graded 4 to 4+ out of a maximum of 5 in both upper and left lower extremities. Laboratory investigations revealed pulmonary function tests showing severe irreversible airflow obstruction with the following findings on pulmonary function tests consistent with severe COPD: forced expiratory volume in 1 second (FEV1) was 0.54 L/s (25% predicted) and forced vital capacity (FVC) of 1.37 L (52% predicted). Arterial blood gases demonstrated compensated hypercapnic respiratory failure (PaCO 2 at 54 mmHg) and hypoxemia (PaO 2 at 58 mmHg). An echocardiogram showed a normal systolic ejection fraction at 76% with mild diastolic dysfunction. In December 2008, the patient underwent 10 weeks of in-patient rehabilitation. Her severe COPD was optimized with inhaler therapy consisting of bronchodilators and inhaled corticosteroids. Oxygen therapy was utilized during rehabilitation exercises and ambulation, with the goal being to keep percutaneous oxygen saturation above 92% during activities and rehabilitation. Following initial slow progress due to the patient's generalized deconditioning, low endurance and stump contracture, her motivation and endurance gradually improved. She was then able to fully participate in the rehabilitation program. She attended two physiotherapy sessions per day (approximately 60 minutes in length each time). Her pre-prosthetic rehabilitation program included general upper- and lower-extremity strengthening and conditioning exercises. Oxygen supplementation during exercise and ambulation greatly facilitated the rehabilitation. We were able to improve the stump contracture from 15 degrees to 10 degrees, and she was able to hop with the aid of a walker. She was casted for a custom trans-tibial patellar tendon-bearing prosthesis with a 1.5-mm silicone liner (ICEROSS) and sleeve suspension system with a dynamic solid ankle cushion heel (SACH) foot. With further gait training, she was able to ambulate 200 feet with the aid of a walker and was discharged to home. At the time of hospital discharge, kinematic data were collected using the VICON motion analysis system to capture the kinematics of the lower limbs and the spatio-temporal parameters of her gait. The patient walked independently with supplemental oxygen using a two-wheeled walker. It was unknown how much weight bearing occurred through the upper extremities during the level walking trials. Her cadence and velocity were very slow compared to 76- to 87-year-old community-dwelling older adults (Table 1 ), but there was no difference between the left and right step time and step length [ 10 ]. However, the time spent in the right single support of the gait cycle was considerably less than the time spent on the left single support of the gait cycle. At a follow-up visit at six months, the patient had returned to her previous activities. She lived independently, ambulated and performed activities of daily living with the use of her prosthesis. Competing interests The authors declare that they have no competing interests. Consent Written informed consent was obtained from the patient for publication of this case report. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Authors' contributions JS analyzed and interpreted the patient data regarding the hospital progress and rehabilitation. AA supervised the prosthetic fitting and rehabilitation. AA and SS were major contributors in writing the manuscript. All authors read and approved the final manuscript.

CC BY

no

2022-01-12 15:21:44

J Med Case Reports. 2010 Dec 22; 4:410

oa_package/c5/29/PMC3016304.tar.gz

PMC3016305

21143898

Substrains of the Spontaneously Hypertensive rat (SHR), a putative animal model of Attention-Deficit/Hyperactivity Disorder (ADHD), have demonstrated increased sensitivity to many drugs of abuse, including psychostimulants. Therefore, it was suggested that studies in SHR may help elucidate ADHD and comorbidity with substance use disorder (SUD). However, the drug intake profile of the SHR in the most relevant animal model of drug addiction, the self-administration (SA) test, and its response on the conditioned place preference (CPP) paradigm are not yet determined. In the present study, we employed SA and CPP tests to investigate the reinforcing effects of the psychostimulant methamphetamine in an SHR substrain obtained from Charles River, Japan (SHR/NCrlCrlj). Concurrent tests were also performed in Wistar rats, the strain representing "normal" heterogeneous population. To address if the presence of ADHD behaviors further increases sensitivity to the rewarding effect of methamphetamine during adolescence, a critical period for the onset of drug abuse, CPP tests were especially conducted in adolescent Wistar and SHR/NCrlCrlj. We found that the SHR/NCrlCrlj also acquired methamphetamine SA and CPP, indicating reinforcing effects of methamphetamine in this ADHD animal model. However, we did not observe increased responsiveness of the SHR/NCrlCrlj to methamphetamine in both SA and CPP assays. This indicates that the reinforcing effects of methamphetamine may be similar in strains and that the SHR/NCrlCrlj may not adequately model ADHD and increased sensitivity to methamphetamine.

Findings Attention-Deficit/Hyperactivity Disorder (ADHD) is a complex neurodevelopmental disorder characterized by the core symptoms such as hyperactivity, inattention and impulsivity [ 1 , 2 ]. It is the most commonly diagnosed disorder of childhood [ 3 ] and also present in about 4%-9% of youths [ 4 , 5 ] and 4% of adults [ 6 ]. ADHD is comorbid with substance use disorder (SUD) [ 7 , 8 ] and epidemiological data not only affirm the ADHD-SUD link but also indicate greater risk for earlier onset of substance abuse among ADHD individuals [ 4 , 9 - 11 ]. The exact etiology, however, cannot be determined for in some respects, there is a lack of appropriate animal models [ 12 ]. The Spontaneously Hypertensive rat (SHR), bred from the normotensive Wistar Kyoto (WKY) rat strain, is the most validated animal model of ADHD [ 13 ]. A number of SHR substrains exist [ 14 ] and Sagvolden et al [ 15 ] asserted that the SHR obtained from Charles River, Germany (SHR/NCrl), with the WKY bred from Harlan, UK (WKY/NHsd) as the reference strain, most excellently represents ADHD. Nevertheless, earlier studies have found increased reactivity of some SHR substrains (e.g. SHR/NTac, SHR/NCrl) to stimulants, opioids, alcohol and other addictive drugs in comparison with other rat strains [see [ 12 ] for review], indicating that the SHR, in general, may be used to investigate the relationship between ADHD and drug addiction. However, the behavioral profiles of the SHR in the most relevant drug addiction assay, the self-administration (SA) paradigm [ 12 ] and in another animal model of addiction, the conditioned place preference (CPP) protocol [ 16 ] are not completely known. In this study, we conducted SA and CPP tests in an SHR substrain obtained from Charles River, Japan (SHR/NCrlCrlj) (via Orient Bio., Korea), to investigate if it shows stable responding for the stimulant methamphetamine. Its response was compared with the Wistar rat, strain representing the "normal" heterogeneous population [ 17 - 19 ]. Adolescence is a risk factor for addiction [ 20 ] and it is known that children with ADHD are more vulnerable to use illicit drugs than children without it [ 21 , 22 ]. We would like to know if we can model this in our study, thus CPP tests were conducted in adolescent Wistar and SHR/NCrlCrlj. All experiments complied with the Principles of Laboratory Animal Care (NIH) and the Animal Care and Use Guidelines of Sahmyook University, Korea. We used adolescent (PND 20-47) and adult (PND 90-110) male outbred Wistar and inbred SHR/NCrlCrlj. They were housed collectively (8 rats/cage in CPP experiments) or individually (SA experiments), in a temperature (22 ± 2°C) and humidity- (55 ± 5%) controlled animal room with a 12 hr/12 hr light/dark (6 AM-6 PM) cycle. They had free food and water access except during tests and on initial lever-training sessions (SA test). Methamphetamine hydrochloride, procured from Korea Food and Drug Administration, was dissolved in saline for use in all experiments. Self-administration tests were carried out in standard operant chambers (Coulbourn Instruments, Allentown, PA, USA) and the methods employed were as outlined previously [ 23 ]. Briefly, after rats acquired stable lever responding, they were implanted with silastic catheters in the right jugular vein. SA tests commenced after they have recovered from surgery. Two levers were present during SA tests and response on the left lever (active lever, FR1) switched on the infusion pump for 10 s, delivering 0.1 ml of 0.25 mg methamphetamine, and also the stimulus light above it (which was lit for 10 s and 20 s more after the end of the infusion). A time-out period was indicated and no priming injections were given at any time. Rats were only allowed up to 30 infusions although lever responses were recorded until the end of the 2-hour SA. Catheter patency was ensured as detailed [ 23 ]. Conditioned place preference tests were conducted in two-compartment polyvinylchloride boxes having distinct visual and tactile cues. The methods were patterned after previous reports [ 23 , 24 ] with some modifications. After determining the rats' initially-preferred compartment, approximately half of the rats per group were assigned to the black compartment as the drug-paired side, while the other half to the other [ 24 ]. If their staying time was less than 200 s, they were excluded from further testing. Conditioning phase followed where animals were paired with methamphetamine (1.25 or 5 mg/kg) or saline in their non-preferred compartment. The rest of the procedures were performed as described [ 23 ]. Post- and preconditioning staying times were determined using automated systems (Ethovision Noldus IT b.v., Netherlands). Two-way ANOVA was used to determine the effects of strains or days or interaction between these factors in SA tests (active vs inactive lever presses, and total number of methamphetamine infusions) and also in CPP tests (strains or treatment effects and interaction between them). Student's t -test was used for further analysis. The significance level was set to p < 0.05. Figure 1A shows that both strains acquired methamphetamine SA. Responses for the active vs inactive lever differed significantly in Wistar (F (1,50) = 59.92, p < 0.001) and in SHR/NCrlCrlj (F (1,50) = 42.26, p < 0.001). However, two-way ANOVA only revealed significant effect of days (F (4,50) = 5.69, p < 0.001), but without any strain × days interaction. In Figure 1B , the number of methamphetamine infusions earned by SHR/NCrlCrlj and Wistar rats during the SA tests is shown. Significant days (F (4,50) = 7.01, p < 0.001) but not strain effects was noted, and there was no strain x days interaction. Figure 2 illustrates that CPP was expressed by both strains in response to methamphetamine conditioning. There was a remarkable effect of treatment (F (2,42) = 6.46, p < 0.05), but no strain effects, and strain × treatment interaction. In Wistar rats, there was CPP to 1.25 mg/kg (t (14) = 1.81, p < 0.05) and 5 mg/kg (t (14) = 2.17, p < 0.05) methamphetamine. CPP was also demonstrated by SHR/NCrlCrlj conditioned with 1.25 (t (14) = 3.33, p < 0.01) and 5 mg/kg (t (14) = 2.96, p < 0.01) methamphetamine. This study showed that the SHR/NCrlCrlj, alike the Wistar rat, readily acquired methamphetamine self-administration. Adolescent SHR/NCrlCrlj also demonstrated conditioned place preference to methamphetamine although in both assays differential strain response was not observed. It indicates that the reinforcing/rewarding properties of methamphetamine are similar in both strains and more importantly, that the SHR/NCrlCrlj may not be more sensitive to methamphetamine reward/reinforcement. The results herein do not conform to previous investigations. Initially, it has been predicted that the SHR would show increased behavioral response to addictive drugs not only for their ADHD-like behaviors, but also as it manifests high levels of impulsivity, novelty-seeking behaviors [ 25 ] and defective/deficient reinforcement processes [ 26 ]. These behaviors predispose greater vulnerability to drug abuse as known in human studies [ 27 , 28 ]. Secondly, our results do not coincide with previous studies (which used locomotor activity as an index to the reinforcing effects of drugs) which demonstrated increased behavioral response of an SHR substrain (SHR/NTac) to amphetamine, methylphenidate and the selective D1 receptor agonist SKF-81297, as compared to Wistar Kyoto rats (WKY/NTac) [ 29 - 32 ]. However, one study has found similarity of the effects of amphetamine or GBR-12909, a dopamine reuptake inhibitor, in another SHR substrain (SHR/Cbp) and in WKY/Cbp [ 33 ]. The present inconsistency could be due to the following factors: (1) difference in procedures ( i.e . locomotor activity and SA tests measure different aspects of reinforcement mechanism) and (2) difference in SHR substrains used ( i.e . the SHR substrain used herein may deviate phenotypically or genetically from SHR/NCrl bred in the USA or elsewhere) [ 15 ]. Nevertheless, it has to be considered that stimulants act differently in exerting their reinforcing effects [ 34 ]. This factor may also have influenced the present results. At any rate, additional investigations should be performed to resolve this discrepancy. Our CPP results do not model increased vulnerability of ADHD children to drugs of abuse in comparison with "normal" children [ 21 , 22 ]. However, our data may represent a previous study which showed no difference in the rates of drug abuse or dependence to individual substances in both ADHD and non-ADHD controls [ 35 ]. Of interest is that our CPP data agree with those obtained in SA test. This is notable considering that CPP and SA are believed to be dissimilar forms of drug reward [ 36 - 38 ]. In summary, we have shown that the SHR/NCrlCrlj also shows stable responding and CPP to methamphetamine. However, we have not observed increased sensitivity of this substrain to methamphetamine reward/reinforcement relative to Wistar rats, showing that it may not fully represent ADHD and increased vulnerability to methamphetamine. Concerning methamphetamine-induced CPP in SHR/NCrlCrlj, it cannot be attributed to an anxiolytic effect of repeated methamphetamine treatment [see [ 39 ]], as drug-induced anxiolysis (which may influence place preference) is not always apparent with drug-induced decreases in central serotonin [ 40 ]. It could not be due to "response to novelty" [ 36 ] as the CPP procedure was modified to minimize the effect of this confound. In future studies, it would be worthwhile to compare the behavioral responses of the SHR/NCrlCrlj or SHR/NCrl (of Charles River, Germany) with its most appropriate control strain, the WKY/NHsd, in both SA and CPP assays. Such work may not only determine the reliability of the present conclusion, but may also give clues on the contribution of neurobiological differences (seen in ADHD and non-ADHD individuals) in response to the reinforcing effects of drugs of abuse. Competing interests The authors declare that they have no competing interests. Authors' contributions ID and JHC conceived and designed the study. ID, HSA and JYC performed the SA and CPP experiments. ID performed the statistical analysis while SCY, JHR JHC provided insights on interpreting the results. All authors participated in drafting and in writing the final version of the manuscript.

Acknowledgements This research was funded by grants from the Korean Food and Drug Administration (09172- 응용연 671) and Sahmyook University.

CC BY

no

2022-01-12 15:21:44

Behav Brain Funct. 2010 Dec 9; 6:72

oa_package/0e/2d/PMC3016305.tar.gz

PMC3016306

21134248

Background Low back pain is a major cause of sickness absence and work disability. The majority of workers who take sick leave due to back pain return to work quickly, but many will continue to experience pain [ 1 ] and between 18% and 44% will have a further episode of absence within the year [ 2 ]. Previous studies have described the occupational factors that contribute to work disability as 'blue' and 'black' flags [ 3 ]. Blue flags refer to workers' perceptions of their work and workplace that can impede recovery, for example a belief that work is harmful. Black flags are actual workplace or work-related influences, for example managers' negative attitudes. Shaw and colleagues [ 3 ] have recently identified seven of these flags as 'core' workplace variables to be used in the assessment of occupational obstacles, one being the ability to modify or adjust work. In the UK, a recent review [ 4 ] urged that workplaces 'consider potential adjustments which could enable employees to remain in or return to work while recovering from ill-health'. Work modifications may include changes to the workplace, equipment, work design and organisation, working conditions and/or work environment [ 5 ]. There is evidence that modified work can reduce or avoid sickness absence, increase return to work rates and job retention and decrease the recurrence of symptoms [ 6 - 9 ]. Modifications should be temporary; the aim should be for a gradual return to normal duties. Advice on the management of workers with musculoskeletal disorders is available to employers in the UK through the Health and Safety Executive [ 10 ] and the Faculty of Occupational Medicine [ 11 ]. In addition to the availability of modified work, these also support early reporting of symptoms, prompt assessment and management of obstacles to recovery, and good communication between the 'key players'. More recently, combined clinical and occupational guidance has been published [ 12 ]. Previous research has identified that the involvement of clinicians and therapists in advising on work modifications can be beneficial [ 13 ]. However, it is not known if such approaches are common in the UK nor which employers implement them, and there is little evidence that UK General Practitioners (GPs) and clinicians are involved. Current research suggests the structures to support joint working do not exist [ 4 , 14 , 15 ]. Responsibility therefore lies with the employer, employee, and any occupational health service provided by the employer (the latter are not routinely available in the UK). Although there has been some quantitative research conducted into workplace interventions for low back pain, there is little from the patient perspective. A greater understanding of experiences in attempting to remain at work with back pain may inform future workplace management of this condition. The aim of this study was therefore to explore employed patients' experiences and perceptions of work, prior to attending a rehabilitation programme.

Methods Research design, sample and recruitment Data were collected through individual semi-structured interviews with a convenience sample of low back pain patients who had been referred to multidisciplinary rehabilitation. The purpose of the interview was to enable each participant to report their individual experience of working with back pain. Participants were recruited to the study by clinicians from a multidisciplinary back pain rehabilitation team during routine initial assessment, following referral by the patient's GP or other healthcare professional. The eligibility criteria were the participant was 1) fluent in English 2) had been offered a programme of rehabilitation and 3) had responded positively to a written statement 'I am concerned about my ability to work due to low back pain' completed at the start of their initial assessment. A total of 25 patients participated in the study. Thirteen were female, twelve male. They represented both private and public sector; professional, skilled, semi-skilled and unskilled workers, manual and non-manual occupations. Twenty worked for large enterprises (>250 employees), three worked for small enterprises (<50 employees) and two were self-employed. Of the twenty who were employed by large enterprises, ten worked in the private sector, ten in the public sector. Six were off sick at the time of the interview. Six had never taken sick leave for back pain and eleven had taken no sick leave for back pain in the previous six months. The mean age was 44.7 yrs (range 22-58 yrs), and mean history of back pain 6.8 yrs (range 3 mths-35 yrs). Further detail of the research design has been published elsewhere [ 16 ]. Ethical approval was granted by the Nottingham 1 Research Ethics Committee. Data collection Interviews took place during July and August 2008, either at the participant's home, workplace, or at the office base of the rehabilitation team. Interviews lasted approximately 45 minutes and were digitally recorded. Written consent was obtained at the interview. A list of topic areas was developed through a review of the literature, by discussion with two user representatives, and between the authors. Topics included the experience of working with back pain and the help received in managing their symptoms at work and in remaining at work. The list was prepared as a guide for the interviews to ensure that the same topics were covered in each, but open questions allowed participants to add further individual experiences and observations; amendments and additions to the guide were made in response to new topics arising as the interviews progressed. Data analysis All of the interviews were conducted and recorded by the interviewer and transcribed verbatim. A qualitative software package NVivo8, (QSR International Pty Ltd) was used to manage the data which were analysed thematically [ 17 ]. Initial codes were refined following constant comparison of the transcripts. Themes were identified and analysed by repeated study of the scripts and discussion with the research team as the data collection proceeded. Two of the researchers (CC and PJW) then reviewed and agreed the themes and findings.

Results A number of themes and sub-themes were identified, as shown in Table 1 . These themes are illustrated in the text with quotations from the participants. Work modifications: assistance from Occupational Health A service for employers rather than employees The majority of the participants in this study worked for large employers (>250 employees) who are more likely to provide an occupational health service. Some of those who had accessed occupational health reported positive experiences and examples of practice which reflected current occupational guidelines [ 10 - 12 ]. However, several participants were unsure whether there was such a service, or what it might offer them. It was usually accessed through referral by (and at the discretion of) the line manager; a service that the employee might be 'sent to' or that the employer was 'willing' for the employee to access. Agreeing to attend was seen as a necessary procedure to be followed, for example: 'I've been to occupational health at work, and basically been compliant throughout the whole thing.' (10 Male) The view that occupational health was employer-orientated could result in a lack of trust. Employees might have doubts over confidentiality, or whether it might affect their job security if a judgement was made that they were not fit to work. This participant describes why she had chosen not to use a telephone help-line: 'It says that it's private and confidential, but I do know for a fact that it goes back to your managers. Which to me is wrong.' (25 Female) Occupational health was generally perceived as an absence management procedure, associated with 'return to work' interviews following a period of sick leave, rather than a service for supporting people to remain at work. Two participants on sick leave for six weeks at the time of the interview expected to be referred for their first consultation on their return, not before. In the time between her clinical assessment and the study interview the following participant had been retired on ill health by her employers, having been off sick for a year. Modified work did not appear to have been considered by occupational health, despite her motivation to return to work: 'Well, I had to go and see a private doctor from XXXX and then another one. I had to see two separate ones. And they did all the same that everyone else has done. They all say that it wouldn't be advisable for me to do the job... probably they are right because I'm still a little bit nervous in case that goes again. Nobody's ever told me it won't, you know, so I suppose.. if it did happen, and they'd let me go back to work - everybody's frightened of suing. I think I could have probably gone back to it myself...' (18 Female) Of those who had received consultations, these were generally conducted away from the work-site. Only one participant described a visit by an occupational physician to look at his work environment; another participant, a staff nurse, had been promised a visit but it had not taken place. Assessment of participants' ability to do their job was generally through discussion rather than observation; some also reported a limited test of physical function such as bending. One participant questioned the validity of an assessment which had been conducted by telephone. Advice may be over-cautious From the descriptions that participants gave, the advice received as a result of the consultations varied in its adherence to occupational guidelines [ 10 - 12 ]. Guidelines recommend temporary modifications to enable a graded return to normal duties, without raising fears about further pain, or causing 'damage' to the back. However it seemed that some participants had been advised to avoid certain duties rather than being gradually exposed to normal activities. This was particularly the case in tasks thought to be more closely associated with back pain, and where back pain was perceived to have started at work. Two participants had become involved with Occupational Health following accidents at work. One had had to contact them himself, but appreciated their support and found them effective in advising him on a phased return to full duties, although the underlying message seemed to be that he should be careful about lifting and six months later he was still on 'light' duties: 'Occupational Health came in at work, and they said "no, don't do anything, just do 'light' duties" - you know - computer stuff - recommended to HR what I should do and things - and said keep on light duties for another - month I think he said and then he's going to come in and assess the jobs and things and say whether or not he thinks that they're suitable.' (6 Male) The other participant was still on 'light' duties over a year later and was in the process of applying for disability benefits. In her case their advice on modifications had also helped her remain in employment, but not return to full duties, and implementation seemed to largely rest with her manager. Another participant had referred herself to Occupational Health as she was keen to return to work after four months sick leave. Their advice on modifications had helped her to return to work, but again, implementation of their advice largely rested with the manager, and advised restricted lifting with no apparent indication of when this arrangement should change: 'It's like I had to refer myself to OH - and that's the only time I've got anywhere. They said I am better off going back to work if at all possible. They actually wrote and said that I should be fine going back to work, part-time........I'm not allowed to pick - 10 kg or something - the (occupational health) doctor put it on the letter.' (26 Female) Less common was the experience this participant reported by a care worker who had felt reassured by her consultation that modifications were not required: 'I saw (occupational health physician) and you know he talked through everything with me, examined me, and he wrote a letter to my manager and sent me a copy, and said that I could carry on with normal work activities. He felt that my back wouldn't stop me doing anything, but if I did something to aggravate it, it wouldn't make it worse - I'd just be in extra pain for a few days .' (5 Female) However, this participant had taken minimal sick leave for her back pain. She had been given this advice through a consultation triggered by a period of 8 weeks sick leave due to depression, not back pain. Influence may vary and may be dependent on perceived causation There were different experiences in the extent to which the advice of occupational health would be taken up by employers. For example, this participant had been on sick leave for over six months as she was struggling to drive to work because of her back pain. She had attended more than one occupational health consultation. Her public sector employers were either unable or unwilling to act on the advice they had been given. At the time of the interview she was involved in legal proceedings against her employers over the application of reasonable adjustments as defined by the Disability Discrimination Act [ 18 ] and had been off sick for seven months. She describes one of the consultations: 'He said "the sensible solution would be to relocate her to an office closer to her home, otherwise the problem will not go away". That was the opinion. I had a meeting briefly after the report and she (her manager) said "well, there's nothing I can do, there's no jobs there". Problem is, employers can just ignore what they say - and they have done, throughout the whole of this.' (12 Female) The two participants whose back pain had followed workplace accidents indicated that the response of their employers was associated with the perceived cause of their pain: 'But he's been actually pretty good (Occupational Health Physician) - he's given quite good advice I think and given them a bit of a kick as well actually - when he came in he just looked around and said "Oh God that's awful - you shouldn't be doing that" - and took the boss into the office - and it was quite nice really that there was.. ' There was someone looking out for you?' Yes.' (6 Male) And 'I think they're more understandable (understanding) because it's been done at work. They're more lenient. I think if I'd have done this - say I was gardening at home and I'd done it, then I think they'd have been more inclined to have let you go, more so than try and help you to work through it.' (24 Female) Modifications left to manager to implement It appeared rare for occupational health personnel to meet with anyone other than the patient. Usually the patient was left to act as a conduit between occupational health, their employer and their GP. In the two instances where workplace injuries were perceived to have taken place, occupational health did meet with the manager/employer; otherwise communication between the parties was by written report or letter. However written communication does not provide an opportunity for all those involved to clarify or discuss any advice or recommendations given, and how it might be implemented or evaluated. This participant described how there seemed to be an expectation from both the physician and the manager that the participant was responsible for the transmission of information: 'He said "email me your latest thing from the last meeting you had, I'll look at it, review it and then forward it to her (the manager ) and then she can read that then she's got everything there". So he was - I think he's of the opinion- that he wants to check what I'm saying and make sure that things are recommended correctly, rather than them asking me, I say something and it goes wrong. So I said "the best answer is to go through the right channel".' (6 Male) This participant highlights a similar lack of clear, effective communication: 'I haven't been back to see her since that initial consultation. It was a series of consultations and on one of them the boss wanted to sit in, I had no objection, but the Occupational Health officer did, so - it didn't happen and the boss wasn't pleased about that and she basically gave me a good grilling - "well, what did she say - what are you going to do - what's going to happen?" ' (10 Male) Work modifications: assistance from employers/managers As we have described, Occupational Health had played a limited role in modifying work for the participants in this study. Day to day management of the employee therefore largely rested with the worker themselves, their colleagues and their supervisors or line managers. It was common for the interviewees to talk about support of their colleagues, but perceptions and experiences of managers were mixed. Managers were usually responsible for applying the policies and procedures of absence management, including return-to-work interviews, health and safety management and referral to occupational health. Help depends on the individual manager Some participants had received help from their managers in making minor adjustments which had enabled them to remain at work. One 22 year old described how she had recently started work as a member of a small team in the postal department of a large company before her symptoms became more troublesome. She had changed from her normal occupation (fashion design) due to a combination of stress and back pain, where modifications had been unavailable. Her manager and colleagues in this new job had agreed that when she was in more discomfort they would take a greater share of the heavier manual handling tasks, and were happy for her to take more of their share of computer-based tasks in return. This informal and flexible arrangement had enabled her to feel productive rather than a burden to her colleagues. Similarly, a librarian described how temporary work adjustments agreed with her manager had meant that she was able to reciprocate: 'Well it's a team effort really - I'm doing things that other people aren't. If anybody needs to do my allocation of shelving I'm doing something for them in return.' (21 Female) However, if duties were reduced indefinitely, with no extra cover, workers might feel that they were burdening their colleagues. There were doubts as to how long their colleagues support might continue. This participant had injured her back at work, and modified duties had been arranged, but she felt that she was not fulfilling her part in the team: 'I feel as though I'm useless. I just poodle about doing what I can, where I can. And the men go "Oh, bloody come out the way" if I try and do something. But they're not going to carry on doing that are they?' (24 Female) This sense might be heightened by the possibility that their colleagues could question the validity of their pain: 'I know this manager will understand, but because we work as a team its like - do they think I'm swinging the lead? But it's like letting the team down, because you want to be able to do your quota, not put more strain on the other side of the team.' (15 Male) Inability or unwillingness of employers to address low staffing levels could limit attempts to modify their workload. For this participant, low staffing levels were compounded by a culture that made it difficult to ask for modifications in the form of postural changes. This participant had to maintain an uncomfortable sitting position: 'During the day, can you get up and move if you wanted to?' 'Oh yeh, if I wanted to yeh, but it's the pressure of sort of having to - if I was to do that it'd be "where are you going? You've got work to get done, you ain't got time to go talking' and stuff like that!" "Oh I don't mind you moving around, but you get those ten units done for the post". So you've still got to sit still.' Although he felt that his employers would be more amenable to modifying his job through the purchase of equipment: 'Oh there'd be no problem in buying a different chair - they're very good like that.' (19 Male) Whereas other office-based employees described receiving workstation assessments and modifications, this participant had not found her employers (a multinational company) at all helpful in providing her with suitable display screen equipment: 'They don't like to spend money where they think they can get away with it. I mean the chair - I just really, really forced the issue, because I said to the manager - I just can't cope with coming to work, sitting in a chair that's causing me more pain when I get home. And even now, the lap top - this is not the one I originally had - and I did say to them when the other one broke down - maybe now I'll get the monitor and a keyboard separate - "Oh well, we've got a spare one floating round at xxxxx Road, you'll have to have that lap top".'(25 Female) This office worker again had experienced little support from his manager: 'I did inform the boss about it and - because one time I was lifting from the ground and I felt something jolt in my back and I was in agony and he just said "well why didn't you report straight away?" - because I didn't do that and - he was trying to blame me, that I'd done it elsewhere - "it's not our fault".' (7 Male) It appeared to be his reduced hours, rather than modified work that had enable him to remain in work: 'I haven't been off sick with my back - because I do part-time anyway so I try and take it easy during the day - and then I can keep at work.' (7 Male) Whereas this participant's employers had agreed to her taking regular breaks from sitting which had not affected her productivity: 'Generally sitting does cause a problem, although I've got a special chair, and wrist rests and arm rests and foot rests! And I do get up every hour and walk round, and every half hour when it's really painful. Everybody's aware of why I'm doing it. Even though I'm away from my desk for 15-20 minutes an hour, I'm still exceeding my targets. So it's not impacting - I mean people who are there at their desk all the time are not hitting their targets so - I'm constantly exceeding mine.' (4 Female) May be over-cautious in their support Some managers could be overprotective, perhaps due to a sense of responsibility and their own anxieties about back pain, and encourage participants to modify their workload. Participants reacted differently in these situations; some seemed relieved that their problems were being taken account of: 'They've been very good. My immediate manager has been excellent. He's been very good. If I go in and say I can't manage it, it's "well, leave it then".' (24 Female) While others were less inclined to accept: 'She is very good. She says to me today "what are you carrying that ladder for?" I said - "feel the ladder, it's a lightweight ladder, two step". She says "Oh but you shouldn't have been carrying it". I says "I've got to do my job, you've got to let me do my job. If I can't do my job, there's no point in my being here".' (26 Female) Lack of adequate help in effective work modifications could lead to further sickness absence, even with the best of intentions. Another participant had been signed off for six weeks following a previous attempt to remain at work on 'light' duties, which failed when he went straight back to his usual duties without a gradual return. Managers with experience of back pain perceived to be more sympathetic As back pain is a common health condition, it is quite likely that managers will also have some experience of back pain. Participants generally felt that their manager was more sympathetic as a result of their own experience of pain: 'I spoke to my boss - he said "yes, take it easy". My boss, he's got long term back pain, and last time he was off with his back he had to wear a support belt and everything, and he understands what it's all about.' (3 Male) There was a sense that other managers may not be as tolerant of workers with back pain: 'I'm lucky that my line manager he has a back problem as well so he's knows what I go through.' (4 Female) Work modifications: patient control Easier to modify workload if in control In this study some participants were able and/or had chosen to modify their own duties and/or hours on an informal basis, either by themselves, or by involving their colleagues. This council worker, with a long history of back pain had pursued a combination of self management, taking a few days off work and accessing private manual therapy to remain in work. The nature of his job meant that he was able to adjust his tasks and workload to remain at work most of the time: 'As I say, I have flare-ups, but because I manage my own day, if it ain't the best then I'll stay in the office all day. It just means I'm not climbing in and out of vans all day.' (15 Male) The ability to modify his workload was also a key factor in work retention for this finance consultant. His flare-ups were now becoming more frequent, but he had been able to manage these by working flexibly: 'No I haven't taken sick leave. I work with it in the sense that say, well, I can't get into the car, go to the office, go up the stairs, I will stay here, do some calculating, phone calls. So that's how I manage it. You could argue that the way I work is self-employed.' (23 Male) Some participants reported quite minor alterations to their working methods that had helped them to manage the more physically demanding parts of their jobs, as this care assistant describes: 'So I do alter the way I do that a little bit. If I'm moving footplates on wheelchairs, everybody else just bends over - I actually get down on the floor on my knees, and they've provided me with a kneeling pad so I'm not hurting my knees on the floor.' (5 Female) This building trade worker also described how he had been able to slow down his pace of work: 'I take me time more. I used to go like a bull at a gate, so now I take me time a lot more. It has helped me back. Other than that, nothing else has changed.' (20 Male) The pros and cons of working for oneself However, working for an unsympathetic boss, and the inability to control his workload had led this participant to start up his own business: 'I'm going to pace meself with what jobs I'm doing. Not take me time as such, cos I'm always used to working at nine hundred mile an hour - but I'll be able to limit meself - do a couple of jobs a day instead of six, seven, eight jobs a day.' (22 Male) Other participants agreed that there were advantages to being self-employed. One participant with a three year history of back pain had been working part-time as a freelance IT consultant, as well as running his own property development company. Because of difficulty managing his back pain, he had given up the consultancy in order to concentrate on the latter. Although he was worse off financially, it meant that he was able to have more control over his daily routine, and delegate to his employees. 'If I hadn't been self-employed - because of the property business that I've got - but if I was actually working for somebody - I'd probably be unemployed by now.' (2 Male) Fewer options if working alone Another participant worked in catering both privately, and for an agency. To some extent she could choose how much work to take on, however once she had accepted a booking she was generally working alone without the possibility of adjustments. She considered that asking clients for help would have lessened their confidence in her ability to complete the job. 'When you're actually doing a job and you're doing a good job, if you let them start seeing well I'll have to sit down for five minutes they'll think -" oh she's not very reliable, we won't book her again we'll get somebody else".' (14 Female) Colleague support Others were able to ask colleagues for help on an informal basis when their symptoms were more troublesome. This seemed to work well when the help was available from a team, as this participant describes: 'Oh they're very good. If there's days when I can't bend down - or I sit there in the chair like this - they do things for me.' (11 Female) However for those who worked with just one other colleague, such informal arrangements appeared to carry greater risks to job retention: 'I've always gone to work, and who I work beside has been brilliant - you know when my back's been playing up he'll say don't lift those up, I'll do that . If he wasn't there I wouldn't be able to do it.' (17 Female)

Discussion Although most people who experience back pain remain at work, or return to work within a few weeks, we do not know if they are successfully managing their duties. Some may work at reduced capacity, rely on the help of colleagues, remain on adjusted duties or hours, have periods of absence for a secondary illness such as depression, take early retirement, or change occupation. If modifications are unavailable or ineffective, healthy and productive work may prove unsustainable. In this study, we found that most participants were making informal modifications to help them remain at work, either independently or jointly with their colleagues and line manager. Those who could manage their own workload or a choice of tasks had an obvious advantage. Some of these modifications were simple, and used flexibly when the need arose. Only a minority of the participants in this study had received support through occupational health services, often following a period of sickness absence. Self-referral was unusual. Experiences of occupational health varied; modifications may not have been considered or have been inappropriate or ineffective. Implementation largely rested with the line manager; there were few examples of face-to-face communication between all the parties concerned, leaving the details to the interpretation of the manager and the employee. Those whose symptoms had followed a workplace accident seemed to have received more attention, perhaps due to employers' fears of compensation claims. Such concerns may lead to an over-cautious approach. In the UK only a minority of the workforce can access occupational healthcare compared with countries within the European Union. In 2006 the Faculty of Occupational Medicine of the Royal College of Physicians reported that some member states (e.g. the Netherlands, France, Belgium, and Finland) achieve 90% coverage [ 19 ]; UK coverage was estimated at 34%. Only Greece at 28% had lower coverage, although the nature of services differs between the states. In the UK the extent of the service is determined by the costs that employers are willing/able to bear. Employees' perceptions of the confidentiality and affiliation of occupational health are also important. Reducing the number of sickness absence days that 'trigger' a referral to occupational health may lead to more effective management of musculoskeletal disorders [ 20 ], however if the service is viewed solely in connection with disciplinary procedures, employees may be reluctant to access it. Previous UK research suggests that the implementation of occupational health guidelines, particularly prompt intervention, may be hindered by organisational obstacles such as strict referral criteria [ 21 ]. Line managers have a vital role in supporting employees with health conditions such as low back pain. Their beliefs and attitudes, and the support and guidance available to them, can either facilitate or impede the employee. A recent study of line manager competencies [ 22 ] recognises that line managers are 'the key to work adjustments and implementation of work redesign initiatives' and that they require support in this role. The report concludes that managers do not need to be knowledgeable about health conditions to be effective, however, it would seem from our findings that some basic understanding of pain mechanisms may be helpful in clarifying whether tasks are 'harmful' to the back. In this study, participants considered that managers were more sympathetic if they had also experienced back pain. However, sympathy in itself did not necessarily lead to appropriate management. If a manager believes that pain should be avoided, and that heavy work is inherently dangerous, their approach may be overcautious and result in permanent restrictions. The ease with which work modifications can be made has been described as 'adjustment latitude' [ 23 ]. There is a risk however, that if workers are able to, and choose to avoid certain tasks because they think they are unsafe or will make their condition worse, then it may become a permanent arrangement and lead to reduced capacity. These 'representations' (thoughts, attitudes and beliefs) that an individual has of their condition are one of the key factors in the 'margin of manoeuvre' model described by Durand et al [ 24 ]. The findings of our study suggest that the representations held by managers and other stakeholders are also important. Much research has studied the effect of fear-avoidance beliefs of patients, GPs and other clinicians [ 25 , 26 ]. Those of employers, line managers and work colleagues are of equal importance, but feature less in the literature. In our study, participants' experiences of line managers were mixed. In her study, Foster concluded that 'employees are reliant upon the goodwill of individual line managers for successful adjustments, turning what should be a legal obligation into a personal lottery' [ 27 ]. Research conducted for the British Occupational Health Foundation [ 22 ] found that the relationship with the manager prior to sickness absence had a bearing on return to work, and suggested that the attitudes of managers were perceived by employees as varying according to the health condition. In a study of university employees, Munir et al [ 28 ] found that only 50% of those with chronic health conditions had disclosed their condition to their boss. Employees with back pain who are concerned about being seen as fraudulent or unreliable may be unwilling to disclose their condition [ 16 ]. Their need to maintain an identity of independence and ability, and/or not wanting to appear pre-occupied with their pain may be a barrier to seeking support [ 29 ] and they may perceive themselves as primarily responsible for managing their condition at work [ 22 , 30 ]. However, interventions designed to empower employees with chronic diseases suggest that it is possible to train employees to negotiate work accommodations [ 31 ]. Modifications should be temporary and involve a gradual return to full hours and duties. However the line manager may then be faced with conflicting demands if productivity levels are subsequently reduced. The effect on other workers also has to be considered. The participants in our study did not want to be a 'burden' to their colleagues, and felt more comfortable about receiving help if they were able to reciprocate in some way. Some workplaces are better able to offer modifications than others due to staffing levels and the variety of work tasks. Other research has shown that fewer options are available when the work is highly specialised, or physically demanding [ 32 ]. It may be difficult for employers to see the long term 'business case' for offering modifications and as organisations become 'leaner' there is a risk that lower staffing levels increase individual workloads with fewer options for adjustment. In our study, where modifications had been made by employers, more attention seemed to be paid to adjusting equipment, such as seating, rather than grading tasks, such as lifting, which were more likely to be avoided. Employees need to be able to consider a wide range of different types of modification, but the study by Foster [ 27 ] concluded that managers were more likely to favour the provision of equipment, rather than adjustments to work itself that might involve changes to employment conditions. Limitations and strengths This was a convenience sample of employed patients who had been referred for back pain rehabilitation. The majority had taken sick leave, some for several weeks. The participants may therefore not be representative of those who are managing their back pain more successfully at work and so limit the extent to which the findings can be generalised to a wider population. However most participants did work for large employers and might have been expected to have received greater involvement from Occupational Health in work modification. There were comparatively fewer participants recruited who were self-employed or working in small or medium sized enterprises (<250 employees). This may reflect the UK economy where although small enterprises (<50 employees) account for the majority of businesses, large employers (public and private) account for the majority of the workforce. Alternatively it may be that the pressures of working for a small employer impose actual or perceived obstacles to accessing healthcare and taking part in a research study.

Conclusions In this study, work modifications seemed to be mainly casual arrangements made between the employee, line manager and colleagues. Occupational health was usually involved only after substantial work absence. There appeared to be insufficient expertise among managers and occupational health in adjusting work, little indication of joint planning, and overcautious approaches were common.

Background Research indicates that work modifications can reduce sickness absence and work disability due to low back pain. However, there are few studies that have described modified work from the perspective of patients. A greater understanding of their experiences may inform future workplace management of employees with this condition. Methods Individual semi-structured interviews were conducted with twenty-five employed patients who had been referred for back pain rehabilitation. All had expressed concern about their ability to work due to low back pain. Data was analysed thematically. Results Many participants had made their own work modifications, which were guided by the extent of control they had over their hours and duties, colleague support, and their own beliefs and attitudes about working with back pain. A minority of the participants had received advice or support with work modifications through occupational health. Access to these services was limited and usually followed lengthy sickness absence. Implementation largely rested with the manager and over-cautious approaches were common. Conclusions There was little evidence of compliance with occupational health guidance on modified work. There appears to be insufficient expertise among managers and occupational health in modifying work for employees with low back pain and little indication of joint planning. On the whole, workers make their own modifications, or arrange them informally with their manager and colleagues, but remain concerned about working with back pain. More effective and appropriate application of modifications may increase employees' confidence in their ability to work.

Competing interests The authors declare that they have no competing interests. Authors' contributions CC, PJW and AD designed the study. CC conducted and transcribed the interviews and drafted the manuscript. CC conducted the initial data coding and analysis. PJW checked and reviewed the coding and analysis. PJW and AD revised the manuscript. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2474/11/277/prepub

Acknowledgements This work was supported by Arthritis Research UK [Primary Care Fellowship Grant ref. 17891]. The authors would also like to acknowledge the following: The patients who participated; the Nottingham Back and Pain Team who recruited the participants; the members of the steering group who contributed to the study: Ms Janet Clifford, Dr Alison Hammond, Mr George Morris, Dr Kate Radford, Dr Tracey Sach.

CC BY

no

2022-01-12 15:21:44

BMC Musculoskelet Disord. 2010 Dec 6; 11:277

oa_package/d1/66/PMC3016306.tar.gz

PMC3016307

21143887

Background Breathlessness is a common symptom which occurs in a wide range of clinical conditions and affects many individuals [ 1 ]. Some two-thirds of patients experiencing breathlessness will have a respiratory or cardiac condition [ 2 ]. Breathlessness also often occurs in cancer, affecting 50-70% of patients [ 3 ]. Both frequency and severity of breathlessness increase towards the end of life [ 3 - 5 ]. The terms breathlessness and dyspnoea are often used interchangeably and definitions focus on sensations of difficulty, discomfort and distress in breathing experienced by those affected [ 1 , 3 , 6 ], with perhaps the most authoritative definition coming from the American Thoracic Society in 1999: '...a subjective experience of breathing discomfort that consists of qualitatively distinct sensations that vary in intensity. The experience derives from interactions among multiple physiological, psychological, social, and environmental factors, and may induce secondary physiological and behavioural responses' [ 1 ]. The subjective nature of breathlessness has led to comparisons with pain as a symptom [ 3 ] -and indeed breathlessness is the most common symptom after pain for which patients seek medical help [ 2 ]. Breathlessness can have a significant impact on daily living and health-related quality of life for patients and families [ 1 , 3 , 7 - 9 ]. The experience of breathlessness may markedly reduce patients' functional ability [ 1 , 3 , 8 ], and fear of this symptom can itself lead to activity avoidance [ 10 , 11 ]. Strong associations between breathlessness and fatigue have also been noted [ 1 , 12 , 13 ], which can further reduce functional ability. There is a strong relationship between breathlessness and anxiety or panic [ 7 , 8 , 10 , 11 , 14 ], with the latter often resulting in patients who experience breathlessness undergoing emergency hospital admissions. The nature of the relationship between breathlessness and anxiety or panic is complex [ 1 , 10 ] and Bailey's [ 7 ] work in chronic obstructive pulmonary disease suggests that anxiety may be a consequence of breathlessness, rather than the reverse. Severe breathlessness may also be associated with depression, again the inter-relationship being complex [ 14 ]. Breathlessness is distressing for patients and their families [ 7 , 9 , 15 ] and may result in a heavy physical, social and emotional burden on carers [ 3 , 7 , 8 ]. Carers of breathless patients have reported feelings of helplessness, fear, isolation and frustration [ 3 , 7 , 8 ]. Social isolation is often reported in breathless individuals and their family members [ 3 , 8 ]. Importantly, breathlessness may have prognostic significance. Abidov et al [ 16 ] identified increased mortality in patients with known or suspected cardiac disease in whom breathlessness was present, while others [ 13 ] comment on the relationship between breathlessness and poor prognosis in non-heart failure populations. Associations between breathlessness and mortality have also been reported in general and elderly population studies [ 17 , 18 ]. Breathlessness is a common and distressing symptom in the advanced stages of both malignant and non-malignant disease. It is characteristic, for example, of chronic obstructive pulmonary disease (COPD), often causing marked impairment in activity [ 19 ] and confinement to the home [ 20 ]. COPD is a progressive, incurable illness and a leading cause of disability and mortality worldwide [ 21 , 22 ]. Despite this, there is evidence that services for people with COPD are poor and needs (including palliative care needs) are unmet: indeed, breathlessness in COPD has been described as 'invisible', partly because of the lack of attention it receives from health services [ 23 , 24 ]. Breathlessness is also a common and devastating symptom in advanced cancer, with prevalence increasing from referral to palliative care services (15-55.5%) to the last week of life (18-79%) and severity reported to be moderate-severe in 10-63% of cases [ 25 ]. It causes major impairment in function and quality of life (QoL) [ 26 ]. To date, however, there is limited evidence that palliative care interventions have improved breathlessness or that either medical or nursing care have provided effective assistance to individuals with this symptom [ 27 ]. Despite the wide range of conditions that cause distressing levels of breathlessness, including primary and secondary cancer, COPD, cystic fibrosis, interstitial lung disease (ILD) and chronic heart failure (CHF) it is still very difficult to manage [ 23 , 28 ].

Method Against this background we conducted an integrative review of pharmacological and non-pharmacological interventions for breathlessness within a wide range of cardiorespiratory diseases. For the purpose of this study, pharmacological interventions were defined as those classified as medicinal products under the EU Directive 2001/83/EEC, i.e. any substance or combination of substances which may be administered to human beings with a view to making a diagnosis or to restoring, correcting or modifying physiological function in human beings. Non-pharmacological interventions are those that are not classified as medicinal products. This integrative review was undertaken both to provide a clear account of the current body of work on the subject and as a preliminary step in developing further interventions for the management of breathlessness and other respiratory symptoms. Given the large number of systematic reviews available, and the limited clinical utilisation of many of the interventions found (with the exception of a few pharmacological interventions such as steroids and beta-agonists), the decision was taken to focus and limit the integrative review to only systematic reviews. We searched the following electronic databases for systematic reviews of randomised controlled trials, quasi-experimental trials, and uncontrolled trials where there were group comparisons of both pharmacological and non-pharmacological interventions for breathlessness in non-malignant disease: • Cochrane Library (Cochrane Airway Review Group) • AMED (1985-June 2007) • CINAHL (1982-June 2007) • EMBASE (1988-June 2007) • Ovid MEDLINE (1950-June 2007) • PsycINFO (1985-June 2007) Searching was carried out with a combination of assigned index (MESH) terms and text words, with truncations where appropriate and different spellings (e.g. dyspnea AND dyspnoea). Between July 2007 and September 2009, while the data of the review was collated and analysed, regular further searches were carried out using the key words in order to identify any new published systematic reviews, adding them to the main body of the current review. Databases were searched using the following search strategy 1. asthma 2. COPD 3. Chronic AND obstruction AND pulmonary AND disease 4. 2 or 3 5. Heart AND failure 6. Cystic AND fibrosis 7. interstitial MESH lung disease 8. pulmonary AND arterial AND hypertension 9. bronchiectasis 10. 1 OR 4 OR 5 OR 6 OR 7 OR 8 OR 9 11. dyspnea OR dyspnoea 12. difficult AND breath 13. breathlessness 14. short AND breath 15. short of breath 16. 11 OR 12 OR 13 OR 14 OR 15 17. Systematic AND review 18. 10 AND 16 AND 17 Systematic reviews identified in this way were reviewed independently for inclusion by two of the authors (AM, RD). Any disagreements were resolved by discussion. Many reviews did not explicitly measure breathlessness as an outcome, and for others it was not the primary outcome. For this reason reviews were eligible for inclusion in our review if they reported research on adult participants using not only our primary outcome measure of breathlessness (e.g. Borg and/or modified Borg tests, verbal categorical scales, visual analogue scales), or some other measure of respiratory symptoms (e.g. including cough, wheeze in a composite score), but also if they reported other frequently used breathlessness related outcome measures such as lung function or physiological changes (e.g. spirometry, exercise tolerance, arterial blood gases, pulse oximetry). Although these secondary outcomes may act as proxies for breathlessness, they do have some limitations. For example, lung function only moderately relates to the symptom experience of patients, and exercise tolerance may be as much a consequence of fatigue as breathlessness. Moreover, many reviews also reported 'single symptom scores', which although included several symptoms, among them breathlessness too, yielded only a single composite score. Many reviews also reported upon improvements in other key outcomes (e.g. survival, health-related quality of life) that may be linked or influenced by levels of breathlessness, and these have been included within this review. Trials of any non-pharmacological intervention were included, as well as any drug/pharmacological intervention (other than radiotherapy and anti-cancer chemotherapy), given by any route or in any dose. Reviews were excluded if they were not available in English, or were wholly or primarily concerned with the treatment of children (i.e. aged < 18 years), infections, sleep apnoea, or cough. They were also excluded if respiratory symptoms were not included as study outcomes. Data were extracted by one of the authors (RD) using a data extraction form designed for the review and was independently verified by one other author (AM). A third author (AC) had the role of arbitrator in cases of disagreement. Data extracted included the nature of the intervention, the number of trials included in the review, the aggregated sample, primary and secondary outcome measures, the condition treated, and the main conclusions. Following data extraction, systematic reviews identified were categorised by condition treated (asthma, COPD, bronchiectasis, interstitial lung disease, bronchitis, bronchoconstriction, pulmonary arterial hypertension, breathlessness (generic), and respiratory failure) and by treatment type (i.e. non-pharmacological/pharmacological).

Results We initially identified 3359 citations, with an additional 436 resulting from the updated search, giving a total of 3,795 citations. Of these, 3576 were excluded based on the title/abstract as they were irrelevant studies, reviews but not systematic reviews, duplicates, laboratory studies or individual trials which were reported in an identified systematic review. From the 219 systematic reviews identified, 66 were further excluded primarily because they were either paediatric reviews or reviews of medical devices. Of the 153 reviews that met the inclusion criteria, 59 addressed non-pharmacological interventions and 94 addressed pharmacological interventions. A summary of key findings from the review is presented below; further details of individual systematic reviews are set out in the Tables presented in Additional Files 1 , 2 , 3 . Additional file 1 presents data from systematic reviews of interventions for asthma; Additional file 2 presents data from systematic reviews of interventions for chronic obstructive pulmonary disease (COPD), and Additional file 3 presents data from systematic reviews of interventions for other respiratory diseases. Within this integrative review we have sought to draw together data from existing systematic reviews. However, few of the papers included meta-analyses. This was due to the quality of some of the studies included, variation of the interventions between and within categories, variability of the reported data and the provision of unreported data. It was therefore not possible, for the purposes of this integrative review, to summarise or draw conclusions regarding effect sizes and other statistical parameters. Asthma (non-pharmacological treatments) A total of 34 systematic reviews were identified. Six were excluded because they referred only to children. Of the remaining 28, 26 were published in the Cochrane Database of Systematic Reviews. Three included data on breathlessness [ 29 - 31 ], while all others assessed breathlessness as part of a composite score of 'asthma symptoms'. In some individual studies these symptom scores used lung function as a proxy for symptoms, others included the Asthma Quality of Life Questionnaire that included symptom categories, while others used the Asthma Symptom Checklist as an outcome. Furthermore, many of the original trials included a 'respiratory symptom' score, of which breathlessness/dyspnoea was only part. Treatments assessed within the reviews included: acupuncture [ 32 ], breathing exercises/training [ 30 , 33 , 34 ]; dietary measures [ 35 - 39 ]; Heliox [ 29 , 31 ]; homeopathy [ 40 ]; control of allergens [ 41 , 42 ]; humidity control/ionisers [ 43 , 44 ]; education programs [ 45 - 47 ]; manual therapy[ 48 ]; primary care clinics [ 49 ]; psychological interventions [ 50 , 51 ]; individualised care plans [ 52 ]; and non-invasive positive pressure ventilation (NIPPV) [ 53 ], which enhances ventilation by compensating for fatigued ventilator muscles. Of those reviews with positive findings, four reported a single study only [ 35 , 49 , 53 , 54 ]. Holloway & Ram [ 34 ] identified seven randomised or quasi-randomised controlled trials of breathing exercises for asthma, concluding that trends for improvement in QOL were encouraging, although no firm conclusions could be drawn, as only 2 of the 7 studies reviewed showed positive results. Gibson et al [ 45 ] concluded from evidence from 12 RCTs of limited (information only) education programs that symptom perceptions may improve, although actual asthma symptoms showed no significant differences. Yorke et al's [ 50 ] review of 15 RCTs of psychological interventions was unable to draw firm conclusions, but following pooling of data, reported two trials as showing positive effects on QOL after cognitive behavioural therapy (CBT). Biofeedback also appeared to have a positive effect on peak expiratory flow (PEF) in two studies. Gibson et al's [ 47 ] large review of 36 randomised trials concluded that education in asthma self-management involving self-monitoring by PEF or symptoms, together with regular review and written action plans, does improve health outcomes in adults with asthma. While the systematic reviews show that the vast majority of non-pharmacological interventions tested to date have largely been ineffective in relation to asthma symptoms, some effect has been reported in trials of calorie controlled diet [ 35 ], inspiratory muscle training [ 30 ], manual therapy (particularly massage with relaxation) [ 48 ] and psychoeducational interventions [ 50 ]. The role of Heliox is unclear, as one review showed no effect on symptoms [ 29 ], while another showed that 3 of 15 trials reviewed had a positive effect [ 31 ]. Selenium supplementation has shown positive results, but there is only one trial available in the literature that included a small sample of 24 patients [ 54 ]. The conclusions of two trials was that there was no current evidence of the effectiveness of the 'Alexander technique' for chronic asthma [ 55 ] or that feather bedding and pillows reduced asthma symptoms compared with man-made fibres [ 56 ]. Asthma (pharmacological treatments) In all, 69 systematic reviews were identified. Eighteen were excluded because they referred only to children, did not specify symptoms as outcomes, or were reviews of studies of vaccines or comparisons between devices. Of the 51 reviews included, 50 were accessed via the Cochrane Database of Systematic Reviews. Seven included data on breathlessness [ 57 - 63 ] while all others assessed breathlessness as part of a single score of 'asthma symptoms'. Again, there was heterogeneity between systematic reviews in the specific outcomes measured within these symptom scores. Treatments included: anti-leukotrienes [ 64 , 65 ]; beta-agonists (inhaled/non-inhaled) [ 61 - 63 , 66 - 75 ]; allergen immunotherapy [ 76 ]; anti-IgE [ 77 ]; antibiotics [ 78 ]; anti-cholinergics [ 57 ]; azathioprine [ 79 ]; beclomethasone [ 80 - 82 ]; budesonide [ 83 , 84 ]; caffeine [ 85 ]; beta-blockers [ 58 ]; chloroquine [ 86 ]; corticosteroids (inhaled/non-inhaled) [ 59 , 60 , 87 - 91 ]; cyclosporine [ 92 ]; fluticasone [ 93 - 96 ]; anti-gastro-oesophageal reflux treatment [ 97 ]; gold [ 98 ]; inhaled/non-inhaled magnesium sulphate [ 98 , 100 ]; macrolides [ 101 , 102 ]; methotrexate [ 103 ]; oxatomide [ 104 ]; and tailored interventions based on sputum eosinophils [ 105 ]. Many reviews report some level of benefit from or equivalence between treatments under review. Clearly steroids (both oral and intramuscular) improve lung function [ 60 ], while long-acting beta agonists (LABA) with inhaled corticosteroids (ICS) [ 68 ] demonstrate improved outcomes in lung function and symptoms compared to LABA and antileukotrienes, short-acting beta agonists (SABA) [ 75 ] or theophylline [ 74 ]. Treatments from large scale reviews that report positive effects most strongly include: allergen immunotherapy (57 trials) [ 76 ], beclomethasone (60 trials) [ 82 ], budesonide (43 trials) [ 84 ], fluticasone (75 trials) [ 96 ], ICS compared with sodium cromoglycate (25 trials) [ 88 ], LABA (67 trials) [ 62 ], and LABA vs. SABA (31 trials) [ 75 ]. Small reviews of caffeine (6 trials) [ 85 ], inhaled magnesium sulphate (6 trials) [ 99 ], intravenous magnesium sulphate (7 trials) [ 100 ], and tailored interventions based on sputum eosinophils, particularly those with severe asthma and frequent exacerbations (3 trials) [ 105 ] also suggested therapeutic benefits. Caffeine (taken orally) has demonstrated a modest, though short-term (up to 4 hours), effect as a bronchodilator. A number of reviews identified the positive 'steroid sparing' or reduced rescue medication effect of treatments such as Anti-IgE (specifically omalizumab) (14 trials) [ 88 ] and LABA [ 82 - 84 , 86 ] (combined total of 78 trials). One review (67 trials) [ 62 ], while reporting the positive steroid sparing effects of LABA, also raised some important safety concerns. These concerns related to a significant increase in asthma related deaths or life threatening experiences for patients treated with LABA, mainly among African-American patients and those not on inhaled corticosteroids. SABA, while generally less effective than LABA, showed in one review (49 trials) [ 70 ] that regular use of inhaled SABA can decrease symptoms and rescue medication. A modest effect has been shown with regards to antileukotrienes [ 64 ], allergen immunotherapy [ 76 ], cyclosporine [ 91 ], gold [ 98 ], and anticholinergics [ 57 ]. However, this modest effect in combination with serious adverse events and/or the need for close monitoring, make therapeutic options such as immunotherapy [ 74 ], cyclosporine [ 92 ] and gold [ 98 ], problematic and inappropriate for clinical use. Two trials found no evidence to indicate that either colchicines [ 106 ] or dapsone [ 107 ] were effective in the management of steroid dependent asthmatic patients. COPD (non-pharmacological treatments) A total of 28 systematic reviews were identified. One was excluded due to lack of focus on respiratory symptoms. Fifteen of the remaining reviews were published in the Cochrane Database of Systematic Reviews. Twenty also included specific data on breathlessness [ 108 - 123 ], while all others assessed breathlessness as part of a composite COPD 'symptom' score. Treatments included: action plans [ 124 ]; oxygen/Heliox [ 108 , 110 , 111 , 113 , 121 , 125 ]; physical therapy [ 109 ]; home care/hospital at home [ 126 , 127 ]; non-invasive positive pressure ventilation (NIPPV)/non-invasive ventilatory support (NIVS) during exercise [ 119 , 128 ]; nutritional supplements [ 129 ]; inhalers [ 130 ]; rehabilitation [ 112 , 131 ]; education/support [ 120 , 132 ]; inspiratory muscle training [ 114 - 116 ]; physical activity [ 117 ]; psychological interventions [ 118 ]; and surgery [ 122 ]. The majority of reviews (n = 14) were based on fewer than 10 trials: six of these had aggregated samples of more than 150 participants [ 108 - 111 , 119 , 130 ]. Positive effects were reported for breathlessness with domiciliary oxygen (6 trials [ 125 ]), and short term ambulatory oxygen (31 trials [ 113 ]). People with COPD are prone to developing alveolar hypoventilation during exacerbations, which may be contributed to by the administration of high inspired oxygen concentrations. Nevertheless, there was no evidence to indicate whether different oxygen therapies in the pre-hospital setting have an effect on outcome for people with acute exacerbations of COPD [ 133 ]. Positive effects were also reported on varied outcomes, including breathlessness, for NIVS during exercise (7 trials) [ 119 ], and lung volume reduction surgery for end-stage COPD (19 trials) [ 122 ], and for outcomes other than breathlessness for oxygen during training (5 trials) [ 111 ] home care/hospital at home (4 trials [ 126 ], 7 trials [ 127 ]) and individualised action plans (3 trials) [ 124 ]. A large review of pulmonary rehabilitation (31 trials) [ 112 ] found that it relieves breathlessness and fatigue, improves emotional function, and enhances sense of control; a large review of the chronic care model in COPD (32 trials) [ 120 ] found reductions in hospitalization rates and emergency/unscheduled visits, and shorter length of hospital stay, although the result specifically for breathlessness/symptoms was unclear in most trials. Four reviews of inspiratory muscle training concluded that it can improve breathlessness and health-related quality of life in adults with COPD (19 trials [ 114 ], 25 trials [ 115 ], 15 trials [ 116 ], 15 studies [ 123 ]) and an important addition to pulmonary rehabilitation (15 trials) [ 116 ]. However, limitations have been identified within the literature and Shoemaker et al [ 123 ] found that it was unclear whether any improvements were mediated through improved muscle strength and endurance. Ram et al [ 134 ] reviewed 14 trials concerning NIPPV for respiratory failure due to exacerbations of COPD. If applied intermittently and for short periods, NIPPV may be sufficient to reverse respiratory failure, and it was found to have a positive effect on a range of outcomes including mortality, need for intubation, and treatment failure. The authors conclude that trials demonstrate the benefits of NIPPV as first line intervention as an adjunct to usual medical care. Bausewein et al [ 28 ] reviewed non-pharmacological interventions for breathlessness in COPD, cancer, chronic heart failure, interstitial lung disease, and motor neurone disease. Strong evidence was identified to show that neuro-electrical muscle stimulation (NMES) and chest wall vibration (CWV) could relieve breathlessness in COPD, and there was moderate evidence in favour of walking aids in COPD and breathing training in studies of mixed cardio-pulmonary disease and COPD. The authors found only low strength evidence of relief from breathlessness with acupuncture or acupressure, and no evidence for the effect of music. They noted that only a few studies included participants with conditions other than COPD. Insufficient data were available to evaluate evidence for a range of other interventions, including relaxation, counselling and support, case management and psychotherapy. COPD (pharmacological treatments) In all, 22 systematic reviews were identified. Three were excluded (lack of focus on respiratory symptoms/non-intervention studies). Two of the remaining 19 reviews were located from sources other than the Cochrane Database. Sixteen included data on breathlessness [ 135 - 149 ] while the others assessed breathlessness as part of a single symptom score. Treatments included: bronchodilators [ 135 , 140 , 141 , 144 , 147 , 148 ]; LABA and LABA in addition to steroids [ 136 , 142 , 149 ]; vaccines [ 137 , 139 ]; inhaled/non-inhaled corticosteroids [ 143 , 146 , 150 ]; SABA [ 138 , 145 ]; methylxanthines [ 151 ]; and mucolytics [ 152 ]. Seven reviews were based on evidence from fewer than 10 trials [ 135 , 136 , 138 - 140 , 147 , 151 ]. One of these referred to an aggregated sample of fewer than 150 participants [ 138 ]. In one review the size of the aggregated sample was unclear [ 148 ]. Positive conclusions were drawn in respect of some aspect of intervention(s) for COPD in over half of the reviews, including: combined corticosteroid and LABA in a single inhaler (6 trials) [ 136 ]; influenza vaccine (6 trials) [ 137 ]; ICS (47 trials) [ 150 ]; ipratroprium vs. LABA (7 trials) [ 140 ]; LABA (23 trials) [ 142 ]; mucolytic agents (although their effect was small) (26 trials) [ 152 ]; oral corticosteroids (OCS) (24 trials) [ 143 ]; oral theophylline with a modest effect on lung function but largely ineffective in relation to breathlessness (20 trials) [ 144 ]; SABA (13 trials) [ 145 ]; systemic corticosteroids (10 trials) [ 146 ]; tiatropium (9 trials) [ 147 ]; bronchodilators, i.e. β-agonists, anti-cholinergics and theophyllines (33 trials) [ 148 ]; and LABA (12 trials) [ 149 ]. A large review of 47 trials (13,139 patients) [ 150 ] showed that ICS used for 2-6 months resulted in small symptom improvements, while use for more than 6 months had limited effect on symptoms. Jennings et al [ 153 ] reviewed 18 trials concerning the use of opioids for dyspnoea in both cancer and non-malignant conditions (e.g. chronic heart failure, COPD, interstitial lung disease, pulmonary fibrosis), concluding that opioids have a positive effect on the sensation of breathlessness. Data did not support the use of nebulised morphine in the management of breathlessness. Bronchiectasis (non-pharmacological treatments) Searches identified three systematic reviews. One of these was excluded because it focussed on surgical intervention only. The remaining reviews (both from the Cochrane Database) featured nurse specialist care (1 trial) [ 154 ] and physical training (2 trials) [ 155 ]. Bradley et al [ 155 ] included data on breathlessness. Both reviews were based on aggregated samples of less than 80 participants. French et al [ 154 ] reviewed one trial of nurse-led care (no significant changes in outcome measures or significant difference compared to doctor-led care). Bradley et al [ 156 ] included two trials of physical therapy, concluding that inspiratory muscle training improved endurance exercise capacity. Bronchiectasis (pharmacological treatments) Searches identified four systematic reviews from the Cochrane database (2 trials [ 156 ]; 2 trials [ 157 ]; 3 trials [ 158 ]; 9 trials [ 159 ]) that found evidence of effectiveness. Two reviews [ 156 , 157 ] were based on samples of more than 54; in a third review [ 158 ] it was not possible to determine the size of the aggregated sample. One review [ 156 ] included data on breathlessness. Wills & Greenstone [ 156 ] concluded that dry powder mannitol was shown to improve tracheobronchial clearance in bronchiectasis. Ram et al [ 157 ] found that ICS may improve lung function in bronchiectasis, but that larger trials were required to guide practice. Evans et al [ 159 ] found a small benefit from prolonged antibiotic use to interrupt the cycle of bacterial colonisation, inflammatory change and progressive lung damage, but also advocated further, more powerful trials. Crockett et al [ 158 ] concluded that there was insufficient evidence to evaluate mucolytics for bronchiectasis. There were also seven systematic reviews that found no evidence of effectiveness of pharmacological treatments in the management of bronchiectasis: for oral methylxanthines [ 160 ]; LABAs [ 161 ]; SABAs [ 162 ]; anticholinergic therapy [ 163 ]; leukotriene receptor antagonists [ 164 ]; oral steroids [ 165 ]; and pneumococcal vaccines [ 166 ]. Pulmonary sarcoidosis, pulmonary fibrosis, interstitial lung disease Five systematic reviews (Cochrane Database) were considered, three of which included data on breathlessness [ 167 - 169 ]. Paramothayan et al [ 170 ] reviewed 13 trials of corticosteroids for pulmonary sarcoidosis, concluding that oral steroids may be of use in some circumstances. Paramothayan et al [ 168 ] further reviewed five trials and found limited evidence to support the use of immunosuppressive and cytotoxic therapy for this condition. Davis et al [ 167 ] reviewed three trials and found little to justify routine use of non-corticosteroid agents for idiopathic pulmonary fibrosis (IPF), and Richeldi et al [ 171 ] found no evidence for an effect of corticosteroid treatment in patients with IPF. Polosa et al [ 169 ] reviewed a single small trial of nebulised morphine in interstitial lung disease (six participants), which reported no improvement in maximal exercise performance or reduction in breathlessness during exercise. Chronic bronchitis Staykova et al [ 172 ] reviewed nine trials of prophylactic antibiotic use in chronic bronchitis/COPD, concluding that this treatment does significantly reduce days of illness due to exacerbations. Routine use is not recommended due to concerns about antibiotic resistance and it is noted that available data are over 30 years old. Exercise-induced broncho-constriction Three reviews were considered from the Cochrane Database [ 173 - 175 ]. Spooner et al [ 173 ] (24 trials) compared the effects of inhaled mast-cell stabilisers with single dose SABAs or anticholinergics prior to exercise, concluding that all provided relief against broncho-constriction. Overall, SABAs were more effective than mast-cell stabilisers, which were in turn more effective than anticholinergics. Spooner et al [ 174 ] (20 trials) concluded that a single dose of a mast-cell stabiliser (nedocromil sodium) before exercise reduces the severity and duration of exercise-induced bronchoconstriction. Kelly et al [ 175 ] (8 trials) found no difference between the effect of nedocromil sodium and sodium cromoglycate on lung function following exercise. Pulmonary arterial hypertension (PAH) Three reviews were considered from the Cochrane Database [ 176 - 178 ]. All included data on breathlessness. Liu & Chen [ 176 ] reviewed five trials that evaluated endothelin receptor antagonists (ERAs), finding that, in conjunction with conventional therapy, their use improved Borg dyspnoea scores and exercise capacity amongst patients with idiopathic PAH. Paramothayan et al [ 177 ] reviewed nine trials that considered the effects of prostacyclin with conventional therapy, identifying some short-term benefits (e.g. on exercise capacity). Kanthapillai et al [ 178 ] reviewed four trials that considered the effects of the vasodilator sildenafil, concluding that the benefits observed in small trials required further validation in more adequately-sized studies. Cystic fibrosis Reid et al [ 179 ] reviewed two trials concerning the effect of inspiratory muscle training (IMT) on patients with cystic fibrosis. Individual study results were found to be inconclusive for improvement in inspiratory muscle strength, though one study demonstrated improvement in inspiratory muscle strength endurance. The authors concluded that the benefit of IMT in cystic fibrosis for outcomes in inspiratory muscle strength was supported by weak evidence, but that its impact on exercise capacity, dyspnoea and quality of life was unclear.

Discussion Improvements are needed in the management of breathlessness in both malignant and non-malignant conditions. Asthma affects an approximately 5.2 million people in the UK [ 180 ], with an estimated 42% of them facing significant challenges in their daily lives due to breathlessness [ 181 ]. COPD is projected to rank as the world's third leading cause of death by 2030 [ 182 ], while the reported frequency of breathlessness in cancer varies depending upon the stage of disease [ 183 ]. As disease becomes advanced, so the incidence of breathlessness tends to increase [ 3 - 5 ]. In advanced cardiorespiratory disease, for example, refractory breathlessness is described as a 'devastating' symptom with few effective palliative options [ 184 ]. Solano et al [ 185 ] found that breathlessness was one of three 'particularly universal' symptoms (with pain and fatigue) that affected more than 50% of patients with far advanced cancer, AIDS, heart disease, COPD and renal disease. Despite this high prevalence and level of severity, breathlessness and other related symptoms are under-assessed and under-treated [ 185 ]. Our review of systematic reviews suggests that while breathlessness in some non-malignant conditions has been intensively addressed, mostly through a small range of mainly pharmacological approaches, there are other conditions that have received little or no attention. The majority of systematic reviews we identified have been conducted on interventions for asthma and COPD. In comparison, no other category of disease produced more than six reviews for pharmacological and non-pharmacological approaches combined. The emphasis in systematic reviews of pharmacological interventions for asthma is clearly on beta-agonists and corticosteroids, with the use of these for relevant asthma-related conditions and in combination with other appropriate pharmacological agents receiving wide support [ 59 , 60 , 62 , 69 , 71 - 75 , 88 , 91 , 96 ], though with some safety concerns [ 62 ]. Reviews that address the 'steroid-sparing' or rescue medication reduction features of various approaches to the pharmacological treatment of asthma are an important sub-group of studies, given the known side-effect profile of this class of drugs [ 62 , 64 - 66 , 68 , 70 - 73 , 75 , 77 , 82 - 84 , 87 , 88 , 92 , 93 , 96 , 98 , 103 - 105 ]. Systematic reviews of non-pharmacological interventions for asthma are limited by a lack of good quality evidence from randomised controlled trials. The largest review of non-pharmacological interventions, which assessed the effects of asthma self-management programmes with regular health practitioner review did, however, conclude that this approach, particularly when it includes training to use written action plans to adjust medication, can improve health outcomes in adults with asthma [ 47 ]. Bronchodilators, including beta-agonists, are the focus for the majority of reviews of pharmacological interventions for COPD. Positive results were identified among these, and for a number of other agents, including corticosteroids, vaccine, and mucolytics, although balancing potential benefits against known risks remains important. Non-pharmacological interventions with positive outcomes appear to be more established in COPD than in asthma, at least as far as number of trials is concerned. In a number of cases successful interventions were broadly rehabilitative in nature: home care from outreach nursing programmes was shown to have benefits in terms of survival and health related quality of life [ 127 ], hospital at home was an effective option for some patients attending emergency departments with an acute exacerbation of their condition [ 127 ], pulmonary rehabilitation was found to relieve breathlessness and fatigue, improve emotional function, and enhance sense of control [ 112 ], and use of two or more components of the chronic care model resulted in reduced hospitalisations and emergency visits and reduced length of hospital stay [ 120 ]. Benefits for a number of outcomes in COPD, including breathlessness, were also found to be provided by IMT within four studies [ 114 - 116 , 124 ], though one study found that it was unclear whether any such improvement is mediated through improved muscle strength and endurance [ 123 ]. Inspiratory muscle training was also found to improve breathing, exercise capacity and quality of life in Bradley et al [ 155 ] review of physical training for bronchiectasis, although a similar review in patients with cystic fibrosis showed inconclusive results [ 179 ]. In contrast to reviews of asthma treatments, data on breathlessness were included in the majority of reviews of interventions for COPD, both pharmacological and non-pharmacological. Given the distressing nature of breathlessness from the point of view of those who experience it, it is surprising that more studies do not include a specific measure of this symptom. Breathlessness and dyspnoea characterise a range of qualitatively distinct experiences of breathing discomfort, and therefore, outcome measures should arguably take account of patients' perceptions and the language patients use to describe the symptom [ 186 - 188 ]. However, as Bausewein et al [ 189 ] point out, no one of the many assessment tools available is comprehensive enough as a single instrument to measure the sensation of breathlessness, its effect on quality of life, and response to treatment. Indeed, this review suggests that the research may emphasise physiological lung changes rather than the subjective experience of a very distressing symptom. It also suggests that findings may reflect lung function rather than symptom experience or symptom distress. A considerable proportion of the original trials reported in the systematic reviews included either proxy measures for breathlessness (e.g. lung function), used unvalidated measures or used single composite scores for 'respiratory symptoms' that differed both within and between systematic reviews, so providing only indications for the level of improvement related to breathlessness per se. This is quite problematic as the 'validity' and sensitivity of such methods to accurately measure a complex, subjective symptom experience is questionable and can lead to biases, as there is a significant issue with the interpretation of the data, which clinicians often find confusing. Future research should address these measurement issues and use data from validated, patient-reported outcome measures (PROMS) of breathlessness as the primary outcome (e.g. the MRC dyspnoea scale [ 190 ] or Dyspnoea-12 [ 188 ]), in adequately powered trials. With the exception of research into treatments for asthma and COPD, the majority of intervention studies, as reflected in our review, continue to focus on pharmacological approaches. We have included two reviews of non-pharmacological interventions for bronchiectasis, but in all other disease categories we found no further examples. More research should focus in the future on the management of breathlessness in respiratory diseases other than asthma and COPD. As pharmacological treatments do not completely manage breathlessness and they have the added burden of side effects, it seems imperative that non-pharmacological interventions are seen as appropriate complementary interventions to standard pharmacological management, or even an alternative method in selected patients. Despite good evidence from the literature, many non-pharmacological interventions have not become mainstream and are not used in the rehabilitation of patients with respiratory diseases. New ways of thinking about evaluation research, and new combinations of pharmacological and non-pharmacological interventions may be needed for symptoms like breathlessness that continue to impose a heavy burden of distress across a wide range of disease types. There are also a number of non-pharmacological studies that have shown some promise, although conclusions are difficult due to the small sample sizes they have used and limited replication. Such interventions include oxygen, self-management approaches, breathing exercises, use of fans, acupuncture/acupressure, calorie controlled diet, selenium supplementation, approaches to reduce anxiety and panic, or those interventions that may be theoretically useful but for which no studies have been carried out so far (e.g. Alexander technique or other complementary therapies). Future research should focus on such promising approaches and provide a good evidence-base for practice.

Conclusions While the reviews have covered in excess of 2,000 trials, the management of breathlessness remains far from ideal and patients suffer unnecessarily, particularly in the advanced stages of their illness. The development of society guidelines and agreed standards of care should provide a good way for clinicians to navigate across the diversity of potential treatments for breathlessness. Such guidelines should incorporate both pharmacological and non-pharmacological interventions where appropriate. More research is needed in areas of potential benefit, with particular emphasis in the accurate and reliable assessment of breathlessness and in ways that could improve the quality of life as well as health care provision for patients experiencing this distressing symptom. The recent development of clinical guidelines for the management of breathlessness/dyspnea provides some direction for the clinicians in this complex area of symptom management [ 191 ].

Background Breathlessness is a debilitating and distressing symptom in a wide variety of diseases and still a difficult symptom to manage. An integrative review of systematic reviews of non-pharmacological and pharmacological interventions for breathlessness in non-malignant disease was undertaken to identify the current state of clinical understanding of the management of breathlessness and highlight promising interventions that merit further investigation. Methods Systematic reviews were identified via electronic databases between July 2007 and September 2009. Reviews were included within the study if they reported research on adult participants using either a measure of breathlessness or some other measure of respiratory symptoms. Results In total 219 systematic reviews were identified and 153 included within the final review, of these 59 addressed non-pharmacological interventions and 94 addressed pharmacological interventions. The reviews covered in excess of 2000 trials. The majority of systematic reviews were conducted on interventions for asthma and COPD, and mainly focussed upon a small number of pharmacological interventions such as corticosteroids and bronchodilators, including beta-agonists. In contrast, other conditions involving breathlessness have received little or no attention and studies continue to focus upon pharmacological approaches. Moreover, although there are a number of non-pharmacological studies that have shown some promise, particularly for COPD, their conclusions are limited by a lack of good quality evidence from RCTs, small sample sizes and limited replication. Conclusions More research should focus in the future on the management of breathlessness in respiratory diseases other than asthma and COPD. In addition, pharmacological treatments do not completely manage breathlessness and have an added burden of side effects. It is therefore important to focus more research on promising non-pharmacological interventions.

Competing interests The authors declare that they have no competing interests. Authors' contributions Conception of study, development of protocol, study coordination: AM; Searches: RD, RW; Data Extraction, Data Verification, Arbitration: AM, RD, RW, AC, JS; Data Synthesis/Analysis: CB, RW, AM, AC, JS; Data discussions: All authors; Drafting paper: AM, RW, CB. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2466/10/63/prepub Supplementary Material

Acknowledgements This review was partly funded by The Breathlessness Research Charitable Trust, UK and the National Cancer Research Institute.

CC BY

no

2022-01-12 15:21:44

BMC Pulm Med. 2010 Dec 9; 10:63

oa_package/33/5d/PMC3016307.tar.gz

PMC3016308

21126354

Background Utilization of lignocellulosic biomass for biofuel production requires the hydrolysis of cellulose and other cell-wall polysaccharides to their component monosaccharides. This process is affected by many structural and compositional characteristics of the biomass, including the presence of lignin, a phenolic polymer composed of three major types of building blocks: p -hydroxyphenyl (H), guaiacyl (G) and syringyl (S) units. The association between lignin and the recalcitrance of biomass materials has long been recognized in forage feeding and tree pulping practices, and led to earlier lignin-engineering efforts aimed at improving feedstock performance in these processes [ 1 ]. For example, it has been shown that lignin content (the total amount of lignin in the tissue) greatly influences forage digestibility [ 2 ], and that lignin composition (the relative ratio of its component subunits) strongly affects the efficiency of the chemical pulping process. Specifically, transgenic poplar trees with higher S/G ratios show a greatly enhanced pulping efficiency [ 3 ]. A recent analysis of transgenic alfalfa revealed that high lignin content is correlated with the recalcitrance of cell-wall materials to enzymatic saccharification during biofuel production [ 4 ]. There was little variation in the S:G ratios of these alfalfa lines and as a result, the effects of lignin composition on the efficiency of biomass to biofuel conversion remain to be determined. To investigate the effect of S-lignin composition on biomass degradability, we chose to analyze two A. thaliana lines in which the activity or expression of ferulate 5-hydroxylase (F5H), a key enzyme required for the synthesis of S-lignin monomer, is eliminated or enhanced, respectively. Both lines have been generated previously and characterized in detail [ 5 - 8 ]. The fah1-2 mutant is defective in F5H and does not deposit S lignin, whereas overexpression of F5H under the control of the cinnamate 4-hydroxylase (C4H) promoter in the C4H:F5H transgenic line results in lignin with an S-unit content in excess of 90% [ 6 ]. These lines represent two extremes in lignin composition and are each distinct from the wild type, which deposits a G/S copolymer with an S-subunit content of approximately 20 mol%. Despite their lignin difference, these two lines show similar growth to that of wild-type Arabidopsis. In the biomass to biofuel conversion processes, pretreatment is generally used to increase the accessibility of cell-wall polysaccharides to enzymes. Pressure-cooking in liquid hot water (LHW) has been shown to be a cost-effective pretreatment to enhance the enzymatic digestibility of cellulose in a variety of feedstocks [ 9 - 12 ]. In this study, we assessed the cell-wall degradability of mutant and transgenic A. thaliana lines by enzyme hydrolysis without pretreatment, after LHW pretreatment, or after LHW pretreatment followed by hot water washing. Our results indicate that increasing the proportion of S subunits in Arabidopsis lignin decreases the recalcitrance of cell walls to enzymatic hydrolysis.

Methods Materials Generation of the A. thaliana fah1-2 and C4H:F5H lines has been described previously [ 6 , 7 ]. The genetically modified and wild-type A. thaliana (Columbia) plants were grown side by side at 22°C under a 16-hour photoperiod. Mature stems were harvested by removing siliques and leaves. Spezyme CP cellulase preparation from Trichoderma reesei containing exo-, endo- and β-glucosidase activities; batch number 3016295230) was provided by Genencor, Danisco Division (Palo Alto, CA, USA). Novozyme 188 (β-glucosidase from Aspergillus niger ; catalogue number c6150) was purchased from Sigma Chemical Co. (St. Louis, MO, USA). Lignin analysis Lignin composition was determined using the derivatization followed by reductive cleavage (DFRC) method [ 23 ]. Compositional analysis The composition of the plant samples was analyzed using standard National Renewable Energy Laboratory procedures [ 24 ]. To calculate polysaccharide composition, monomer sugars were analyzed by high-performance liquid chromatography (HPLC) after acid hydrolysis of the samples. HPLC analysis of liquid samples was performed on a system consisting of a solvent-delivery system (9010 Gradient HPLC Pump, Varian/Agilent, Santa Clara CA, USA), an autosampler (717 Plus; Waters Corp., Milford, MA, USA), a carbohydrate analysis column (Aminex HPX-87H; Bio-Rad, Hercules, CA, USA); a refractive index detector (2414; Waters Corp.) a dual-wavelength absorbance detector (2487; Waters Corp.); and an integrator (HP3396G; Hewlett Packard, Santa Clara CA, USA). The mobile phase was 5 mmol/l H 2 SO 4 filtered through a 0.2 μm nylon filter (Millipore Corp., Billerica, MA, USA) and degassed. The mobile phase flow rate was 0.6 ml/min and the column temperature was maintained at 60°C by a column heater (CH-30; Eppendorf) with a temperature controller (TC-50; Eppendorf, Hauppauge, New York, USA). Cell-wall degradability analysis The stem materials were ground for passage through a 20 mesh (841 μm) screen for cell-wall degradability analyses. Pretreatment was carried out by pressure cooking 50 mg samples in a metal tube containing 1.5 ml water at 200°C (30 seconds of heat-up time followed by a 10 minute hold). Each tube was placed in a fluidized sand bath (Tecam ® SBL-1; Cole-Parmer, Vernon Hills, IL). The pressure within the tubes was held at the saturation vapor pressure of water to keep the water in a liquid state [ 9 , 11 - 13 ]. The samples were cooled before the addition of 1.5 ml of 100 mmol/l citrate buffer pH 4.8, bringing the final volume to 3 ml (~2% solids (w/v)). For hot-water washing, samples were washed twice with 3 ml of 70°C water. After the second wash, no glucose was detected. The enzyme hydrolysis for all the conditions tested was based on initial solids loading and glucan concentration. Commercial cellulase (Spezyme CP) at 0.2 filter paper units (FPU)/ml or 50 FPU/g glucan (90 mg protein/g glucan) and β-glucosidase (Novozyme 188) at 0.35 cellobiase units (CBU)/ml or 105 CBU/g glucan (34 mg protein/g glucan) were added, and hydrolysis was carried out for different lengths of time at 50°C and pH 4.8 in an incubator shaker (New Brunswick Scientific, Edison, NJ, USA). The ratio of enzyme to solids was equivalent to 10 FPU/g total solids, 21 CBU/g total solids and 25 mg protein/g total solids. Enzyme hydrolysis of 50 mg untreated samples (also at ~2% solids (w/v), 50°C and pH 4.8) was carried out under similar experimental conditions. SEM analysis Stem cross-sections were adhered to a glass slide with epoxy adhesive (J-B Weld; J-B Weld Co., Sulphur Springs, TX, USA) and subjected to different treatments. Pretreatment was performed as described in the previous section. When applied, enzyme hydrolysis was carried out with a two-fold increase in enzyme loading to compensate for the particle size differences between the samples used for SEM and cell-wall degradability analysis. Subsequently, the stem cross-sections were fixed in two steps: first with a mixture of 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 mol/l cacodylate buffer, pH 7.4 for 1 hour and then with 1% OsO 4 in 0.1 mol/l cacodylate buffer, pH 7.4 for 30 minutes. After critical point drying, the samples were sputter-coated with gold, and viewed under SEM (Nova nanoSEM; Fei Co., Hillsboro, OR, USA). Protein and phenolic measurements The protein content of the commercial enzyme preparations was determined using a commercial kit (Pierce BCA Protein Assay Kit; product number 23225; Thermo Scientific, Rockford, IL, USA). Phenolic compounds were assayed using Prussian blue [ 25 ]. The diluted sample liquid (3 ml aliquots) was transferred to a 10 mm cuvette, then 200 μL of 0.008 mol/l K 2 Fe(CN) 6 were added, followed by the immediate addition of 200 μL of 0.1 mol/l FeCl 3 in 0.1 mol/l HCl. Absorbance was read at 700 nm after 5 min at room temperature, against a tannic acid standard.

Results Composition analysis Consistent with previously published results [ 6 ], the lignins of fah1-2 and C4H:F5H plants are dominated by G and S subunits, respectively (Table 1 ). In contrast to the drastic differences in lignin composition, other cell-wall components of fah1-2 and C4H:F5H plants are relatively unchanged (Table 1 ). Whereas fah1-2 is similar to the wild type in glucan, xylan and acetyl content, C4H:F5H has slightly lower values for these components. Arabinan was not detected in any of the samples analyzed. Enzyme hydrolysis of untreated samples As a first step to evaluate the possible effect of S lignin composition changes on cell-wall degradability, mature stems from wild-type, fah1-2 and C4H:F5H plants were subjected to enzyme hydrolysis without any pretreatment. There was limited cellulose hydrolysis in untreated stems (Figure 1A ). After 24 hours of incubation, only about 30% of the maximum theoretical glucose yield was obtained for wild-type samples. The low- and high-S lignin samples have similar hydrolysis kinetics and glucose yield to those of the wild type, indicating that the structural changes to lignin in those plants do not have a significant effect on enzymatic hydrolysis of cellulose in untreated cell-wall materials. Enzyme hydrolysis with LHW pretreatment We further tested the cell-wall degradability of the lines by applying LHW pretreatment before enzyme hydrolysis. Regardless of genotype, the hydrolytic rate and glucose yield of pretreated samples was greater than for the corresponding untreated samples (Figure 1B ). Incubation of LHW-pretreated wild-type samples with enzymes for 1 hour released the same amount of glucose (20% of total glucan) as 6 hours of hydrolysis on untreated samples. More importantly, significant differences in cellulose hydrolysis were observed for LHW-pretreated fah1-2 and C4H:F5H samples (Figure 1B ).No differences in glucose yield between the three genotypes were observed after 1 hour of incubation, but after 3 hours of incubation, C4H:F5H yielded more glucose and fah1-2 less glucose, compared with wild type. After 24 hours, the glucose yield of the C4H:F5H sample was nearly 90%. Scanning electronic microscopy Pretreatment causes structural and/or chemical changes of biomass materials, making cellulose more accessible to hydrolytic enzymes [ 13 ]. To determine whether any anatomical changes occurred during LHW pretreatment and enzyme hydrolysis, we performed scanning electron microscopy (SEM) on stem cross-sections that had been subjected to different treatments. The wild-type, fah1-2 and C4H:F5H stems had similar vascular patterning (Figure 2 ). LHW pretreatment broke the pith cells and caused the detachment of the phloem from the xylem. This happened in all three genotypes, but fah1-2 had more intact pith cells compared with the wild type and C4H:F5H. In samples that underwent enzyme hydrolysis without pretreatment, both phloem and pith cells were completely disrupted, and there were no significant differences between the different genotypes. Interestingly, the stem cross-sections of C4H:F5H that had undergone LHW pretreatment plus enzyme hydrolysis were found to have collapsed, which was not observed in wild-type and fah1-2 samples after the same treatment. Our enzyme hydrolysis results suggest that LHW pretreatment is differentially effective against A. thaliana samples with altered lignin composition. Enzyme hydrolysis after LHW pretreatment and hot-water washing It has been shown that hot-water washing of LHW pretreated poplar can further improve its enzymatic digestibility [ 11 ]. Therefore, we also tested the cell-wall degradability of the A. thaliana samples after this additional step (Figure 1C ). Unlike poplar, hot-water washing did not have a significant effect on the total glucose yield as calculated based on the initial glucan content of the Arabidopsis samples tested; however, the initial rate of cellulose hydrolysis of hot-water washed C4H:F5H samples was almost double that of unwashed samples. By contrast, hot-water washing has little effect on the initial hydrolytic rate of the wild-type and fah1-2 samples. Phenolic inhibitors analysis The previous results suggest that hot-water washing may remove inhibitors of cellulase and/or β-glucosidase activity from C4H:F5H samples. To investigate this possibility, we analyzed the total content of phenolics in the liquid fraction after LHW pretreatment. All three genotypes had a similar concentration of total phenolics, approximately 0.3 mg/ml tannic acid equivalent (Table 2 ). This concentration is lower than the level known to significantly inhibit cellulose, although deactivation of these enzymes, particularly β-glucosidase may occur over a 24-hour period [ 14 , 15 ].

Discussion Lignin is a major contributor to the recalcitrance of biomass, and has been a target for feedstock improvement through genetic engineering. It has been demonstrated in several plant species that a reduction in lignin content using transgenic approaches enhances cell-wall degradability; however, significant improvement in conversion efficiency has often been accompanied by abnormal plant growth and development [ 1 , 4 ]. By contrast, plants seem to be amenable to wide ranges in lignin composition changes, including variation in the content of conventional monomers and the incorporation of atypical precursors [ 6 , 16 - 18 ]. Our findings of improved cell-wall degradability in Arabidopsis stems with high S-lignin content demonstrate the potential of lignin composition modification for the improvement of cellulosic feedstock performance. LHW treatment had a dramatic effect on enzyme hydrolysis of biomass from the high S-lignin line. As much as 90% of the maximum theoretical glucose yield was achieved for C4H:F5H tissue, whereas less than 60% was obtained from wild-type material exposed to the same treatment. SEM analysis detected no obvious anatomical differences between the low and high S-lignin samples after LHW pretreatment. By contrast, after LHW pretreatment and enzyme hydrolysis, stem cross-sections from the high S-lignin stems had a distinct deformity, presumably due to enhanced hydrolysis of the cell wall. How alteration of lignin composition relates to the observed increase in the effectiveness of LHW pretreatment is unclear. Recently, it was proposed that during high-temperature pretreatment, lignin is melted and relocalized to outer surface of the cell wall, increasing the accessibility of the cellulose within [ 19 , 20 ]. The S:G ratio is known to have profound effects on lignin structure. Whereas G-rich lignin has a branched structure, S-rich lignin is more linear and has a lower degree of polymerization [ 21 ]. It is tempting to speculate that S-rich lignin may have a lower melting point and is more easily relocated than G-rich lignin and, thereby, leads to improved enzymatic digestibility. The observation that hot-water washing after LHW pretreatment significantly increases the initial saccharification rate of the high-S sample suggests removal of some inhibitory compounds. It has been shown LHW pretreatment of wet cake (solids left after corn is fermented to ethanol) releases some phenolics and water soluble xylo-oligosaccharides that can inhibit cellulases and β-glucosidases [ 14 ]. However, in this study, the concentration of phenolics in the liquid after LHW pretreatment was below the level reported to cause significant inhibition. Moreover, we did not see any difference in the concentration of phenolics between S-deficient, S-rich and wild-type A. thaliana tissue samples. Therefore, it is unlikely that the observed increase in saccharification rate was due to the removal of inhibitors. The relative influence of phenolic molecules on enzyme activity becomes more pronounced as the ratio of phenols to protein increases at higher solid and lower protein loadings than used in this study. Although protein loadings are currently in the range of 2 to 10 mg/g lignocellulose solids, which represent up to a five-fold decrease in enzyme from only 5 years ago, an even greater reduction is needed for economically viable processes [ 13 , 22 ]. It will be important to extend this study to other biomass feedstocks in the future. One of the factors that needs to be considered is the native lignin composition of different plant species, and possibly also the interaction of lignin with the polysaccharide components of the cell wall. It is likely that the dramatic increase in cell-wall degradability observed in Arabidopsis might be less apparent in plant species with high native S-lignin levels, such as hybrid poplar. However, significant increases in lignin extractability during the pulping process has been observed in S-lignin-enriched transgenic poplar [ 3 ], suggesting that the magnitude of S lignin increase may still contribute significantly to higher cell-wall degradability in this important biomass feedstock.

Conclusions The available genetically modified A. thaliana plants with different lignin composition and structure provided an opportunity to evaluate the possible effect of lignin modification on cell-wall recalcitrance. Our study revealed that high levels of S-lignin have a positive effect on the effectiveness of LHW pretreatment and enzymatic hydrolysis, at least in Arabidopsis . This effect might result from the physicochemical changes of lignin brought about by more linear structure of S subunits. In the future, it will be important to determine how widespread is this phenomenon and to elucidate its underlying mechanism.

Background Lignin is embedded in the plant cell wall matrix, and impedes the enzymatic saccharification of lignocellulosic feedstocks. To investigate whether enzymatic digestibility of cell wall materials can be improved by altering the relative abundance of the two major lignin monomers, guaiacyl (G) and syringyl (S) subunits, we compared the degradability of cell wall material from wild-type Arabidopsis thaliana with a mutant line and a genetically modified line, the lignins of which are enriched in G and S subunits, respectively. Results Arabidopsis tissue containing G- and S-rich lignins had the same saccharification performance as the wild type when subjected to enzyme hydrolysis without pretreatment. After a 24-hour incubation period, less than 30% of the total glucan was hydrolyzed. By contrast, when liquid hot water (LHW) pretreatment was included before enzyme hydrolysis, the S-lignin-rich tissue gave a much higher glucose yield than either the wild-type or G-lignin-rich tissue. Applying a hot-water washing step after the pretreatment did not lead to a further increase in final glucose yield, but the initial hydrolytic rate was doubled. Conclusions Our analyses using the model plant A. thaliana revealed that lignin composition affects the enzymatic digestibility of LHW pretreated plant material. Pretreatment is more effective in enhancing the saccharification of A. thaliana cell walls that contain S-rich lignin. Increasing lignin S monomer content through genetic engineering may be a promising approach to increase the efficiency and reduce the cost of biomass to biofuel conversion.

Competing interests Michael Ladisch is Chief Technology Officer at Mascoma Corporation. Clint Chapple is the inventor on patents related to engineering S lignin content in plants. Authors' contributions XL carried out the lignin analysis, participated in SEM imaging, and drafted the manuscript. EX participated in SEM imaging and helped to draft the manuscript. EX and MS carried out cell-wall degradability analyses. YK and MS carried out cell-wall compositional analysis and phenolic measurements. XL and CC conceived the study. CC, RM and ML participated in its design and coordination, and helped to draft the manuscript. All authors read and approved the final manuscript.

Acknowledgements We thank Junko Maeda, Xingya (Linda) Liu, Rick Hendrickson and Thomas Kreke for their technical assistance. We also thank Chia-Ping Huang at Purdue Life Science Microscopy Facility for assistance with SEM imaging. Finally, we are grateful for the enzymes provided by Genencor. This work was supported by USDA IFAFS contract #00-52104-9663 and DOE grant #DE-FG02-06ER64301.

CC BY

no

2022-01-12 15:21:45

Biotechnol Biofuels. 2010 Dec 2; 3:27

oa_package/0d/f7/PMC3016308.tar.gz

PMC3016309

21245926

Introduction The liver is one of the important organs in our bodies, playing a vital role in glucose homeostasis, the synthesis of bile acids for the metabolism of cholesterol, and the secretion of proteins to aid clotting [1] . Additionally, the liver is primarily responsible for the detoxification of foreign substances, including a variety of environmental toxicants, alcohol, cigarette smoke, and drugs [1] . Hepatocytes are the principal cells in the liver, comprising over 80% of its mass and performing several characteristic functions of this organ. Liver culture systems such as hepatocyte monolayers (HMs) and collagen sandwiches (CSs) are routinely used to test adverse effects of drugs and environmental toxicants. In HMs, hepatocytes are cultured on a single collagen gel. Such cells progressively lose their phenotypic characteristics over time [2] . In CS systems, hepatocytes are maintained between two collagen gels. Hepatocytes in CS cultures remain stable over extended periods of time, and maintain differentiated hepatic functions [3] , [4] . While morphological and physiological characteristics of hepatocytes in CS cultures have been studied extensively, comprehensive transcriptional studies of these culture systems do not appear to have been reported. Therefore, in an earlier study, we performed a systematic temporal study of genome-wide gene expression programs in HMs and in CS cultures over an eight-day period [5] . We used Gene Set Enrichment Analysis (GSEA) [6] to compare the transcriptional programs in the two culture systems. Our results demonstrated that gene expression in hepatocytes in CS cultures steadily and comprehensively diverges from that in HMs [5] . Gene sets up-regulated in CS cultures included several hepatic functions, such as metabolism of lipids, amino acids, carbohydrates, and alcohol, and synthesis of bile acids. Monooxygenases such as Cytochrome-P450 enzymes did not show any change between the culture systems after one day, but exhibited significant up-regulation in CS cultures after three days and later in comparison to HMs. This analysis did not consider the fact that a cell's response to its environment is governed by an intricate network of molecular interactions. These interactions dynamically change in response to a myriad of cues. Therefore, discovering the set of molecular interactions that are active in a given cellular context is a fundamental question in computational systems biology [7] . In the current work, we reanalyze the CS-HM transcriptional data in the light of an underlying molecular interaction network. We propose a novel approach called “Contextual Biological Process Linkage Network” (CBPLN) that focuses on computing which processes in the cell are perturbed in a particular context and how these processes are linked to each other. Our approach is predicated on the belief that high-level linkages between pathways and processes make identification of important biological trends more tractable and intuitive than through interactions between individual genes and molecules alone. Our method requires three inputs: -values representing the statistical significance of the differential expression of each gene (upon comparing a treatment to a control), which we refer to hereafter as expression -values , a functional or physical interaction network connecting genes and proteins, and a dataset of functional annotations for genes and proteins. We extend the method developed by Dotan-Cohen et al. [8] to detect directed linkages between gene sets in the context of a functional interaction network. Given two biological processes and and the sets of genes that are members of each, these authors computed the number of genes annotated by that are themselves not annotated by and interact with at least one gene annotated by . They estimated the statistical significance of this count using the one-sided version of Fisher's exact test. Similar methods developed by Pandey et al. [9] , [10] for regulatory and physical interaction networks are aimed at discovering chains of significantly linked biological processes. In this work, we extend the ideas of Dotan-Cohen et al. to incorporate gene expression measurements to determine which inter-process links are significantly perturbed between the measured conditions. Informally, we compute a score for a link from process to process based upon the expression -values of pairs of interacting genes, where one gene belongs to process and the other to process . Our score takes estimates of confidence in the interactions into account. High-confidence interactions with highly perturbed incident genes make large contributions to the score. We estimate the statistical significance of the score by computing an empirical distribution of scores under two different hypotheses. The first hypothesis tests the dependence of the score on the particular set of genes annotated by , i.e., it asks if we would observe a particular score from process to even if we selected the genes annotated by uniformly at random from the set of all annotated genes. This test directly extends the approach used by Dotan-Cohen et al. The second hypothesis tests the dependence of the score on the specific interactions in the network, i.e., it asks if we would observe the score from to even with an interaction network drawn from a distribution of networks with the same node degrees. Under either hypothesis, we report the significance of the link, after multiple testing correction, as a -value. Hereafter, we refer to this quantity as the link -value , to distinguish it from the expression -values that are inputs to our method.

Materials and Methods Measuring perturbation from gene expression data We applied Linear Models for Microarray Data (LIMMA) [72] to the DNA microarray data to compute expression -values indicating the differential expression of each gene for each of the four contrasts shown in Table 1 . Scoring a link between a pair of processes We first present the approach developed by Dotan-Cohen et al. to identify linkages between biological processes [8] . Given an intracellular interaction network for an organism and Gene Ontology annotations for the genes in those networks, Dotan-Cohen et al. compute what they term a Biological Process Linkage Network ( BPLN ). Informally, given two biological processes, they defined the first process as being linked to the second process if genes annotated by the first process interact with a significant number of genes annotated by the second process. By definition, such links are directed. The resulting output of the algorithm by Dotan-Cohen et al. is, for each ordered pair of processes, the probability that the first process is linked to the second. Formally, let be the set of all biological processes. We seek to ask “Given two processes , is process linked to process ?” More specifically, of the genes that are neighbors of those annotated by , are many more annotated with than would be expected by chance? Let be the set of all genes in an organism. Let be the set of genes annotated by process , and let the universe , be the set of all genes annotated by at least one process in . Let denote an undirected interaction graph where is the set of undirected edges , each representing an interaction between genes . We define the set as the set of genes that meet the following criteria: gene neighbors at least one gene annotated with gene is not annotated with . In other words, Next, we define , i.e., the set of genes that are neighbors of genes annotated with , are not annotated with themselves, and are annotated with process . We define the link -value as the probability that, if we selected a set of genes uniformly at random from , the set would contain or more genes. We can compute this link -value as the tail of a hypergeometric distribution: If this link -value is significant at some cutoff , we conclude that process is linked to process . Extending the score to include transcriptional data and interaction weights With this background, we extend the formulation of BPLN to take transcriptional measurements and interaction weights into account. For each interaction in the graph , we use to denote its weight. The larger the weight of an interaction, the larger is our belief that and indeed interact functionally in the cell. We define a scoring function that maps genes to a non-negative real number representing their degree of perturbation in a given biological context (e.g., CS day 8 versus HM day 8). In this work, we compute as absolute value of the logarithm of the LIMMA -value of the gene. Given processes and , we first define a score . The function measures the contribution of the neighbors of annotated with term based on their perturbation. Ideally, if at least one neighbor of that is annotated with is highly perturbed, we desire that take a high value. On the other hand, if no such neighbor of is highly perturbed, we desire that take a small value. Naturally, the weights of the interactions should also play a role in . Accordingly, we define i.e., is the maximum weighted score of all neighbors of node that are annotated with process . We define the contextual linkage score between processes and as the following: Figure 11 contains a toy example that illustrates these concepts. Thus, a node annotated by makes a large contribution to the contextual linkage score if shows a high amount of perturbation in a particular context and if the neighbors of annotated by also show a high amount of perturbation. If we have many such nodes , then itself will be large. Note that if we set for all and if all edges have weight , then is equal to the size of , identical to the score computed by the original BPLN algorithm. In this formulation, some pairs of processes may have a high contextual linkage score even if all genes were perturbed by the same amount. To account for this possibility, we compute a normalized score , where is a background score computed in the same manner as , but, after setting the gene perturbation score equal to the average expression for all genes in . Thus, represents the score for the link between processes and if all genes had the average expression score. Assessing the statistical significance of links Since the contextual linkage score is a weighted generalization of the statistic measured by Dotan-Cohen et al. it is unclear how to compute its statistical significance analytically. Therefore, we use two different approaches in order to assess the significance of the observed score empirically. The first approach is an empirical version of the test performed by Dotan-Cohen et al. [8] : what is the probability that we would observe a score or more if we were to randomly select the nodes annotated with ? Specifically, we repeatedly select a set of size uniformly at random without replacement from and calculate for each of these random selections. After performing the step 10,000 times, we return the fraction of random scores that are larger than the observed value of as the link -value . Two different processes may annotate some genes in common. To preserve this property even in the random selections of the set over different processes, we adopt the following approach: we construct a bipartite graph in which a node is a gene or a biological process and an edge connects a gene to a biological process it is annotated with. We randomly permute the labels of the genes in this graph. To generate a random set of size , we simply select the genes annotated with in the bipartite graph with randomized gene labels. These steps create a randomized set of annotations that satisfy two properties: (a) every process annotates the same number of genes as in the original set of annotations, and (b) if genes are annotated by each process in a set of processes , then these processes co-annotate exactly genes in the randomized dataset as well. The second approach accounts for the role played by the interactions between the genes in and genes in . Therefore, we generate a graph with the property that each node has the same degree in and . We measure the contextual linkage score between and with respect to . We generate 10,000 to build a null distribution for the contextual linkage score, and compute the link -value as before. To construct , we follow the “edge-swap” approach [73] . We begin with the set of edges and modify the edges in with pairwise edge swaps. For each edge swap, we first select a pair of edges . We then select, with equal probability, either or (i.e., the edges created by swapping the endpoints of the original pair of edges) as a candidate edge pair. If either candidate edge already exists in or creates a self-loop, we retain the original pair of edges in , i.e., we do not perform the edge swap. Otherwise, we remove the original edges from and insert the new edges into . In total, we perform edge-swap events to create a randomized graph , where is a user-defined parameter. In this work we used . We use the method of Benjamini and Hochberg [20] to correct for testing multiple hypotheses, while ensuring that the corrected link -values are monotonic [74] . For either approach, if , we say that term is linked to in the given biological context.

Results and Discussion Input Data Gene Expression Data We used the Affymetrix Rat Genome 230 2.0 GeneChip to measure genomewide transcriptional profiles in rat hepatocytes grown in monolayers and in collagen sandwiches [5] . This dataset is available in MIAME-compliant format in the Gene Expression Omnibus (accession number GSE20659). We marked the day when we deposited the second layer of collagen in CS cultures as day zero. On days one, two, three and eight after deposition of the second layer of collagen, we measured data in triplicate in hepatocytes in each culture system. Functional Linkage Network Existing databases of protein interactions contain very few experimentally detected Protein-Protein Interactions (PPIs) for rat: seven different widely-used sources [11] – [17] contained a total of just 1,274 non-redundant rat PPIs spanning 974 proteins. Therefore, we decided to use the rat functional linkage network predicted by the STRING system [18] . The interaction type in STRING is a functional association , which the authors define as “the specific and meaningful interaction between two proteins that jointly contribute to the same functional process.” Apart from incorporating experimental interaction data, STRING uses multiple methods to predict possible functional linkages including interolog-based interaction transfer, similar transcriptional response across a variety of conditions (co-expression), text-mining, and gene families that share above-random similarities in their evolutionary histories. STRING includes a scheme to score each predicted interaction in the range 150–1000 against a common reference of functional partnership based on the KEGG database [19] . STRING version 8.3, released on May 26, 2010 contains 975,454 predicted interactions among 15,178 rat proteins. We used the subset of these interactions with a weight of at least 500; there were 204,992 such interactions among 9,925 proteins. We selected 500 as a cutoff based on the reasoning that interactions with at least this weight were more likely to connect genes belonging to the same process than to connect genes belonging to different processes. When we further pruned the network to include genes with at least one annotation (see below), we obtained 47,002 interactions among 4,714 genes. Functional Annotations In our earlier work [5] , we used GSEA to compare the two culture systems at each of the four time points; Table 1 lists the contrasts we analyzed. This analysis provided insights into the temporal patterns of up- and down-regulation in the gene sets in the Molecular Signature Database (MSigDB) [6] . In that work, we focused our analysis on gene sets that showed monotonically diverging patterns of expression between CS and HM cultures. In the current paper, we use the curated (c2), motif (c3), and Gene Ontology (c5) collections of gene sets in MSigDB as our set of functional annotations. We focus on establishing linkages among the subset of 18 up-regulated gene sets from the previous study; Table 2 lists these sets along with a short description of each. Overview of Results We considered only those links with a link -value of at most , after using the method of Benjamini and Hochberg [20] to adjust for testing multiple hypotheses. We further restricted our attention to pairs of gene sets for which at least 10 genes exclusively in the second set of the pair interacted with genes in the first set, reasoning that fewer interacting genes might not yield robust link -values. We compared the number of links computed by using each hypothesis test. We also compared these values to the number of links in the (context-free) BPLN computed using the method of Dotan-Cohen et al. [8] . Tables 3 and 4 display the results of the comparisons. Several salient trends emerged. First, in Table 3 , irrespective of the hypothesis test used, the number of links increased with time. This phenomenon parallels our earlier observation that the transcriptional programs of hepatocytes in CS cultures steadily diverged from that in HMs. Second, the size of the intersection between the two sets of links also increased with time, as did the Jaccard similarity coefficient of the two sets (i.e., the size of the intersection divided by the size of the union). Further, for each day, the number of links deemed to be significant by both hypothesis tests was itself statistically significant, based on Fisher's Exact Test (see File S1 ). These trends suggest that once the transcriptional programs of the two culture systems have diverged (day 2 and later), both hypothesis tests find very similar sets of process pairs to be significantly linked at the 0.01 level. However, the number of common links is very close to the number of links identified by the second hypothesis test, indicating that the second test is more conservative than the first in deciding whether a link is statistically significant. We observed similar results when we repeated these analyses with other cutoffs on the link -value (0.005, 0.05, and 0.1) (see File S2 ). Third, normalizing the linkage score (see “Methods”) pruned out a small number of links. Finally, the overlap between the intersection of the results from both hypothesis tests and the BPLN was small in days 1 and 2 and more substantial in days 3 and 8 ( Table 4 ), although the overlap was still statistically significant by Fisher's Exact Test (see File S1 ). These data suggest that only a subset of the links in a BPLN may have some relevance to the particular biological conditions being investigated. By incorporating measurements of gene expression, CBPLNs can identify those inter-process links that correspond to the phenotypic differences observed in the two conditions being compared (e.g., hepatocytes in CS versus HM). Although both hypothesis tests find very similar sets of process pairs to be significantly linked at the 0.01 level, especially in later days, we found that the actual link -values computed for each process pair were not very highly correlated to each other (see File S3 ). Based on these results, we decided to consider a linkage between a pair of gene sets only if this link was significant at the 0.01 level with both hypothesis tests with normalization. The resulting CBPLNs are displayed in Figures 1 , 2 , 3 , 4 . For reference, we have displayed the BPLN in Figure 5 . We discuss the properties of these CBPLNs in the rest of the paper. We focus primarily on the day 8 CBPLN ( Figure 4 ), noting that many of the features we discuss are also apparent in the day 3 CBPLN ( Figure 3 ). When we discuss some pairs of linked gene sets, we refer to the underlying functional interaction network connecting the genes in those sets. We start by discussing properties of liver-specific genes, focusing particularly on the regulation of these genes by the transcription factor HNF1. Then, we discuss the role of lipid homeostasis and bile acid synthesis in the liver. Finally, we summarize the different interpretations of the links in CBPLNs. We stress that the formulation of linkage between processes and is asymmetric. Hence, by definition, links in the CBPLN are directed, i.e., a CBPLN may contain a link between and and between and . Liver Specific Genes The 251 genes in the HSIAO_LIVER_SPECIFIC_GENES gene set are expressed selectively in the liver, as determined by Hsiao et al. [21] from a compendium of gene expression in normal human tissues created with the goal of defining a reference for basic organ systems biology. Genes in this set are members of a spectrum of biological processes, including fatty acid metabolism, metabolism of xenobiotics, blood coagulation, and response to wounding. Not surprisingly, this gene set occupies a central place in the CBPLN on day 8 ( Figure 4 ); it has the highest number of outgoing and incoming links. Outgoing links include connections to glycolysis and gluconeogenesis, alcohol metabolism process, metabolism of xenobiotics by cytochrome P450s, the Peroxisome Proliferator-Activated Receptor (PPAR) signaling pathway, lipid metabolic processes, the urea cycle, and bile acid biosynthesis, among others. In turn, the gene sets such as V$HNF1_Q6 and NUCLEAR_RECEPTORS are linked to HSIAO_LIVER_SPECIFIC_GENES. Some links involving HSIAO_LIVER_SPECIFIC_GENES are unidirectional on day 2 or day 3 ( Figures 2 and 3 ) but bidirectional on day 8 ( Figure 4 ), e.g., to HSA03320_PPAR_SIGNALING_PATHWAY and metabolism of fatty acids, bile acids, and alcohol. Such features suggest that CBPLNs may be representing cellular signals emanating from a subset of liver specific genes to other processes and subsequent feedback from the other processes to liver specific genes. Overall, these results suggest that CBPLNs can assist in the sub-division of liver-specific genes into more refined categories, based not only on the functions of the genes themselves, but also on how they are regulated and what other processes they may control. We discuss one specific link next that illustrates this property. Liver Specific Gene Sets Regulated by HNF1 Hepatic nuclear factor 1 (HNF1), also known as albumin proximal factor, is a transcription factor required for the expression of several liver-specific genes including albumin [22] . The protein functions as a homodimer and binds to the inverted palindrome 5′-GTTAATNATTAAC-3′ . The promoter regions of genes in the MSigDB set V$HNF1_Q6 match this binding site for HNF1 [23] . In our previous study [5] , we noted the monotonic up-regulation of this gene set in CS cultures when compared to HMs. This gene set has an overlap of 25 genes with the gene set HSIAO_LIVER_SPECIFIC_GENES. We concluded that HNF1 monotonically up-regulates the expression of liver-specific genes in CS cultures but not in HMs. CBPLNs assist us in elaborating upon these earlier observations. We studied the link between V$HNF1_Q6 and HSIAO_LIVER_SPECIFIC_GENES in the day 8 CBPLN by examining the functional interactions in the STRING database connecting genes in V$HNF1_Q6 to genes in HSIAO_LIVER_SPECIFIC_GENES. Figure 6 displays a layout of this network. Visual examination of Figure 6 indicates that the linkage between these two gene sets is driven by the genes F2, Plg, CYP2E1, Nr1h4, Lipc, and their interactors, with weaker contributions arising from Hnf1a and Hnf4a. Note that F2, Plg, CYP2E1, Nr1h4, and Lipc are members of both gene sets while Hnf1a and Hnf4a are members of V$HNF1_Q6. We discuss a subset of these proteins next, highlighting liver-specific processes they participate in. HNF1a and HNF4a Hepatocyte nuclear factor 4 (Hnf4a) is a nuclear receptor implicated in the regulation of numerous genes associated with hepatic function [24] – [26] , gluconeogenesis [27] , and activation of the metabolism of xenobiotics, including drugs and pharmaceuticals [28] . It is known that both the HNF4 protein and HNF1 protein can transactivate the HNF1 gene [29] . Although both genes are not very highly up-regulated, their interactions with liver-specific genes Apoa2, Serpina1, Pcbd1, Slc2a2, Slc10a1, Fabp1, and Pck1 suggest the activation of many liver-related pathways. Blood clotting (Plg and F2) Plasminogen (Plg) is a secreted protein that is proteolysed to plasmin and angiostatin. Plasmin dissolves fibrin in blood clots while angiostatin inhibits angiogenesis. In Figure 6 , the significantly up-regulated genes that Plg interacts with include the serpin peptidase inhibitors Serpina1 and Serpinf2, kallikrein B (Klkb1), and coagulation factor XII (F12). Another important protein in Figure 6 is the prothrombin precursor (Coagulation factor II, F2), which interacts with F10, Fga, Fgg, Fn1, Proc, Serpina5, Serpind1, and Vtn. Most of the interactions involving Plg and F2 have been included in STRING via the KEGG pathway for complement and coagulation cascades [19] . The complement system and blood coagulation are a closely interacting pair of proteolytic cascades in blood plasma that are activated after injury [30] . The blood coagulation cascade culminates in the formation of thrombin, the enzyme responsible for the conversion of soluble fibrinogen to the insoluble fibrin clot. Metabolism of xenobiotics (CYP2E1) Cytochrome P450, family 2, subfamily E, polypeptide 1 (CYP2E1) encodes a member of the cytochrome P450 superfamily of enzymes. Cytochrome P450s proteins are monooxygenases, which carry out the liver's prominent role in xenobiotic metabolism and synthesis of cholesterol, steroids and other lipids. CYP2E1 is an important member of this family, implicated in the metabolism of exogenous compounds such as benzene, carbon tetrachloride, ethylene glycol, and substances found in cigarette smoke as well as endogenous compounds including ethanol, acetone, and acetal [31] – [33] . In Figure 6 , CYP2E1 interacts with C2, Cyb5a, CYP4F1, Ephx1, and Mgst1. The interactions of CYP2E1 with Cytochrome P450 4F1 (CYP4F1), Epoxide hydrolase 1 (Ephx1), and Microsomal glutathione S-transferase 1 (Mgst1) are included in the KEGG pathways for metabolism of xenobiotics by Cytochrome P450s and for Arachidonic acid metabolism, which are sources of interactions for STRING. Further support for the role played by CYP2E1 comes from the links to HSA00980_METABOLISM_OF_XENOBIOTICS_BY_CYTOCHROME_P450 from V$HNF1_Q6 and HSIAO_LIVER_SPECIFIC_GENES in the day 8 CBPLN ( Figure 4 ). These links are mediated by the functional interactions between CYP2E1 and members of the alcohol dehydrogenase and glutathione s-transferase gene families (data not shown). Lipid Homeostasis and Bile Acid Synthesis Two of the most important functions that hepatocytes in the liver carry out are lipid homeostasis and bile acid synthesis. These two functions are intrinsically linked. As illustrated schematically in Figure 7 , the liver produces bile acids, which are secreted into the small intestine, where they allow for breakdown of dietary fats and uptake of fatty acids. Subsequently, the liver re-mobilizes these fatty acids throughout the body via lipoproteins [34] . Lipoproteins circulate fatty acids and cholesterol through the body in a cycle that begins with the liver's secretion of fatty acid-rich very low-density lipoproteins (VLDLs) and ends with the liver's uptake of cholesterol-rich high-density lipoproteins (HDLs) [35] . The liver then recycles these cholesterols or converts them into bile acids. Our results capture the high-level relationships between these processes, as displayed in the sub-CBPLNs involving nuclear receptors, the PPAR signaling pathway, bile acid biosynthesis, and fatty acid metabolism ( Figures 8A–8D ). Before we examine some of these links in more detail, we stress that the links in CBPLNs (e.g., the bi-directional links between HSA03320_PPAR_SIGNALING_PATHWAY and HSA00120_BILE_ACID_BIOSYNTHESIS) must be interpreted with caution. Both HSA03320_PPAR_SIGNALING_PATHWAY and HSA00120_BILE_ACID_BIOSYNTHESIS are up-regulated in CS cultures in contrast to HMs ( Fig. 8C and Fig. 8D ). Bile acids directly induce the expression of PPAR [36] , which supports interpreting the observed link from HSA00120_BILE_ACID_BIOSYNTHESIS to HSA03320_PPAR_SIGNALING_PATHWAY as a regulatory one. On the other hand, although it is tempting to infer that the reverse of that link, from HSA03320_PPAR_SIGNALING_PATHWAY to HSA00120_BILE_ACID_BIOSYNTHESIS, also implies the PPAR pathway up-regulates bile acid biosynthesis, such a conclusion may be incorrect. Since the up-regulation trends arise from the comparison of CS cultures to HMs, it is possible that bile acid production in CS cultures is constant (or even decreasing) over time and that bile acid levels in HMs are decreasing. In fact, when we compare the expression values of these two gene sets exclusively within the CS cultures, we observe that there is no statistically significant change between the expression levels of the bile acid biosynthesis genes between days 3 and 8, and that there is a barely statistically significant up-regulation of the genes in the PPAR signaling pathway between the same two days (data not shown). Moreover, PPAR has been shown to directly inhibit production of Cholesterol 7 -hydroxylase (CYP7A1) [37] , [38] . CYP7A1 is the rate-limiting enzyme in the classical pathway of bile acid synthesis from cholesterol [35] . Therefore, while we can conclude from the CBPLN that HSA03320_PPAR_SIGNALING_PATHWAY may regulate HSA00120_BILE_ACID_BIOSYNTHESIS, the mode of regulation (e.g., induction or inhibition) requires more detailed study. We also note modest changes in the interconnections between the gene sets in Figures 8A–8D over the time-course. One example is the disappearance of the link from HSA03320_PPAR_SIGNALING_PATHWAY to HSA00120_BILE_ACID_BIOSYNTHESIS from day 1 to day 2, followed by the reappearance of this link at day 3. We attribute this behavior to a spurious report of the link as significant at day 1, since we believe our methods may be over-sensitive when very few genes are significantly perturbed in a given contrast (as was the case for day 1). We are currently investigating ways to improve the robustness of our methods in reporting links for such scenarios. Two other noticeable changes over the time series have immediate biological interpretations. First, the link from NUCLEAR_RECEPTORS to HSA03320_PPAR_SIGNALING_PATHWAY appears at day 2, which we interpret as a regulatory relationship reflected in the underlying functional interaction network and the corresponding up-regulation of the two gene sets. Second, the link from HSA00071_FATTY_ACID_METABOLISM to HSA03320_PPAR_SIGNALING_PATHWAY also appears at day 2, which we interpret in light of feedback in the fatty acid metabolic pathway. In the rest of this section, we discuss the linkages between these three gene sets, anchoring our discussing on the underlying functional interaction networks on day 8 ( Figures 9 and 10 ). We divide our discussion into three parts: interactions of nuclear receptors with cytochrome P450 enzymes, the role played by PPAR , and the regulation of fatty acid metabolism. Interactions of nuclear receptors with cytochrome P450s In Figure 9 , the nuclear receptors that contribute to the linkage between NUCLEAR_RECEPTORS and the HSA03320_PPAR_SIGNALING_PATHWAY are Hepatocyte Nuclear Factor 4 (Hnf4a), Liver Receptor Homolog-1 (Nr5a2/Lrh1), Liver X Receptor (Nr1h3/Lxra), PPAR , Nuclear Orphan Receptor (Nr1h2/OR-1), and Retinoic acid receptors , , and (RXRa and RXRb, RXRg). The dense network of interactions involving PPAR , RXRa, RXRb, and Nr1h3 have been incorporated into STRING from curated pathway databases such as REACTOME [11] . All these nuclear receptors exhibit increasing perturbation over time, and interact with CYP7A1, a cytochrome P450 enzyme that is a member of the PPAR signaling pathway. Note that CYP7A1 itself shows no significant perturbation until day 8. We discuss the support in the literature for a subset of the interactions with CYP7A1. HNF4 has been shown to bind to the promoter regions of CYP7A1, resulting in up to a nine-fold increase in production of the CYP7A1 protein in vitro [39] . The literature suggests tenuous regulatory connections between liver receptor homolog 1 (LRH-1, or Nr5a2) and CYP7A1. In vitro studies have shown that Nr5a2 both promotes and represses the expression of CYP7A1 [40] , [41] . In a recent study, a knockout of Lrh-1 ( Nr5a2 ) performed selectively in cells that developed into mouse hepatocytes demonstrated that the absence of Nr5a2 had little effect on expression of CYP7A1 [42] . Liver X receptors regulate cholesterol and lipid homeostasis in multiple tissues via two isoforms: LXR (Nr1h3), which is highly expressed in liver, and LXR which is more abundant in adipose tissue, gut, kidney, and macrophages [43] . In contrast to the connection between LRH-1 and CYP7A1, LXR is well known to activate transcription of CYP7A1 in the presence of cholesterol [44] . Thus, it is surprising that we did not observe significant perturbation in expression of CYP7A1 until day 8. However, in vitro studies indicate that CYP7A1 protein exhibits low turnover [35] , raising the possibility that the hepatocytes in both cultures had ample amounts of the proteins up to day 3. Another set of contributions to the linkage between these two gene sets come from interaction of the nuclear receptors Hnf4a and Nr5a2/Lrh1 with sterol 12 -hydroxylase (CYP8B1), a member of the PPAR signaling pathway. CYP8B1 catalyzes a fate-determining reaction in which cholesterol is ultimately converted into the primary bile acid cholic acid, rather than chenodeoxycholic acid [35] . The study of selective knockout of Lrh-1 ( Nr5a2 ) in mice [42] showed that, in contrast to the effect on the expression of CYP7A1 , the knockout caused a significant drop in expression of CYP8B1 , demonstrating a very strong regulatory relationship between Nr5a2 and CYP8B1 [42] . Additionally, strong experimental support for Hnf4a promotion of CYB8B1 expression exists [45] . Thus, the expression of CYP8B1 also increases over time, although it lags the expression of its regulatory receptors Hnf4a and Nr5a2. Nr5a2 is also predicted to interact with 27-hydroxylase (CYP27A1), a mitochondrial cytochrome P450 enzyme that is responsible for a step in the conversion of cholesterol to approximately 25% of the bile acids in mouse [35] . We observe an increase in the perturbed expression of CYP27A1 concomitant to but lagging that of Lrh-1 ( Nr5a2 ). The knockout of Lrh-1 led to significantly decreased expression of CYP27A1 [42] , supporting the interaction of these two genes. The role of PPAR Next, we focus on the role played by PPAR in the linkage between nuclear receptors and the PPAR signaling pathway. PPARs are a class of nuclear receptors responsive to fatty acid ligands. PPARs have been divided among three known subtypes, , , and , with each subtype occurring in distinct tissues and effecting differing biological responses. Liver cells express PPAR , which is responsible for the regulation of fatty acid uptake and catabolism [46] , [47] . In our data, only PPAR shows increasing expression in CS cultures, compared to HMs; the other PPARs are not significantly different between the two culture systems. In Figure 9 , the significantly perturbed members of the PPAR pathway that PPARa interacts with include Scd1, Fabp1, Apoa2, Lpl, Acox1, Cpt1a, and CYP7A1. PPAR has been shown to promote expression of these genes by binding to their upstream Peroxisome Proliferator Regulatory Element (PPRE) regions as a heterodimer with RXR (reviewed in [48] ). We note that RXR shows significant up-regulation in CS versus HM, as well ( Fig. 9 ). RXR has been shown to be particularly highly expressed in the liver [49] . RXR , however, tends to have low expression levels across all tissues [49] . The significant up-regulation of RXR in CS versus HM is somewhat puzzling, given that RXR tends to be exclusively expressed in the brain, anterior pituitary, and skeletal muscle [49] – [51] , where it is responsible for triglyceride uptake and metabolism [52] . We discuss a subset of the interactions involving PPAR next. Stearoyl-Coenzyme A desaturase 1 ( 9-desaturase, Scd1) is the main hepatic isoform of SCD. Scd1 helps catalyze the rate-limiting step in the synthesis of monounsaturated fatty acids, particularly the production of palmitoleic acid and oleic acid from palmitic acid and stearic acid, respectively [48] , [53] . LXR indirectly regulates transcription of Scd1 through activation of transcription of sterol regulatory element binding protein (SREBP) 1c [54] , [55] , an activator of Scd1 transcription [56] , [57] . Additionally, LXR directly activates Scd1 transcription through an upstream response element [58] . PPAR has also been demonstrated to directly activate transcription of Scd1 [59] . Thus, our observation of increasingly significant changes in expression for LXR and PPAR , and a similar trend in Scd1, runs in accordance with previous studies. The interaction of Fatty Acid Binding Protein 1 (Fabp1, L-FABP) with PPAR through protein-protein contacts is thought to promote the expression of proteins involved in fatty-acid oxidation and gluconeogenesis [60] , [61] . Included among these genes is Fabp1 . Thus, it regulates its own expression through PPAR . Regulation of fatty acid metabolism by nuclear receptors The genes in NUCLEAR_RECEPTORS are responsible for initiating cellular responses to a wide variety of conditions and for starting appropriate signal cascades. The nuclear receptors in HSA03320_PPAR_SIGNALING_PATHWAY are the specific subset responsible for initiating the signaling cascade leading to the breakdown of long chain fatty acids [48] . The gene set HSA00071_FATTY_ACID_METABOLISM contains the full contingent of genes responsible for the catabolism of fatty acids. HSA03320_PPAR_SIGNALING_PATHWAY acts as a bridge between the two general classes of genes, NUCLEAR_RECEPTORS and HSA00071_FATTY_ACID_METABOLISM. Figure 9 shows the interactions of individual genes in NUCLEAR_RECEPTORS with those in HSA03320_PPAR_SIGNALING_PATHWAY responsible for the upstream processes of fatty-acid catabolism, including uptake, such as L-FABP (Fabp1) and early-stage fatty-acid -oxidation in the peroxisome, such as acyl-Coenzyme A oxidase 1 (Acox1) [48] . Figure 10 shows the individual genes in HSA03320_PPAR_SIGNALING_PATHWAY that interact with those in HSA00071_FATTY_ACID_METABOLISM responsible for later stages of -oxidation in the mitochondria, such as acetyl-Coenzyme A acyltransferase 2 (Acaa2) and hydroxyacyl-Coenzyme A dehydrogenase/3-ketoacyl-Coenzyme A thiolase/enoyl-Coenzyme A hydratase (Hadhb) [48] . Thus, the signals from NUCLEAR_RECEPTORS are transferred to HSA00071_FATTY_ACID_METABOLISM via the subset of nuclear receptors that are members of the PPAR signaling pathway, a chain of events that we are able to recover in the CBPLNs. Interpretation of Links in CBPLNs Keeping the examples of the previous sections in mind, we now discuss how links in CBPLNs might be interpreted. Regulatory relationship Gene set may contain genes whose products regulate genes and/or their products in gene set . An example is the linkage from NUCLEAR_RECEPTORS to other gene sets such as HSA03320_PPAR_SIGNALING_PATHWAY and genes involved in cellular lipid metabolism; many liver-specific nuclear receptors such as LXR and HNF4 regulate critical hepatic processes. Multi-input motif Multiple gene sets may link to a gene set , suggesting that the expression of genes in is regulated by genes in multiple other sets. Such a phenomenon is called a “multi-input motif” in the case of a gene being regulated by multiple transcription factors [62] . An example is HSIAO_LIVER_SPECIFIC_GENES and links to this gene set from V$HNF1_Q6 and NUCLEAR_RECEPTORS. Feedback Links that exist in both directions between and may suggest that regulates and that receives a feedback signal from . This phenomenon may be observed within CBPLNs when the link is unidirectional at some time points and bidirectional in later time points. A specific example is the linkage between bile acid biosynthesis and HSA03320_PPAR_SIGNALING_PATHWAY, which is unidirectional on day 2 ( Figure 8B ) but bidirectional on days 3 and 8 ( Figures 8C and 8D ). Downstream in the signal flow A link from process to process and another from and process may suggest that lies downstream of . An instance of this feature is the link from NUCLEAR_RECEPTORS to HSA03320_PPAR_SIGNALING_PATHWAY and the link from HSA03320_PPAR_SIGNALING_PATHWAY to HSA00071_FATTY_ACID_METABOLISM in Figure 8B . Multi-functional gene set A gene set that has many incoming links and/or many outgoing links might be an example of a multi-functional gene set. A prominent example in our CBPLNs is the central HSIAO_LIVER_SPECIFIC_GENES gene set. As we remarked earlier, the links incident on this gene set suggest what other processes the genes in HSIAO_LIVER_SPECIFIC_GENES may regulate or be connected to. Clearly, such a feature depends on how a gene set is defined. For example, many biological processes in the Gene Ontology such as “response to stress” are themselves composed of well-defined and functionally-coherent processes. Similarly, the genes that are perturbed by a particular stimulus may participate in a wide variety of processes. CBPLNs can situate such genes in a rich context within the underlying network of molecular interactions. Conclusions We have presented an approach that represents cellular responses at the granularity of biological processes and connections among them. Our approach extends the work of Dotan-Cohen et al. [8] by integrating transcriptional data (the “context”) with functional interaction networks. We focused our analysis on nearly 20 MSigDB gene sets we had identified as up-regulated in hepatocyte cultures in an earlier study. CBPLNs revealed numerous meaningful connections between different biological processes and gene sets, which we were successful in interpreting within the context of liver metabolism. Links and local network features in CBPLNs are generalizations of diverse physiological phenomena such as regulation, feedback, and downstream signal flow from the gene/protein level to the scale of biological processes. Our approach is a complement to a suite of methodologies that integrate physical, signaling, regulatory, and functional networks with measurements of molecular profiles such as transcriptional, proteomic, or metabolic data to compute the response network , which may be defined as the sub-network of interactions that are perturbed in a particular condition. A wide variety of methods have been developed for computing such response networks [63] – [67] . Response networks are typically interpreted by computing which biological processes are enriched in them. In contrast, rather than compute the entire response network, we focus on discovering connections between perturbed biological processes. Since response networks can include genes without any annotations, they can be used to predict biological processes to which unannotated genes belong [68] . In contrast, only genes annotated to some biological process can contribute to CBPLNs. A detailed comparison of CBPLNs to response networks and the development of methods that combine both approaches will be the focus of future research. Generalizing our approach to the entire spectrum of MSigDB gene sets or to the set of all biological processes in the Gene Ontology raises several interesting challenges. First, gene sets can have considerable overlap, leading to redundant links. Second, scaling this approach up to thousands of gene sets may result in tens to hundreds of thousands of links that are deemed to be statistically significant. This deluge of links will be hard to interpret. Third, it will be challenging to computationally scale our permutation-based sampling to the large number of process pairs we will have to test. We are currently investigating these issues. In this work, we computed CBPLNs for two conventional hepatocyte culture systems. Three dimensional liver mimics [69] , [70] and microscale co-culture systems [71] have shown improved retention of hepatic phenotype over conventional systems. In the future, we plan to apply CBPLNs to liver mimics and co-culture systems in order to obtain insights into the inter-cellular signaling mechanisms that confer improved hepatic phenotype. More generally, our approach may provide a novel route to explore, analyze, and interpret cellular responses to internal and external cues.

Results and Discussion Input Data Gene Expression Data We used the Affymetrix Rat Genome 230 2.0 GeneChip to measure genomewide transcriptional profiles in rat hepatocytes grown in monolayers and in collagen sandwiches [5] . This dataset is available in MIAME-compliant format in the Gene Expression Omnibus (accession number GSE20659). We marked the day when we deposited the second layer of collagen in CS cultures as day zero. On days one, two, three and eight after deposition of the second layer of collagen, we measured data in triplicate in hepatocytes in each culture system. Functional Linkage Network Existing databases of protein interactions contain very few experimentally detected Protein-Protein Interactions (PPIs) for rat: seven different widely-used sources [11] – [17] contained a total of just 1,274 non-redundant rat PPIs spanning 974 proteins. Therefore, we decided to use the rat functional linkage network predicted by the STRING system [18] . The interaction type in STRING is a functional association , which the authors define as “the specific and meaningful interaction between two proteins that jointly contribute to the same functional process.” Apart from incorporating experimental interaction data, STRING uses multiple methods to predict possible functional linkages including interolog-based interaction transfer, similar transcriptional response across a variety of conditions (co-expression), text-mining, and gene families that share above-random similarities in their evolutionary histories. STRING includes a scheme to score each predicted interaction in the range 150–1000 against a common reference of functional partnership based on the KEGG database [19] . STRING version 8.3, released on May 26, 2010 contains 975,454 predicted interactions among 15,178 rat proteins. We used the subset of these interactions with a weight of at least 500; there were 204,992 such interactions among 9,925 proteins. We selected 500 as a cutoff based on the reasoning that interactions with at least this weight were more likely to connect genes belonging to the same process than to connect genes belonging to different processes. When we further pruned the network to include genes with at least one annotation (see below), we obtained 47,002 interactions among 4,714 genes. Functional Annotations In our earlier work [5] , we used GSEA to compare the two culture systems at each of the four time points; Table 1 lists the contrasts we analyzed. This analysis provided insights into the temporal patterns of up- and down-regulation in the gene sets in the Molecular Signature Database (MSigDB) [6] . In that work, we focused our analysis on gene sets that showed monotonically diverging patterns of expression between CS and HM cultures. In the current paper, we use the curated (c2), motif (c3), and Gene Ontology (c5) collections of gene sets in MSigDB as our set of functional annotations. We focus on establishing linkages among the subset of 18 up-regulated gene sets from the previous study; Table 2 lists these sets along with a short description of each. Overview of Results We considered only those links with a link -value of at most , after using the method of Benjamini and Hochberg [20] to adjust for testing multiple hypotheses. We further restricted our attention to pairs of gene sets for which at least 10 genes exclusively in the second set of the pair interacted with genes in the first set, reasoning that fewer interacting genes might not yield robust link -values. We compared the number of links computed by using each hypothesis test. We also compared these values to the number of links in the (context-free) BPLN computed using the method of Dotan-Cohen et al. [8] . Tables 3 and 4 display the results of the comparisons. Several salient trends emerged. First, in Table 3 , irrespective of the hypothesis test used, the number of links increased with time. This phenomenon parallels our earlier observation that the transcriptional programs of hepatocytes in CS cultures steadily diverged from that in HMs. Second, the size of the intersection between the two sets of links also increased with time, as did the Jaccard similarity coefficient of the two sets (i.e., the size of the intersection divided by the size of the union). Further, for each day, the number of links deemed to be significant by both hypothesis tests was itself statistically significant, based on Fisher's Exact Test (see File S1 ). These trends suggest that once the transcriptional programs of the two culture systems have diverged (day 2 and later), both hypothesis tests find very similar sets of process pairs to be significantly linked at the 0.01 level. However, the number of common links is very close to the number of links identified by the second hypothesis test, indicating that the second test is more conservative than the first in deciding whether a link is statistically significant. We observed similar results when we repeated these analyses with other cutoffs on the link -value (0.005, 0.05, and 0.1) (see File S2 ). Third, normalizing the linkage score (see “Methods”) pruned out a small number of links. Finally, the overlap between the intersection of the results from both hypothesis tests and the BPLN was small in days 1 and 2 and more substantial in days 3 and 8 ( Table 4 ), although the overlap was still statistically significant by Fisher's Exact Test (see File S1 ). These data suggest that only a subset of the links in a BPLN may have some relevance to the particular biological conditions being investigated. By incorporating measurements of gene expression, CBPLNs can identify those inter-process links that correspond to the phenotypic differences observed in the two conditions being compared (e.g., hepatocytes in CS versus HM). Although both hypothesis tests find very similar sets of process pairs to be significantly linked at the 0.01 level, especially in later days, we found that the actual link -values computed for each process pair were not very highly correlated to each other (see File S3 ). Based on these results, we decided to consider a linkage between a pair of gene sets only if this link was significant at the 0.01 level with both hypothesis tests with normalization. The resulting CBPLNs are displayed in Figures 1 , 2 , 3 , 4 . For reference, we have displayed the BPLN in Figure 5 . We discuss the properties of these CBPLNs in the rest of the paper. We focus primarily on the day 8 CBPLN ( Figure 4 ), noting that many of the features we discuss are also apparent in the day 3 CBPLN ( Figure 3 ). When we discuss some pairs of linked gene sets, we refer to the underlying functional interaction network connecting the genes in those sets. We start by discussing properties of liver-specific genes, focusing particularly on the regulation of these genes by the transcription factor HNF1. Then, we discuss the role of lipid homeostasis and bile acid synthesis in the liver. Finally, we summarize the different interpretations of the links in CBPLNs. We stress that the formulation of linkage between processes and is asymmetric. Hence, by definition, links in the CBPLN are directed, i.e., a CBPLN may contain a link between and and between and . Liver Specific Genes The 251 genes in the HSIAO_LIVER_SPECIFIC_GENES gene set are expressed selectively in the liver, as determined by Hsiao et al. [21] from a compendium of gene expression in normal human tissues created with the goal of defining a reference for basic organ systems biology. Genes in this set are members of a spectrum of biological processes, including fatty acid metabolism, metabolism of xenobiotics, blood coagulation, and response to wounding. Not surprisingly, this gene set occupies a central place in the CBPLN on day 8 ( Figure 4 ); it has the highest number of outgoing and incoming links. Outgoing links include connections to glycolysis and gluconeogenesis, alcohol metabolism process, metabolism of xenobiotics by cytochrome P450s, the Peroxisome Proliferator-Activated Receptor (PPAR) signaling pathway, lipid metabolic processes, the urea cycle, and bile acid biosynthesis, among others. In turn, the gene sets such as V$HNF1_Q6 and NUCLEAR_RECEPTORS are linked to HSIAO_LIVER_SPECIFIC_GENES. Some links involving HSIAO_LIVER_SPECIFIC_GENES are unidirectional on day 2 or day 3 ( Figures 2 and 3 ) but bidirectional on day 8 ( Figure 4 ), e.g., to HSA03320_PPAR_SIGNALING_PATHWAY and metabolism of fatty acids, bile acids, and alcohol. Such features suggest that CBPLNs may be representing cellular signals emanating from a subset of liver specific genes to other processes and subsequent feedback from the other processes to liver specific genes. Overall, these results suggest that CBPLNs can assist in the sub-division of liver-specific genes into more refined categories, based not only on the functions of the genes themselves, but also on how they are regulated and what other processes they may control. We discuss one specific link next that illustrates this property. Liver Specific Gene Sets Regulated by HNF1 Hepatic nuclear factor 1 (HNF1), also known as albumin proximal factor, is a transcription factor required for the expression of several liver-specific genes including albumin [22] . The protein functions as a homodimer and binds to the inverted palindrome 5′-GTTAATNATTAAC-3′ . The promoter regions of genes in the MSigDB set V$HNF1_Q6 match this binding site for HNF1 [23] . In our previous study [5] , we noted the monotonic up-regulation of this gene set in CS cultures when compared to HMs. This gene set has an overlap of 25 genes with the gene set HSIAO_LIVER_SPECIFIC_GENES. We concluded that HNF1 monotonically up-regulates the expression of liver-specific genes in CS cultures but not in HMs. CBPLNs assist us in elaborating upon these earlier observations. We studied the link between V$HNF1_Q6 and HSIAO_LIVER_SPECIFIC_GENES in the day 8 CBPLN by examining the functional interactions in the STRING database connecting genes in V$HNF1_Q6 to genes in HSIAO_LIVER_SPECIFIC_GENES. Figure 6 displays a layout of this network. Visual examination of Figure 6 indicates that the linkage between these two gene sets is driven by the genes F2, Plg, CYP2E1, Nr1h4, Lipc, and their interactors, with weaker contributions arising from Hnf1a and Hnf4a. Note that F2, Plg, CYP2E1, Nr1h4, and Lipc are members of both gene sets while Hnf1a and Hnf4a are members of V$HNF1_Q6. We discuss a subset of these proteins next, highlighting liver-specific processes they participate in. HNF1a and HNF4a Hepatocyte nuclear factor 4 (Hnf4a) is a nuclear receptor implicated in the regulation of numerous genes associated with hepatic function [24] – [26] , gluconeogenesis [27] , and activation of the metabolism of xenobiotics, including drugs and pharmaceuticals [28] . It is known that both the HNF4 protein and HNF1 protein can transactivate the HNF1 gene [29] . Although both genes are not very highly up-regulated, their interactions with liver-specific genes Apoa2, Serpina1, Pcbd1, Slc2a2, Slc10a1, Fabp1, and Pck1 suggest the activation of many liver-related pathways. Blood clotting (Plg and F2) Plasminogen (Plg) is a secreted protein that is proteolysed to plasmin and angiostatin. Plasmin dissolves fibrin in blood clots while angiostatin inhibits angiogenesis. In Figure 6 , the significantly up-regulated genes that Plg interacts with include the serpin peptidase inhibitors Serpina1 and Serpinf2, kallikrein B (Klkb1), and coagulation factor XII (F12). Another important protein in Figure 6 is the prothrombin precursor (Coagulation factor II, F2), which interacts with F10, Fga, Fgg, Fn1, Proc, Serpina5, Serpind1, and Vtn. Most of the interactions involving Plg and F2 have been included in STRING via the KEGG pathway for complement and coagulation cascades [19] . The complement system and blood coagulation are a closely interacting pair of proteolytic cascades in blood plasma that are activated after injury [30] . The blood coagulation cascade culminates in the formation of thrombin, the enzyme responsible for the conversion of soluble fibrinogen to the insoluble fibrin clot. Metabolism of xenobiotics (CYP2E1) Cytochrome P450, family 2, subfamily E, polypeptide 1 (CYP2E1) encodes a member of the cytochrome P450 superfamily of enzymes. Cytochrome P450s proteins are monooxygenases, which carry out the liver's prominent role in xenobiotic metabolism and synthesis of cholesterol, steroids and other lipids. CYP2E1 is an important member of this family, implicated in the metabolism of exogenous compounds such as benzene, carbon tetrachloride, ethylene glycol, and substances found in cigarette smoke as well as endogenous compounds including ethanol, acetone, and acetal [31] – [33] . In Figure 6 , CYP2E1 interacts with C2, Cyb5a, CYP4F1, Ephx1, and Mgst1. The interactions of CYP2E1 with Cytochrome P450 4F1 (CYP4F1), Epoxide hydrolase 1 (Ephx1), and Microsomal glutathione S-transferase 1 (Mgst1) are included in the KEGG pathways for metabolism of xenobiotics by Cytochrome P450s and for Arachidonic acid metabolism, which are sources of interactions for STRING. Further support for the role played by CYP2E1 comes from the links to HSA00980_METABOLISM_OF_XENOBIOTICS_BY_CYTOCHROME_P450 from V$HNF1_Q6 and HSIAO_LIVER_SPECIFIC_GENES in the day 8 CBPLN ( Figure 4 ). These links are mediated by the functional interactions between CYP2E1 and members of the alcohol dehydrogenase and glutathione s-transferase gene families (data not shown). Lipid Homeostasis and Bile Acid Synthesis Two of the most important functions that hepatocytes in the liver carry out are lipid homeostasis and bile acid synthesis. These two functions are intrinsically linked. As illustrated schematically in Figure 7 , the liver produces bile acids, which are secreted into the small intestine, where they allow for breakdown of dietary fats and uptake of fatty acids. Subsequently, the liver re-mobilizes these fatty acids throughout the body via lipoproteins [34] . Lipoproteins circulate fatty acids and cholesterol through the body in a cycle that begins with the liver's secretion of fatty acid-rich very low-density lipoproteins (VLDLs) and ends with the liver's uptake of cholesterol-rich high-density lipoproteins (HDLs) [35] . The liver then recycles these cholesterols or converts them into bile acids. Our results capture the high-level relationships between these processes, as displayed in the sub-CBPLNs involving nuclear receptors, the PPAR signaling pathway, bile acid biosynthesis, and fatty acid metabolism ( Figures 8A–8D ). Before we examine some of these links in more detail, we stress that the links in CBPLNs (e.g., the bi-directional links between HSA03320_PPAR_SIGNALING_PATHWAY and HSA00120_BILE_ACID_BIOSYNTHESIS) must be interpreted with caution. Both HSA03320_PPAR_SIGNALING_PATHWAY and HSA00120_BILE_ACID_BIOSYNTHESIS are up-regulated in CS cultures in contrast to HMs ( Fig. 8C and Fig. 8D ). Bile acids directly induce the expression of PPAR [36] , which supports interpreting the observed link from HSA00120_BILE_ACID_BIOSYNTHESIS to HSA03320_PPAR_SIGNALING_PATHWAY as a regulatory one. On the other hand, although it is tempting to infer that the reverse of that link, from HSA03320_PPAR_SIGNALING_PATHWAY to HSA00120_BILE_ACID_BIOSYNTHESIS, also implies the PPAR pathway up-regulates bile acid biosynthesis, such a conclusion may be incorrect. Since the up-regulation trends arise from the comparison of CS cultures to HMs, it is possible that bile acid production in CS cultures is constant (or even decreasing) over time and that bile acid levels in HMs are decreasing. In fact, when we compare the expression values of these two gene sets exclusively within the CS cultures, we observe that there is no statistically significant change between the expression levels of the bile acid biosynthesis genes between days 3 and 8, and that there is a barely statistically significant up-regulation of the genes in the PPAR signaling pathway between the same two days (data not shown). Moreover, PPAR has been shown to directly inhibit production of Cholesterol 7 -hydroxylase (CYP7A1) [37] , [38] . CYP7A1 is the rate-limiting enzyme in the classical pathway of bile acid synthesis from cholesterol [35] . Therefore, while we can conclude from the CBPLN that HSA03320_PPAR_SIGNALING_PATHWAY may regulate HSA00120_BILE_ACID_BIOSYNTHESIS, the mode of regulation (e.g., induction or inhibition) requires more detailed study. We also note modest changes in the interconnections between the gene sets in Figures 8A–8D over the time-course. One example is the disappearance of the link from HSA03320_PPAR_SIGNALING_PATHWAY to HSA00120_BILE_ACID_BIOSYNTHESIS from day 1 to day 2, followed by the reappearance of this link at day 3. We attribute this behavior to a spurious report of the link as significant at day 1, since we believe our methods may be over-sensitive when very few genes are significantly perturbed in a given contrast (as was the case for day 1). We are currently investigating ways to improve the robustness of our methods in reporting links for such scenarios. Two other noticeable changes over the time series have immediate biological interpretations. First, the link from NUCLEAR_RECEPTORS to HSA03320_PPAR_SIGNALING_PATHWAY appears at day 2, which we interpret as a regulatory relationship reflected in the underlying functional interaction network and the corresponding up-regulation of the two gene sets. Second, the link from HSA00071_FATTY_ACID_METABOLISM to HSA03320_PPAR_SIGNALING_PATHWAY also appears at day 2, which we interpret in light of feedback in the fatty acid metabolic pathway. In the rest of this section, we discuss the linkages between these three gene sets, anchoring our discussing on the underlying functional interaction networks on day 8 ( Figures 9 and 10 ). We divide our discussion into three parts: interactions of nuclear receptors with cytochrome P450 enzymes, the role played by PPAR , and the regulation of fatty acid metabolism. Interactions of nuclear receptors with cytochrome P450s In Figure 9 , the nuclear receptors that contribute to the linkage between NUCLEAR_RECEPTORS and the HSA03320_PPAR_SIGNALING_PATHWAY are Hepatocyte Nuclear Factor 4 (Hnf4a), Liver Receptor Homolog-1 (Nr5a2/Lrh1), Liver X Receptor (Nr1h3/Lxra), PPAR , Nuclear Orphan Receptor (Nr1h2/OR-1), and Retinoic acid receptors , , and (RXRa and RXRb, RXRg). The dense network of interactions involving PPAR , RXRa, RXRb, and Nr1h3 have been incorporated into STRING from curated pathway databases such as REACTOME [11] . All these nuclear receptors exhibit increasing perturbation over time, and interact with CYP7A1, a cytochrome P450 enzyme that is a member of the PPAR signaling pathway. Note that CYP7A1 itself shows no significant perturbation until day 8. We discuss the support in the literature for a subset of the interactions with CYP7A1. HNF4 has been shown to bind to the promoter regions of CYP7A1, resulting in up to a nine-fold increase in production of the CYP7A1 protein in vitro [39] . The literature suggests tenuous regulatory connections between liver receptor homolog 1 (LRH-1, or Nr5a2) and CYP7A1. In vitro studies have shown that Nr5a2 both promotes and represses the expression of CYP7A1 [40] , [41] . In a recent study, a knockout of Lrh-1 ( Nr5a2 ) performed selectively in cells that developed into mouse hepatocytes demonstrated that the absence of Nr5a2 had little effect on expression of CYP7A1 [42] . Liver X receptors regulate cholesterol and lipid homeostasis in multiple tissues via two isoforms: LXR (Nr1h3), which is highly expressed in liver, and LXR which is more abundant in adipose tissue, gut, kidney, and macrophages [43] . In contrast to the connection between LRH-1 and CYP7A1, LXR is well known to activate transcription of CYP7A1 in the presence of cholesterol [44] . Thus, it is surprising that we did not observe significant perturbation in expression of CYP7A1 until day 8. However, in vitro studies indicate that CYP7A1 protein exhibits low turnover [35] , raising the possibility that the hepatocytes in both cultures had ample amounts of the proteins up to day 3. Another set of contributions to the linkage between these two gene sets come from interaction of the nuclear receptors Hnf4a and Nr5a2/Lrh1 with sterol 12 -hydroxylase (CYP8B1), a member of the PPAR signaling pathway. CYP8B1 catalyzes a fate-determining reaction in which cholesterol is ultimately converted into the primary bile acid cholic acid, rather than chenodeoxycholic acid [35] . The study of selective knockout of Lrh-1 ( Nr5a2 ) in mice [42] showed that, in contrast to the effect on the expression of CYP7A1 , the knockout caused a significant drop in expression of CYP8B1 , demonstrating a very strong regulatory relationship between Nr5a2 and CYP8B1 [42] . Additionally, strong experimental support for Hnf4a promotion of CYB8B1 expression exists [45] . Thus, the expression of CYP8B1 also increases over time, although it lags the expression of its regulatory receptors Hnf4a and Nr5a2. Nr5a2 is also predicted to interact with 27-hydroxylase (CYP27A1), a mitochondrial cytochrome P450 enzyme that is responsible for a step in the conversion of cholesterol to approximately 25% of the bile acids in mouse [35] . We observe an increase in the perturbed expression of CYP27A1 concomitant to but lagging that of Lrh-1 ( Nr5a2 ). The knockout of Lrh-1 led to significantly decreased expression of CYP27A1 [42] , supporting the interaction of these two genes. The role of PPAR Next, we focus on the role played by PPAR in the linkage between nuclear receptors and the PPAR signaling pathway. PPARs are a class of nuclear receptors responsive to fatty acid ligands. PPARs have been divided among three known subtypes, , , and , with each subtype occurring in distinct tissues and effecting differing biological responses. Liver cells express PPAR , which is responsible for the regulation of fatty acid uptake and catabolism [46] , [47] . In our data, only PPAR shows increasing expression in CS cultures, compared to HMs; the other PPARs are not significantly different between the two culture systems. In Figure 9 , the significantly perturbed members of the PPAR pathway that PPARa interacts with include Scd1, Fabp1, Apoa2, Lpl, Acox1, Cpt1a, and CYP7A1. PPAR has been shown to promote expression of these genes by binding to their upstream Peroxisome Proliferator Regulatory Element (PPRE) regions as a heterodimer with RXR (reviewed in [48] ). We note that RXR shows significant up-regulation in CS versus HM, as well ( Fig. 9 ). RXR has been shown to be particularly highly expressed in the liver [49] . RXR , however, tends to have low expression levels across all tissues [49] . The significant up-regulation of RXR in CS versus HM is somewhat puzzling, given that RXR tends to be exclusively expressed in the brain, anterior pituitary, and skeletal muscle [49] – [51] , where it is responsible for triglyceride uptake and metabolism [52] . We discuss a subset of the interactions involving PPAR next. Stearoyl-Coenzyme A desaturase 1 ( 9-desaturase, Scd1) is the main hepatic isoform of SCD. Scd1 helps catalyze the rate-limiting step in the synthesis of monounsaturated fatty acids, particularly the production of palmitoleic acid and oleic acid from palmitic acid and stearic acid, respectively [48] , [53] . LXR indirectly regulates transcription of Scd1 through activation of transcription of sterol regulatory element binding protein (SREBP) 1c [54] , [55] , an activator of Scd1 transcription [56] , [57] . Additionally, LXR directly activates Scd1 transcription through an upstream response element [58] . PPAR has also been demonstrated to directly activate transcription of Scd1 [59] . Thus, our observation of increasingly significant changes in expression for LXR and PPAR , and a similar trend in Scd1, runs in accordance with previous studies. The interaction of Fatty Acid Binding Protein 1 (Fabp1, L-FABP) with PPAR through protein-protein contacts is thought to promote the expression of proteins involved in fatty-acid oxidation and gluconeogenesis [60] , [61] . Included among these genes is Fabp1 . Thus, it regulates its own expression through PPAR . Regulation of fatty acid metabolism by nuclear receptors The genes in NUCLEAR_RECEPTORS are responsible for initiating cellular responses to a wide variety of conditions and for starting appropriate signal cascades. The nuclear receptors in HSA03320_PPAR_SIGNALING_PATHWAY are the specific subset responsible for initiating the signaling cascade leading to the breakdown of long chain fatty acids [48] . The gene set HSA00071_FATTY_ACID_METABOLISM contains the full contingent of genes responsible for the catabolism of fatty acids. HSA03320_PPAR_SIGNALING_PATHWAY acts as a bridge between the two general classes of genes, NUCLEAR_RECEPTORS and HSA00071_FATTY_ACID_METABOLISM. Figure 9 shows the interactions of individual genes in NUCLEAR_RECEPTORS with those in HSA03320_PPAR_SIGNALING_PATHWAY responsible for the upstream processes of fatty-acid catabolism, including uptake, such as L-FABP (Fabp1) and early-stage fatty-acid -oxidation in the peroxisome, such as acyl-Coenzyme A oxidase 1 (Acox1) [48] . Figure 10 shows the individual genes in HSA03320_PPAR_SIGNALING_PATHWAY that interact with those in HSA00071_FATTY_ACID_METABOLISM responsible for later stages of -oxidation in the mitochondria, such as acetyl-Coenzyme A acyltransferase 2 (Acaa2) and hydroxyacyl-Coenzyme A dehydrogenase/3-ketoacyl-Coenzyme A thiolase/enoyl-Coenzyme A hydratase (Hadhb) [48] . Thus, the signals from NUCLEAR_RECEPTORS are transferred to HSA00071_FATTY_ACID_METABOLISM via the subset of nuclear receptors that are members of the PPAR signaling pathway, a chain of events that we are able to recover in the CBPLNs. Interpretation of Links in CBPLNs Keeping the examples of the previous sections in mind, we now discuss how links in CBPLNs might be interpreted. Regulatory relationship Gene set may contain genes whose products regulate genes and/or their products in gene set . An example is the linkage from NUCLEAR_RECEPTORS to other gene sets such as HSA03320_PPAR_SIGNALING_PATHWAY and genes involved in cellular lipid metabolism; many liver-specific nuclear receptors such as LXR and HNF4 regulate critical hepatic processes. Multi-input motif Multiple gene sets may link to a gene set , suggesting that the expression of genes in is regulated by genes in multiple other sets. Such a phenomenon is called a “multi-input motif” in the case of a gene being regulated by multiple transcription factors [62] . An example is HSIAO_LIVER_SPECIFIC_GENES and links to this gene set from V$HNF1_Q6 and NUCLEAR_RECEPTORS. Feedback Links that exist in both directions between and may suggest that regulates and that receives a feedback signal from . This phenomenon may be observed within CBPLNs when the link is unidirectional at some time points and bidirectional in later time points. A specific example is the linkage between bile acid biosynthesis and HSA03320_PPAR_SIGNALING_PATHWAY, which is unidirectional on day 2 ( Figure 8B ) but bidirectional on days 3 and 8 ( Figures 8C and 8D ). Downstream in the signal flow A link from process to process and another from and process may suggest that lies downstream of . An instance of this feature is the link from NUCLEAR_RECEPTORS to HSA03320_PPAR_SIGNALING_PATHWAY and the link from HSA03320_PPAR_SIGNALING_PATHWAY to HSA00071_FATTY_ACID_METABOLISM in Figure 8B . Multi-functional gene set A gene set that has many incoming links and/or many outgoing links might be an example of a multi-functional gene set. A prominent example in our CBPLNs is the central HSIAO_LIVER_SPECIFIC_GENES gene set. As we remarked earlier, the links incident on this gene set suggest what other processes the genes in HSIAO_LIVER_SPECIFIC_GENES may regulate or be connected to. Clearly, such a feature depends on how a gene set is defined. For example, many biological processes in the Gene Ontology such as “response to stress” are themselves composed of well-defined and functionally-coherent processes. Similarly, the genes that are perturbed by a particular stimulus may participate in a wide variety of processes. CBPLNs can situate such genes in a rich context within the underlying network of molecular interactions. Conclusions We have presented an approach that represents cellular responses at the granularity of biological processes and connections among them. Our approach extends the work of Dotan-Cohen et al. [8] by integrating transcriptional data (the “context”) with functional interaction networks. We focused our analysis on nearly 20 MSigDB gene sets we had identified as up-regulated in hepatocyte cultures in an earlier study. CBPLNs revealed numerous meaningful connections between different biological processes and gene sets, which we were successful in interpreting within the context of liver metabolism. Links and local network features in CBPLNs are generalizations of diverse physiological phenomena such as regulation, feedback, and downstream signal flow from the gene/protein level to the scale of biological processes. Our approach is a complement to a suite of methodologies that integrate physical, signaling, regulatory, and functional networks with measurements of molecular profiles such as transcriptional, proteomic, or metabolic data to compute the response network , which may be defined as the sub-network of interactions that are perturbed in a particular condition. A wide variety of methods have been developed for computing such response networks [63] – [67] . Response networks are typically interpreted by computing which biological processes are enriched in them. In contrast, rather than compute the entire response network, we focus on discovering connections between perturbed biological processes. Since response networks can include genes without any annotations, they can be used to predict biological processes to which unannotated genes belong [68] . In contrast, only genes annotated to some biological process can contribute to CBPLNs. A detailed comparison of CBPLNs to response networks and the development of methods that combine both approaches will be the focus of future research. Generalizing our approach to the entire spectrum of MSigDB gene sets or to the set of all biological processes in the Gene Ontology raises several interesting challenges. First, gene sets can have considerable overlap, leading to redundant links. Second, scaling this approach up to thousands of gene sets may result in tens to hundreds of thousands of links that are deemed to be statistically significant. This deluge of links will be hard to interpret. Third, it will be challenging to computationally scale our permutation-based sampling to the large number of process pairs we will have to test. We are currently investigating these issues. In this work, we computed CBPLNs for two conventional hepatocyte culture systems. Three dimensional liver mimics [69] , [70] and microscale co-culture systems [71] have shown improved retention of hepatic phenotype over conventional systems. In the future, we plan to apply CBPLNs to liver mimics and co-culture systems in order to obtain insights into the inter-cellular signaling mechanisms that confer improved hepatic phenotype. More generally, our approach may provide a novel route to explore, analyze, and interpret cellular responses to internal and external cues.

Conclusions We have presented an approach that represents cellular responses at the granularity of biological processes and connections among them. Our approach extends the work of Dotan-Cohen et al. [8] by integrating transcriptional data (the “context”) with functional interaction networks. We focused our analysis on nearly 20 MSigDB gene sets we had identified as up-regulated in hepatocyte cultures in an earlier study. CBPLNs revealed numerous meaningful connections between different biological processes and gene sets, which we were successful in interpreting within the context of liver metabolism. Links and local network features in CBPLNs are generalizations of diverse physiological phenomena such as regulation, feedback, and downstream signal flow from the gene/protein level to the scale of biological processes. Our approach is a complement to a suite of methodologies that integrate physical, signaling, regulatory, and functional networks with measurements of molecular profiles such as transcriptional, proteomic, or metabolic data to compute the response network , which may be defined as the sub-network of interactions that are perturbed in a particular condition. A wide variety of methods have been developed for computing such response networks [63] – [67] . Response networks are typically interpreted by computing which biological processes are enriched in them. In contrast, rather than compute the entire response network, we focus on discovering connections between perturbed biological processes. Since response networks can include genes without any annotations, they can be used to predict biological processes to which unannotated genes belong [68] . In contrast, only genes annotated to some biological process can contribute to CBPLNs. A detailed comparison of CBPLNs to response networks and the development of methods that combine both approaches will be the focus of future research. Generalizing our approach to the entire spectrum of MSigDB gene sets or to the set of all biological processes in the Gene Ontology raises several interesting challenges. First, gene sets can have considerable overlap, leading to redundant links. Second, scaling this approach up to thousands of gene sets may result in tens to hundreds of thousands of links that are deemed to be statistically significant. This deluge of links will be hard to interpret. Third, it will be challenging to computationally scale our permutation-based sampling to the large number of process pairs we will have to test. We are currently investigating these issues. In this work, we computed CBPLNs for two conventional hepatocyte culture systems. Three dimensional liver mimics [69] , [70] and microscale co-culture systems [71] have shown improved retention of hepatic phenotype over conventional systems. In the future, we plan to apply CBPLNs to liver mimics and co-culture systems in order to obtain insights into the inter-cellular signaling mechanisms that confer improved hepatic phenotype. More generally, our approach may provide a novel route to explore, analyze, and interpret cellular responses to internal and external cues.

Conceived and designed the experiments: CDL TMM. Performed the experiments: CDL. Analyzed the data: CDL PR TMM. Contributed reagents/materials/analysis tools: CDL TMM. Wrote the paper: CDL PR TMM. Designed and developed the software used in analysis: CDL. The liver plays a vital role in glucose homeostasis, the synthesis of bile acids and the detoxification of foreign substances. Liver culture systems are widely used to test adverse effects of drugs and environmental toxicants. The two most prevalent liver culture systems are hepatocyte monolayers (HMs) and collagen sandwiches (CS). Despite their wide use, comprehensive transcriptional programs and interaction networks in these culture systems have not been systematically investigated. We integrated an existing temporal transcriptional dataset for HM and CS cultures of rat hepatocytes with a functional interaction network of rat genes. We aimed to exploit the functional interactions to identify statistically significant linkages between perturbed biological processes. To this end, we developed a novel approach to compute Contextual Biological Process Linkage Networks (CBPLNs). CBPLNs revealed numerous meaningful connections between different biological processes and gene sets, which we were successful in interpreting within the context of liver metabolism. Multiple phenomena captured by CBPLNs at the process level such as regulation, downstream effects, and feedback loops have well described counterparts at the gene and protein level. CBPLNs reveal high-level linkages between pathways and processes, making the identification of important biological trends more tractable than through interactions between individual genes and molecules alone. Our approach may provide a new route to explore, analyze, and understand cellular responses to internal and external cues within the context of the intricate networks of molecular interactions that control cellular behavior.

Supporting Information

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e15247

oa_package/93/64/PMC3016309.tar.gz

PMC3016310

21126334

Background It is well established that the incidence, prevalence and presentation of mental disorders differ by gender, ethnicity and age. For example, men suffer from higher rates of alcohol dependency and antisocial personality disorder, and women have higher rates of depression, anxiety and somatic complaints [ 1 , 2 ]. Black and minority ethnic groups (BME) have a higher reported incidence of psychotic disorders [ 3 ], and are more likely to experience compulsory admission to psychiatric hospitals than whites [ 4 ] There is a higher incidence of schizophrenia in men compared with women [ 5 ] and men have poorer outcomes [ 6 ] but women are more likely to present with late onset psychotic disorders [ 7 ]. Alzheimer's Disease is more prevalent in women (reflecting the high proportions of women in the older adult population in industrialised countries)[ 8 ]. The validity of such findings is predicated on the assumption that recruitment of study participants is not overly influenced by sampling bias. However, there is evidence that some groups are under-represented in mental health research [ 9 ] whereby insufficient numbers of participants are recruited to adequately represent a particular group of patients. For example, the National Survey of American Life: a study of racial, ethnic and cultural influences on mental disorders and mental health [ 10 ] found that initial refusal to participate was higher in African Caribbean participants, and the authors cite fears and suspicions concerning questions about possible immigration status as a reason for this. We also found that in a preliminary analysis of the Aetiology and Ethnicity of Schizophrenia and other Psychoses (AESOP) dataset [ 11 ], ethnic group and gender interacted to predict consent to participate; Black Caribbean men, and Black African women were more likely to refuse to take part in mental health research than their white British counterparts (Sloan & Morgan, personal communication). There are a number of possible reasons why some groups are under-represented in mental health research. Firstly, until recently, investigators have tended to aim for homogeneity of study populations to avoid potential confounding. In addition women of childbearing age are also often routinely excluded from aetiological and intervention studies (e.g. neuro-imaging studies or drug trials) because of fears that if they are pregnant, or conceive during the study, the foetus will be put at risk [ 12 ], though there is evidence that this is changing [ 13 , 14 ]. Secondly, certain groups are less likely to access mental health services and will therefore not be available for studies that recruit through service contacts, e.g. BME patients and young men are less likely to access mental health services [ 15 ]. Also, gender and BME specific pathways to mental health care may also inhibit recruitment of these groups e.g. BME groups are more likely to have contact with mental health services via the criminal justice system [ 11 ]. The stigma associated with mental illness [ 16 ] may also affect willingness to participate in mental health research, both for participants who are mentally ill and healthy controls, particularly those from BME groups, or older potential participants[ 17 ]. Older adults may also be more physically frail, limiting their ability to attend research appointments and they are more likely to have chronic physical diseases which may mean they cannot be recruited into studies due to the exclusion criteria. There is increased recognition of the importance of generalisability for study findings and the US National Institutes of Health (NIH) published amended guidelines in 2001 on the inclusion of women and minorities in research [ 18 ]. These guidelines state 'It is the policy of NIH that women and members of minority groups and their subpopulations must be included in all NIH-funded clinical research, unless a clear and compelling rationale and justification establishes to the satisfaction of the relevant Institute/Centre Director that inclusion is inappropriate with respect to the health of the subjects or the purpose of the research'. It is therefore vital that researchers know about potential barriers to participation in these groups and strategies that can effectively be used to overcome these barriers. To our knowledge there has been no systematic review of barriers to participation in mental health research for different groups or how researchers have tried to overcome them. We therefore aimed to review the current literature on the nature of barriers to participation across different mental health studies with a focus on whether there are specific gender, age and ethnicity related barriers and to examine the evidence on the effectiveness of strategies used to overcome barriers.

Methods The research literature was searched using bibliographical electronic databases: Psyc-INFO, Medline and Embase, (1990-2008). The inclusion criteria were English language reports on barriers to recruitment in mental health research and strategies to address these on adult participants. The exclusion criteria were articles with a primary focus on eating disorders or substance abuse, non-empirical research articles and book chapters. "These databases were searched using search terms Mental disorders (Mapped term: exploded), and Recruitment$ and Research$ resulting in a retrieval of 661 articles across the three databases. Then, to narrow the search, the word barriers$, and finally, the term participation$ was added to the above search terms. All of the articles retrieved (n = 157), using the above search terms, were subject to an initial screen. This involved reviewing the title and abstract of the retrieved article for subject relevance. After the search was conducted a primary researcher (CS) was responsible for selecting articles for review, and a second researcher (AW) repeated this process to check all relevant articles had been included. Where there was disagreement a senior author (LH) examined the paper and consensus was achieved after discussion. A circulatory approach to the review process was adopted [ 19 ], whereby the author moved between searching the literature base, analysing relevant studies to identify further studies and writing up; this is done so that the review remains firmly grounded in the available literature. Conceptual and methodological literatures are not readily subject to meta-analysis. It was anticipated that this review would include both qualitative and quantitative research therefore a more open and qualitative analysis of the results was considered appropriate [ 20 ]. We excluded articles relating to substance misuse only and eating disorders only (n = 11), articles unrelated to mental health research (n = 46), non-empirical research articles or book chapters (n = 20), and articles with no focus on recruitment (n = 34). A scan of the reference list, for potential additional articles of relevance was carried out for each paper and six additional papers were selected for initial inclusion in this review (see Figure 1 ).

Results Forty-nine papers met the inclusion criteria of this review and were examined in detail (see additional file 1 ). The papers included in this review were diverse in type using a broad range of methodologies and participant populations. Papers are concerned with Dementia (n = 19), Schizophrenia (n = 7), depression (n = 15), Bipolar Disorder (n = 2), or mental illness in general (n = 8); one paper included patients with schizophrenia, depression, and bipolar disorder and therefore is included in each category. The methodologies used were qualitative (including surveys (n = 12), focus groups (n = 2), and semi-structured interviews (n = 8)), descriptions of different recruitment strategies in clinical trials (n = 7), and discussion of issues of recruitment and/or comparison of recruitment strategies within the same study (n = 26). One paper [ 21 ] compared the recruitment strategies of two RCTs on treatment models for depression. Additional files 2 , 3 , 4 , 5 , and 6 provide details of included studies for the different disorders, with information on the barriers identified in each study and the country from which papers originate. Figure 2 provides details of barriers and facilitators identified by the included studies. Identified Barriers The barriers identified by researchers were broad ranging and included fear, suspicion and/or distrust of researchers [ 22 - 30 ], concerns about confidentiality [ 31 ], transportation difficulties [ 23 , 32 - 34 ], severity of illness[ 35 - 39 ], lack of financial reward [ 28 , 40 ], an increase in age - associated illness [ 35 , 37 , 41 - 45 ], inconvenience [ 23 , 33 , 46 - 48 ], fear of relapse as a result of participation [ 31 , 49 ], and the stigma of mental illness [ 24 , 26 , 26 , 26 , 27 , 44 , 50 - 53 ]. Also discussed were barriers that are not explicitly linked to the population being studied but rather the researchers themselves, which include competing academic centres studying the same group (which potentially increases participant refusal in one project due to participation in another), tensions between academic institutions and community centres, interdisciplinary differences [ 54 ], and relying on referrals from clinicians who have misconceptions about the research design and consequently have difficulty identifying and explaining the study to prospective patients [ 55 ]. In efforts to recruit minority ethnic groups specifically, a number of studies identified the stigma of mental illness [ 24 , 26 , 26 , 27 , 51 , 53 ] and distrust of researchers [ 10 , 22 , 56 ] as significant barriers. Results from a focus group with caregivers of patients with Alzheimer's disease indicated that the primary barrier for white participants was 'inconvenience' whereas for African American families it was more a general distrust of research [ 23 ]. Interviews with Chinese American Alzheimer caregivers also suggested that the social stigma associated with the disease was a barrier to research participation [ 51 ]. Language barriers [ 34 ] have also been cited as barriers to recruitment in minority ethnic groups. Immigration status proved to be a barrier to participation when attempting to recruit Mexican and Puerto Rican patients [ 26 ] and African American and African Caribbean migrant and second and older generation populations [ 10 , 26 ]. Loue and Sajatovic (2008) [ 26 ]discussed immigration status as also being a potential inhibiter to both initially contacting and remaining in contact with some patients as they were attempting to remain undetected by other authorities. In addition, the fear of being asked about immigration status within the context of a survey served as a challenge to initial recruitment [ 10 ]. Older participants are more difficult to recruit in some studies, and this can be due to a higher likelihood of physical illness[ 41 ]. Many of the studies done specifically with an older population also found that the difficulty of accepting a diagnosis of Dementia was a common barrier [ 22 , 23 ]. Through interviews it was found that caregivers of people with dementia were also concerned that the research activities would be harmful and cause excessive worry for the patient [ 51 ]. Rather than age itself serving as a barrier to recruitment, studies in relation to this factor were largely concerned with barriers to recruiting older adults into dementia research. It is interesting to note that one study of younger adults experiencing their first episode of mental illness also highlighted the difficulty of accepting a diagnosis as well as a decreased need for treatment as barriers [ 57 ]. In terms of gender, one study found that males were harder to recruit because they were less likely to endorse a diagnosis of depression because of the associated stigma [ 44 ]. However more men than women admitted to a psychiatric hospital recorded that they saw no reason to refuse participation when asked to consent to a range of hypothetical studies [ 28 ]. Recruitment strategies to overcome barriers to participation The recruitment strategies employed and discussed by researchers can be classified into three broad categories; suggestions for recruitment based on focus groups, interviews, and surveys with patients and others; author's opinions on what techniques they thought were helpful in their own recruitment; and strategies that were actually tested and measured in terms of effectiveness in increasing recruitment. It should be noted that strategies were not developed to recruit more effectively by age, gender or ethnicity specifically; instead more general strategies were described though some had a focus on minority recruitment. Patient, caregiver, and professional suggestions on components essential for effective recruitment elicited from focus groups, interviews and surveys included: involving care givers [ 23 ]; emphasis on possible psycho-social benefits to participants including 'social and emotional support' from the research staff [ 23 , 31 , 58 , 59 ]; bilingual staff [ 51 ]; familiarity with researchers [ 31 , 60 ] and transportation assistance and incentives [ 28 , 31 , 32 , 61 ]. Some strategies are specifically relevant to certain contexts. For example, a study in the United States of America (USA) involved semi-structured interviews with patients who had and had not participated in mental health research to determine their main motivations and barriers to participation [ 31 ]. A strong motivating factor for patients with a diagnosis of schizophrenia was the offer of free medication; patients who were least happy with their current condition were more motivated to participate as they hoped that their symptoms would be alleviated. Some researchers have provided suggestions on what recruitment techniques they believe were helpful in their research. These range from using bilingual staff (where studies have participants with other first languages) [ 26 , 26 , 34 , 62 ]; targeted marketing material [ 10 , 33 , 40 , 62 , 62 , 62 ]; assistance with travel costs and incentives [ 34 , 46 , 57 , 63 , 64 ]; flexible meeting times and locations [ 26 , 34 , 57 ], and avoiding the use of mental illness terms where possible to minimise the effect of the associated stigma [ 26 , 40 , 65 ]. Several studies describe different forms of outreach work designed to engage and consequently recruit ethnic minority participants. These range from hiring a specific 'outreach worker' (a person living within the community) [ 24 ] to assist researchers in meeting potential participants and advise study researchers on how to appropriately communicate, to (in certain contexts) ongoing consultation with community leaders [ 66 ]. Targeted marketing in the local communities was also employed [ 44 , 50 ]. While our review did not focus on retention of study participants, some authors commented on high retention rates which they attributed to certain strategies including the collection of alternative contacts for highly mobile subjects, birthday cards, and gifts for subjects [ 26 , 34 ]. One study used a group session to inform potential participants about a study [ 67 ]. The 45-minute presentation to potential participants described the nature of the RCT of supported employment. To be eligible for recruitment, potential participants had to attend four of these sessions; this was to ensure informed and committed people were recruited. The project met its recruitment target and had a good retention rate. However none of these strategies were formally evaluated. A small number of studies have evaluated the efficacy of researcher/participant ethnic matching [ 10 , 21 , 68 ] and found little effect on recruitment rates. For example, Thompson et al (1996) found that ethnic matching of researchers to potential participants did not influence rates of refusal of interview completion for African American psychiatric inpatients. Recruitment data from a randomised trial to evaluate the effectiveness of a social model of care for patients with depression, anxiety or heavy drinking showed that ethnically matched recruiters were no more effective than experienced researchers with previous experience of conducting work with minorities and community member recruiters, with recruitment rates of 64%, 70%, and 77% respectively [ 21 ]. Other methods to increase minority recruitment that researchers have tried to evaluate included the distribution of educational pamphlets, a study newsletter and compensation for transportation costs in a study on dementia in black patients. When these strategies were introduced participation rates in a registry for Alzheimer's disease increased from 60 black patients (out of 607 potential participants) in year one to 150 in its second year [ 32 ]. Fritsch et al (2006) [ 69 ] increased minority recruitment rates (17% to 36%) over a 6 month period by exposing potential participants to a piece of educational theatre on Alzheimer's disease.

Discussion The main findings in terms of barriers identified ranged from practical factors (including transport, lack of financial reward, inconvenience, language difficulties, and immigration status), to more complex internal barriers (ranging from distrust of research and concerns about confidentiality, stigma of mental health, reluctance to accept a psychiatric diagnosis and fear of relapse). Strategies to overcome barriers to recruitment ranged from transport assistance and monetary incentives, bilingual staff, and inclusion of caregivers. Such practical solutions such as flexible meeting times, locations, and monetary incentives are relatively easy to implement in order to address barriers pertaining to transport, inconvenience, and financial difficulties. However other more complex barriers associated with the health, beliefs, and the culture of patients and/or carers were identified, which may be more difficult to overcome. We will therefore discuss these barriers in more detail. Fear and Concerns about Confidentiality Through interviews with patients who had not consented to participate in a study of schizophrenia, Kaminsky et al (2003) [ 31 ] found that refusal was based on a fear of not knowing what was involved in the research, concerns about confidentiality of information and concerns that their personal information may be misused. Therefore clear, comprehensive explanations of the procedures involved in the study by researchers could serve to lessen any initial distrust. The recruitment targets were met in a study that utilised a research induction group to facilitate recruitment [ 67 ]. Within this forum there is the potential to provide sufficient information and respond to any concerns prospective participants have. The mistrust and scepticism of mental health research found in ethnic communities [ 70 ] also inhibits prospective patients from volunteering to participate in research projects. This is in line with findings that ethnic minorities have a greater distrust of medical research in general [ 22 , 23 , 70 ]. Given that it is still discussed in the research and lay literature[ 71 ], this is perhaps a legacy from medical research projects such as the Tuskegee study, in which black men were not offered efficacious treatments for syphilis [ 72 ]. This scepticism of mental health research includes concerns about confidentiality of information shared [ 31 ] and perhaps the more deeply entrenched feelings of 'mental illness rooted in white oppression' [ 27 ]. The majority of studies were conducted in the USA and there did not appear to be significant differences in perceived barriers and facilitators to participation across countries. However we noted that three of the five studies based in England discussed the reluctance of clinicians to refer participants (due to a lack of confidence, skill, or misconception about research) as a significant barrier[ 73 - 75 ], in a way that was not explicitly discussed in studies conducted in the USA. This may indicate that research is more embedded in clinical practice there, or that less attention is paid to clinicians as potential barriers. Stigma of Mental Illness The stigma associated with mental illness has been widely researched and seen as a factor effecting service engagement [ 16 , 76 ]. Stigma has also been cited as a factor affecting lower service use among ethnic minority groups [ 77 - 79 ]. The stigma attached to mental illness experienced by ethnic minorities such as African American [ 23 ] and Chinese American [ 51 ] communities were perceived to be a strong barrier to participation in mental health research. As ethnic minorities are less likely to access mental health services, reliance on clinical referrals and recruitment through services limits the potential pool of ethnic minority participants. Stigma and general distrust can also stem from a lack of understanding about the illness; some researchers who had difficulty recruiting Latino caregivers of persons with dementia cited a general lack of awareness and knowledge of dementia in the community as a significant barrier [ 62 ]. Other researchers deliberately avoided using terms associated with 'mental illness' and other 'stigmatizing diagnostic classifications' in their marketing and recruitment material, though ethically such techniques may be problematic if they are not transparent about the study aims. A lack of knowledge about older age disorders has been highlighted as a potential barrier to participation in research into dementia [ 69 ]. Fritsch et al (2006) [ 69 ] significantly increased recruitment rate of African Americans into Alzheimer's disease research through exposure to live educational theatre on the topic. This type of community education can serve to lessen the stigma attached to a diagnosis of Alzheimer's disease. Acceptance of Diagnosis Patients and family members' acceptance of a diagnosis is an important factor in gaining consent for participation. Studies of older adults with dementia or Alzheimer's Disease [ 22 , 23 ] found a reluctance to accept the diagnosis was a significant barrier. Not only did it prove to be a barrier at a patient level but clinicians themselves also resisted formally diagnosing Chinese American patients in an effort to reduce the perceived stigma of the diagnosis for that population [ 51 ]. Clinicians in this study reasoned that there was no need to give a potentially stigmatizing diagnosis if the elder was well cared for and not exhibiting any dangerous or violent behaviour. This consequently has implications for recruitment of these types of patients. A lack of insight or understanding, or an emotional need to deny a diagnosis, was found not only in elderly populations and their families but also with adolescents experiencing their first episode of psychosis or depression [ 57 ]. In one study males were more difficult to recruit due to their non-acceptance of their diagnosis of depression [ 44 ]. This is consistent with research that suggests males experience depression privately, unshared with others and attempt to alleviate it with little external help [ 80 ]. Facilitators to Research Participation Incentives to research participation such as the offer of free medication are relevant in countries where patients have difficulty in accessing and paying for health care. Even in countries with free healthcare people may see research as another avenue for help, as found by a United Kingdom (UK) based online survey on the views of patients regarding research participation (see http://www.healthtalkonline.org ) [ 81 ]. This study also found that, of those people who were approached about mental health research, those who believed that their mental health may alter or be at risk of deteriorating as a result of participation were far less likely to participate in experimental studies and drug trials. Results from studies that employed racial matching did not clearly indicate an improved consent rate from ethnic minority participants. Experience or 'cultural competence' appears to be more effective. However language specific marketing and use of researchers fluent in the minority groups' language is inevitably an effective component of recruitment [ 26 , 34 , 50 , 51 , 62 ]. Feedback from focus groups about research with older adults with dementia [ 23 , 50 ] highlight how important it is that family members and carers are actively involved in recruitment and research procedures, as the decision of consent will in large part be influenced by them. It is particularly important to include caregivers in the explanation and consent process as some caregivers have expressed worry that research participation could be potentially harmful to a patient [ 51 ]. A major motivating factor in participation in research appears to be the perception that the research may help others [ 31 , 50 , 61 , 82 ]. It follows that receiving research feedback from a study can reinforce this and show participants how they are contributing to the development of knowledge in that particular area [ 83 ]. Most studies agree that this is good practice and UK research ethics committees now routinely ask if this is planned. It is less clear how much time and resources should be allocated to providing detailed feedback. This is a longer-term strategy which aims to engage with the communities and educate them in what the research and their previous participation achieved; in the hope that this will lead to higher participation rates in future studies. This wider form of community awareness and engagement has been advocated by many authors [ 26 , 50 , 51 , 69 ] and is seen as crucial to recruitment. Participants also advocate this. In a study on schizophrenia, participants expressed a strong preference for being thoroughly debriefed about the purpose of each task at the conclusion of the study [ 61 ]. Outlining to a participant that they will receive some form of immediate feedback in the form of a debriefing may therefore increase the likelihood of consent. Implications Research findings will not be generalisable if particular groups of patients are under-represented; research on the effectiveness of medication for certain ethnic groups for example has been limited [ 84 ] making it difficult to identify whether dosage should be altered or whether different drugs should be used. Study design ideally needs to reflect the population under investigation and where recruitment of large subgroups is not possible for practical reasons this should be addressed in the analysis e.g. by probability weighting in cost-effectiveness analyses. Studies could also use more comprehensive datasets (e.g. administrative data) which could be helpful in triangulating research findings. Grant bodies could also try to ensure that planned research addresses the evaluation of recruitment strategies to ensure relevant groups are adequately represented. Limitations of Review This literature search did not include hand-searching of relevant journals and formal rating of methodological quality was beyond the scope of this review - the diversity of the papers considered made this impractical. We found few studies that actually tested the effectiveness of a strategy making it difficult to attribute successful recruitment to a particular method. The lack of control groups in the majority of studies and comparisons of rates of recruitment could be affected by confounding factors. The literature to date illustrates how the majority of recruitment methods have not been formally evaluated and there is therefore a real gap in our understanding of barriers to participation. Much of the research in this area is at an exploratory stage only. Future Research Directions Many researchers described barriers as discussion points or as a set of limitations after the results rather than something specifically considered in the study design. Where researchers did try to address potential barriers in their recruitment strategies, they did not formally evaluate these in any way. In view of the very limited evidence base on recruitment strategies to overcome barriers to participation in mental health research, we would recommend that future feasibility and pilot studies should include systematic evaluation of different recruitment strategies before starting a major study. Such development work is recommended by the UK Medical Research Council [ 85 ] and has been used successfully in other medical research settings [ 86 ]. We would also recommend that researchers clearly describe whether their sample is representative of the population of interest as recommended by the extended CONSORT guidelines[ 87 ]. Further research on age related barriers would be beneficial as little information was found on this. Other factors that may also be important barriers to participation such as level of education and socio-economic deprivation could also be explored in future research.

Conclusions There is little evidence on which recruitment strategies are effective for increasing rates of participation but studies did identify clear barriers which could be addressed by future researchers. For example, addressing difficulties with transportation is a clear and practical way to facilitate recruitment. Stigma, fear, and distrust were consistently found to be barriers across studies and consequently attempts to address these would also presumably increase recruitment. Transparency of the research project and a clear explanation of what is expected of participants may go a long way in dispelling fear and distrust. In addition, it is also important and worthwhile to be inclusive of caregivers and family members as in many cases they will be important contributors to decision making in participants' lives. This may also serve to ease some anxieties that prospective participants and their families may have about the impact of the research on the participants' health, and potential benefits. These could include allocation of different strategies in a randomised controlled trial would be beneficial. For example, a comparison of different marketing strategies and recruitment materials (e.g. information sheets) could be done whereby one set of materials uses less mental illness terminology that is potentially stigmatising, to determine if this increases recruitment rates. We would also recommend that researchers clearly describe whether their sample is representative of the population of interest. It is important for researchers to be aware and to try to recruit under-represented groups in future studies to ensure the validity of reported findings are not potentially undermined by sampling bias.

Background It is well established that the incidence, prevalence and presentation of mental disorders differ by gender, ethnicity and age, and there is evidence that there is also differential representation in mental health research by these characteristics. The aim of this paper is to a) review the current literature on the nature of barriers to participation in mental health research, with particular reference to gender, age and ethnicity; b) review the evidence on the effectiveness of strategies used to overcome these barriers. Method Studies published up to December 2008 were identified using MEDLINE, PsycINFO and EMBASE using relevant mesh headings and keywords. Results Forty-nine papers were identified. There was evidence of a wide range of barriers including transportation difficulties, distrust and suspicion of researchers, and the stigma attached to mental illness. Strategies to overcome these barriers included the use of bilingual staff, assistance with travel, avoiding the use of stigmatising language in marketing material and a focus on education about the disorder under investigation. There were very few evaluations of such strategies, but there was evidence that ethnically matching recruiters to potential participants did not improve recruitment rates. Educational strategies were helpful and increased recruitment. Conclusion Mental health researchers should consider including caregivers in recruitment procedures where possible, provide clear descriptions of study aims and describe the representativeness of their sample when reporting study results. Studies that systematically investigate strategies to overcome barriers to recruitment are needed.

Competing interests The authors declare that they have no competing interests. Authors' contributions CS and AW conducted the literature search and retrieved relevant articles. AW analysed and interpreted the content of the papers. LH and CM made substantial contributions to the interpretation of the data and revised the manuscript critically for important intellectual content. All authors helped draft the document, read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-244X/10/103/prepub Supplementary Material

Acknowledgements This study was funded by the National Institute for Health Research, 'Biomedical Research Centre for Mental Health', Institute of Psychiatry and South London and Maudsley NHS Foundation Trust. The funding body had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; or in the decision to submit the manuscript for publication. The authors declare that they have no competing interests.

CC BY

no

2022-01-12 15:21:45

BMC Psychiatry. 2010 Dec 2; 10:103

oa_package/b1/59/PMC3016310.tar.gz

PMC3016311

21129214

Background A novel influenza of swine origin was first detected in Mexico during March and early April 2009 as increasing incidence of atypical respiratory disease in localised areas in Mexico was reported. Details of the epidemiology, spread, and risk factors for infection and death have been reported for early spread of the disease in Mexico [ 1 , 2 ]. Although initially thought to be the result of an extended seasonal influenza outbreak, the high level of hospitalisation and severe cases of pneumonia in young and otherwise healthy adults was unusual. In Oaxaca on 15 April 2009 health officials were notified of a suspected case of atypical pneumonia; the patient died within a few days. Investigation of this case identified a novel agent, later identified as a non-typeable strain of influenza A. On 23 April the Public Health Agency of Canada and the Communicable Diseases Center (CDC) in Atlanta confirmed that a common novel influenza A virus had been detected in two Mexican samples; the one from Oaxaca, and another from La Gloria, Veracruz, and was similar to a strain isolated from patients in California [ 3 ]. A week later on 29 April the World Health Organisation (WHO) announced a global pandemic alert level Phase 5 [ 4 ], indicating sustained human-to-human transmission in one WHO region of the world, and this was later raised to a global Phase 6 pandemic on 11 June 2011 [ 5 ], which was the pandemic alert level at the time of our study. With uncertainty regarding the virulence and transmissibility of the pandemic in the early stages, and immense media scrutiny and reporting, there was widespread public fear [ 6 ]; and media reports of panic, especially in Mexico [ 7 ]. Even the most trusted source of global health information; WHO, was being reported in the media as warning that "all of humanity is under threat" [ 8 ]. High levels of fear and concern persisted in Mexico due to concerns about the severity of the illness, uncertainty surrounding its mortality rate, the susceptibility of younger and healthy people, and potential for contagion and stigma. Pandemic context at the time of this study Our study was conducted from 1 to 30 November 2009. On 21 September 2009 the Government of Mexico announced that the country was at an "intermediate warning" level for influenza caused by A/H1N1 indicating that people should strengthen measures to promote health [ 9 ]. During the period from the start of the outbreak until 19 September, 220 people had died and 26,865 had been infected throughout the country, and 3,486 people had died and 296,000 had been infected, globally. The Health Secretary, José Ángel Córdova, reported that infection levels had accelerated in the States of Nuevo León, Baja California, Sinaloa in Mexico City, Tlaxcala and Oaxaca. At the time of the study, Mexico had experienced three peaks in infection rates; the first from 23 to 30 April, the second between 26 June and 24 July, and the third in mid September. Background research Research in the area of psychological response of family members of patients has generally focused on the psychological assessment of family members or informal caregivers of patients admitted to ICUs [ 10 ], and has included assessments of post traumatic stress symptoms [ 11 ], psychological impacts of being involved in making end-of-life decisions and interventions to support families [ 12 ], and assessment of psychological or physical health of caregivers of patients on prolonged mechanical ventilation and the chronically critically ill [ 13 , 14 ]. Another source of research literature on the psychological response of caregivers has focused on the longer term mental and physical health burden on caregivers providing care for patients with long term conditions, such as HIV/AIDS, cancer, or dementia [ 15 ]. Therefore studies of caregiver psychological response are highly varied, both in terms of psychosocial impacts specific to different types of conditions (e.g. acute trauma and possible situations surrounding that, or fatal illness), their temporal features and outcomes with regard to care-giving (long-term care and eventual death, potential for recovery), and the psychological assessments used. In addition, there are differences in the time frames in which psychological symptoms are assessed in such studies; typically ranging from hospitalisation to 6-9 months post-discharge for ICU-related studies. However, one common aspect of most ICU studies of this nature is that they usually assess one or more of the following; stress, depression, and/or anxiety. A recent review of symptoms experienced by family members of patients in ICUs [ 10 ] identified common risk factors for stress, anxiety and depression from 18 core studies. In terms of demographic risk factors, being female was a risk factor for most types of stress (including acute stress disorder and post-traumatic stress disorder (PTSD)) and depression, and being a spouse was a common risk factor for depression and anxiety. Azoulay et al [ 11 ], in conclusion of their study of PTSD in family members of ICU patients, commented on the high levels of PTSD and the need for preventative and early intervention strategies. They suggest that high rates of anxiety and depression in family members may increase the risk of PTSD reaction and call for a need to identify factors detectable at the time of the ICU stay and associated with increased vulnerability in family members. In our study, our focus was to evaluate levels of perceived stress, depression and death anxiety in the primary family caregiver of patients hospitalised and admitted to the ICU with suspected A/H1N1. Our aim was to empirically document the nature of the psychological impact of this epidemic in Oaxaca, to screen the primary family caregiver for adverse psychological symptoms, and to analyse data to identify risk factors for these adverse reactions. In addition, if evidence of adverse psychological response was found, we sought to develop appropriate resources to assist this population to cope with and reduce such responses, and in doing so, possibly lower levels of acute stress and likelihood of development of PTSD. In this article we will report an overview of the levels of psychological response reported during the screening of these family caregivers; identify risk factors that are associated with an elevated adverse response; and reference the extent of this response by comparing our findings to comparable data collected by the research team at the same hospital, and from comparable ICU studies reported in the literature, as well as established normative population and community data for our psychological assessment tools. We will then provide a brief overview of the early psychological support intervention offered to family caregivers.

Methods Participant selection The research team assessed the psychological response of the family primary caregiver of all patients admitted to hospital, by ambulance, with respiratory distress and hospitalised in the ICU with suspected influenza A/H1N1 in the General Hospital of Zone 1 (HGZ 1) of the Mexican Institute for Social Security (IMSS) in Oaxaca, during 1-30 November 2009. Due to the infectious nature of the medical condition determining ICU admission and the need for stringent infection control only one relative of the patient is authorised to have contact with the patient in ICU. This primary caregiver is allowed access to communicate with the patient and to attend to their personal care. The authorised caregiver was the one approached to take part in the study. Participation in the study was voluntary and anonymous. The only exclusion criterion for participation was a prior psychiatric diagnosis. The study was approved by the Research and Ethics Committee of the HGZ 1 of the IMSS and all participants provided written informed consent. Materials The psychological response of the family primary caregiver was assessed using three established assessment tools: - Perceived Stress Scale (PSS-10) [ 16 ]. The PSS-10 is a 10-item self report scale used to measure global perceived stress, it has been found useful as a predictor of physical symptoms and health outcomes. The scale assesses the respondent's appraisal of his/her life as unpredictable, uncontrollable, and overloaded during the preceding month. Scores range from 0 to 40 with higher scores indicating a higher risk factor for future distress. - Center for Epidemiologic Studies Depression Scale (CES-D) [ 17 ]. The CES-D is a 20-item self-report scale developed for the general population to measure the frequency of depressive symptoms during the past week. It is not a clinical diagnostic tool, but has been used widely as a useful screening tool. It has excellent reliability (α coefficients, 0.85-0.91) and validity. Responses are rated on a four-point scale to yield total scores in the range 0 to 60. Higher scores indicate a greater risk of depression, with scores ≥16 indicating an increased risk of clinical depression and, possibly, mortality [ 18 ]. - Death Anxiety Questionnaire (DAQ) [ 19 ]. The DAQ is a 15-item self-report scale that measures attitudes towards one's own death and dying, including fear of the unknown, fear of suffering, fear of loneliness and fear of personal extinction. Death anxiety can be interpreted as an additional form of general anxiety or distress in the context of our study. Death anxiety has been linked to self-esteem and well-being, personality, valuing life, cultural values and differences and religiosity [ 20 ]. These scales were chosen for a range of reasons; the CES-D had been used in other clinical studies in ICUs to assess responses of caregivers and others [ 10 ], there were established normative data from populations and community based samples for all scales [ 17 , 19 , 21 ], and these scales had been used successfully in studies of the psychological impacts of a range of other medical conditions and situations previously conducted by the research team, allowing for direct comparisons to be made to these data [ 22 ]. Procedure A single interviewer collected data from all participants, in the period shortly after the patient was diagnosed and hospitalised in the hospital ICU. Participants were presented with each question and set of response options by the interviewer, and the interviewer noted each response and subsequently scored the data for each participant. Demographic data were also collected; age, gender, family relationship to patient and education level. Statistical analysis Exploratory data analysis was conducted using frequency distribution for categorical variables and graphs and summary statistics for continuous variables. Continuous data for the psychological response scale variables were examined using regression analysis and checked for homogeneity of variance. Skewed distributions were natural logarithm transformed before simple and multiple linear regression analysis was performed. Statistical analyses were undertaken using the statistical package STATA, version 10 (2008; Stata Corporation, College Station, TX, USA). Statistical significance was taken as p ≤ 0.05.

Results Characteristics of the sample During the study period 36 patients were hospitalised and admitted to the ICU, and all were subsequently confirmed as having A/H1N1. Only one family member in the role of primary caregiver refused to participate in the study, and no family members in the primary caregiver role had a prior psychiatric diagnosis. Therefore the final study sample size was 35. Table 1 summarises the characteristics of the sample. Three quarter of the study participants were female (74.3%), 43% were in a spousal relationship with the admitted patient, and around a third (34.3%) had a university-level education. The mean age of participants was 32 (range 20-55). The mean scores for perceived stress, depression, and death anxiety were 16.7, 16.4 and 15.1, respectively. These data were categorised, using established cut-off scores, and are shown in Figure 1 for the three assessment scales. From Figure 1 it can be seen that the majority of participants reported no stress or depression (60% and 57%, respectively) and around a third of participants' responses were categorised as 'low' for stress and depression (37% and 34%, respectively). High levels of stress and depression were noted for a small proportion of participants (3% for both measures). Although the term 'low' has been used for categorisation of depression it should be noted that this represents the cut-off score of 16, above which individuals are regarded as being at higher risk of clinical depression i.e. 43% of the sample had a score above this cut-off. High levels of death anxiety were reported by 17% of participants, with the majority reporting moderate levels of death anxiety. Regression Analyses Univariate analysis, conducted using the continuous psychological response data, identified that the following were significantly associated with perceived stress (coefficient, R 2 and p-value): female gender (16.4, 0.23, 0.003), non-spousal family relationship (0.48, 0.83, < 0.001), and increasing age in years (17.6, 0.59, < 0.001). Simple regression analysis also indicated that the following were significantly associated with depression (coefficient, R 2 , p-value): female gender (2.5, 0.22, 0.004), non-spousal family relationship (2.60, 0.55, < 0.001), increasing age in years (0.08, 0.85, < 0.001), and university-level education (2.65, 0.34, < 0.001); and for death anxiety: female gender (16.1, 0.26, < 0.001), non-spousal family relationship (15.9, 0.57, < 0.001), increasing age in years (0.44, 0.83, < 0.001), and university-level education (15.8, 0.34, < 0.001). These results are summarised in Table 2 . Multivariate analysis, summarised in Table 3 , indicated that the following were significantly associated with perceived stress (coefficient; 95% CI, p-value): increasing age in years (0.34; 0.24-0.44, < 0.001) and non-spousal relationship (6.79; 2.81-10.77, 0.002); depression (coefficient; 95% CI, p-value): increasing age in years (0.05; 0.04-0.07, < 0.001), non-spousal relationship (0.80; 0.20-1.39, 0.010) and female gender (4.85; 0.47-9.22, 0.031); and death anxiety: increasing age in years (0.32, 0.23-0.41, < 0.001), and university-level education (5.99, 2.44-9.54, 0.002). Comparison data The research team has used the same methodology and assessment measures in small studies, also at the Oaxaca General Hospital, evaluating the psychological responses of relatives and patients to three other medical conditions or situations, those being: relatives of patients admitted to the Intensive Care Unit (ICU) (Vargas-Mendoza & Aguilar, unpublished data), patients who encountered foetal death (Vargas-Mendoza & Pacheco-Chávez, unpublished data), and patients undergoing haemodialysis in ambulatory care [ 22 ]. A further aim of the current study was to compare the psychological response to A/H1N1 with data gathered in these other studies. Categorical data from these studies have been summarised, alongside findings from the current study, in Table 4 . Chi square statistical tests (Fishers exact), have been used to test for statistically significance differences. Comparisons with similar studies conducted at the Oaxaca General Hospital indicated that there was a statistical difference between the levels of perceived stress of family members of patients admitted to the ICU for A/H1N1 and for other reasons. Comparing the pattern of response it appears that the perceived stress levels in relation to A/H1N1 were lower. There did not appear to be statistical differences between the current A/H1N1 study data and equivalent data collected for other medical conditions in relation to depression or death anxiety.

Discussion This small study has provided evidence of a moderate psychological response in the family members of patients hospitalised in the ICU for A/H1N1. In the context of a novel influenza pandemic we have not found evidence that the level of response has been as excessive or alarming, as might have been predicted from reports in the media, and we found no evidence of panic or an 'epidemic of panic'. We note that the majority of family members reported sub-threshold levels of stress and depression (60% and 57%, respectively); however 43% of participants reported levels of depression above the established cut-off score for higher risk of clinical depression. This was higher than levels recorded for caregivers of patients who had been mechanically ventilated for > 48 hours in ICU (30% at 2 months) [ 13 ] but much lower than caregivers of chronically critically ill patients when in ICU (75% at ICU enrolment) and similar to levels at 2 months post discharge (43%) [ 14 ]. In their review of ICU studies, McAdam and Puntillo [ 10 ] conclude that depression affected 15%-35% of patient family members, however, they too are comparing studies with inconsistent time frames used to examine symptoms, different medical conditions, and assessment instruments. With regard to reported levels of stress, the PSS-10 has not been used by researchers evaluating stress in family members of patients in ICUs or generally, and therefore, this comparison is not available. In our study the mean stress score for the sample was equivalent to that recorded in a large heterogeneous European Spanish "stressed" sample of people coping with a range of adversities [ 23 ], suggesting that our sample could also be regarded as 'stressed'. Reported normative data for PSS-10 for normal healthy adults range from mean scores of 14.2 (SD 6.2) for those aged 18-29 years, up to 11.9 (SD 6.9) for those aged 55-64 years [ 21 ]. Our study sample mean of 16.7 (SD 7.9) would appear to be elevated compared to healthy norms. Levels of death anxiety were generally much higher (often double) mean scores reported for heterogeneous groups [ 19 ], with just under three quarters of family members (71%) reporting moderate levels of death anxiety and 15% reporting high death anxiety. Given that all patients in this study had been admitted to the hospital ICU via ambulance in a state of respiratory distress it is possibly understandable that this would stimulate thoughts of potential death of the patient and bring feelings of one's own mortality into consciousness. Simple univariate statistical analysis has shown associations between heightened psychological response in family members who are female, older, with higher levels of education and who are in a non-spousal relationship with the admitted patient. However, when subjected to multivariate analysis the most consistent association, across all measures, was an increased psychological response with increasing age. The reason for this finding is unclear. In general, caregiver age has not been reported as a significant risk factor in studies of stress and depression in caregivers in acute clinical settings [ 10 ]. It is likely that older participants are more likely to be older than the patient (the maximum age in our sample was 55), so one possible explanation is that the emotional response to the potential loss of someone younger may be more acute. The finding that increasing age is associated with increasing death anxiety is also an interesting finding. The generally accepted relationship between age and death anxiety is that death anxiety decreases across life span, although there is evidence that this decrease occurs from midlife, i.e. beyond the age of the majority of our study sample [ 24 ], and the impact of sudden mortality salience on death anxiety does not appear to be studied. In addition to the effect of increasing age in the multivariate analysis, higher stress response and depression was noted for those in non-spousal relationships with the patient, i.e. in our sample these were mothers, daughters, brothers and sisters. This finding differs from other studies of family members of patients in ICUs where family relationship has been found to be a risk factor for adverse psychological response; in those studies spousal relationship has been found to be associated with higher depression [ 10 ]. It is interesting to note that family relationship was found not to be associated with death anxiety in the multivariate analysis. Here, higher levels of education and being female were the factors most strongly associated with higher reported death anxiety. Higher death anxiety is generally noted in females [ 20 ]. Comparing our current study data with similar prior studies conducted at the Oaxaca General Hospital it was interesting to note that, with a degree of confidence, the levels of stress reported in family members in response to patients admitted to the ICU with suspected A/H1N1 was lower than equivalent data reported by family members of patients admitted to the ICU for other medical conditions. Although one needs to be cautious when interpreting data based on small samples, this finding does add support to a lack of evidence of an extreme psychological response or 'panic' in association with pandemic A/H1N1, in the country most severely impacted. Limitations and strengths This study has a number of limitations that need consideration. Firstly, it is based on a small sample of primary family caregivers and therefore the findings can only be regarded as indicative. Also the psychological assessments undertaken provide a single snapshot of how family caregivers were feeling close to the time of admission of the patient to the ICU, and do not therefore provide an indication of longer term psychological trajectories. There was also no opportunity to control for extraneous factors, that may have influenced caregivers' psychological condition, e.g. other life events, physical health status, and therefore it is not possible to identify psychological response attributable to the patient's condition and to pandemic A/H1N1 per se. Despite these limitations, participants did not have a history of psychiatric illness, and with regard to the main aim of the study; which was to identify if there was evidence for an adverse psychological response in family caregivers of patients admitted to ICU for A/H1N1 to support provision of psychological support, the evaluation that was undertaken adequately suited this purpose. Clinical Outcome In reviewing data from our study we believed that there was evidence of moderate psychological response and that this confirmed the need for a level of psychological support to the families of patients hospitalised for A/H1N1. Therefore, in response we developed a psychological support strategy based on four principles, as follows: 1. Provide supportive information . The threat of pandemic influenza for our patients and their families can be a stressful event. It is important that they receive timely and adequate information concerning how to take care of, and protect, their loved ones. Such information increases a sense of control and self-efficacy and enables them to respond and support their loved one and other family members at this difficult time. 2. Acknowledge their psychological response . In addition, we need to let family members know that if someone close to them is sick it is normal to have a range of feelings such as feeling concerned by newscasts and media reports; feeling anxious, irritable or impatient; or losing the ability to concentrate on tasks. 3. Confront stress . Advise families to continue with normal life, to take time to eat, exercise, and rest, and to keep busy and focus on daily activities. Avoid drugs and alcohol. Stay in touch with friends and family and pay attention to television and radio reports that provide information on how to stay healthy and safe. Encourage them to talk to someone about their feelings if they are fearful or concerned. 4. Consider the response of children: To help children we advise that family members express what they feel and explain that people may feel concerned and that it is normal when they have stress. Give them information they can understand. Tell them that you will protect them so that they feel reassured. Encourage them to make drawings and paintings. These projects help to express what they feel. Touch and embrace them frequently. Keep to your routines with laughter and games. Teach them protective behaviours to protect them of infectious diseases; such as washing hands.

Conclusion This study sought to evaluate the psychological response of family primary caregivers of patients hospitalised in the ICU for suspected influenza A/H1N1 to establish whether there was evidence of an adverse psychological response, to identify risk factors for such a response, and to assess if the level of response was sufficient to support development of a specific package of psychological support for these individuals. Our data provided evidence of elevated perceived stress, depression, and death anxiety, particularly in caregivers who were older, or female, or in non-spousal relationships with the patient, and were in excess of levels that would have been predicted from normative population data and were generally comparable, or slightly lower, that levels reported elsewhere in ICU caregiver studies. Consequently we have developed a simple low level psychological support intervention as a form of psychological first aid to reduce acute stress and other adverse psychological reactions in these caregivers, and hopefully to reduce the likelihood of the development of PTSD.

Background The A/H1N1 pandemic originated in Mexico in April 2009, amid high uncertainty, social and economic disruption, and media reports of panic. The aim of this research project was to evaluate the psychological response of family primary caregivers of patients hospitalised in the Intensive Care Unit (ICU) with suspected influenza A/H1N1 to establish whether there was empirical evidence of high adverse psychological response, and to identify risk factors for such a response. If such evidence was found, a secondary aim was to develop a specific early intervention of psychological support for these individuals, to reduce distress and possibly lessen the likelihood of post-traumatic stress disorder (PTSD) in the longer term. Methods Psychological assessment questionnaires were administered to the family primary caregivers of patients hospitalised in the ICU in the General Hospital of Zone 1 of the Mexican Institute for Social Security (IMSS), Oaxaca, Mexico with suspected influenza A/H1N1, during the month of November 2009. The main outcome measures were ratings of reported perceived stress (PSS-10), depression (CES-D), and death anxiety (DAQ). Data were subjected to simple and multiple linear regression analysis to identify risk factors for adverse psychological response. Results Elevated levels of perceived stress and depression, compared to population normative data, and moderate levels of death anxiety were noted. Levels of depression were similar to those found in comparable studies of family members of ICU patients admitted for other conditions. Multiple regression analysis indicated that increasing age and non-spousal family relationship were significantly associated with depression and perceived stress. Female gender, increasing age, and higher levels of education were significantly associated with high death anxiety. Comparisons with data collected in previous studies in the same hospital ICU with groups affected by a range of other medical conditions indicated that the psychological response reported in this study was generally lower. Conclusions Data indicated that, contrary to widely publicised reports of 'panic' surrounding A/H1N1, that some of those most directly affected did not report excessive psychological responses; however, we concluded that there was sufficient evidence to support provision of limited psychological support to family caregivers.

Key Messages ▪ When screened shortly after patient admission to ICU, family caregivers of patients with suspected A/H1N1 reported moderately elevated levels of stress and depression and high levels of death anxiety. ▪ Comparisons with published ICU studies and additional data from the same hospital suggested that caregivers of ICU patients with suspected A/H1N1 did not report higher levels of adverse psychological response than caregivers of patients admitted for other medical reasons. ▪ Older caregivers and those in non-spousal relationships with the patient were at higher risk of elevated stress and depression. ▪ Data supported the need for some low level psychological support for caregivers of A/H1N1 patients in the ICU. ▪ Even though this sample was highly A/H1N1 pandemic-affected, there was no evidence to support the media image of a panic-stricken public. List of Abbreviations A/H1N1: Influenza A, variant H1N1 the pandemic strain of influenza; CES-D: Center for Epidemiologic Studies Depression Scale; DAQ: Death anxiety questionnaire; HGZ1: General hospital Zone 1; ICU: Intensive Care Unit; IMSS: Mexican Institute for Social Security; PSS-10: Perceived Stress Scale (10 item); PTSD: post-traumatic stress disorder. Competing interests The authors declare that they have no competing interests. Authors' contributions JE-R and JEV-M conceived the study. All authors were involved in study development under the leadership of JE-R. JEV-M supervised the data collection and psychological assessment, MM-G, CM-Z helped with caring for the families and patients and supervised AE-C who conducted the interviews and initial data analysis, JE-R and JEV-M drafted the first manuscript and translated it into English, MT developed the draft and final version of the manuscript, assisted with analysis and interpretation of the data, and is the corresponding author, KA conducted the statistical analysis, and contributed to the data interpretation and draft manuscript. All authors reviewed the final version of the manuscript. Authors' Information JE-R is Coordinator of Health Research in the IMSS and is Professor Investigator in the Faculty of Medicine at the Benito Juarez University of Oaxaca. His area of research is mental health. JEV-M is a Clinical psychologist and Chief of Psychology Services at IMSS, Honorary President of Oaxaqueña Association of Psychology, and is interested in the mental health implications of medical conditions. MM-G and CM-Z are internists and doctors of internal medicine services and AE-C is an MD with interest in health research; all are at IMSS and are in the Faculty of Medicine at the Benito Juarez University of Oaxaca. MT and KA are researchers in the Disaster Response and Resilience Research Group of the School of Medicine at the University of Western Sydney; they are working on population threat perception to pandemic and the psychosocial impacts of emergency disease outbreaks in humans and animals. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-244X/10/104/prepub

Acknowledgements The authors would like to acknowledge Dr. Luciano Galicia Hernandez and Dr. Gerardo Soria Cuevas, Directors of IMSS, Oaxaca for supporting the development of this research.

CC BY

no

2022-01-12 15:21:45

BMC Psychiatry. 2010 Dec 3; 10:104

oa_package/1d/81/PMC3016311.tar.gz

PMC3016312

21134296

Background Schizophrenia is a serious mental disorder characterized by a number of symptoms. To evaluate the effects of treatment for schizophrenia, it is important to assign quantitative values to the symptoms. Many rating scales have been used to evaluate various symptomatic domains in schizophrenia [ 1 ]. This has led to confusion regarding the suitability of the different scales available, not only in relation to evaluation and treatment of the disease but also in research and clinical studies of the effects of medication. Currently, consensus is lacking about which rating scales are appropriate to evaluate schizophrenia. Evaluation scales that are relevant, quick, user-friendly, graduated at equal intervals and with high linearity are needed to facilitate measurement-based treatment of schizophrenia. The Brief Psychiatric Rating Scale [ 2 ] is one of the standard instruments used most frequently in daily practice for evaluating the severity of schizophrenia. Also popular are the Clinical Global Impression-Schizophrenia Scale [ 3 ], the Positive and Negative Syndrome Scale [ 4 ], the Scale for Assessment of Positive Symptoms [ 5 ] and the Scale for Assessment of Negative Symptoms [ 6 ]. Although the BPRS includes 18 items and the allocation of marks is defined clearly, as all items have the same range of marks (i.e., 1-7), it is not unusual to find that scores for the BPRS differ widely from those for the CGI-SCH. Ideally, scores from one scale could be mapped directly onto the other, making it possible to compare individuals evaluated with one scale or the other. We decided to investigate this divergence analytically, looking at the clinical weight of respective symptoms (the relative magnitudes of symptoms in schizophrenia) and the issue of scale nonlinearity. In the present study, we investigated whether there was a linear relationship between the scores of the two scales, to observe whether linearity of the BPRS to the CGI-SCH could be influenced by changing point allocation of the BPRS through devising an example of a possible modified BPRS subscale. The aim of the present study is 4-fold: 1) to investigate the linearity of the BPRS in relation to its items and mark allocation by examining the reasons for the incongruity between BPRS scores and clinicians' impressions of symptom severity in schizophrenic patients; 2) to determine a mathematical expression that represents the relationship between the BPRS and the CGI-SCH more precisely; 3) to seek which symptoms are important from a clinical standpoint; and 4) if possible, to construct an example of a possible modified BPRS subscale that is expected to have improved correlation with the CGI-SCH scores compared with the full BPRS within the limitations of the data obtained in this trial.

Methods Participants This was a retrospective study of outpatients and inpatients treated at the Tokyo Women's Medical University, Miyazaki Hospital and Depression Prevention Medical Center, Kyoto Jujo Rehabilitation Hospital, Japan, who met the DSM-IV-TR [ 7 ] criteria for schizophrenia. A total of 150 patients (74 males, 76 females) with a mean age of 44.5 years (range, 17-83) were included in this study. Fifty patients were suffering their first episode of schizophrenia or attending for initial treatment (Group A) and 100 were randomly selected during either the acute or chronic phase of schizophrenia (Group B). The study involved a retrospective chart review and was approved by the ethics committee of our institution. Research design All patients were evaluated and rated from their medical records using the BPRS and the CGI-SCH during the same session, but at initial consultation for Group A and at a random treatment session for Group B. In this study, we utilized the CGI-SCH as a scale that substituted for the evaluation made by the patients' psychiatrists under the tentative assumption that the CGI-SCH had perfect linearity and that it represented the precise clinical global impression of the treating psychiatrists. If the linearity of the BPRS to the CGI-SCH was not initially apparent, we aimed to derive a mathematical equation to represent more precisely the relationship between the scales, clarifying which symptoms were important in evaluating schizophrenia and how we could improve the correlation between the BPRS and the CGI-SCH. Two experienced psychiatrists shared their evaluations, and the scores for the BPRS and the CGI-SCH were presented graphically, making it possible to examine whether the two demonstrated a linear relationship. At this stage, we examined the distribution on the scatter plot of the two scales and expressed the relationship in a precise mathematical equation. Next, backward stepwise regression was performed, with the CGI-SCH as a dependent variable and with all 18 items of the BPRS as independent variables. An F-value of less than 2.000 was used to identify variables for removal. In addition, backward stepwise regression with F-value at the same condition was performed using three small sets of variables based on derived scores. These independent variable groups were: positive symptoms (conceptual disorganization, grandiosity, hostility, suspiciousness, hallucinations, and excitement); negative symptoms (emotional withdrawal and blunted affect); and general psychopathological symptoms (somatic concern, anxiety, guilt, tension, bizarre behavior, depressed mood, motor retardation, uncooperativeness, unusual thought content, and disorientation) with reference to three domains of the PANSS. Variables showing a positive association with the CGI-SCH were derived from these three domains, and multivariate regression analysis was performed using the selected items as independent variables and the CGI-SCH as a dependent variable. By convention in stepwise regression, even if p values exceed 0.05, it is permissible to adopt the variables if those symptoms are judged as clinically important, as long as the p values do not exceed 0.20. However, we selected the variables positively associated with the CGI-SCH within the condition that p values were less than 0.05. In the stepwise and multivariate regression analyses, we often obtained variables inversely associated with the CGI-SCH. In this study, we adopted a way to remove them. Furthermore, utilizing the results of multivariate regression, we allocated marks in proportion to the magnitude of the multiple regression coefficient of each variable so that each symptom was allocated different marks proportional to the positive multiple regression coefficient. With regard to linearity, we then examined the distribution on the scatter plot of the BPRS and the CGI-SCH scores. We obtained the Pearson's r coefficient as an indication of the degree of linearity of the relationship between the two scales, r-squared being one of values used to estimate how much the fit of model shrinks (by observing how r-squared decreased), before and after exclusion of the various items and modification of the mark allocation. Furthermore, we examined whether the selection of specific items and/or changing the distribution of the marks enhanced the correlation of the BPRS with the CGI-SCH and how much the r-squared decreased. On the basis of the results, we constructed an example of a possible modified BPRS subscale, "the modified seven-item BPRS", which would be expected to have a higher correlation with the CGI-SCH within the limitations of the applicability for our data at this stage. We used SPSS for Windows, version 14 [ 8 ] for the stepwise regression analysis, Stata Release 10.0 [ 9 ] for the multivariate regression analysis, and Microsoft Excel 2003 [ 10 ] for plotting the graph.

Results Figure 1 shows the relationship between the 18-item BPRS score and the CGI-SCH score. Although there was a rough correlation, a curve with upper convexity was obtained, and the straight-line relationship that had been thought to exist between the two scales was not apparent. Because the shape of the curve was similar to a logarithmic curve, we performed a logarithmic transformation of the 18-item BPRS total score. The curve was then modified to an almost linear distribution, which was described by the equation [CGI-SCH] = 7.1497 × log 10 [18-item BPRS] - 6.7705 (p < 0.001; Figure 2 ). Pearson's r coefficient for the relationship between the 18-item BPRS and the CGI-SCH was 0.7926 (p < 0.001) and r-squared (that of multivariate regression using the full item of BPRS) was 0.7560. The results of backward stepwise analysis for correlation are shown in Table 1 (p < 0.001). According to the results of other backward stepwise regressions for variables divided into three groups, eight items were selected (p < 0.001; Table 2 ). 'Conceptual disorganization' (p < 0.001), 'hostility' (p < 0.001), 'hallucinations' (p < 0.001), 'emotional withdrawal' (p < 0.001), 'anxiety' (p < 0.001), 'motor retardation' (p < 0.001), 'uncooperativeness' (p = 0.004) and 'unusual thought content' (p < 0.001) were significantly correlated with the CGI-SCH. The selection of the above eight variables from the 18-item BPRS gave Pearson's r of 0.8185 and r-squared of 0.7198. Using these eight items as independent variables that were expected to be important for the correlation between the BPRS and the CGI-SCH, multivariate regression analysis was performed (Table 3 ). After further selection of the seven variables from the above eight variables at the multivariate regression, "the seven-item BPRS" was obtained that comprised these positively associated seven items. Pearson's r for the relationship between "the seven-item BPRS" and the CGI-SCH was 0.8315 and r-squared was 0.7036 (p < 0.001). Furthermore, because we were able to consider the weight of the multiple regression coefficient as the clinical weight, the standard deviation of each variable was assumed to be almost the same, and by allocating marks to each respective item of "the seven-item BPRS" in proportion to the magnitude of the multiple regression coefficient, Pearson's r was increased further to 0.8339 (between "the modified seven-item BPRS" and the CGI-SCH; p < 0.001; Figure 3 ) and r-squared did not change (0.7036). As a result, the distribution on the scatter plot of the two scales changed from that shown in Figure 1 to that shown in Figure 3 , yielding a more linear relationship between "the modified seven-item BPRS" and the CGI-SCH than was the case between the 18-item BPRS and the CGI-SCH. Pearson's r was increased after the series of manipulations, and r-squared decreased by slow degrees, although statistical significance was not apparent for this change. By multiplying each multiple regression coefficient by 40, we composed an example of a possible modified BPRS subscale: "the modified seven-item BPRS" (of tentative meaning, given the limits of our data at this stage) (Figure 4 ).

Discussion The BPRS is one of the most frequently used instruments for evaluating the psychopathology of patients with schizophrenia. Although its psychometric properties in terms of reliability, validity and sensitivity have been extensively examined [ 11 ], patients are examined by clinicians with different observer ratings using different criteria. On the other hand, assessment with the CGI-SCH is based on a score of 1-7, making it simple and relevant. The CGI-SCH may be as sensitive as the BPRS in detecting efficacy differences between antipsychotic drugs [ 12 ], but it is necessary that treatment response be interpreted in the context of patient characteristics [ 13 ]. However, patients with different characteristics but with similar scores are often treated similarly in clinical trials. Therefore, training is required for performing a standardized evaluation [ 14 ]. Other user-friendly assessments include the Revised Global Outcome Assessment of Life in Schizophrenia (Revised GOALS) [ 15 ], the Investigator's Assessment Questionnaire (IAQ) [ 16 ] and the Targeted Inventory on Problems in Schizophrenia (TIP-Sz) [ 17 ], although they have some limitations in terms of methodology. We also believe that other important aspects of illness management should be supplemented with appropriate subjective scales as necessary [ 18 ]. Nonetheless, there is no consensus among clinicians regarding the most suitable scale. To address this perplexing issue, more advanced investigations are necessary to devise rating scales using some form of statistical method. Leaving aside the debate over whether psychopathological severity or state can be expressed in evaluation scales such as the BPRS or the CGI-SCH and accepting the need and utility of such instruments, we focused here on improving the BPRS scale. We examined whether the more detailed assessment afforded by its items and individual point allocations could be made proportional to the simpler and more global CGI-SCH scale. Many previous attempts have been made to evaluate the adequacy of the BPRS from the viewpoint of which items should be selected because of their relevance. However, no study has approached this issue by addressing how the degree of linearity of the BPRS can be changed by modifying not only its constituent items but also their weighting, using stepwise regression and multivariate regression analysis. In the present study, we first examined whether the BPRS and the CGI-SCH showed a mutual linear relationship. By plotting the BPRS scores and the CGI-SCH scores in the form of a graph, we compared their respective distributions. The scatter plot of the 18-item BPRS and the CGI-SCH yielded a curve with upper convexity, thus demonstrating that the relationship between the two scales was not linear (see Figure 1 ). Because the shape of the curve had upper convexity similar to a logarithmic curve, we performed a common logarithmic transformation on the 18-item BPRS score. Then, we were able to obtain a possibly more precise equation as a logarithmic form shown in Figure 2 . From this result, we presumed that there was a possibility that an increase in the logarithm of the total score for all symptoms might be roughly proportional to the global increase in symptom severity observed clinically in schizophrenic patients. We recognize, however, that this model has applicability to only this trial at this stage. We inferred that the logarithmic relationship between the single score scale, the CGI-SCH, and the plural score scale, the BPRS, might be an important tool in our determination of the severity of illness. We then investigated whether modifying the constituent items and/or allocation of marks could affect the linearity of the BPRS, at least, within this trial itself, looking at the correlation of the BPRS with the CGI-SCH in terms of Pearson's r and r-squared, which express one of the degrees of the fit between the two scales. To evaluate the clinical severity of schizophrenia, we substituted the CGI-SCH score for the clinical impression. Partly because the values of Pearson's r were slightly higher between "the seven-item BPRS" (constructed by selection of specific items) and the CGI-SCH than between the 18-item BPRS and the CGI-SCH without considerable decreases of r-squared, the shape of the scatter plot between the two scales became more linear than that before the selection. We inferred that there was a possibility that the selection of these items from 18 items is related to the linearity of the BPRS. The clinical weights might be related to the heightened values of Pearson's r between "the modified seven-item BPRS" (constructed by changing the allocation of marks) and the CGI-SCH, as the shape of the scatter plot between the two scales became more linear than before. We presumed that there was a possibility that the weighting was also associated with the linearity of the BPRS. Furthermore, by assigning different weights to each item proportional to the respective regression coefficients, we were able to compose a possible modified BPRS subscale, "the modified seven-item BPRS", by assuming that the magnitude of each regression coefficient represented the respective clinical weight of each item. This scale is only an example of a possible modified BPRS subscale that we are able to assume within our data, and the number of items decreased from 18 to 7. Historically, a widely used algorithm employing a stepwise method was first proposed by Efroymson [ 19 ], and supplementary articles were later reported by Hocking [ 20 ] and others. In the field of psychiatry, stepwise methods have been used for predicting the quality of life of schizophrenic patients by reference to schizophrenia symptoms [ 21 ], for estimating predictive values of neurocognition in schizophrenic patients [ 22 ], and for estimation of the relationship between executive functions and positive symptoms in schizophrenia [ 23 ]. However, some problems with stepwise and multivariate regression analysis have been reported. To compare the relative magnitudes of variables, the partial regression coefficients are often normalized using their respective standard deviations. However, the predictor variable is at least partially redundant with other predictors and the regression coefficient is influenced by the range of the predictor variable [ 24 ]. In addition, the relative importance of predictor variables is a tenuous concept, and comparison of the importance of predictors is not always the best approach in multiple regression. As the individual items of the BPRS had the same range of marks (1-7), we considered that there would not be crucial differences in the sizes of standard deviations for predictor variables in this study. With this assumption, we considered that, for practical purposes, the magnitudes of the standardized and unstandardized β might be regarded as almost equivalent. For these reasons, we utilized the magnitude of the unstandardized β (multiple regression coefficient) to modify the distribution of marks of the BPRS and to design a tentative BPRS subscale. If supplemented with this adjustment, the scatter plot representing the relationship between "the modified seven-item BPRS" and the CGI-SCH showed a distribution proportional to the scatter plot connecting the score of "the seven-item BPRS" multiplied by the unstandardized β (multiple regression coefficient) for each item and the score of the CGI-SCH. This is because both have almost the same significance on the graph. Additional improvements in fit may be possible. The limitations of the present study should be noted. The first was the use of the CGI-SCH as a scale that substituted for the evaluation made by the patients' psychiatrists. There is no evidence that the CGI-SCH has perfect linearity and this was merely an assumption to allow modification of the BPRS under a determinate condition. For the CGI-SCH, only a certain degree of reliability has been reported [ 3 , 12 , 13 ]. Nonetheless, we thought that this kind of simplification was unavoidable and the trade-off necessary, even if this assumption would sacrifice rigor to some extent in exchange for examining the degree of an abstract value such as 'linearity.' Second, there is no evidence supporting the assumption that the BPRS score and the CGI-SCH score obtained retrospectively by coding of the symptoms reported in the clinical chart would be comparable to the data obtained from trained BPRS raters monitored for inter-rater reliability and performing standardized interviews to probe for presence and severity of a complete list of symptoms. The quality of the clinical chart is notoriously variable, so there may exist errors and distortions from missing symptoms and falsely rating symptoms as absent when reviewing a chart, because of the failure of the clinician to mention them in the chart, which would have been detected in a structured, face-to-face interview. The 0.7926 value of Pearson's r might be to some extent considered high. However, we presume that this was because the study was retrospective. Therefore, some items of the BPRS might not have been marked, thus minimizing the distribution of the BPRS score. In effect, the results of this paper may be applicable only within our own data at this stage (including the derived stepwise regression, multivariate regression and, particularly, "the modified seven-item BPRS") and there is no guarantee that the results would be comparable to prospective research. From this standpoint, our report might be regarded as one of these experimental case studies. At any rate, prospectively randomized trials are needed in future studies. Third, through the manipulations employed here, the degree of change in Pearson's r was rather ambiguous. The increases appear slight (as a total, from 0.7926 to 0.8339; particularly, in the final manipulation, from 0.8315 to 0.8339). Moreover, it is considerably uncertain whether statistical significance exists. From another viewpoint, despite the fact that the scale was simplified, and in particular, the number of items decreased, the degree of correlation (Pearson's r) stayed at the same level or increased just slightly. Although this might indicate that a simplified subscale might be valuable in comparison with the full scale, and it might be useful to clinicians for shortening time of rating, the reproducibility of items and point allocation is quite uncertain in this model. We believe that a study of this theme in the future is desirable. Fourth, as for r-squared, in general, the more variables we exclude from the model, the more r-squared tends to decrease. The selection of the subset for which the decrease of r-squared is smallest is preferred so that the loss of model fit would be minimal. The r-squared of our data ranged from about 0.70-0.75. The size of these numbers is not low, but they may not be sufficiently high even with the moderate degree of decrease. For example, 0.7560 for the full item BPRS; 0.7524 for the results of stepwise regression using the full items of BPRS; 0.7198 for the selected eight items from stepwise regression using three small sets; and 0.7036 for the selected seven items from multivariate regression (all p values of respective analysis of variance were less than 0.001). This means that the selection of items might have caused shrinkage of the model. Fifth, the selection of items and modification of point allocation may have contributed some artifacts of multicollinearity. There are likely to be intercorrelations among the data. In this study, variables that were inversely correlated with the CGI-SCH score, indicating that the more severe the BPRS item, the lower the CGI-SCH score (a phenomenon which was a departure from the clinicians' experiences), were simply excluded from the model in an ad hoc procedure. This ignored the fact that the selection of the other predictors from among the list of candidates depended on the presence of the excluded variable. Additional unknown and complicated factors are predicted to exist as well, for example, that both inpatients and outpatients were evaluated by the CGI-SCH, and that the results might have been negatively influenced by differences in cognitive ability [ 25 ]. The treatment of negative coefficients is a crucial weakness of our model. Sixth, above all, the results are not likely to be reproducible. If we performed the same procedure on new data, it is very likely that different symptoms would be selected, and different point allocations would probably be assigned to individual items. We infer that a possible way to remedy this problem, even if partially, might be to perform a number of prospective trials in line with our method, and then summarize and calculate an average on items and point allocation. If these scales are composed as a summary, they might be less problematic than that of our trials. However, even in such scales, there would still be no assurance that they would have a greater degree of reproducibility. Therefore, the extent to which the results of this paper could be applicable may be quite limited: at the extreme, only within our present data. For this reason, future studies are necessary. The true aim of this manipulation was not always to determine the best subset and/or point allocation, but to consider a specific example of a possible modified scale. Therefore, "the modified seven-item BPRS" is merely a tentative idea at this stage, to propose a new viewpoint of the importance of point allocation in the BPRS. Needless to say, the present study has many limitations, and is thus only a first step from which further studies may learn. We believe that improving evaluation scales to make them more linear could minimize distortions in evaluation for severity of illness, including over- and under-diagnosis and estimations for efficiency and effect in clinical research. We anticipate that our present results will serve as a useful reference for clinicians attempting to devise an evaluation scale, and that further research will focus on the optimal number of items, the fittest items for selection, and the allocation of marks in rigorous methodology to maximize the linearity of the BPRS.

Conclusions Within the limits of our data, although there was a rough correlation, the linear relationship that had been thought to exist between the 18-item BPRS and the CGI-SCH was not apparent. Also, a roughly logarithmic relationship was assumed between the two scales. In addition, not only specific items but also their weightings were considered to be important in the realization of a linear relationship between the BPRS and the CGI-SCH and in the further improvement of the BPRS as a diagnostic scale.

Background Although the Brief Psychiatric Rating Scale (BPRS) is widely used for evaluating patients with schizophrenia, it has limited value in estimating the clinical weight of individual symptoms. The aim of this study was 4-fold: 1) to investigate the relationship of the BPRS to the Clinical Global Impression-Schizophrenia Scale (CGI-SCH), 2) to express this relationship in mathematical form, 3) to seek significant symptoms, and 4) to consider a possible modified BPRS subscale. Methods We evaluated 150 schizophrenia patients using the BPRS and the CGI-SCH, then examined the scatter plot distribution of the two scales and expressed it in a mathematical equation. Next, backward stepwise regression was performed to select BPRS items that were highly associated with the CGI-SCH. Multivariate regression was conducted to allocate marks to individual items, proportional to their respective magnitude. We assessed the influence of modifications to the BPRS in terms of Pearson's r correlation coefficient and r-squared to evaluate the relationship between the two scales. Utilizing symptom weighting, we assumed a possible BPRS subscale. Results By plotting the scores for the two scales, a logarithmic curve was obtained. By performing a logarithmic transformation of the BPRS total score, the curve was modified to a linear distribution, described by [CGI-SCH] = 7.1497 × log 10 [18-item BPRS] - 6.7705 (p < 0.001). Pearson's r for the relationship between the scales was 0.7926 and r-squared was 0.7560 (both p < 0.001). Applying backward stepwise regression using small sets of items, eight symptoms were positively correlated with the CGI-SCH (p < 0.005) and the subset gave Pearson's r of 0.8185 and r-squared of 0.7198. Further selection at the multivariate regression yielded Pearson's r of 0.8315 and r-squared of 0.7036. Then, modification of point allocation provided Pearson's r of 0.8339 and r-squared of 0.7036 (all these p < 0.001). A possible modified BPRS subscale, "the modified seven-item BPRS", was designed. Conclusions Limited within our data, a logarithmic relationship was assumed between the two scales, and not only individual items of the BPRS but also their weightings were considered important for a linear relationship and improvement of the BPRS for evaluating schizophrenia.

Competing interests The authors declare that they have no competing interests. Authors' contributions JS performed the evaluation of the patients and the statistical analysis, and wrote the manuscript. SM also performed the evaluation of the patients and revised the manuscript. JI was responsible for checking the methodology of the study and evaluating the results of the statistical analysis. In addition, all authors read and approved the final version of the manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-244X/10/105/prepub

Acknowledgements The authors wish to acknowledge Katsuji Nishimura, Takao Kanai, Ken Inada and Kaoru Sakamoto for providing us with very useful advice in this study.

CC BY

no

2022-01-12 15:21:45

BMC Psychiatry. 2010 Dec 7; 10:105

oa_package/98/a6/PMC3016312.tar.gz

PMC3016313

21144025

Background The prevalence of overweight and obesity has significantly increased over the past several decades with roughly 34% of U.S. adults currently being obese [ 1 , 2 ]. In persons with serious mental illness (e.g., schizophrenia, bipolar disorder), overweight and obesity are at epidemic levels that are higher than in the overall U.S. population [ 3 - 5 ]. Obesity significantly increases the risk of morbidity and mortality from cardiovascular disease mainly through the effects of obesity on hypertension, hyperlipidemia and diabetes mellitus [ 6 , 7 ]. Furthermore, many classes of psychotropic drugs are associated with weight gain. [ 3 , 8 , 9 ]. Given the long-term needs of most patients with serious mental illness to continue psychotropic medications, interventions to reduce obesity and cardiovascular risk are urgently needed. National guidelines for weight loss emphasize a multifaceted approach to include reduced energy intake, improved dietary patterns and increased physical activity [ 10 - 13 ]. Randomized controlled behavioral weight loss intervention trials showing efficacy for weight loss in the general population have systematically excluded individuals with chronic mental illness from participating [ 14 - 19 ]. In addition, behavioral lifestyle interventions in persons with serious mental illness pose several challenges since these individuals may have competing demands on a daily basis. Person with serious mental illness are often dealing with active mental health symptoms, housing issues, substance abuse or other social issues that may take precedence over behavioral change for weight loss. Furthermore, the high prevalence of cognitive deficits in this population can impede individuals from successfully performing activities of daily living [ 20 ]. Thus, traditional behavioral interventions that have been shown to be effective in populations without mental illness should be tailored to meet the specific needs of persons with serious mental illness [ 14 ]. There have been few behavioral interventions that have targeted weight loss in persons with serious mental illness. Most previous studies have applied strategies similar to those used in trials for the general population and have focused exclusively on dietary intake with a few incorporating some physical activity component. Furthermore, the interventions in this population have been small (< 60 participants), without a control comparison group and have used various methods to induce behavior change [ 21 - 28 ]. Nevertheless, the literature indicates that selected persons with serious mental illness can benefit from short-term behavioral weight loss interventions [ 17 ]. Given the health benefits of exercise and healthy dietary patterns, randomized controlled trials of comprehensive weight loss interventions (dietary and physical activity components) are needed among persons with serious mental illness. This protocol describes a randomized, controlled trial of a comprehensive behavioral weight loss and weight maintenance program among persons with serious mental illness that attend psychiatric rehabilitation day programs across Maryland. Psychiatric rehabilitation programs (PRP) are outpatient facilities where persons with serious mental illness attend up to several days a week and receive vocational and other services. Psychiatric rehabilitation programs may be opportune and efficient settings for testing, implementing and disseminating interventions. Rehabilitation programs have classroom and communal space suitable for group weight management and exercise sessions. The potential sustainability of a weight loss intervention at psychiatric rehabilitation programs may be more feasible than in other types of mental health settings because consumers regularly attend and the centers already offer other types of classes. In addition, most rehabilitation programs serve persons with a range of serious mental illness diagnoses, including schizophrenia; thus, interventions targeted to these settings could generalize to broad populations with serious mental illness.

Methods Study Design Summary The core design is a randomized, two-arm, parallel, multi-site clinical trial. Inclusion/exclusion criteria were chosen to enroll a population with chronic mental illness who are overweight or obese and who may safely participate in a weight loss intervention including moderate intensity exercise. Interested mental health consumers are being screened for eligibility at 10 psychiatric rehabilitation centers across Maryland. Three hundred and twenty adult participants completing screening are targeted for enrollment and randomization to the ACHIEVE (Achieving Healthy Lifestyles in Psychiatric Rehabilitation) intervention or usual care. Intervention participants receive group and individual weight management sessions and group physical activity classes. In the initial 6-month intervention phase, study interventionists lead the sessions and also train designated rehabilitation program staff. A 12-month maintenance intervention phase follows in which interventionists continue leading some sessions and trained rehabilitation staff assume responsibility for delivering part of the intervention. Interventionists also provide education to rehabilitation program kitchen staff to increase options for healthy meals served on-site. Follow-up data collection occurs at 6, 12, and 18 months from baseline. The study was approved by the Johns Hopkins Institutional Review Board (protocol number NA00015231). Specific Aims Primary Aim Determine the effect of the ACHIEVE intervention on weight loss at 6 and 18 months. The hypothesis is that the ACHIEVE intervention group will have greater weight loss than the control group. Secondary Aims Determine the effect of the ACHIEVE intervention on the following outcomes at 6 and 18 months: physical fitness by submaximal bicycle ergometer; waist circumference; blood pressure; lipids; Framingham cardiovascular risk score; health status with SF-12 and depression with CES-D. The hypothesis is that the ACHIEVE intervention group will have greater improvement of these outcomes compared to the control group. In addition, costs per participant will be assessed and a cost effectiveness analysis will be performed. Study sites and population The study sites will include 10 community psychiatric rehabilitation programs across urban and suburban Maryland. Each location shares certain features: (1) adult psychiatric rehabilitation day program with consumers attending regularly at least 3 days a week; (2) space for on-site physical activity classes; and (3) meals served to consumers. The trial is enrolling men and women, age 18 and older who are overweight (body mass index ≥ 25.0 kg/m 2 ) or obese and attend the psychiatric rehabilitation programs. By definition of their psychiatric rehabilitation attendance, these participants have a serious mental illness diagnosis. Persons with serious mental illness that receive care from rehabilitation programs' on-site mental health clinics are also eligible for recruitment. The inclusion and exclusion criteria for the trial are listed below. Inclusion criteria ■ Age 18 and older; ■ Overweight, defined by Body Mass Index at least 25.0 kg/m 2 ; ■ Able and willing to give informed consent and participate in the intervention; ■ On the same psychiatric medications within the 30 days before baseline weight (dose changes allowed;) ■ Able to attend at least 2 intervention sessions per week (one weight management and one physical activity session) during initial 6-month phase; Exclusion criteria ■ Contraindication to weight loss - Receiving active cancer treatment (radiation/chemotherapy) - Liver failure - History of anorexia nervosa; ■ Cardiovascular event (unstable angina, myocardial infarction) within previous 6 months; ■ Prior or planned bariatric surgery; Use of prescription weight loss medication or over-the-counter orlistat within 3 months if participant does not agree to stop taking it; ■ Twenty pound or greater weight loss in 3 months prior to baseline, as documented by staff measurement; ■ Inability to walk to participate in exercise class; ■ Pregnant or planning a pregnancy during study period. Nursing mothers would need approval from physician; ■ Alcohol or substance use disorder either: 1) active and determined to be incompatible with participation in the intervention through discussion with program staff; or 2) new abstinence from alcohol or substance use disorder in past 30 days; ■ Planning to leave rehabilitation center within 6 months or move out of geographic area within 18 months; ■ Investigator judgment (e.g., for concerns over safety, adherence or follow-up); ■ Weight greater than 400 pounds. Recruitment Strategies Most recruitment activities occur on-site at the psychiatric rehabilitation centers and affiliated psychiatric clinics. The primary means of recruitment is direct communication with consumers. Regular mental health consumer meetings at the rehabilitation programs provide opportunities for study staff to present the study. During presentations, study staff often show an informational video and may incorporate an exercise class demonstration. Posters are displayed and fliers and brochures are distributed. In addition, study staff make presentations to rehabilitation program staff and outpatient mental health clinic staff. Rehabilitation program staff and outpatient clinic staff may mention the study to their clients. Potential participants also are identified by working with rehabilitation staff and reviewing together the list of rehabilitation program attendees, using a HIPAA waiver. Study staff are available on-site at the rehabilitation programs to discuss the study during the times when consumers attend. Potential participants are also given a phone number and a postcard to mail back to reach study staff to indicate their interest. To track recruitment, a roster of consumers attending the center is provided by each rehabilitation program. These rosters assist in defining and counting the number of consumers contacted and enable the characterization of enrollees and non-enrollees. Data Collection and Measurements Before formal screening for the trial begins, oral consent from rehabilitation center attendees is obtained and their weight and height at study baseline is measured for calculation of body mass index; weight will be measured again at 6 and 18-month follow up data collection points. Measuring weight of all consumers at the day program, regardless of their interest in joining the weight loss intervention study, allows for an understanding of: (1) the natural history of weight gain in this population without intervention and (2) how the environmental component of the intervention (i.e., increasing options for healthy meals on-site) may affect weight at the centers in a pre/post fashion. The goal will be to measure height and weight for as many attendees as possible. For the formal trial, all participants provide written informed consent using procedures reviewed and approved by Johns Hopkins Institutional Review Boards (IRB). A two-stage consent process is used with the first consent obtained to conduct screening procedures and a second consent obtained after baseline data collection and prior to randomization. An evaluation of ability to give consent is also administered for each participant before screening consent which includes answering questions about the goal of the study, what they will be asked to do, and what risks may be involved if they join the study. If a participant is deemed not able to give consent, he/she may not join the study. Participant eligibility is determined by the completion of several screening measures. The Rose Angina Questionnaire and a checklist of medical conditions are implemented to determine health status for a moderate intensity exercise program [ 29 ]. If potential participants have a positive Rose Questionnaire, a prior history of cardiovascular disease or have diabetes mellitus, they require an approval letter from their primary care physician in order to enroll in the study. In addition, primary care physicians for all participants are contacted about the study to ensure there are no contraindications to weight loss or participation in moderate intensity physical activity. Alcohol and substance abuse are assessed with the Addiction Severity Index-Lite [ 30 ]; medications are reviewed; and plans to remain at the rehabilitation center during the duration of the study to participate in the trial are discussed with the consumer. Data collection visits occur on-site at psychiatric rehabilitation centers and are conducted by trained data collectors certified in study measures. Measurements are conducted using standardized operating procedures and quality control methods. Table 1 summarizes the data collection measures and schedule. Follow-up measures are collected at 6, 12, and 18 months after baseline. Randomization and Blinding After baseline data collection and before the intervention begins, participants meet with study staff and the intervention arms of the study are described in detail. The study coordinator confirms that the participant meets eligibility criteria, and that all required baseline data have been collected. Participants interested in enrollment are consented individually and randomized to either the intervention or the control group. The study coordinator ascertains and communicates treatment assignments to participants [ 31 ]. Randomization is stratified by gender and site using blocks of sizes 2 and 4 in random order to create the randomization sequence for each stratum. Due to the nature of the intervention, both participants and interventionists will be aware of the assignment. The following mechanisms are in place for data collection staff to be masked to treatment assignment: 1) designating and tracking unmasked study staff; 2) excluding data collection staff from any part of intervention delivery; 3) performing outcome assessments in separate rooms than the intervention; and 4) reminding participants not to share their group assignment. Until the trial end, investigators, staff and participants are masked to outcome data with the exception of the trial statistician and data analyst. In addition, the primary outcome variable, weight, is subject to very little measurement bias [ 32 ]. Intervention The ACHIEVE intervention incorporates concepts from social cognitive theory, behavioral self-management and the relapse prevention model [ 33 - 36 ]. The theoretical base of the ACHIEVE Trial fits well within the psychiatric rehabilitation framework which emphasizes tenets of intrinsic skills building and environmental supports [ 37 , 38 ]. Motivational interviewing provides an important framework for helping participants problem solve and set goals for weight loss. The ACHIEVE intervention operationalizes these models by providing frequent and extended contacts, opportunities for group interactions and social support, goal setting and self-negotiation, problem solving, and examples of new behavioral options. The intervention was developed from a comprehensive lifestyle intervention tested in the NHLBI sponsored PREMIER Trial (A Trial of Lifestyle Interventions for Blood Pressure Control), which has proven efficacious for weight loss by incorporating dietary and physical activity components [ 16 , 19 , 39 ]. To the PREMIER foundation of individual and group counseling sessions, on-site physical activity sessions were added to take advantage of the psychiatric rehabilitation environment as an opportunity for both skills modeling and attaining most of the weekly recommended physical activity. To further tailor the intervention, the neurocognitive deficits in working memory, verbal memory and executive function that are common in persons with serious mental illness were addressed [ 40 , 41 ]. Successful didactic interventions in schizophrenia emphasize learning a few, specific and narrow skills repeatedly, breaking material into small units, learning aides to reduce requirements on memory and attention, repetition of content, and rehearsing behavioral skills [ 42 ]. One successful approach for coping with cognitive deficits is the use of compensatory environmental strategies, which are adaptations in the environment designed to bypass neurocognitive impairments and improve adaptive functioning [ 43 ]; examples include signs, labels, and devices designed to cue and sequence appropriate behaviors and structure [ 44 ]. All aspects of the ACHIEVE intervention were tailored to meet the specific needs of a psychiatric rehabilitation population. Table 2 outlines the cognitive adaptations in the intervention. The main intervention goals of ACHIEVE include: (1) reducing caloric intake by avoiding sugar drinks and "junk food," (2) eating 5 fruits and vegetables a day, (3) choosing smart portions and snacks, and (4) increasing caloric expenditure through participation in 3 moderate intensity aerobic exercise sessions per week at the psychiatric rehabilitation program [ 10 ]. Table 3 reflects the ACHIEVE intervention characteristics for participants randomized to the intervention group. In the first individual session, the interventionist begins a partnership with the participant and assesses his/her readiness to change and understanding nutritional principles. Behavior goals are set and in subsequent sessions the interventionist uses feedback and motivational interviewing techniques to assess the participant's current progress and to help move towards the next goal. These sessions allow the interventionist to tailor the intervention to individual needs. Group weight-management classes occur three times per month during the intensive phase of the intervention and each month the classes are focused on one main topic, such as increasing fruit and vegetable consumption. The sessions begin with a discussion to support and review the concepts discussed and practiced since the prior session. A portion of the weekly session is devoted to didactic information about healthy eating or physical activity education, which is supported by food models and posters, a self-monitoring worksheet, hands-on activities related to the monthly topic, or food tastings. Participants set individual behavioral goals based on the material presented that week. Group physical activity sessions are held three days per week at the rehabilitation center (e.g., in a multipurpose area) and led by a trained exercise leader from the study staff. A progressive exercise program starts at a level appropriate for sedentary adults: 10-minute warm-up; 10 minutes of moderate intensity physical activity and 5-minute cool down [ 45 ]. The exercise duration gradually increases each week until participants are completing 40 minutes of moderate intensity physical activity and 10 minutes of warm-up and cool-down. Participants are encouraged to incorporate daily physical activity outside of the group exercise class and may set goals that reflect this effort. In addition to the organized sessions, participants meet with the intervention staff monthly for an individual weight loss counseling session. This brief activity provides immediate feedback on weight loss progress. If participants lose weight, they are asked what worked for them. If they gain weight, staff work to problem-solve and assist in working towards a behavioral goal or setting a more realistic goal. Self-monitoring and positive reinforcement are important aspects of successful weight loss trials. Participants are asked to fill out a "Tracker" as a self-monitoring tool outside of group sessions. Each tracker is used for one week; participants record the number of servings of fruits and vegetables, and respond yes or no to: exercising for 30 minutes; drinking sugar drinks; eating junk food; smart portions or smart snacks. The Tracker provides a behavioral cue to participants. An incentive program rewards participation in class and individual sessions with choices of varying priced items (e.g., gym suit, store gift card) after a specified number of stickers have been earned. In order to earn a sticker in exercise class the participant must remain standing and engaged in class from the first minute of warm up through the last minute of cool down. For weight management group and individual visits, participants earn a sticker for being present in class throughout the entire duration of the session time. For the maintenance intervention period, rehabilitation center staff assumes the responsibility for much of the exercise portion of the intervention in a stepped process over two 6-month phases. Designated rehabilitation center employees are trained by intervention study staff and provided with exercise videos made by the study team in an effort to mimic the instructor led exercise class as much as possible. The rehabilitation staff take responsibility for encouraging attendance and participation, starting the video, overseeing the safety of the class, and recording attendance data. Intervention study staff are available for consultation as needed to offer more support during this phase. This transition occurs in order to facilitate the rehabilitation center's ownership of the program, with the goal of increasing the likelihood that the center will continue to offer components of the intervention after the study is complete. Intervention Delivery ACHIEVE interventionists are skilled facilitators with experience in behavior change and group and individual-level counseling. Interventionists have a skill level that would be typical for a community health educator with a bachelor's degree; exercise leaders have at least one year of experience in leading an exercise class and/or are a certified exercise instructor. Intervention staff are trained to deliver any or all components of the intervention in order to maximize the resources of this multi-site trial. Manualized procedures and standardized materials are used to ensure consistency of the intervention including standardized formats for the group exercise classes. Staff members are regularly observed as part of ongoing staff training and fidelity assurance. Psychiatric rehabilitation program leadership and staff at each site support the intervention and are involved in the study on multiple levels. Program leadership from each site work with the study team so that intervention classes fit into the overall center schedule, and collaborate on participants' individual rehabilitation plans. Each site designates at least one employee to become trained to conduct group physical activity classes using an exercise video. Resources included in measuring the costs of the intervention delivery include the number of staff, and the duration of each activity. For each study site, staff record one week of data at intervention months 2, 3, 5, 6, 12, 15 and 18. Staff time includes time with participants for intervention sessions, time spent preparing for sessions, training and intervention-related meetings. All research related activities are excluded from the cost analysis. Control Group Participants in the control group receive standard nutrition and physical activity written information at baseline. Health classes are held quarterly for control participants with content unrelated to weight loss (e.g., cancer screening, oral health). Environmental Nutrition Intervention In order to support intervention group participants' ability to select healthy foods, interventionists provide consulting services to rehabilitation program kitchen staff. The consultation sessions help kitchen staff identify healthier food choices for meals served on-site within site budget and regulatory constraints (e.g., federal food guidelines). Interventionists work with rehabilitation staff to identify goals and then offer options to improve food choices such reducing high sugar foods, working on appropriate portion sizes and modifying vending machine offerings. Rehabilitation staff choose goals and which options they will incorporate. A random sample of menus at baseline, 6 months, and 18 months are collected and evaluated for nutritional content using ESHA software (The Food Processor, 2009, Salem, OR) [ 46 ]. Data Analysis Randomly assigned intervention group (i.e., behavioral intervention or control) is the main independent variable for intent-to-treat analysis [ 47 ]. The co-primary outcomes are weight loss at 6 and 18 months. To evaluate the efficacy of the intervention for each of these outcomes, generalized estimating equations are used [ 48 ]. These models account for the longitudinal nature of the trial and incorporate baseline, and 6, 12 and 18-month measurements and will account for study site and other baseline characteristics found not to be balanced by randomization. For secondary outcomes that are categorical, logistic regression GEE models are used according to the same principles outlined above for continuous outcomes. In addition to the analyses that preserve the intention to treat principle, analyses on subgroups defined post-randomization are exploratory. These include analyses in participants who attended the majority of weight management and exercise intervention sessions. Although second generation antipsychotics and other psychotropic medications can induce substantial weight gain [ 3 , 9 ], we expect that both study groups will be equivalent in their distribution of these medication due to the process of randomization. Several analytical approaches are planned to address three main potential effects of antipsychotic and other concomitant medication on study outcomes; these include: (1) imbalances in medication use between the intervention and control groups, despite randomization procedures, (2) variation in intervention efficacy by medication or medication class, and (3) changes in drug use after randomization. The analytic approach to handle missing data will be anchored on the assumption that data is missing at random (MAR), where the probability of missing can depend on all observed information such as measured weights and covariates but does not depend on variables that are not recorded. The analysis model will include parameters for visit specific means for each treatment group, baseline covariates associated with study retention, and use an unstructured covariance structure. Missing at random is almost never strictly correct, but careful modeling should make the missing data process as close to MAR as possible. Primary analyses will be conducted under the assumption of MAR; sensitivity analyses will be based on sensible "missing not at random" scenarios to evaluate the robustness of the inferences under the MAR assumption. For the environmental nutrition intervention, the menus are collected at each data collection time point and analyzed for nutrients using ESHA software. The mean number of calories, macronutrients, and micronutrients at each site are determined and t-tests are used to determine significant differences at each time point. Given the expected variability between sites in making changes to menus, differences in menu changes will be assessed within each site and then overall. To support the long-term goal to integrate the ACHIEVE program into psychiatric rehabilitation centers, a cost analysis is planned. The primary analysis assesses the direct cost per participant of intervention implementation from the perspective of a future payer (e.g., Medicaid). A second analysis assesses costs from the societal perspective projecting cardiovascular risk factor changes 10 years into the future. Sample size and power The main objective of this trial is to detect weight loss having public health significance. Previous work has indicated that 4-5 pounds of weight loss should reduce systolic blood pressure by ~3 mmHg, which has been estimated to reduce stroke mortality by 6-8%; cardiovascular heart disease mortality by 4-5%; and to reduce risk of incident hypertension by 20% [ 18 , 49 ]. A Monte Carlo simulation study was used to assess the power to detect a clinically meaningful effect on weight loss at months 6 and 18 under a range of conservative assumptions about the effect size, standard deviation, and follow-up with potentially clustered sites [ 50 ]. It was assumed that a 4.5 lb difference in weight at 18 months between intervention and control groups would be observed and that the difference at 6 months would be larger. For power calculations, we assumed a standard deviation of change in weight of 12 lbs and that follow-up would be 80% complete. Under these assumptions, for two-sided 0.05-level tests of the null hypothesis, the study should provide approximately 86% power for detecting a difference of 4.5 lbs with SD = 12 lbs. In addition, the study will have the same power to detect the a similar effect size with a smaller sample size if we achieve a higher follow-up rate.

Discussion Despite successful behavioral weight loss interventions in the general population, few randomized controlled trials of comprehensive behavioral weight loss interventions among persons with serious mental illness have been performed [ 51 ]. Given the high prevalence of obesity and cardiovascular risk factors, effective weight loss programs are needed in this vulnerable population. The ACHIEVE investigators led a previous pilot weight loss study (n = 52) in two psychiatric rehabilitation programs and demonstrated preliminary success with high levels of recruitment, retention and pre/post weight loss of 4.8 pounds [ 52 ]. The ACHIEVE Trial will definitively test the effectiveness of this innovative, practical intervention to realize and sustain weight loss in overweight and obese persons with serious mental illness. If successful, the intervention will be a model program that should provide important health benefits by reducing cardiovascular disease risk for persons with serious mental illness, and with appropriate resources, could be disseminated widely. This study compares the effectiveness of a multifaceted weight loss intervention to a standard care group among persons who often have cognitive impairments and other comorbidities. Behavioral weight loss trials have shown efficacy for weight loss in other populations. For example, the ACHIEVE intervention was modeled after the PREMIER Trial, a comprehensive lifestyle intervention that incorporated education and counseling for diet and physical activity; the trial was proven effective for weight loss in the general population [ 39 ]. The Trial of Nonpharmacologic Intervention in the Elderly (TONE) study demonstrated significant weight loss (3.5-4.5 kg average reduction) among adults age 60-80 years over a 30-month follow-up period [ 53 ]. Similarly, initial 1-year results from the Look AHEAD (Action for Health in Diabetes) trial have shown that older adults (> 65 years) attend more lifestyle intervention sessions and participate in more physical activity than their younger counterparts [ 54 ]. At the end of a 6-month follow-up period, participants in The Weight Loss Maintenance trial demonstrated significant weight loss across racial and gender groups; weight loss was greatest among non-African American men and least among African American women [ 55 ]. Although there have been few behavioral weight loss intervention trials among persons with serious mental illness, previous work suggests that short-term weight loss can be achieved in this population [ 51 , 56 ]. The magnitude of weight change in ACHIEVE and other trials for persons with serious mental illness could be lower than seen in other studies and populations. If true, this may be due in part to participants having difficulty incorporating targeted behaviors from weight management sessions or lacking resources to buy lower-calorie foods. Other barriers to weight loss may include persistent mental health symptoms and frequent hospitalizations. However, the ACHIEVE Trial is unique in that interventionists provide frequent and extended contacts at locations participants regularly attend. In contrast, previous lifestyle interventions in populations without mental illness often have less frequent in-person interaction and require participants to go to other locations for intervention groups and data collection. The frequent contacts in a familiar setting in ACHIEVE may help overcome barriers from cognitive limitations and/or mental health symptoms and subsequently foster significant weight loss. The multiple components of the intervention are designed to include a variety of methods to induce behavior change through repetitive and on-going activities (e.g., group and individual sessions, rewards, food models, daily record trackers). The ACHIEVE Trial is one of the first weight loss trials that incorporates tailored weight management sessions and on-site exercise classes to persons with serious mental illness. This multi-site study will include a diversity of racial/ethnic groups, suburban and urban areas across Maryland, younger and older adults, and persons with varying severity and types of psychiatric disease. Thus, the results should be applicable to a wide range of persons with serious mental illness. One challenge of the ACHIEVE Trial is the extensive support and buy-in from staff at the psychiatric rehabilitation centers required for success. ACHIEVE interventionists and data collectors need the center's physical space and other resources such as time to consult with staff in order to implement the intervention and collect data. Even with enthusiasm from psychiatric rehabilitation programs, intervention implementation may still be challenging because of certain program constraints. In the current funding environment, many mental health programs are under significant financial stress and have high staff turnover. If the ACHIEVE intervention proves effective, there will be strong justification for mechanisms to sustain the program at current sites and disseminate it to other centers. Resource data collected during the trial will inform future costs of continuing the intervention. Practical considerations for intervention sustainability and dissemination are complex and include how cardiovascular disease prevention fits into centers' priorities and what funding the rehabilitation programs would have to conduct the intervention. Centers likely would need dedicated resources or reimbursement mechanisms to contract with experienced interventionists and/or to invest in training psychiatric rehabilitation staff to conduct appropriate intervention components. The ACHIEVE Trial tests an evidence-based approach to the problem of obesity in persons with serious mental illness. The study will provide knowledge about how to accomplish weight loss through an appropriately tailored intervention delivered in a psychiatric rehabilitation setting. Furthermore, the results from this study will inform future work in healthy lifestyle interventions for cardiovascular disease prevention in populations with chronic mental illness.

Background Overweight and obesity are highly prevalent among persons with serious mental illness. These conditions likely contribute to premature cardiovascular disease and a 20 to 30 percent shortened life expectancy in this vulnerable population. Persons with serious mental illness need effective, appropriately tailored behavioral interventions to achieve and maintain weight loss. Psychiatric rehabilitation day programs provide logical intervention settings because mental health consumers often attend regularly and exercise can take place on-site. This paper describes the Randomized Trial of Achieving Healthy Lifestyles in Psychiatric Rehabilitation (ACHIEVE). The goal of the study is to determine the effectiveness of a behavioral weight loss intervention among persons with serious mental illness that attend psychiatric rehabilitation programs. Participants randomized to the intervention arm of the study are hypothesized to have greater weight loss than the control group. Methods/Design A targeted 320 men and women with serious mental illness and overweight or obesity (body mass index ≥ 25.0 kg/m 2 ) will be recruited from 10 psychiatric rehabilitation programs across Maryland. The core design is a randomized, two-arm, parallel, multi-site clinical trial to compare the effectiveness of an 18-month behavioral weight loss intervention to usual care. Active intervention participants receive weight management sessions and physical activity classes on-site led by study interventionists. The intervention incorporates cognitive adaptations for persons with serious mental illness attending psychiatric rehabilitation programs. The initial intensive intervention period is six months, followed by a twelve-month maintenance period in which trained rehabilitation program staff assume responsibility for delivering parts of the intervention. Primary outcomes are weight loss at six and 18 months. Discussion Evidence-based approaches to the high burden of obesity and cardiovascular disease risk in person with serious mental illness are urgently needed. The ACHIEVE Trial is tailored to persons with serious mental illness in community settings. This multi-site randomized clinical trial will provide a rigorous evaluation of a practical behavioral intervention designed to accomplish and sustain weight loss in persons with serious mental illness. Trial Registration Clinical Trials.gov NCT00902694

Abbreviations CES-D: refers to Center for Epidemiologic Studies Depression Scale; SF-12: refers to Short Form 12 Health Survey; HIPAA: refers to Health Insurance Portability and Accountability Act; GEE: refers to Generalized estimating equation. Competing interests The authors declare that they have no competing interests. Authors' contributions GLD conceived the design of the study. EG and NYW participated in the analytic and statistical analysis plans. KDF participated in the cost-analysis plans. GJJ participated in developing the exercise intervention and bike test measures. FBD participated in the study design. CAA, JVG, SSS, AD participated in the design and implementation of the environmental nutrition intervention. RWG participated in the intervention design. JF designed the data entry and documentation system. MO led the intervention staff. OF and LC directed data collection. JBC participated in the coordination of the trial. SSC, GLD, and LJA drafted the article. All authors edited and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-244X/10/108/prepub

Acknowledgements Funding for this study is provided by the National Institute of Mental Health, Grant R01MH080964

CC BY

no

2022-01-12 15:21:45

BMC Psychiatry. 2010 Dec 13; 10:108

oa_package/1b/d8/PMC3016313.tar.gz

PMC3016314

21144063

Background Though many chronic factors placing an individual at increased risk for suicide are well established, the acute factors that lead a person to make a suicide attempt (SA) are not known. Chronic risk factors include suicidal ideation (SI), history of suicide attempts, severe psychopathology, history of psychiatric hospitalization, substance abuse, and poor social supports[ 1 , 2 ]. Among these, SI and history of previous SA are most prominent and most relied upon in general clinical practice[ 3 - 7 ]. At present, however, no instruments are well established for the prediction of imminent SA [ 7 ]. Moreover, current measures of suicidality, including the Suicide Assessment Scale,[ 8 - 10 ] Suicide Intent Scale,[ 11 , 12 ] and Motto and Bostrom's proposed scale,[ 13 ] rely heavily on self-report of overt suicidal thoughts and plans. However, acutely suicidal individuals often deny or hide their suicidal intent,[ 14 , 15 ] and the presence of a plan for suicide is a poor predictor of attempt, as many attempters report only fleeting ideation and no premeditated plan[ 4 ]. In fact, a recent study reported an average interval of only 10 minutes between the onset of SI and the actual suicidal act[ 16 ]. Past research suggests that transition from SI to SA may be triggered by specific affective, behavioral, and cognitive factors [ 17 - 19 ]. However, the exact nature of these "trigger" factors or whether they constitute a distinct "trigger state" is not known. Esposito et al.,[ 17 ] reported that in adolescents, after controlling for depression, only anger and affect dysregulation differentiated multiple from single suicide attempters. Nock and Kazdin[ 18 ] have identified negative automatic thinking as a risk factor for suicide attempts. This type of cognition might be related to the "diffuse ruminative thought process"[ 20 ] characteristic of psychosis. Indeed, Radomsky et al.,[ 21 ] showed that 30.2% of patients with psychosis make a suicidal attempt at some point in their life. Furthermore, although controversial, a growing body of evidence links panic attacks to suicidal behavior in patients with depression [ 22 , 23 ]. It has been reported that this link persists even when controlling for depression, substance abuse and sociodemographic characteristics[ 22 , 23 ]. Weissman et al.,[ 24 ] found that 20% of subjects with panic disorder and 12% of those with panic attacks had made suicide attempts. Finally, Schnyder et al.,[ 25 ] observed that panic and self-report of "loss of control" seems to be a distinct state that occurs before individuals attempt suicide, while Busch et al., [ 15 ] found in an acute psychological autopsy study of 76 completed inpatient suicides, that nearly 80 percent both denied suicidal ideation in the days before their suicides and, using items from the Schedule for Affective Disorders and Schizophrenia (SADS), met criteria for severe to extreme anxiety or agitation, and Hendin et.al., [ 26 ] identified acute high affective intensity, in particular desperation, as the distinguishing feature of suicide completers in a case controlled psychological autopsy study. In the course of our work on psychotic panic,[ 27 ] we have encountered a distinct psychopathologic state or syndrome related to panic and psychosis,[ 27 , 23 ] fitting with the findings of Hendin, Busch, and Snyder described above, which is reported by many suicide attempters as occurring immediately prior to their suicide attempt. In accordance with the aforementioned literature and our own observations, we have therefore hypothesized that this syndrome may serve as a "suicide trigger state" (ST state) mediating the transition to active suicide attempt in the potentially suicidal patient. Thus, identification of the proposed ST state in a high-risk population may be a powerful tool for the prediction of acute suicide risk. Analysis of our data is suggestive of a state is marked by "ruminative flooding" (a confusing, uncontrollable and overwhelming profusion of negative thoughts) coupled with an acute, "frantic hopelessness", in which not only is there a fatalistic conviction that life cannot improve, but also an oppressive sense of entrapment and imminent doom. This builds to an intolerable, confused state in which patients feel that suicidal action is the only conceivable route of escape. In this state of severe distress, many patients have also reported the experience of "near-psychotic somatization" characterized by a concrete/somatic experience of thought, (e.g., thoughts creating head pressure) as well as somatic distortions (e.g., a subjective experience of a change in bodily size or shape). In order to characterize the proposed ST state we have developed the Suicide Trigger Scale (STS), a rating scale that contains items testing for the above symptoms. Importantly, the STS does not rely on self-report of suicidal ideation. In this pilot study we aim to test the reliability and construct validity of the ST state as assessed by the STS-2, using statistical analysis of its coherence, internal structure, and relationship to a known validated instrument (the SCL-R 90). Further, we will assess the STS-2's relation to suicidal risk by examining the associations of scores on the scale and its individual components with a reported history of suicide attempt among patients with suicidal ideation.

Methods Participants The study was approved by the Beth Israel Institutional Review Board. Inclusion criteria were admission to psychiatric inpatient unit, chief complaint of suicidal wish/ideation upon admission, age ≥ 18 years, ability to understand and answer instrument questions, and literacy in the English language. The exclusion criteria were substance abuse in the 6 months prior to current hospitalization and a diagnosis of mental retardation or dementia. No other psychiatric diagnoses were exclusion criteria. Subjects were recruited from the population of psychiatric patients receiving treatment at Beth Israel Medical Center's two non-dual diagnosis inpatient psychiatric units during the period of September 2006 through July 2008. During this time, of 2230 psychiatric admissions, a total of 141 (6.3%) met inclusion criteria, agreed to participate, signed the informed consent forms and provided sufficient data to be used in the study. Of these 130 (92.2%) completed all items on the STS-2 and 104 (73.8%) also completed the SCL-90R. Suicide attempt history was considered definitive if it was confirmed by participants' clinicians' consensus recorded in the chart at the time of their discharge. Suicide attempt history is obtained by policy as part of the admission assessment for all psychiatric inpatients at Beth Israel Medical Center. Due to administrative issues unrelated to this project, only 41 charts were available for the retrospective review of suicidal ideation and behavior. Demographic and clinical data are presented in Table 1 . Axis I diagnosis was unavailable for 15 subjects due to unavailability of their charts for review. The demographic characteristics of our population are comparable to those of large clinical trials such as the STAR*D,[ 28 , 29 ] demonstrating similar proportions of males and females and similar distributions of age and level of education, though in our sample a substantially higher percentage was identified as Hispanic while a lower percentage was identified as Caucasian. This difference reflects the demographics of the local population at large [ 30 ]. Procedure and Instruments The participants were approached by research assistants who explained the purpose of the study, the nature of the scales, the measures taken to ensure confidentiality of the disclosed information and subjects' right to refuse or stop participation. After signing informed consent forms, subjects were given the self-administered STS-2 and SCL-90R to complete. The scales were administered in no particular order. Research volunteers collected demographic information from patient charts after the questionnaires were completed. Diagnoses and medication information were obtained from the medical charts of the psychiatric unit. Suicide Trigger Scale version 2 (STS-2) The STS-2 (additional file 1 ) is a 39 item scale with 3 response categories (0 = not at all, 1 = somewhat, 2 = a lot) and is derived from STS-1 [ 31 ]. The STS-1 was originally given to 36 subjects on the same acute psychiatric units as STS-2 and re-administered 7-14 days later to those 13 who were still hospitalized (Cronbach's alpha 0.86;test re-test reliability 0.911)[ 31 ]. The scores had normal distribution. Exploratory factor analysis with the STS-1 revealed 4 factors with eigenvalues greater than 1. These were labeled Dread and Doom (Factor 1), Changes in Body (Factor 2), Head Pressure (Factor 3), and Hopelessness (Factor 4). After a consensus development meeting, the STS-1 was then revised by removing non-contributory items and adding new clinically-derived items to capture more symptoms of dissociation, somatization, head pain, and dread. The result was the 39-item STS-2. The Symptom Checklist -90-Revised (SCL-90-R) The SCL-90-R is a well-established 90-item scale with 5 response categories (0 = 'not at all' to 4 = 'very much') that assesses the presence and intensity of a wide variety of psychological symptoms [ 32 ]. The total score and 9 sub-scales were used in the analyses. The sub-scales of the SCL-90-R are Anxiety, Depression, Obsessive-Compulsive, Interpersonal Sensitivity, Somatization, Phobic Anxiety, Psychoticism, Hostility, and Paranoid Ideation, and have all been found to have high reliability with Cronbach's alphas ranging from 0.8 to 0.9, one-week test-retest reliability ranging from 0.8 to 0.9, and convergent validity with the Minnesota Multiphasic Personality Inventory (MMPI) [ 32 ]. Item 59, which assesses the presence of "thoughts of death," was also used in the analysis. Statistical Analysis Reliability was assessed through Cronbach's alpha, which was used as a measure of internal consistency. Construct validity was assessed through a variety of statistical methods, including principal component analysis to explore the internal structure of the STS, Receiver Operator Characteristic (ROC) analysis with Fisher's exact test for cut-score to demonstrate clinical significance, and logistic regression analysis to examine which items of the STS-2 appeared to be most associated with suicidal action. Additionally, concurrent validity was assessed with correlation coefficients between STS-2 and SCL-90R scores and sub-scores. Internal Structure of the STS-2 Principal components analysis (PCA) with component rotation was used to assess the internal structure of the STS[ 33 ]. Because PCA requires pairwise-complete observations to calculate the correlation matrix that determines the factor loadings only data from those subjects (N = 130) who completed every item of the STS-2 could be used. (See Table 1 for comparison of PCA subjects and the total sample.) Three methods were used in succession to decide the number of components to be extracted in PCA: on first pass, eigenvalues >1, on second pass Scree plot, and finally, interpretability of components was used to eliminate components marginal on scree plot. Following PCA, component rotation was performed by both Varimax rotation and Promax rotation, both with Kaiser Normalization. Varimax rotation preserves orthogonality of components while maximizing the variance of factor loadings on each component. The aim of this technique is to produce conceptually coherent, maximally independent, component subscales. Promax rotation does not preserve orthogonality, but aims to maximize component coherence and thus their semantic interpretability. Clinical Significance of the STS-2 - Construct Validity Clinical significance of the STS-2 was assessed using ROC analysis of the STS-2 scores in discriminating past suicide attempters from those who had not made any suicide attempts[ 34 ]. ROC was performed on the unscaled STS to determine both Area Under the Curve (AUC) as a measure of the scale's robustness, and an optimal cut-score, the statistical significance of which was measured using Fisher's exact test. As the distributions of STS-2 scores in the PCA group and the subgroup chart-reviewed for suicide attempt history were very close (mean(standard deviation); 38(18) vs. 42(15), respectively), ROC analysis was also performed on the principal components produced in the Varimax PCA analysis to measure their robustness as discriminators between suicide attempters and non-attempters. In addition, logistic regression analysis[ 35 ] was used to assess which individual items appeared to be most strongly associated with suicidality. Logistic regression analysis was used to produce a coefficient for each item of the STS-2 based on a separate regression of SA onto it. The resulting odds ratio is interpreted as the change in log-odds of SA when that item score increases by one. Concurrent Validity Finally, scores on the STS-2 and its principal components were correlated with total and subscale scores on the SCL-90R as a measure of concurrent validity. Bonferroni correction for multiple (n = 30) comparisons was used to correct the threshold for statistical significance.

Results The scale showed a normal distribution of scores (p-values of the Shapiro-Wilk test of normality were 0.974 and 0.18 for the SA and non-SA groups respectively). For the 130 subjects who completed the STS-2, there was a mean score of 34 and standard deviation of 16. Reliability The STS-2 showed high internal consistency with a Cronbach's alpha of 0.949. Four items (#13 trouble falling asleep, #16 panic attack, #29 ideas turning over and over, and #30 feeling doomed) were demonstrated to decrease Cronbach's alpha. Of these only one, 'doom', loaded strongly on our final principal component solution (see Table 2 ). Internal Structure Principal component analysis extracted 8 components with eigenvalues > 1, together accounting for 66% of the variance in the STS scores. The Scree plot suggests the use of one to three principal components (see Figure 1 ). However, the one-component solution lacked semantic coherence, while the three-component solution yielded two components approximately equivalent to the two-component solution followed by a minimally contributory and semantically incoherent third component. Thus the solution with two principal components accounting for 44% of the variance (37% and 7%, respectively), was found to best fit the data and was used as the basis for subsequent analysis. Based on the two factor solution, we characterized the two principal components as follows: Principal Component 1 : Ruminative Flooding (thought experienced as a confusing and uncontrollable of flood of ruminative ideas) and Near-Psychotic Somatization (distorted/bizarre somatic perception and concrete/somatic experience of thought). Principal Component 2 : Frantic Hopelessness (acute, fatalistic conviction that one's situation is hopeless and life cannot improve compounded by a fearful and oppressive sense of entrapment and doom). The Varimax solution, which maintains component orthogonality, is very similar to the Promax solution presented here in Table 2 . Inspection of the graphs of ordered factor loadings suggested an item loading cut-off value of 0.6 for both principal components (see Figure 2 ). The graphs show clusters of items loading similarly on a given factor, and inspection of items with similar loading values reveals generally similar content. Items describing a sense of entrapment (# 4,14,26,36) had substantial loadings (0.4-0.6) on both components but did not meet the cut-off threshold. Clinical significance - Construct Validity ROC analysis of the STS-2 raw scores (N = 36) showed significant and robust detection of a reported history of suicide among suicidal ideators with an AUC of 0.724 and asymptotic significance of 0.027. Analysis of the ROC curve suggests an optimal cut-score of 48 (approximately one standard deviation above the sample mean). Sensitivity for a cut-off total STS-2 score of 48 is 0.667, specificity is 0.704 and the 1-sided p-value of the Fisher exact test is significant at the 0.02 level (see Figure 3 ). ROC analysis of subscales ROC analysis of both Promax and Varimax 2-component solutions found statistically significant (asymptotic p = 0.002) prediction of suicide attempt history in the second component, (Frantic Hopelessness) with AUCs of 0.83 and 0.82, respectively. This finding correlates well with the results of the logistic regression on the individual items discussed below. Regression analysis Logistic regression was performed to determine the association between each STS-2 item and the reported history of suicide attempt (N = 36). Regression coefficients and uncorrected p-values for STS-2 individual items regressed onto reported history of SA are presented in Table 3 . Although logistic regression analysis of the individual items of the STS-2 against history of SA found no statistically significant results after Bonferroni correction for multiple comparisons (required p value <0.00128), this criterion may be excessively stringent [ 36 ]. The items with the highest coefficients were all descriptive of one of three themes: ruminative flooding, doom/hopelessness, and entrapment. Item #33 (can stop thoughts that are troubling) had the highest odds ratio (16.01). In other words, subjects who endorsed a score of 2 ("a lot") were approximately 16 times more likely to have had a previous suicide attempt than subjects who endorsed a score of 1 ("somewhat"). Likewise, 9 items describing ruminative flooding (Items #2, 3, 9, 12, 13, 20, 29, 32, and 33) had a mean regression coefficient of 0.97 (corresponding to an OR of 2.64). Contrary to expectations, items describing near-psychotic somatization (Items #5, 11, 18, 19 and 24) produced negative coefficients in the regression analysis (albeit only at an uncorrected trend level of significance). Thus in our sample population of psychiatric inpatients, more bizarre somatic experience corresponded to a decreased likelihood of having made a past suicide attempt. Integration of Principal Component and Regression Analyses Several of the best-performing items in regression analysis loaded strongly (factor loading values ≥ 0.5) on the principal components. Furthermore, items with relatively high regression coefficients (> 1.0) had a strong mean loading of 0.46 on Principal Component 2 (which was a robust detector of past SA), but a weak mean loading (0.15) on Principal Component 1 (which performed poorly as a detector of past SA under ROC analysis). In combination with the heavy loading of somatic symptoms on Component 1, this appears to account for Component 1's poor performance as a predictor of suicide attempt history on ROC analysis. Concurrent and External validity of the STS-2 One hundred and four (104) subjects completed both the SCL-90-R and the STS-2. Correlations between STS-2 total score and principal component 1 and 2 scores were calculated and correlated with the SCL-90-R total scores, the nine subscales and Item 59 - "Thoughts of death or dying". There was a high correlation between total scores on the STS-2 and the SCL-90; r = 0.77. High correlations were found for all subscales, principally for depression and anxiety. The lowest correlation coefficient was found for Item 59. However this is most likely an artifact of the low range of scores possible for a single item as compared to a subscale, which makes it more susceptible to noise. The results are shown in Table 4 below. All correlations were significant to p < 0.001, (equivalent to p < 0.03 after Bonferroni correction for multiple comparisons). Substantial numbers of high STS-2 scores were found in all demographic and diagnostic subgroups, demonstrating that the instrument measures a state that is not demographically bound, and is distinct from panic, mood, and psychotic disorder. Table 5 shows the mean scores on the STS-2 across demographic and diagnostic variables as well as the percentage and N of each demographic subgroup of the entire sample that scored above the cut-score. While substantial differences may be noted between different demographic subgroups, a substantial proportion (> 20%) of each subgroup reported a score greater than the cut-score. Comparison of demographic and diagnostic categories by Fisher exact test demonstrated no significant differences at the p < 0.05 level, providing preliminary evidence of external and divergent validity.

Discussion The results of this preliminary investigation are limited by its retrospective design, reliance on self-report, relatively small size of the whole sample and of an even smaller subgroup of subjects with data on past suicide attempts. Thus, our findings should be viewed as exploratory in nature and are not intended to demonstrate causality or define a definitive component structure. Nonetheless, the high Cronbach's alpha suggests that the STS-2 defines a coherent psychopathological clinical state, and principal component analysis, though underpowered by a factor of two, is suggestive of two principal components. The first component was termed Ruminative Flooding and Near-Psychotic Somatization, while the second was termed Frantic Hopelessness. Items describing entrapment and dread loaded strongly though below the cut-off level for both components, and were found in regression analysis to be highly sensitive to past SA. We conceptualize entrapment and dread as elements of Frantic Hopelessness. High scores on the STS-2 demonstrated significant sensitivity and specificity in distinguishing suicidal ideators with a history of attempt from those without. Finally there were high correlations between scores on the STS-2 and the SCL-90-R assessment of general psychopathology, as well as the depression and anxiety subscales of the SCL90-R, consistent with the conception of the suicide trigger state as a syndrome of disordered thought and affect. Our findings appear to be the first quantitative description of a discrete psychopathologic state other than suicidal ideation, and distinct from Axis I diagnosis, that demonstrates a differential association with suicidal action. Our data supports our hypothesis that this state is associated with suicidal action, but cannot demonstrate causality. Further investigation is warranted to determine whether this state indeed serves as an acute trigger state for suicidal actions or, alternatively, serves as a marker of a trait susceptibility to taking suicidal action. Our results indicate that items encoding Ruminative Flooding and Frantic Hopelessness, including those describing entrapment and dread, were particularly associated with history of suicide attempt and thus may play a more prominent mediating role in the precipitation of suicidal action. Combining the results from all our statistical analyses, our data paint a picture of a panic-like state characterized by disturbed thought process (rumination, perceptual distortion, near-psychotic somatization), and a pathological cathexis of thought content and affective arousal which we term 'frantic hopelessness.' In this state, hopelessness is acutely sharpened to a sense of doom, entrapment and dread. The robustness of the second principal component of the STS-2 (Frantic Hopelessness) in distinguishing ideators with history of attempt from those without is consistent with the literature that identifies hopelessness as a primary risk factor for suicide attempt[ 37 ]-[ 38 - 40 ]. It might be argued that indeed our results no more than recapitulate Beck's finding that hopelessness is a strong predictor of suicidality. We suggest however that the coherence of the STS-2 demonstrated by its high Cronbach's alpha combined with the scale's inclusion of many items which are clearly distinct from hopelessness on face value, argues for a unique clinical syndrome broader in scope than hopelessness alone as described by Beck. Furthermore, the second principal component, while including elements akin to canonically described hopelessness, is distinct not only by virtue of existing within the context of this syndrome, but also because it contains items - such as doom (#30), fatigue (#1), and cognitive oppression (#32) - which lend it an acute, fatalistic and oppressive quality not previously described. This finding however is limited by lack of power for a definitive factor analysis. Though Cronbach's alpha was high, two items, doom (#30) and panic attacks (#16) reduced this metric. That Cronbach's alpha was decreased by item 30 "Doom" could suggest that doom does not belong to the syndrome. However, Cronbach's alpha was not decreased by semantically similar items, or by other items that loaded most heavily on the Frantic Hopelessness component. An alternative explanation may be that 'doom', a somewhat literary word, was not familiar in the vocabulary of some subjects, and perhaps more so given the high proportion of Hispanic subjects, many of whom may not have been raised in an English-speaking environment. Similarly, item 16 "panic attack" may have reduced Cronbach's alpha because it relies upon subject familiarity or comfort with this technical term, which may not be as common in the lay vocabulary as, for example, "depression." Further, the high correlation of the total STS-2 scores and the two principal components with the SCL90-R Anxiety Subscale is consistent with the literature supporting panic and anxiety disorders as risk factors for suicide attempt [ 23 , 41 , 42 , 4 ]. Our finding that those items in the first principal component which are descriptive of Ruminative Flooding (such as racing and too many thoughts) generally produced fairly high regression coefficients (mean value 0.97) is consistent with the findings of Morrison and O'Connor[ 19 , 43 ] who identify ruminative thought as a suicide risk factor. The high correlation between STS-2 and SCL-90R total scores is in agreement with the literature that finds general severity of psychopathology to be a risk factor for suicide[ 4 , 44 , 45 ]. The marked variability of SCL-90R Item 59 (thoughts of death or dying) in a sample population of patients presenting with SI highlights the limited reliability of patient self report of SI. The comparatively low correlation between scores on item 59, which should, a priori, be high for suicidal ideators, and scores on the STS-2 items most predictive of past SA as grouped in Component 2, highlights the importance of a clinical measure which does not rely on overt self-report of suicidality. Our results also present the unexpected finding that items of the STS-2 that describe near-psychotic somatization (which could be interpreted as variants of somatic and dissociative symptoms of panic attack) appear to correlate negatively - though not significantly - with history of SA. This is contrary to the literature linking suicide risk to panic attacks, and overall severity of psychopathology and psychoticism[ 21 , 24 , 45 ]. While our data are not sufficiently powered to demonstrate this, inspection of score distributions across different axis I diagnoses suggests that schizoaffective subjects were more heavily represented among those with history of SA but had lower scores on the STS-2 somatization items, while subjects scoring highest on somatization items were rather those with combined depression and anxiety diagnoses. Possibly this is merely an artifact of small sample size and sample population. We speculate however, that among those subjects with primary anxiety diagnoses, somatization is a marker of concern for bodily integrity (as in the hypochondriac) and may protect against self-harm behaviors [ 46 , 47 ]. As highlighted, our study has a number of limitations. In summary, while the study has the advantage of comprising a demographically and diagnostically balanced population, it is limited in sample size and was not sufficiently powered to reliably detect differences between subgroups. Furthermore, the sample size is too small for a definitive factor analytic study and thus the factor structure should be considered preliminary. The limitations imposed on the secondary analyses by small sample size were magnified by the lack of availability of complete clinical data for many subjects due to lack of chart availability, such that Axis I diagnosis unknown for 15 subjects and suicide attempt history was only known for 39 subjects. Though there were no significant differences between the subject group as a whole and the subgroup of subjects whose charts were available for review of SA history in terms of ethnic group composition, or scores on the STS-2, a significantly higher proportion of the entire group carried bipolar and psychotic disorder diagnoses than in the chart-reviewed subgroup (approximately 40% vs. 25%, p = 0.04). The cultural diversity of the sample may also affect the results in ways which the current study is unable to account for due to cultural mediation of symptomatology; somatic symptoms in particular may exhibit culturally mediated differences in salience, semantic significance, and prognostic value [ 48 , 49 ]. A further limitation common to studies of infrequent phenomena such as suicide is its retrospective design, and, in particular, its reliance on self-report as the only measure of suicide attempt history. As with all self-report instruments, there is risk that subjects did not understand all of the scale items, answer accurately, or without bias.

Conclusions Within the study limitations, our findings suggest that the STS-2 describes a novel and coherent syndrome of psychic experience, separate from suicidal ideation and DSM-IV axis I diagnosis, which demonstrates an association with report of past suicidal action. This state consists of ruminative flooding, near-psychotic somatization and frantic hopelessness. Scores on the STS-2 can distinguish between suicidal ideators who report having made an attempt in the past from those who deny past suicide attempts. There is a great need for a reliable and valid instrument that would enable health care professionals to identify patients at increased risk of acting on their ideations and to pre-empt serious suicide attempts, particularly in those patients at greater risk for "low plan" or impulsive suicide or those who deliberately conceal or unconsciously repress suicidal ideation[ 14 , 15 ]. Thus, an assessment that does not rely heavily on the self-reported cognitions of patients would be of particular value. The lack of emphasis on suicidal ideation and plan in the STS-2 could make it particularly suited to this task, as these features may be absent, outside of conscious awareness, or may be intentionally underreported. Future larger studies utilizing prospective approaches, larger samples, and corroborated suicidal events are therefore needed to substantiate the current results and establish the STS-2 as a predictor of suicidal action. Future studies should also explore the influence of culture, gender, and primary psychiatric diagnosis on STS global scores and subscales, to demonstrate its ability to predict suicide acutely and prospectively and to further elucidate which elements of the state are most predictive of suicide attempts.

Background This study aims to develop the construct of a 'suicide trigger state' by exploring data gathered with a novel psychometric self-report instrument, the STS-2. Methods The STS-2, was administered to 141 adult psychiatric patients with suicidal ideation. Multiple statistical methods were used to explore construct validity and structure. Results Cronbach's alpha (0.949) demonstrated excellent internal consistency. Factor analyses yielded two-component solutions with good agreement. The first component described near-psychotic somatization and ruminative flooding, while the second described frantic hopelessness. ROC analysis determined an optimal cut score for a history of suicide attempt, with significance of p < 0.03. Logistic regression analysis found items sensitive to history of suicide attempt described ruminative flooding, doom, hopelessness, entrapment and dread. Conclusions The STS-2 appears to measure a distinct and novel clinical entity, which we speculatively term the 'suicide trigger state.' High scores on the STS-2 associate with reported history of past suicide attempt.

Competing interests The authors declare that they have no competing interests. Authors' contributions ZY drafted the manuscript and contributed the design and completion of the data analyses. CK assisted in the drafting of the manuscript, performance of the statistical analyses, as well as the coordination of the study. MSJ designed and performed the principal statistical analyses. DE and LJC provided substantial editorial input in the drafting of the manuscript. IIG conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-244X/10/110/prepub Supplementary Material

Acknowledgements We would like to acknowledge the substantial efforts of the research volunteers who collected and tabulated the data for this study, Serena Fox, MD who helped coordinate their efforts, and Ramin Mojtabai MD, PhD, MPH for his invaluable counsel and editorial support in the drafting of the manuscript. This research was supported in part by the Hope for Depression Research Foundation, the Empire Clinical Research Investigator Program, the Family Center for Bipolar Disorder, and the Zirinsky Mood Disorders Center. This research was presented in part at the following meetings: Yaseen Z, Johnson M, Galynker I. Construct Validity of a Suicide Trigger State. The 162nd Annual Meeting of the American Psychiatric Association, San Francisco, CA (2009) Yard S, Tecuta L, Blumenfeld A, Mojtabai R, Cohen L, Galynker I: Reliability and Validity of the Para-Psychotic Symptoms Scale. The 160th Annual Meeting of the American Psychiatric Association, San Diego, CA (2007).

CC BY

no

2022-01-12 15:21:45

BMC Psychiatry. 2010 Dec 14; 10:110

oa_package/c3/d5/PMC3016314.tar.gz

PMC3016315

21159169

Background Addressing stress during pregnancy and the postpartum period is important as these periods are physically, psychologically and socially distinct periods in a women's lifetime during which mothers experience concerns about the health of their child, their own health, changes in their bodies and the subsequent effect on changes in their marital relationship. Additionally, worries regarding economic insecurity, breastfeeding, and bonding with the infant can exacerbate the stress often experienced in this period [ 1 , 2 ]. First-time mothers have the added stressors of adapting to their new role as mothers and the insecurities associated with their ability to nurture an infant for the first time [ 3 - 5 ]. These stressors have a significant impact on the mother's psychological well being such as prenatal and postpartum depression, especially when stressors are perceived as stressful [ 6 ]. Perceived stress is a person's appraisal of certain life events as potentially threatening. This perception is reached in light of the person's ability to cope with such events [ 7 , 8 ]. Therefore people evaluate potentially stressful life events differently. The ability to accurately measure perceived stress is essential in order to treat and evaluate the effectiveness of interventions and treatments. There are few validated tools for the measurement of stress during pregnancy or during the postpartum period [ 9 - 11 ]. One commonly used scale is the 10-item Cohen Perceived Stress Scale (PSS-10). The PSS-10 has been used to research stress among different population groups including healthy university students, drug addicts, elderly populations, as well as pregnant and postpartum women. The Cohen PSS was originally developed in 1983 as a 14-item Likert type questionnaire in order to measure one's own perception and appraisal of life events as stressful [ 12 ]. A shorter version emerged when the psychometric properties of the PSS-14 scale were assessed using data from phone interviews with 2387 male and female US residents from all ages, ethnicity and household income [ 13 ]. Four questions were dropped from the PSS-14 when it was found that they did not load on either of the two factors obtained using exploratory factor analysis for the PSS-14. The PSS-10 was found to have adequate reliability and validity and a slightly higher internal reliability than PSS-14 (Alpha coefficient of 0.78 vs. 0.75). Exploratory factor analysis of PSS-10 uncovered the same two- factor structure as PSS-14. The first factor included questions reflecting negative feelings (being upset, angry, or nervous) and inability to handle stress while the second factor included questions expressing positive emotions and ability to act in stressful situations. This has been confirmed in more than one study that examined psychometrics of self-administered PSS-10 among different populations [ 14 , 15 ]. The 10 items in the scale inquire about feelings and thoughts that tap the degree to which respondents find their current life situation unpredictable, uncontrollable and stressful. Respondents indicate how often in the past month they have felt or thought a certain way on a 5-point Likert scale (0 = never, 1 = almost never, 2 = sometimes, 3 = fairly often, 4 = very often). The higher the score the higher the perceived stress is. The scale correlates with different psychosocial measures specifically depression, anxiety, and perception of poor health as well as with decreased satisfaction with self, job and life in general [ 13 , 14 ]. The PSS-10 has been translated to different languages including Arabic, Spanish, Turkish, Mexican Swedish, Greek, Bulgarian, Chinese, Thai, Japanese, Persian, and Hungarian [ 16 ]. However, not all translated versions of PSS-10 have been validated. Hamdan-Mansour and Dawani [ 17 ] used an Arabic version of PSS-10 to study stress among university students in Jordan. They reported an internal consistency of 0.68. Using the pilot data of a study investigating stress and health among working Jordanian women, Hattar-Pollara and Dawani [ 18 ] reported a Cronbach alpha of the Arabic translated version of 0.86. The authors of the latter study recommended further testing of the PSS-10 to establish its construct validity and ensure that the scale is culturally sensitive. Although PSS-10 has been used in many studies to measure stress among pregnant and postpartum women [ 19 , 20 ], none of these studies were designed for validation purposes among these populations. However, some of these studies reported a good level of reliability with internal consistency of the PSS-10 scale ranging from 0.71 to 0.83 [ 19 , 21 , 22 ]. The aim of this study was to translate the PSS-10 to classical Arabic and examine its psychometric properties among pregnant and postpartum women. Specifically, the objectives of the study were to assess the reliability of the Arabic PSS-10, its concurrent validity with similar validated scales used in pregnancy and the postpartum periods, and its construct validity by examining its ability to detect meaningful variance between specific groups of the population. A reliable and validated Arabic tool for measuring stress among women in the postpartum period, such as the Arabic version of the PSS-10 is essential for assessing the impact of interventions aimed at decreasing stress levels in the postpartum period among Arabic speaking women in different countries.

Methods Participants Overall 268 women participated in the study. These included 113 women in their third trimester (starting week 28 of pregnancy), and 97 women in the postpartum period (within 6 months after delivery). Moreover, 58 healthy female university students; who were neither pregnant nor mothers; were recruited to act as controls. Pregnant and postpartum mothers were recruited through the clinics of two obstetricians and one paediatrician during prenatal, postpartum or well baby visits. The three clinics chosen provided care to patients from different socio economic backgrounds. Consecutive pregnant and postpartum women attending the chosen clinics were approached and all consented to the study (100% response rate). Female university students were selected using quota sampling from the six academic units at the American University of Beirut, with 10 from each unit. Only two students out of 60 approached refused to participate in the study (97% response rate). Data collection was done over a two- month period. Instrument The first step of the validation process was the translation of the original English version of PSS-10 to classical Arabic by a professional translator. Classical Arabic was chosen to make the tool useful for all Arabic speaking countries, as "spoken" Arabic can be very different between countries. After translation, the scale was then reviewed by a bilingual psychiatrist for appropriateness of language. The reviewed version of the translated PSS-10 was then back translated by the psychiatrist into English and compared to the original one in order to check for consistency. Both translator and psychiatrist were not familiar with the scale. Discrepancies were corrected accordingly. The obtained Arabic PSS-10 was piloted on a small group of mothers (n = 10) to ensure that all the terms employed were understandable and to modify any ambiguity. Finally, a sample of bilingual female university students (n = 10) was asked to complete both the English and the Arabic versions of the PSS-10 [Additional file 1 ] consecutively on the same day. The mean time between the two administrations was 5 minutes. The Spearman correlation coefficient (rho) between the English and the Arabic versions was 0.71. Those 20 women were not included in the final sample used in this study. Moreover, the validity of the PSS-10 was assessed by concurrently administering the Arabic versions of the General health Questionnaire (GHQ-12) to all participants and the Edinburgh Postpartum Depression Scale (EPDS) to pregnant and postpartum women. The GHQ-12 is a short version of the 60-item GHQ developed by Goldberg [ 23 ] to screen for psychological disorders in primary care and community settings. Its use in research and in the clinical settings was previously established in a study conducted by the World Health Organization in 1995 [ 24 ]. The Arabic GHQ-12 has a sensitivity of 83%, a specificity of 80%, and a total discriminatory power of 86% [ 25 ]. The EPDS is a 10-item screening instrument developed to identify postnatal depression in health care settings and it has been used extensively for research [ 24 , 26 ]. The Arabic version of EPDS was validated by Ghubash and Abou-Saleh [ 25 ]. Using a cut-off score of 10 the sensitivity and specificity of the Arabic EPDS were 91% and 84% respectively and the internal reliability using Cronbach α was 0.84. Both EPDS and GHQ are validated rating scales to screen for psychological distress and depression among women in the prenatal and postpartum periods [ 27 , 28 ]. PSS-10 has been reported to correlate positively with GHQ-12 (r = 0.61 with n = 508) [ 15 ] and EPDS (r = 0.52 with n = 130) [ 29 ]. We expected a positive association between the three scales PSS-10, GHQ-12 and EPDS. In addition to comparison with the GHQ-12 and the EPDS, the validity of the PSS-10 was evaluated by comparing the PSS scores among the 3 groups studied while adjusting for variables unbalanced between the three groups and potentially affecting the PSS-10 scores. We expected that pregnant and postpartum women would be more stressed than healthy university students. Administration After informed consent was obtained, the selected participants answered a structured questionnaire that included the three Arabic scales: PSS-10, GHQ-12, and EPDS (for pregnant and postpartum women only). Basic demographic information (such as age and level of education) was solicited. As an indicator of socio economic status (SES), women were asked whether their income was sufficient. Moreover, the women were provided with a list of 7 negative life events and were asked if they experienced any in the past year and how much they were affected (no impact, little impact, moderate impact and severe impact). These events were: divorce/separation in the family, problems at work/university, death of a family member or a close friend, illness of a family member, financial problems, personal problems (that required effort to deal with), and health problems related to pregnancy/delivery. Correlating PSS scores with current stressful life events scores provided an additional means for assessing validity. To check for reliability, a test-retest procedure was performed where 60 participants; 20 from each group, were chosen at random and asked to consent to a retest after one week. Of those, 41 women (20 students, 17 pregnant and 4 postpartum women) accepted (68% response rate). Postpartum women were reluctant to come back to the clinic for a retest after one week. The questionnaire was self-administered; however, in case selected participants were unable to read or had problems reading the scales, the recruiters in each clinic read the scales to them and noted their responses. The study was approved by the Institutional review Board at the American University of Beirut. Data Analysis Descriptive statistics (means with standard deviations or frequency distributions) on age, education, work status, and perceived adequacy of income were calculated for each sample and for the combined samples. Comparisons of such demographic variables between the three samples were done using one-way ANOVA for comparing age, the chi-squared test for comparing work status, and Fisher exact test for comparing education and perceived adequacy of income [Table 1 ]. The internal consistency reliability and the test-retest reliability of the Arabic PSS-10 were assessed using Cronbach's alpha coefficient and Spearman's correlation coefficient (rho) respectively. Internal consistency was evaluated for the combined sample as well as for each of the three women groups (student, pregnant, and postpartum) [Table 2 ]. The test-retest reliability was performed for the combined group of all women who accepted to do the retest. Subgroup reliability measures were computed for the student group and the combined pregnant and postpartum participants since only 4 pregnant women did the retest. Exploratory factor analysis was performed using principal components with varimax rotation [Table 3 ]. Mean PSS-10 and GHQ-12 scores were compared between the three groups using the one-way ANOVA. Post-hoc comparisons were done using Bonferroni's method. Mean EPDS scores were compared between the pregnant women group and the postpartum women group using the independent t-test [Table 4 ]. Questions about stressful life events were coded from 0 (event did not occur or event occurred and the participant said that it had no impact on her) to 3 (event occurred and participant said it had a severe impact on her) with increasing values indicating increased impact of the event on a person's life. A stressful life event score was computed by adding the scores of all the questions about stressful life events. This event score was compared among the three groups using the Kruskal-Wallis test. Post-hoc comparisons were made using Bonferroni's method. Its correlation with the PSS-10 score was computed using Spearman's correlation coefficient. Finally, a multivariable analysis of covariance (ANCOVA) regression model was fit with PSS-10 score as the response variable and group as the independent variable (using two indicator variables as group is a nominal variable) while adjusting for the following covariates: age, educational level, stressful life event score, work status, and perceived income adequacy.

Results The mean age of the 268 participants was 27.6 years with a standard deviation of 5.5 years. The majority (64%) did not work while a high proportion (51%) perceived that their income was almost adequate for their needs. In both the postpartum group and the pregnant group, the majority (55% and 56%) of women were university educated. University students were significantly younger, more educated, and a higher proportion (75%) of them considered their income sufficient (defined as being adequate for their needs "most of the time" or "more than enough") as compared to the two other groups (32% for pregnant women and 41% for women in the postpartum period) [Table 1 ]. Apart from work status, demographic variables were not significantly different between pregnant and postpartum samples. Except for two, all participants were able to fill in the questionnaires by themselves. Excluding those two women from the data analysis did not change the significance of any of the results; hence the two women were kept in the final sample used for this analysis. For the overall sample, Cronbach's alpha for assessing the internal consistency reliability of the Arabic PSS-10 was 0.74. It ranged from 0.71 for postpartum women to 0.75 for pregnant women [Table 2 ]. Factor analysis showed that the scale is composed of two components with eigen values of 3.1 and 1.6 and accounting for 47.3% of the variance (data not shown). Component one consisted of questions 1, 2, 3, 6, 9, and 10 and component 2 consisted of the questions 4, 5, 7, and 8 [Table 3 ]. The test-retest reliability of the Arabic PSS-10, was moderately high with Spearman's correlation coefficient of 0.74. Reliability was higher (0.79) among students as compared to the other two groups (0.63). Time to retesting was one week for university students and varied between 2 to 3 weeks for participants from the other two groups. As for evaluation of validity, the Arabic PSS-10 exhibited significant positive correlations with both GHQ-12 and EPDS [Table 2 ]. Spearman's Rho values were higher for correlations with GHQ12, when considering both the sub-samples and the total samples, and indicated moderate association. EPDS correlated better with PSS-10 in postpartum women than in pregnant women. There was also a significant positive correlation (Spearman's Rho = .30) between PSS-10 and the score of stressful life events. That association was highest among students as compared to the other two groups [Table 2 ]. Examining the sensitivity of PSS-10 to different population groups, the results showed that PSS-10 scores were significantly higher among the students as compared to both pregnant women and postpartum women [Table 4 ]. No significant difference was observed between pregnant and postpartum women. Similar trend was observed for GHQ-12, however differences were not statistically significant (p = 0.06). As for stressful life events scores, both university students and pregnant women scored significantly higher than postpartum women [Table 4 ]. In particular, for each stressful life event, mean PSS-10 scores for those who reported severe impact were significantly higher than all other categories. Moreover, for each event, a higher proportion of university students reported experiencing a severe impact than the two other groups (data not shown). On the other hand, there was no significant difference in the mean EPDS score for the pregnant and postpartum groups. Finally, differences in PSS-10 scores among the three groups became non significant (p = 0.29) after adjusting for age, education, perceived adequacy of income, work status, and total event score using the multivariable ANCOVA regression model.

Discussion This is the first study designed to evaluate the reliability and validity of the Arabic PSS-10 scale. This Arabic translation of the PSS-10 was found to have reasonably adequate psychometric properties. Among university students, the reliability coefficient was similar to that found by Cohen and Williamson [ 13 ]. For pregnant and postpartum women the estimates fell within the range (.71-.82) of previously reported values in other studies on similar groups [ 19 , 22 , 30 ]. The overall test retest reliability of .74 was acceptable and comparable to that obtained for the Spanish version of PSS-10 [ 31 ]. However, reliability was higher among the students as compared to the other two groups possibly due to the difference in time to retest. Results of the factor analysis were similar to those obtained from studies in Turkey [ 15 ] and the US [ 12 ] where the same two factors were detected. Örücü and Demir [ 15 ] had described these two factors as "perceived helplessness" and "perceived self efficacy". As hypothesized, the Arabic PSS-10 had positive associations with two previously validated scales: GHQ-12 and EPDS. There was also a significant positive correlation between PSS-10 and life events scale but that correlation was of a lower magnitude. This might be due to differences in the way people cope with life events, thus people with stressful life events and good coping strategies might have a higher score on life event scale however a lower score on PSS-10. Further, studies on relating stress to life events are recommended taking into account the coping strategies of people. The mean PSS-10 score of the sub-sample of students in this study was similar to what is reported in Arab female university students (22.7 ± 7) by Hamdan-Mansour and Dawani [ 17 ]. Contrary to our expectations, scores of university students on PSS-10 were higher than those of pregnant or postpartum women. One possible reason is that the students' sample unexpectedly experienced significantly higher stressful life events than the other two groups. Further studies for assessing the sensitivity of PSS-10 to distinguish between different stress levels among postpartum and pregnant women need to be pursued. Study Limitations This study was conducted among specific groups of women, the use of PSS-10 in studies about women of different groups or men should also try to address internal consistency and factor structure in those populations. Moreover, women were recruited from three clinics and one private university, and therefore our sample may not be representative of all pregnant, postpartum, and university women. This study was not able to establish the ability of PSS-10 to distinguish between different stress levels and further investigation in that direction is recommended. One other limitation of the study was that retesting among pregnant and postpartum women was done after two to three weeks as compared to one week for students. Events might have happened thereby affecting the scores in the second interview and thus lowering the reliability.

Conclusions Stress is a risk factor for several chronic diseases including hypertension, diabetes, and coronary artery disease. The ability to measure stress reliably would be useful to further characterize the link between stress and health. More importantly it would help evaluate interventions that may decrease stress levels. PSS-10 can be used to estimate perceived stress. It is short and can be easily administered to women coming to seek medical help. The availability of a valid PSS-10 scale in different languages allows for comparisons across studies from different countries and cultures that ultimately help in understanding reactions to stress and its determinants. The translation of PSS-10 into classical Arabic makes this a useful tool for researchers conducting studies that address stress in a variety of Arabic-speaking communities.

Background This study was conducted to evaluate the validity of the Arabic translation of the Cohen Perceived Stress Scale (PSS-10) in pregnant and postpartum women. Methods A sample of 268 women participated. These included 113 women in their third trimester of pregnancy, 97 in the postpartum period and 58 healthy female university students. GHQ-12 and EPDS were also administered to the participants. Internal consistency reliability, assessed using Cronbach's α, was 0.74. Results PSS-10 significantly correlated with both EPDS and GHQ12 (ρ = 0.58 and ρ = 0.48 respectively), and significantly increased with higher scores on stressful life events. PSS-10 scores were higher among university students who also recorded higher stressful life events scores. Conclusion The Arabic translated version of the PSS-10 showed reasonably adequate psychometric properties.

Competing interests The authors declare that they have no competing interests. Authors' contributions MC, HO, and GN, contributed to conceptualization, and write up. ZM and MC, contributed to data collection, analysis and write up. All authors have read and approved the manuscript being submitted. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-244X/10/111/prepub Supplementary Material

Acknowledgements This study was sponsored by the Center for Research on Population and Health at the American University of Beirut, Lebanon, with generous support from the Wellcome Trust.

CC BY

no

2022-01-12 15:21:45

BMC Psychiatry. 2010 Dec 15; 10:111

oa_package/ac/27/PMC3016315.tar.gz

PMC3016316

21176203

Background ADHD is a common, inherited and disabling developmental disorder with early onset. Most often ADHD persists across the life span, affecting 2-4% of adults [ 1 ]. The core symptoms of ADHD are inattention, hyperactivity and impulsivity. Further, deficits in executive functioning are commonplace, such as planning, organising, exerting self-control, working memory, and affect regulation. Therefore, ADHD affects educational and occupational performances, psychological functioning, and social skills. Adults with ADHD are at increased risk for unemployment, sick leave, coexisting disorders, abuse, and antisocial behaviour leading to conviction [ 2 , 3 ]. Nearly 80% of adults with ADHD present with at least one coexisting psychiatric disorder [ 3 , 4 ]. Further, studies display ADHD to be common among prison inmates [ 5 - 9 ]. However, little attention has been paid to profiles of ADHD symptoms and executive functions of prison inmates compared with other groups affected by ADHD, and to controls [ 10 ]. Besides, effects of pharmacological treatment for ADHD among prison inmates remain unexplored. The clinical presentation has shown to change with age, as hyperactivity declines, whereas inattention and executive dysfunction persist, thus representing the core features of adult ADHD [ 11 , 12 ]. However, most previous studies have excluded prison inmates, questioning how relevant these findings are to prison inmates. To gain some more information, we evaluated ADHD and criminality. The first aim of this study was to estimate the prevalence of ADHD among longer-term inmates of a high-security Swedish prison. The second aim was to describe ADHD, coexisting disorders, and executive functions among prison inmates. The final aim was to compare these findings with ADHD psychiatric outpatients and healthy controls. We hypothesized that ADHD would be common among this group comprising mainly longer-term prison inmates, typically convicted of crimes because of violence and drugs. Also, we hypothesized that they would present more severe ADHD symptoms across the lifespan, more common coexisting psychiatric disorders, and poorer executive functions compared with the other groups.

Methods The present study included an estimation of the prevalence of ADHD among longer-term prison inmates. Further, it included a description of ADHD and executive functions among prison inmates compared with ADHD among psychiatric outpatients and healthy controls. The Regional Ethical Board in Stockholm approved the studies. Participants provided written informed consents before study procedures. Participants Norrtälje Prison is a high-security prison placed outside Stockholm, Sweden, serving the entire country, hosting 200 adult male inmates. The prison holds mainly longer-term inmates, typically convicted of crimes because of drugs or violence. Figure 1 shows the study flowchart. Norrtälje Prison hosted 589 inmates between December 2006 and April 2009. Of those inmates, we did not invite 200 for screening, as we could not include them in the following trial because of deportation out of the country after served conviction. Further, we did not approach 74 inmates because of practical reasons, or if we considered them as too mentally affected to take part. Thus, a specially trained correction officer successively approached 315 prison inmates for screening during the study period. Another purpose of screening was to identify subjects for a diagnostic evaluation for ADHD before recruitment for a clinical trial. Therefore, we ended recruitment as we had randomised all 30 subjects for the trial in April 2009. Following the screening survey, we performed extensive diagnostic assessments for ADHD and coexisting disorders among a group of inmates. We selected subjects first according to their origin, as the Stockholm County Council funded the assessments as part of regular clinical practice. Thus, we invited all prison inmates marking adult ADHD by the screening, registered in the Stockholm County, with at least 14 months left to conditional release, and approved by the security officers to stay at the ADHD ward. By this pre-screening, we evaluated if subjects with ADHD would fulfil criteria for taking part in the following clinical trial with methylphenidate (Ginsberg and Lindefors, unpublished data). Subjects with coexisting disorders, such as ASD, anxiety and depression could take part if considered stable by the investigator at the assessment. Further, the general cognitive functioning had to be above the level of mental retardation. In addition, subjects could continue stable pharmacological treatment for coexisting disorders if we did not suspect treatment interfering with methylphenidate. Additionally, subjects had to be free from serious medical illnesses. Thus, after meeting criteria for the following trial and providing a written informed consent, the subject could take part in the diagnostic evaluation. We considered 47 prison inmates for assessment. However, we excluded one subject because of an exclusion criterion, whereas six subjects denied taking part. Of 40 consented subjects, six dropped out during the assessments. Therefore, we finally assessed 34 subjects and could confirm ADHD among 30 of them (Figure 1 ). When appropriate, we extended the evaluation to confirm ASD in consistence with DSM-IV. We defined ASD as fulfilling the criteria for Autistic syndrome, Asperger syndrome or Pervasive developmental disorder, not otherwise specified (PDD-NOS). This evaluation included the Asperger Syndrome Screening Questionnaire (ASSQ) [ 13 ], the Diagnostic Interview for Social and Communication Disorders (DISCO) [ 14 , 15 ], and the Autism Diagnostic Observation Schedule (ADOS), module 4 [ 16 ]. The psychiatric outpatient study group comprised 20 adult men with ADHD, 18 of them with ADHD of the combined type, and two with the predominantly inattentive subtype. We consecutively recruited these subjects to another study [ 17 ] between 2004 and 2006, from the Neuropsychiatric Unit, Karolinska University Hospital; a psychiatric outpatient tertiary unit specialised in ADHD. Notably, the exclusion criteria for taking part were different among psychiatric outpatients, as ongoing pharmacological treatment for coexisting disorders, APD, ASD, 70 > IQ < 85, or pure 'sluggish, inattentive' ADHD [ 18 , 19 ] excluded. Because of different criteria, we expected a difference in IQ between groups. Thus, we controlled for IQ in the statistical analyses of executive functions. The control group [ 17 ] comprised 18 adult healthy males not needing psychiatric care, assessment for learning difficulties or educational support during childhood. Further, they did not need psychiatric care during the present study. We recruited age-matched controls from advertisement on fitness training centres in Stockholm City and among friends of staff-members. Procedures Estimation of ADHD prevalence among longer-term prison inmates WURS is a 61-item self-administered scale for rating frequencies of ADHD childhood symptoms and behaviours retrospectively on a 5-point scale, from 0 = not at all or slightly , to 4 = very much . The subscale WURS-25 provides a total sum score (range 0-100) by summing those 25 items best discriminating between ADHD and controls [ 20 ]. According to the originators, a cut-off score of 36 is 96% sensitive and specific for identifying childhood ADHD among the general population [ 20 ]. The ASRS-Screener comprises the 6 out of 18 most predictive items of the Adult ADHD Self-Report Scale (ASRS) [ 21 ] for defining present ADHD in adulthood. Fulfilling at least 4 out of 6 significant items [ 22 ] on ASRS-Screener defines adult ADHD. Both scales are standard tools in clinical practice, despite the lack of Swedish validations. In this study, we defined adult ADHD as reaching the cut-off levels for WURS-25 and ASRS-Screener, respectively. Assessment for ADHD among prison inmates Board certified psychiatrists and clinical psychologists well experienced in ADHD, conducted the clinical assessments. We confirmed ADHD in accordance to DSM-IV [ 23 ]. The evaluations included a semi-structured clinical diagnostic interview for ADHD based on the DSM-IV-criteria [ 23 ]. Further, ASRS [ 24 ] is an 18-item self-administered scale with appropriate psychometric properties [ 25 ] based on the DSM-IV criteria and adjusted to reflect ADHD symptoms as seen in adults [ 22 ]. We used a non-validated Swedish version of the ASRS [ 24 ] for rating symptom frequencies on a 5-point scale, from 0 = never ; to 4 = very often , providing a total sum score (range 0-72). Whenever possible, we collected collateral information from parents or other significant others by questionnaires, before psychologists or psychiatrists performed interviews. The questionnaires included the Five to Fifteen (FTF) questionnaire [ 26 , 27 ] and the Conners' Brief Parent Rating Scale - Conners' Hyperactivity Index [ 28 , 29 ], respectively. The Five to Fifteen (FTF) questionnaire [ 26 , 27 ] elicits childhood symptoms and developmental problems of ADHD and coexisting disorders in the ages five to fifteen years. The FTF shows acceptable to excellent inter-rater and test-retest reliability and comprises 181 items scored on a 3-point scale, from 0 = does not apply , to 2 = definitely applies . The Conners' Brief Parent Rating Scale - Conners' Hyperactivity Index is validated in several countries. This scale describes ADHD and oppositional defiant symptoms and behaviours in children up to 10 years of age [ 28 ], comprises 10 items, scored 0-3, and provides a total sum score (0-30). We collected additional collateral information by medical records from child- and adolescent psychiatry, school health services, adult psychiatry and forensic psychiatry. Further, we evaluated coexisting disorders by the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID I) [ 30 ], the Hare Psychopathy Check List-Revised (PCL-R), a semi-structured interview defining psychopathy by a total sum-score ≥ 30 [ 31 ], and the self-rated version of the Structured Clinical Interview for DSM-IV Axis II personality disorders, the SCID II Patient Questionnaire (SCID II PQ). We estimated frequencies of personality disorders by increasing the screening cut-off level for each personality disorder by one score. This procedure has shown an acceptable agreement with the SCID II interview [ 32 ]. Furthermore, the evaluation comprised a medical history, physical examination, routine laboratory tests, urine drug screening and a neuropsychological test battery assessing IQ and executive functions. As prison inmates often present learning disabilities such as reading difficulties [ 9 ], we assessed neuropsychological tests not requiring reading, writing or mathematic skills. We estimated IQ by the Wechsler Adult Intelligence Scale-III subtests Vocabulary and Block Design, a dyadic short form correlating 0.92 with WAIS-III FSIQ [ 33 , 34 ]. Neuropsychological tests of executive functions Digit Span [ 33 ] measures verbal working memory (WM) whereas Span Board [ 35 ] measures visuospatial WM. Further, we measured sustained attention, impulse inhibition and other executive functions by the computerized The Conners' Continuous Performance Test II (CCPT) [ 36 ]. The CCPT measure Hit RT reflects basic reaction time , whereas Hit RT SE, Variability, Hit RT block change, Hit SE block change, Hit RT ISI change, Hit SE ISI change and Perseverations reflect variability dependent measures . Finally, Omission errors, Commission errors, Detectability (d'), and Response style (â) reflect accuracy dependent measures . Assessment for ADHD among psychiatric outpatients The diagnostic evaluation comprising neuropsychological tests was similar as among prison inmates. However, we did not assess SCID I, SCID II PQ, or PCL-R among ADHD psychiatric outpatients. Case files provided information on psychiatric comorbidity. Besides, the self-rated Beck Depression Inventory [ 37 , 38 ], the Beck Anxiety Inventory [ 39 ], and the Current ADHD Symptom Scale - Self-Report Form [ 40 ], evaluated present psychiatric symptoms. Healthy controls We interviewed controls for confirming the absence of learning difficulties or psychiatric problems during childhood and the study, respectively. Further, we used the same self-rating scales for present psychiatric symptoms as among the psychiatric outpatients. Finally, the neuropsychological tests were similar as for the other groups. Statistical analysis Descriptive statistics summarised demographic data and clinical characteristics of subjects. We carried out inferential statistics by analyses of variance (ANOVA), Student's t-test or Mann-Whitney U -test for continuous measures, and chi-square test or Fisher's exact test for categorical measures. Further, for comparing between groups on neuropsychological measures, we performed a series of analysis of variance (ANOVA) with Bonferroni corrected post hoc comparisons, whenever main analyses reached significance. In addition, we aimed to control for IQ differences. Thus, we reanalysed measures of executive functions (DS, SB, and CCPT) by performing a series of ANCOVA with the dyadic estimated IQ entered as a covariate. By these analyses, we evaluated if lower IQ among prison inmates could explain their executive dysfunctions. We present statistics from both ANOVAs and ANCOVAs, as most measures of executive functions did not co-vary with IQ. We set the alpha-level at p = .05. Finally, we performed all statistical analyses by SPSS 17.0 and 18.0, respectively.

Results ADHD prevalence Figure 1 presents a flowchart of the study. As calculated from this figure, the total response rate was 62% (194/315). We defined adult ADHD as reaching the cut-off levels for both childhood and adult ADHD. By this procedure, we increased the specificity of the screening survey. When applying our predefinition of adult ADHD, the prevalence rate was 45%, as 88 out of 194 subjects fulfilled this definition (Figure 1 ). Overall, responders were slightly older and served longer convictions compared with non-responders (Table 1 ). However, when we assessed 34 subjects marking ADHD by the screening, we confirmed ADHD among 30 of them. Thus, the screening survey pointed out to be 88% (30/34) specific. Therefore, we imply a more conservative 40% ADHD prevalence (0.88 × 45) among longer-term prison inmates. Clinical characteristics of ADHD among adult male prison inmates This study included an extensive diagnostic evaluation of ADHD and coexisting disorders among a group of prison inmates (Figure 1 ). Table 2 shows the clinical characteristics of those 30 subjects confirmed with ADHD. As shown, almost all subjects confirmed ADHD of the combined type. Further, all subjects presented coexisting disorders. In fact, all 30 subjects presented a lifetime history of SUD, with amphetamine as the most preferred drug among almost two thirds. In general, the subjects showed an early onset of abuse and antisocial behaviour. In addition, lifetime mood and anxiety disorders were obvious among a vast majority and treated among almost half of subjects at the assessment. Besides, almost one fourth confirmed ASD, much more common than we expected. On the other hand, psychopathy was present among only one tenth, which was less than we expected. Further, personality disorders were present among 96% (22/23) of subjects. Among personality disorders, antisocial, borderline, paranoid, narcissistic, or obsessive-compulsive personality disorder were most obvious. Further, there was a striking finding of this study; despite most subjects reported prior need of health services and educational support at school, few received a diagnosis of ADHD during childhood. In summary, prison inmates showed severe symptoms and severities from ADHD, SUD, ASD, personality disorders, mood- and anxiety disorders. Comparisons between ADHD prison inmates, ADHD psychiatric outpatients, and healthy controls As depicted in Table 2 , all three groups were of similar age. Notably, 83% of ADHD prison inmates fulfilled nine-year of compulsory school or less, compared with 30% among ADHD psychiatric outpatients, and 6% among healthy controls, thus reflecting a remarkably lower educational level among prison inmates. Standardised questionnaires The ADHD-prison group rated more ADHD related symptoms and behaviours during both childhood and adulthood, compared with the ADHD-psychiatry group (Table 3 ). By contrast, when parents retrospectively rated childhood symptoms and behaviours, differences between groups were negligible, which we did not expect. Table 3 presents statistics and Figure 2 presents mean values (+/- 2 SE), respectively. Neuropsychological tests The dyadic estimation of IQ displayed similar IQ for controls and the ADHD-psychiatry group; (Controls, n = 18, M = 112 (± 9.65), range 97 - 132); (ADHD-psychiatry, n = 20, M = 108.25 (±11.48), range 89 - 132). On the other hand, IQ was substantially lower among ADHD prison inmates; (M = 95.18 (± 9.99), range 78 - 113). The ADHD-prison group (n = 22) had missing data for eight subjects. We expected significant differences between groups on estimated IQ ( F = 14.76, p < .001, η p 2 = .341) because of different inclusion criteria. In fact, only the ADHD-prison group included subjects with IQ between 70 and 85. As a result, 10% (3/30) of prison inmates presented estimated dyadic IQ within this range, specifically between 78 and 85. Therefore, we excluded those three inmates with IQ < 85 for making inclusion criteria homogenous. However, the ADHD-prison group still showed lower estimated IQ after performing this procedure, compared with both other groups ( F = 10.49, p < .001, η p 2 = .28). Neuropsychological tests of executive functions The ADHD-prison group showed poorer results on several measures of executive functions compared with both other groups, also when controlling for IQ (Table 4 ). On measures of working memory, controls outperformed the ADHD-psychiatry group on both verbal (DS) and visuo-spatial working memory (SB). On the other hand, the ADHD-psychiatry group outperformed the ADHD-prison group on the same measures. However, when controlling for IQ, the differences in working memory between ADHD groups no longer remained, but controls still outperformed both ADHD groups. Thus, both working memory tests showed executive dysfunctions associated with ADHD, also when controlling for IQ. On the Conners' Continuous Performance Test II (CCPT), controls and the ADHD-psychiatry group showed similar results. However, at least one of the other groups outperformed the ADHD-prison group on all four accuracy dependent measures , and in three out of seven variability dependent measures , respectively. On the other hand, there were no significant differences in reaction time between groups (Table 4 and Figure 3 ). Notably, 5 out of 27 (18.5%) subjects among the ADHD-prison group showed remarkably increased values (T-score >200) on Perseverations, a measure considered to reflect flexibility. Therefore, we performed analyses both including and excluding subjects with extreme values. However, we observed similar results on Perseverations also when excluding those subjects, thus implying decreased flexibility among prison inmates with ADHD. Further, estimated IQ did not explain the CCPT results in this study (Table 4 ).

Discussion The present study included an estimation of ADHD prevalence among adult male longer-term prison inmates from a high-security Swedish prison. Further, we evaluated ADHD and executive functions among prison inmates and then compared results with ADHD psychiatric outpatients and healthy controls. We estimated a prevalence rate as high as 40% among these prison inmates. Further, those inmates we later confirmed with ADHD were severely affected and disabled from ADHD and coexisting disorders, such as SUD, ASD, personality disorders, mood- and anxiety disorders. Previous studies reported increased frequencies of major mental disorders, personality disorders, and early adjustment problems among prison inmates, regardless of ADHD [ 41 ]. The present study confirms these observations. In addition, educational level and executive functions were poorer among ADHD inmates compared with ADHD psychiatric outpatients and controls. These findings remained after controlling for IQ. Thus, our findings imply prison inmates with ADHD to present a severely affected group of ADHD. Although ADHD is common among prison inmates, prevalence rates are inconsistent, probably because of different used criteria among different prison populations [ 5 - 9 ]. Further, symptoms of ADHD, such as hyperactivity and impulsivity have shown to decline by age, whereas inattention and executive dysfunction continue [ 12 ]. Besides, most prevalence studies on male prison inmates have been conducted among younger inmates [ 8 ]. Further, knowledge is sparse on clinical features and executive functions among adult male prison inmates confirmed with ADHD [ 6 - 10 ] compared with adult ADHD among other groups and controls. To our best knowledge, this study is the first to report a screening survey for ADHD, followed by extensive evaluations of ADHD and coexisting disorders among adult male longer-term prison inmates. The evaluations incorporated both self-reports and confirming collateral information from parents, medical records and school reports. Additionally, evaluations included a physical examination and neuropsychological assessments. Further, we compared ADHD prison inmates with ADHD psychiatric outpatients and controls for ADHD symptom load, coexisting disorders and executive functions. Prevalence of ADHD among prison inmates As hypothesized, ADHD was prevalent among these adult male longer-term prison inmates with a median age of 31 years. We estimated the prevalence as high as 40%, compared with previous findings by Rösler et al [ 8 ] who reported a prevalence of 45%, though among younger inmates (mean age 19). Thus, our results suggest ADHD to be comparably present among older and younger inmates. Our finding contradicts the common view of ADHD to decline by age. Thus, this symptom reduction by age might not held true for ADHD prison inmates. Further, the total survey response rate was 62%, which we view as acceptable, considering a common mistrust against authorities among prison inmates. However, we have to consider the attrition rate and its impact on the results. We imply that we not exaggerated the ADHD prevalence, as we did not approach inmates who we considered too psychiatric affected to take part. In some of these cases, ADHD might contribute to their psychiatric symptoms. On the other hand, we can not exclude some selection bias at the end of the study period when the study was more commonly known in the Swedish prison and probation service. It might be that some inmates recognised themselves as having ADHD and therefore applied for serving conviction at Norrtälje Prison in hope for treatment. However, as we screened the majority at the beginning of the study period, we imply this potential bias to be of minor importance. In summary, when considering the specificity of the screening procedure, we suggest a 40% ADHD prevalence rate among adult male longer-term inmates from a high-security prison. Clinical characteristics of ADHD This study only partially supported our hypothesis that ADHD prison inmates would present more severe ADHD symptoms across the lifespan, compared with ADHD psychiatric outpatients. The ADHD-prison group reported more ADHD symptoms and behaviours during both childhood and adulthood. However, collateral information from parents on childhood symptoms did not reveal any differences between groups. As a result, subjects rated more childhood symptoms retrospectively compared with parental ratings. This observation contradicts previous findings by Barkley [ 42 ] who displayed adults with ADHD to underreport their symptoms compared with parents. Thus, when considering the negative trajectory of these prison inmates and continuing ADHD symptoms, you would predict symptoms to be obvious during childhood, consistent with self-reports. Further, most subjects reported previous need of health services and educational support during childhood, pointing to obvious difficulties, although not recognised as ADHD. Notably, prison inmates showed a remarkably lower educational level compared with both other groups. Lower IQ levels among these inmates might partially explain these findings. Further, executive dysfunctions may contribute to lower school attendances and performances. In fact, we expect educational underachievement among ADHD also with normal IQ [ 43 ]. Besides, more hindering symptoms from ADHD and coexisting learning disabilities, including dyslexia and externalising symptoms such as ODD and CD, possibly contribute to poorer educational achievements and early dropouts from school. Another explanation might be prison inmates exaggerating their symptoms in hope for methylphenidate treatment. However, parents of both ADHD groups rated similarly on Conners' Hyperactivity Index. This index reflects externalising symptoms besides ADHD, which is notable considering the negative trajectory of our ADHD-prison group. Therefore, self-reported childhood symptoms by prison inmates seem more in line with their negative trajectories across time. Further, symptoms of substance abuse, depression and anxiety could mimic ADHD. However, our inmates were kept from drugs for more than three months, in some cases for years. Further, all coexisting disorders were stable and treated at the assessment, thus implying present symptoms to be ADHD related. To summarise, our findings imply the importance of recognising ADHD early and offering effective treatment immediately. Prospective studies should evaluate if treatment will reduce the risk for serious outcomes. Coexisting disorders As hypothesized, coexisting disorders were common among our prison inmates. In fact, all subjects reported a lifetime history of SUD, with amphetamine as the most preferred drug of choice. Besides, abuse and antisocial behaviour had an early onset, consistent with previous findings [ 44 ]. Additionally, anxiety disorders and depression were common, and half of inmates received treatment at the assessment. Further, all but one subject displayed CD before APD. Notably, psychopathy was present among only one tenth, which was fewer than we expected, as all but one subject displayed APD. However, previous studies reported that most psychopaths fulfil the criteria for APD, whereas the opposite is true for only a minority of inmates. These findings signal that psychopathy would be a more homogeneous disorder than APD [ 31 ]. In addition, Soderstrom used a 3-factor model of PCL-R among forensic subjects for distinguishing psychopathy traits and evaluating if certain traits reflected ADHD [ 45 ]. By this model, he showed that total PCL-R scores, as well as Factor 2 (unemotionality) and Factor 3 (behavioural dyscontrol), reflected ADHD. However, Factor 1 defining exaggerated self-opinion towards others and dishonesty did not reflect ADHD. In fact, the literature considers these interpersonal traits of Factor 1 to be most specific of psychopathy. Besides, we confirmed ASD among almost one fourth of ADHD prison inmates, mainly PDD-NOS. We are not aware of any previous reports estimating the prevalence of ASD among prison inmates. However, Anckarsater [ 46 ] showed that ASD was more common among forensic subjects than among the general population. In that study [ 46 ], PDD-NOS presented the most common ASD, paralleling our findings. In summary, we suggest that ASD is common also among prison inmates. However, studies comprising larger samples need to confirm these preliminary findings. If ASD is common among prison inmates, we need to consider this for successfully meeting the specific needs of these inmates. Previous studies reported that personality disorders are common among different ADHD populations, such as prison inmates [ 9 ]. Recently, Rydén et al observed that personality disorders were common among adults with "pure" ADHD, ADHD combined with bipolar disorder, and bipolar disorder only, although most prevalent among "pure" ADHD (Rydén E, and collaborators, personal communication). For defining personality disorders, they used the same procedure as in the present study. By comparing those, "pure" ADHD with our ADHD prison inmates, most personality disorders implied more common among inmates. However, histrionic, depressive, and schizoid personality disorder implied more common among "pure" ADHD subjects (Rydén E, and collaborators, personal communication). Cognitive abilities The present study supported our hypothesis that ADHD prison inmates would present poorer cognitive abilities compared with ADHD psychiatric outpatients and healthy controls. As expected, the ADHD-prison group showed lower estimated IQ. However, different inclusion criteria could not explain the observed IQ differences between groups, as differences remained when excluding prison inmates with IQ < 85. As presented, both ADHD groups displayed poorer executive functions compared with controls, also when adjusting for IQ. Working memory functions were similar between ADHD groups when adjusting for IQ. Considering the CCPT results overall, controls and the ADHD-psychiatry group showed similar results. Further, at least one of them outperformed ADHD prison inmates on all accuracy dependent measures , and on several variability dependent measures , respectively. On the other hand, reaction time was comparable between groups, thus implying slow reaction time not to be a concern among adult ADHD. Summarising, these findings are in line with theories of ADHD as an executive disorder [ 47 ]. In addition, these findings parallel recent reports by Wood et al [ 48 , 49 ] who suggested that lower IQ does not account for the key cognitive problems noted among ADHD. Further, one striking notion of the present study, was the increased levels of Perseverations on the CCPT, which reflects difficulties in holding back or adjusting non-proper behaviours. Previous studies reported response perseveration among ADHD subjects suffering from CD [ 50 - 52 ], as well as among pathological gamblers [ 53 ]. However, researchers interpreted response perseveration among ADHD in different ways. Quay [ 50 , 51 ] suggested increased thriving for rewards among CD, because of a more actively working behavioural activation (reward) system (BAS) compared with the behavioural inhibition system (BIS). Reverse, he suggested less active BIS compared with BAS among ADHD. Beauchaine interpreted the opposite way [ 54 ], as he suggested less active BAS among CD, resulting in a reward-seeking behaviour as a stimulation seeking. Finally, Seguin [ 55 ], Newman and Wallace [ 56 , 57 ] respectively, suggested deficits in attending for peripheral information, which usually directs the subject changing for a more effective behaviour. Therefore, future studies should explore the cognitive underpinnings of response perseveration, as they remain elusive. Limitations We have to consider several limitations of this study. As the attrition rate of the screening survey was 38%, we must interpret the results with caution. However, we imply that we not exaggerated the prevalence rate of ADHD. Further, both rating scales used for screening lack Swedish validations. Nevertheless, these scales are used as standard tools in clinical practice. Besides, this study included only male longer-term prison inmates, why results can not extend to female inmates or to inmates serving shorter-term convicts. Further, there might have been selection bias when recruiting for the screening survey, mainly at the end of the study period when the study was commonly known in the Swedish Prison and Probation service. It might be that some inmates recognised themselves as suffering from ADHD and therefore applied for serving conviction at Norrtälje Prison in hope for treatment. However, as we screened the majority at the beginning of the study period, we imply this potential bias as minor important. Further, there might have been selection bias because of different inclusion criteria between groups. Therefore, it implies a selection of subjects among ADHD psychiatric outpatients, functioning better than average, as treatment for coexisting disorders excluded for the present study. Actually, in clinical practice, adults with ADHD often receive treatment for common coexisting disorders. On the other hand, the ADHD-psychiatry group may better reflect ADHD among the general population, as considered presenting less severe symptoms and severities compared with psychiatric outpatients. Additionally, there may have been a selection bias when recruiting prison inmates for diagnostic assessments. We noticed a few prison inmates denied taking part in the study in lack of motivation for changing their behaviour, or resistance to stay at the ADHD ward. The ward was apart from other wards for reducing the risk of exposing inmates to illicit drugs. As a result, study subjects received less time for physical exercise and restricted access to some prison programmes, as long as they stayed at the ADHD ward. Therefore, a selection bias towards more motivated prison inmates could have been present. If so, the bias probably worked towards better performances and higher functioning, than the reverse. Finally, the study samples were small. However, results were statistically significant despite small sample sizes. Notably, the strength of this study was the extensive clinical description of ADHD, coexisting disorders and executive functioning among prison inmates, as well as comparisons with ADHD psychiatric outpatients and controls. The extensive diagnostic evaluation included self-reported information, collection of collateral information, physical examination, structured diagnostic interviews and neuropsychological assessments. To our best knowledge, such extensive evaluations of longer-term prison inmates have not previously been reported. We infer our reported findings of ADHD symptom severity, coexisting disorders and executive functioning among prison inmates, are clinical important and relevant. We need to consider these severities when adjusting existing, or designing new ADHD treatment programmes for prison inmates. Further, these extensive evaluations might provide helpful insight for addressing future research on ADHD endophenotypes. Knowledge on endophenotypes may promote individually tailored treatments by identifying who will benefit from treatment. Finally, we will report effects of methylphenidate treatment among these ADHD prison inmates in another paper (Ginsberg and Lindefors, unpublished data).

Conclusions This study suggested ADHD to be present among 40% of adult male longer-term prison inmates. Diagnostic evaluations for ADHD among 30 inmates showed them severely disabled from ADHD and coexisting disorders, such as SUD, ASD, personality disorders, mood- and anxiety disorders. Further, these ADHD prison inmates displayed poorer executive functions, also when controlling for estimated IQ, compared with ADHD psychiatric outpatients and healthy controls. We infer the reported findings of ADHD symptom severity, coexisting disorders and executive functioning among prison inmates, are clinical important and relevant. These findings imply the need for considering these severities when introducing ADHD treatment programmes for prison inmates.

Background ADHD is a common and disabling disorder, with an increased risk for coexisting disorders, substance abuse and delinquency. In the present study, we aimed at exploring ADHD and criminality. We estimated the prevalence of ADHD among longer-term prison inmates, described symptoms and cognitive functioning, and compared findings with ADHD among psychiatric outpatients and healthy controls. Methods At Norrtälje Prison, we approached 315 male inmates for screening of childhood ADHD by the Wender Utah Rating Scale (WURS-25) and for present ADHD by the Adult ADHD Self-Report Screener (ASRS-Screener). The response rate was 62%. Further, we assessed 34 inmates for ADHD and coexisting disorders. Finally, we compared findings with 20 adult males with ADHD, assessed at a psychiatric outpatient clinic and 18 healthy controls. Results The estimated prevalence of adult ADHD among longer-term inmates was 40%. Only 2 out of 30 prison inmates confirmed with ADHD had received a diagnosis of ADHD during childhood, despite most needed health services and educational support. All subjects reported lifetime substance use disorder (SUD) where amphetamine was the most common drug. Mood and anxiety disorders were present among half of subjects; autism spectrum disorder (ASD) among one fourth and psychopathy among one tenth. Personality disorders were common; almost all inmates presented conduct disorder (CD) before antisocial personality disorder (APD). Prison inmates reported more ADHD symptoms during both childhood and adulthood, compared with ADHD psychiatric outpatients. Further, analysis of executive functions after controlling for IQ showed both ADHD groups performed poorer than controls on working memory tests. Besides, on a continuous performance test, the ADHD prison group displayed poorer results compared with both other groups. Conclusions This study suggested ADHD to be present among 40% of adult male longer-term prison inmates. Further, ADHD and coexisting disorders, such as SUD, ASD, personality disorders, mood- and anxiety disorders, severely affected prison inmates with ADHD. Besides, inmates showed poorer executive functions also when controlling for estimated IQ compared with ADHD among psychiatric outpatients and controls. Our findings imply the need for considering these severities when designing treatment programmes for prison inmates with ADHD.

Competing interests YG has been on the speaker's bureau and consultant for Janssen-Cilag, Novartis and Lundbeck A/S. YG has been the principal investigator of two clinical trials sponsored by Janssen-Cilag. NL has been the investigator of a clinical trial sponsored by Janssen-Cilag. TH declares no conflicts of interest. Authors' contributions YG designed the study in collaboration with NL and TH, applied to the Ethical Board in collaboration with NL, prepared the Case Report Forms, conducted clinical assessments, collected data, planned and executed the analyses, interpreted the results in collaboration with TH and NL, and prepared all drafts of the manuscript in collaboration with TH. NL revised the manuscripts critically. TH was responsible for assessments and analyses of the psychiatric outpatients and controls. All authors read, commented on and approved the final manuscript. Authors' information 1 Department of Clinical Neuroscience, Division of Psychiatry, Karolinska Institutet, Stockholm, Sweden, 2 Karolinska Institutet Center of Neurodevelopmental Disorders, Stockholm, Sweden, 3 Department of Molecular Medicine and Surgery, Center for Molecular Medicine, Karolinska Institutet, Stockholm, Sweden . Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-244X/10/112/prepub

Acknowledgements The Swedish Ministry of Health and Social Affairs, and Stockholm County Council, Sweden financially supported this study. The funding sources were not involved in the authors' work. We are grateful to all participants and collaborators from Stockholm County Council and the Swedish Prison and Probation Service who made this work possible. We especially thank Gunnar Johansson at Norrtälje Prison for invaluable help in administering the screening survey, Monica Hellberg for administrative assistance, and co-investigators Michaela Wallensteen, Ann-Charlotte Wiklund, Maria Kristensen, Agneta Ljungberg, Anna Eriksson, Julia Alfredsson, Else Waaler, Pernilla Bothén, and Annelie Holmström for providing data. We thank Martin Grann for valuable comments on the manuscript.

CC BY

no

2022-01-12 15:21:45

BMC Psychiatry. 2010 Dec 22; 10:112

oa_package/97/b9/PMC3016316.tar.gz

PMC3016317

21245927

Introduction Evidence indicates that early nutrition, growth and subsequent health are crucially related. Several studies have demonstrated that early life growth patterns exert programming effects on disease risk in later life, highlighting the key role played by early nutrition [1] , [2] . There is still debate as to when the sensitive periods of early development occur, during which the “programming” takes place [3] and the relative contribution of intrauterine and postnatal growth to subsequent health outcomes needs further clarification. Body composition, in terms of fat mass (FM), may contribute to this “programming” process [4] . Small size at birth as well as rapid catch-up growth during infancy has been associated with an increased risk for developing the metabolic syndrome in adulthood [5] , [6] . It has been recently suggested that relative adiposity, which is a well known risk factor for cardiovascular disease [7] , may develop due to under-nutrition as well as growth retardation [8] . Preterm infants are at increased risk for developing insulin resistance due to the stressful conditions and the cumulative nutritional deficits they experience during early postnatal life. As a consequence, hyperinsulinaemia and a down-regulation of visceral β3-adrenoreceptors may lead to increased intra-abdominal adiposity [8] . Preterm small for gestational age (SGA) infants assessed at term corrected age have been reported to be at risk for developing increased adiposity [9] . In addition, abnormal body composition and altered insulin sensitivity have been found in SGA infants who experienced rapid postnatal weight gain [10] , [11] , [12] . Data on the dynamic features of body weight gain and FM gain that occur during the first months of life in SGA preterm infants is scarce. Therefore, the aim of the present study was to investigate the postnatal weight gain and FM accretion during the first five months of corrected age in a cohort of preterm infants who were categorized according to intrauterine growth pattern and according to postnatal growth.

Materials and Methods Ethics statements The study was approved by the Departmental Ethics Committee, Fondazione IRCCS “Ca' Granda” Ospedale Maggiore Policlinico, and written consent was obtained from both parents. Patients Two hundred and seven preterm infants among all consecutive newborns admitted to the same Institution from January 2007 to June 2009 were enrolled in the study. Inclusion criteria were: birth weight <1500 g, singleton pregnancy. Exclusion criteria were: presence of congenital diseases, chromosomal abnormalities, chronic lung disease (defined by the use of supplemental oxygen at 36 weeks' postconceptional age), severe brain, metabolic, cardiac or gastrointestinal diseases (i.e. necrotizing enterocolitis classified as stage 3 according to the classification of Bell et al. [13] ), being breastfed after term. The reason for choosing relatively strict eligibility criteria relied on the fact that we wanted to investigate the growth and body composition in subgroups of preterm infants not affected by illnesses that could interfere with the variables investigated. Design We conducted a prospective, observational study. Basic subject characteristics (birth weight, length, head circumference, gestational age, gender, being adequate [AGA] or small for gestational age [SGA]) were recorded. Anthropometric parameters (weight, length and head circumference) and fat mass were assessed at term and at 1, 3 and 5 months of corrected age. Gestational age was based on the last menstrual period and first trimester ultrasonogram. Corrected age was calculated using the chronologic age and adjusting for gestational age, that is, for the number of additional weeks from term (40 weeks). Infants with birth weight ≥ or <10 th percentile for gestational age, according to the Fenton's chart [14] , were classified as AGA or SGA, respectively. Growth and fat mass measurements Body weight, length and head circumference were measured according to standard procedures [15] . Subject mass was measured on an electronic scale accurate to the nearest 0.1 g and body length was measured to the nearest 1 mm on a Harpenden neonatometer (Holtain Ltd, UK). Head circumference was measured to the nearest 1 mm with non stretch measuring tape. Growth z-scores were calculated by EuroGrowth 2000 software (Euro-Growth Study Group, Vienna, Austria). Infants with weight ≥or <2 SD at term were classified as infants being non-growth retarded (GR−) or growth retarded (GR+), respectively. The change in weight [100 × (weight at second examination – weight at first examination)/weight at first examination)] between birth and term, between term and 3 months of corrected age and between 3 and 5 months of corrected age were then calculated. FM was assessed using an air displacement plethysmography system (PEA POD Infant Body Composition System, LMI, Concord, CA, USA). A detailed description of the PEA POD's physical design, operating principles, validation and measurement procedures is provided elsewhere [16] , [17] . The PEA POD assesses FM and fat free mass by direct measurements of body mass and volume and the application of classic densitometric principles. Infants were measured in the PEA POD naked. Each PEA POD test took about 3 min to complete. Subject volume was measured in an enclosed chamber by applying gas laws that relate pressure changes to volumes of air in the chamber. Body density was then computed from the measured body mass and volume, and inserted into a standard formula for estimating the percentage of total body FM according to a 2-compartment model. The intra-observer coefficient of variation for the percentage of FM estimates was 0.3%. The change in FM [100 × (FM at second examination−FM at first examination)/FM at first examination)] between term and 3 months of corrected age and between 3 and 5 months of corrected age were then calculated. Nutritional Practices Preterm infants received parenteral and minimal enteral feeding, with expressed breast milk or preterm formula, for a minimum of two weeks. Subsequently, the nutritional regimen up to discharge was either fortified breast milk (2.2 g/100 ml and 82 Kcal/100ml) or preterm formula (2.4 g/100 ml and 80 Kcal/100 ml) when breast milk was unavailable or insufficient. From term up to the fifth month, infants were fed a nutrient-enriched postdischarge formula (protein 2 g/100 ml; energy 75 kcal/100 ml) on demand and were given no other foods. At discharge, parents were instructed to record the daily quantities of milk consumed by the infants in a diary. The average daily energy and protein intakes were then calculated. Statistical analysis Descriptive data are expressed as mean (SD) or number of observations (percentage). Differences among infants in repeated measurements of growth parameters and FM were assessed by an analysis of variance. Significance of multiple comparisons was adjusted by the Bonferroni correction. A χ2 test was used for comparisons between discrete variables. For analysis, infants were categorized as born AGA without growth retardation at term (GR−), born AGA with growth retardation at term (GR+), born SGA. Statistical significance was set at α = 0.05 level. All statistical analyses were performed using SPSS (SPSS, version 12, SPSS Inc., Chicago, IL).

Results Growth and body composition were assessed in 195 (96 males) infants. There was no infant mortality throughout the follow-up. Out of the 207 infants originally recruited, 4 moved away or returned to their country of origin; 8 failed to attend the scheduled appointments. Mean gestational age (weeks) and birth weight (g) were 30.2 (2.3) and 1190 (284). Basic subject characteristics are shown in table 1 . Birth weight was significantly lower in the infants born SGA when compared to the infants born AGA (GR−) and (GR+) whereas gestational age was significantly higher in infants born SGA. Growth At term, and at 3 and 5 months of corrected age, the mean z-score for weight was significantly lower in SGA infants when compared to AGA (GR−) and AGA (GR+) infants ( figure 1 ). The mean weight (g) differences (95% Confidence Interval) between SGA and AGA (GR−) infants at term, 3 and 5 months of corrected age were −837 (−999;−674), −792 (−1125;−459), −613 (−1215; −12), respectively. The mean weight (g) differences (95% Confidence Interval) between SGA and AGA (GR+) infants at term, 3 and 5 months of corrected age were −237 (−392;−83), −368 (−678;−60), −573 (−1227; −79), respectively. The AGA (GR+) infants showed the mean z-score for weight significantly lower at term and at 3 months of corrected age as compared to the AGA (GR−) infants whereas no difference was found at 5 months ( figure 1 ). The mean weight (g) differences (95% Confidence Interval) between AGA (GR+) and AGA (GR−) infants at term and 3 months of corrected age were −599 (−769;−429), −423 (−768;−78), respectively. The mean z-score for length in SGA infants was significantly lower from term to the third month when compared to AGA (GR+) and (GR−) infants ( table 2 ). The mean length (cm) differences (95% Confidence Interval) between SGA and AGA (GR−) infants at term and 3 months of corrected age were −3.6 (−4.5;−2.7), −2.1 (−3;−1), respectively. The mean length (cm) differences (95% Confidence Interval) between SGA and AGA (GR+) infants at term and 3 months of corrected age were −1.7 (−2.6;−0.8), −1.6 (−2.7;−0.5), respectively. No difference in the mean z-score for length was found between groups at 5 months. Mean z-score for head circumference was significantly lower at each study point in SGA infants when compared to AGA (GR+) and AGA (GR−) infants ( table 2 ). The mean head circumference (cm) difference (95% Confidence Interval) between SGA and AGA (GR−) infants at term, 3 and 5 months of corrected age were −1.78 (−2.8;−0.7), −2.2 (−2.27; −0.4), −1.5 (−2.1; −0.7), respectively. The mean head circumference (cm) differences (95% Confidence Interval) between SGA and AGA (GR+) infants at 3 and 5 months of corrected age were −1.2 (−1.8;−0.3), −0.9 (−1.3; −0.5), respectively. Mean z-score for head circumference was significantly lower in AGA (GR+) infants when compared to AGA (GR−) infants at term and at 3 months.. The mean head circumference (cm) differences (95% Confidence Interval) between AGA (GR+) and AGA (GR−) infants at term and at 3 months were −1.6 (−2.7;−0.5), −0.8 (−1.1; −0.2), respectively. However, there was no difference between the two groups at 5 months. In Table 3 the mean changes in weight and FM between each study point according to group categorization are shown. The mean difference (95% Confidence Interval) in weight change between birth and term corrected age was significantly higher in AGA (GR−) infants when compared to AGA (GR+) [35.5 (7.6; 63)] and SGA [36 (9.5; 63)] infants. On the contrary, the mean difference (95% Confidence Interval) in weight change between term and 3 months of corrected age was significantly lower in AGA (GR−) infants when compared to AGA (GR+) [−24.6 (−37; −12)] and SGA [−27.9 (−40; −15.6)] infants. No difference between the AGA (GR−) and AGA (GR+) groups was detected between 3 and 5 months of corrected age whereas the SGA infants showed a significantly higher mean difference (95% Confidence Interval) in weight change than the AGA (GR−) [4.4 (0.8; 8)] and the AGA (GR+) infants [4.1 (0.2; 7.9)]. The mean change difference (95% Confidence Interval) in FM between term and 3 months of corrected age was significantly lower in AGA (GR−) infants when compared to AGA (GR+) [−37.8 (−65; −10)] and SGA infants [−38.6 (−65; −12)] ( table 3 ), whereas no differences between groups were detected between 3 and 5 months. Figure 2 shows the mean % FM at each study point according to categorization. At term % FM was significantly higher in AGA (GR+) infants when compared to AGA (GR−) and SGA infants, whereas no differences between groups were detected at 3 and 5 months. No significant differences in energy and protein intakes between groups were found during the study period ( table 4 ).

Discussion This study investigates longitudinally, the postnatal weight gain and weight gain composition during the first months of corrected age in a cohort of preterm infants who were classified according to their intrauterine growth pattern and according to their postnatal growth. In the present study, SGA infants showed the lowest mean z-score for weight at term in comparison to AGA (GR+ ) and (GR−) infants. Both the impaired intrauterine growth and the cumulative postnatal nutritional deficit explains this finding. As a consequence, infants who were born SGA, although exhibiting an increased growth rate between term and the fifth month, attained mean z-score values for weight that were persistently lower than that attained by infants born AGA (GR+) and (GR−). These results are consistent with previous studies which reported that SGA preterm infants experience a severe extra uterine growth failure [18] – [21] . Bertino et al. [19] have recently reported that being born preterm and small for gestational age exert negative effects on growth assessed at term and at 24 months of corrected age. Jordan et al. [20] observed that at 36 months, SGA infants remained lighter, shorter and had smaller head circumference values than AGA infants even if the postnatal growth rate of SGA infants was higher than that of AGA infants. Moreover, Hack et al. [21] demonstrated that SGA very low birth weight males do not catch up in growth by 20 years of age. In contrast, infants born AGA (GR+), who also showed an increased growth rate between term and the fifth month, successfully achieved similar mean z-score values for weight within the fifth month of corrected age when compared to AGA (GR−) infants. In addition, AGA (GR+) infants also recovered in terms of length and head circumference within the third month of corrected age. These results confirm previous findings reported by our group [22] related to a smaller sample of (GR+) and (GR−) AGA preterm infants. In contrast with our findings, Latan et al. [23] reported postnatal growth retardation in a group of AGA very low birth weight infants, resulting in weight that was below the 10th percentile at two years of age. A possible explanation is that infants having medical complications (i.e. bronchopulmonary dysplasia, severe intraventricular hemorrhage) that could negatively affect postnatal growth after discharge, were enrolled in the latter study. On the contrary, only premature infants without medical complications, that could interfere with subsequent growth, were included in our study. Our results suggest that the potential to rapidly correct anthropometry observed in the AGA (GR+) infants when compare to AGA (GR−) infants may reflect the influence of fetal programming, implying that the trajectory of growth may not be permanently affected by the development of postnatal growth restriction. In the case of AGA (GR+) infants, the lack of impaired intrauterine growth may allow these infants to recover from their postnatal growth restriction. In contrast, the persistence of postnatal GR in the SGA infants may suggest that either these infants have an intrinsic lower growth potential or that the growth constraint experienced during the intrauterine life may delay the occurrence of recovery of growth. Although SGA infants achieved mean z scores for growth parameters above −2 z-scores within the fifth month, they did not attain values similar to that of AGA infants in terms of weight and head circumference. With respect to body composition, both SGA and AGA (GR+) infants showed % FM at term significantly lower than that of AGA (GR−) infants, suggesting that the postnatal GR is accompanied by a relative lack of FM accretion. Nevertheless, the mean FM value presented at term by all the infants enrolled in the study, regardless of categorization, was much higher than that found in full term neonates at birth [24] . The finding of an increased adiposity in preterm infants assessed at term corrected age is consistent with previous reports [9] , [25] . The increased amount of fat accretion has been linked to the energy intake [26] and could also be partially dependent on the several differences between fetal nutrition and postnatal nutrition [27] . The higher fat deposition could also represent an adaptive mechanism to postnatal life, for example, to augment body energy stores and ameliorate thermoregulation [28] . Surprisingly, from term to the third month of corrected age, both SGA and AGA (GR+) infants showed a higher change in FM than the AGA (GR−) infants, so that no difference in percentage of FM between groups was detected at three months. From the third month up to the end of the study, SGA, AGA (GR+) and (GR−) infants showed comparable change in FM. Moreover, the mean FM values attained at three and five months by all infants, regardless of group categorization, were comparable to those of full term breastfed infants [29] . To our knowledge there is a paucity of data related to body composition changes that occur over the first months of life in SGA infants. According to our results, Beltrand et al. [30] reported the restoration of body size and fat stores within the fourth month of age in fetal growth restricted full term infants without detrimental consequences at one year of age on body composition or metabolic profile. Ibanez et al. [31] reported that full term SGA children, who developed a spontaneous catch up growth, at 2 years of age, showed similar body composition when compared to full term AGA infants. However, the authors found a striking shift towards visceral adiposity in full term SGA children between 2 and 4 years of age and a further increase in central adiposity between 4 and 6 years [12] . On the contrary, Willemsen et al. [32] reported a decreased percentage of total body FM in former preterm, short SGA children assessed at 6.8 years of age in comparison to AGA children but a similar body fat distribution both in the SGA and AGA children, suggesting a trend for SGA children towards the development of central adiposity. Meas et al. [33] described a fast progression of adiposity from 22 up to 30 years of age in adults born full term SGA resulting in a higher percentage of total body FM than in subjects born full term AGA. These findings indicate that being born SGA may affect body composition in different ways according to the period of development. In the present study, body composition in preterm infants was assessed in early infancy. Whereas the rapid recovery of FM exhibited by AGA (GR+) infants suggests that the absence of impaired intrauterine growth may allow these infants to recover from their postnatal lack of FM accretion, the accelerated gain in FM experienced by the SGA infants may partly be due to the fact that subjects who have been exposed to impaired fetal growth may be susceptible to gain more fat. The “FM catch up” may represent a compensatory event associated with the degree of impairment of fetal growth [34] . In addition, as energy intake has been advocated as a major determinant in FM accretion [26] , the rapid advances in FM from term to the fifth month in the SGA infants could be explained by the cumulative effect on FM gain caused by the slightly higher energy intakes throughout the study. SGA infants showed slightly higher energy and protein intakes, even though the energy and protein intakes were not significantly different between the groups of infants at each study point. On the contrary, the accumulation and/or the aberrant distribution of fat mass reported in children and adults born SGA may reflect the long-term fetal programming of adipose tissue alterations, both in terms of quantity and functions [35] , and may contribute to the increased risk of developing the metabolic syndrome later in life [36] . The dynamic changes in adiposity that occur during postnatal catch up growth seem to play a critical role in the development of metabolic complications [36] . However, the exact timing of these changes that contributes to the increased later disease risk is still under debate. Ezzahir et al. [34] suggested that the effect of catch-up in body mass index on adiposity in adulthood is mostly detrimental in children born SGA when occurring after 1 y of age. The strength of the present study relies on the fact that it is a longitudinal study conducted in a relatively large cohort of preterm infants, make observing the changes in growth and body composition more accurate. However, as we aimed to investigate the growth and FM gain according to intrauterine growth pattern and according to postnatal growth, data was analyzed at each study point between groups. The limitation of the study is that the length of the follow- up was relatively short. The present study provides preliminary evidence on growth and weight gain composition of preterm infants according to intrauterine growth pattern and according to postnatal growth. Our data suggests that fetal growth pattern influences the potential to rapidly correct anthropometry whereas the restoration of fat stores takes place irrespective of birth weight. Long term follow- up studies are needed to elucidate the metabolic consequences of these findings.

Conceived and designed the experiments: PR FM. Performed the experiments: NL FT OA LM PP. Analyzed the data: MLG AO. Contributed reagents/materials/analysis tools: AO. Wrote the paper: MLG. Responsible for writing the report: PR. Directed the clinical measurement team: PP. Contributed to the revision: MA. Supervised leading aspects: FM. Background Preterm small for gestational age (SGA) infants may be at risk for increased adiposity, especially when experiencing rapid postnatal weight gain. Data on the dynamic features of body weight and fat mass (FM) gain that occurs early in life is scarce. We investigated the postnatal weight and FM gain during the first five months after term in a cohort of preterm infants. Methodology/Principal Findings Changes in growth parameters and FM were prospectively monitored in 195 infants with birth weight ≤1500 g. The infants were categorized as born adequate for gestational age (AGA) without growth retardation at term (GR−), born AGA with growth retardation at term (GR+), born SGA. Weight and FM were assessed by an air displacement plethysmography system. At five months, weight z-score was comparable between the AGA (GR+) and the AGA (GR−), whereas the SGA showed a significantly lower weight.The mean weight (g) differences (95% CI) between SGA and AGA (GR−) and between SGA and AGA (GR+) infants at 5 months were −613 (−1215; −12) and −573 (−1227; −79), respectively. At term, the AGA (GR+) and the SGA groups showed a significantly lower FM than the AGA (GR−) group. In the first three months, change in FM was comparable between the AGA (GR+) and the SGA groups and significantly higher than that of the AGA (GR−) group.The mean difference (95% CI) in FM change between SGA and AGA (GR−) and between AGA (GR+) and AGA (GR−) from term to 3 months were 38.6 (12; 65); and 37.7 (10; 65). At three months, the FM was similar in all groups. Conclusions Our data suggests that fetal growth pattern influences the potential to rapidly correct anthropometry whereas the restoration of fat stores takes place irrespective of birth weight. The metabolic consequences of these findings need to be elucidated.

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e14489

oa_package/5f/4d/PMC3016317.tar.gz

PMC3016318

21245928

Introduction Rising concerns regarding Influenza A (H5N1) and the pandemic of Influenza A (H1N1) 2009 have led to the use of infrared thermal image scanners (ITIS) at some borders for the mass screening of travellers to detect those who might be infected with influenza [1] . ITIS measure body surface temperature rapidly, non-invasively, and with no contact, minimising the risk of contagion. They therefore have the potential to comply with the International Health Regulations' emphasis on containing the spread of disease in ways that avoid unnecessary interference with international traffic and trade [2] . Evaluations of the use of ITIS in clinical settings have been conducted, and have reported sensitivities of 15% to 90% for confirmed fever depending on the cut-off used to define fever [3] , [4] , [5] , [6] . However, these findings may not be applicable to border screening. ITIS measure body surface temperature, not body core temperature, and so ITIS temperature measurements are subject to the influence of a range of human and environmental factors. These include whether a person is sunburnt, has taken antipyretics or has circulatory problems, and also the ambient temperature and humidity. Consequently it is important that the relationship between body surface temperature and body core temperature be evaluated within the environment in which ITIS are to be operated. In the airport setting, thermal scanning of arriving travellers has been used to screen for several different infectious diseases. During the outbreak of Severe Acute Respiratory Syndrome (SARS) ITIS use was documented, however only the numbers of travellers triggering the scanner were reported, without stating the cut-off threshold used for fever or reporting on any subsequent method used to confirm febrile status [7] , [8] , [9] . A trial dengue fever screening programme found that among travellers arriving into Cairns airport [10] 12% (118/963) of travellers who triggered the pre-set alarm threshold were confirmed to be febrile on tympanic temperature measurement. Influenza screening in Singapore found that only 12% of cases of pandemic (H1N1) 2009 infection with onset within 10 days of arrival were detected by ITIS on entry [11] . Proper evaluation of a screening test requires that the ‘gold standard’ test is applied to both test positive and test negative participants in the study. To evaluate the use of ITIS in border screening for influenza, its performance in predicting both fever and also influenza infection is necessary. However to date no studies have been reported that tested ITIS negative travellers for either fever or influenza infection [12] , [13] . We undertook both ITIS and tympanic temperature measurement on, and collected specimens for testing for influenza from, symptomatic and asymptomatic air travellers arriving into Christchurch, New Zealand during the Southern hemisphere winter in 2008. This paper assesses the performance of ITIS in detection of fever and infection with seasonal influenza in these airline travellers.

Methods Study design This evaluation of thermal image scanning was carried out as part of a larger study to measure the prevalence of seasonal influenza infection in arriving airline travellers and the effectiveness of a screening questionnaire for detecting those with influenza infection. The design followed closely a pilot study carried out in 2007 [14] . Participants Three airlines agreed to have their staff distribute a screening questionnaire to travellers (passengers and crew) during flights travelling from Australian airports to Christchurch, New Zealand. The questionnaires were collected by research assistants following immigration processing on arrival in Christchurch. ‘Symptomatic’ travellers were defined as those who reported one or more of the following symptoms: cough, sore throat, sneezing, fever or chills, runny or blocked nose, muscle aches or pains, feeling generally unwell, chest discomfort or breathing difficulties. Measures Symptomatic travellers were all invited to have throat and nose swabs (Copan Italia SPA, Brescia, Italy) taken and their temperature measured. In addition, half the questionnaires were marked and were randomly placed into the sets of questionnaires delivered to the flight crew (the sequence was determined by the RAND function of Microsoft Excel©). Arriving travellers carrying a marked questionnaire were also invited to have swabs and temperature taken. The nurse taking the swabs noted on the request form whether the traveller was symptomatic or asymptomatic. For the 23 working days from 21 August to 12 September 2008, cutaneous temperature from those travellers invited to participate who had given consent was measured using ITIS (ThermaCAMTM E45, FLIR Systems, Sweden) prior to swabs being taken. A focal plane array (160×120 pixels) was used on the front of the face and the side of the face (see Figure 1 ) and the maximum temperature reading for each was recorded. After the swabs were taken, each participant's tympanic temperature was measured using an infrared tympanic thermometer (ThermaScan PRO4000, BRAUN, Germany). The ambient temperature in the arrivals hall was a consistent 20.5°C at all times during data collection. Laboratory Analysis All nasal and throat swab samples were analysed at Canterbury Health Laboratories, Christchurch. A multiplexed tandem polymerase chain reaction (MT-PCR) assay was employed to detect the presence of influenza A and B virus infection, as described by the manufacturer (Easy-Plex Influenza A+B kit, Cat. No. 3005.01, AusDiagnostics Pty Ltd, Sydney, Australia). Data Analysis Stata© 10 was used to analyse the data. The cii command was used to calculate Poisson exact confidence intervals around the proportion of influenza-infected travellers who were febrile. Information about temperature measurements was collected on the swab consent form and linked to the symptom information on the questionnaire using a unique swab identifier. Nine swab results were unable to be linked as their identifier had not been attached to any questionnaire. For these individuals the nurse's note of whether or not they were symptomatic was used to define their symptom status. Analyses were performed to assess the accuracy of ITIS measurements in predicting two different tympanic temperature thresholds: tympanic temperature ≥37.8°C (>100°F – the level used by the Centers for Disease Control in defining ‘influenza-like illness’) [15] tympanic temperature ≥37.5°C (the threshold used in the majority of reports) [12] . Firstly, a Receiver Operating Characteristic (ROC) curve [16] was constructed. ROC curves assess the ability of a test (in this case the ITIS measure) to discriminate between people who have, and who do not have, a condition (fever). The area under the ROC curve for an uninformative test is 0.5. Secondly, a level of ITIS temperature with sensitivity closest to 85% was chosen and the specificity calculated. Finally, the positive predictive value (PPV) of the chosen level of ITIS temperature was estimated. The positive predictive value is the proportion of people who test positive (i.e. are ‘positive’ on ITIS) who actually have the condition of interest. It is not appropriate to calculate the PPV of ITIS measures directly in this sample since it was not a random sample of the population of travellers but was instead ‘enriched’ by including as many symptomatic travellers as were prepared to provide respiratory samples. Therefore, the prevalence of fever (by each definition) in the holders of marked questionnaires was combined with the sensitivity and specificity of ITIS for detecting fever to estimate the PPV in the population of all travellers who arrived on the flights that took part in the study. To assess the utility of fever as a screening test for influenza infection (MT-PCR result), sensitivity, specificity, and population PPV for influenza were estimated for each tympanic temperature threshold, and the ITIS threshold used above. Ethics This study was approved by the New Zealand Health and Disability Multiregion Ethics Committee. Written informed consent was obtained from all participants.

Results Participants In total, 5274 travellers returned a questionnaire during the study period, of whom 823 (15.6%) were symptomatic by our definition. Figure 2 shows the pathway of potential participants through the study. Accuracy of thermal scanning in predicting core temperature Seven participants had a tympanic temperature of ≥37.8°C (2 reported no symptoms and 5 were symptomatic). Five held marked questionnaires, giving a prevalence of fever by this definition of 0.5% (5/1063). Half of the 38 participants with a tympanic temperature of ≥37.5°C were symptomatic. Thirty-two of them held a marked questionnaire, so the prevalence of fever by this definition was 3.0% (32/1063). Figure 3 is a ROC curve showing the ability of ITIS front of face measurement to predict a tympanic temperature of ≥37.8°C. Table 1 shows the test characteristics of ITIS as a predictor of tympanic temperature. For each definition of ‘fever’ (determined by tympanic temperature measurement), and for each site of ITIS measurement (front and side of face), the table shows: the ITIS threshold that gave a sensitivity closest to 85% in our data; the proportion of travellers with an ITIS measure above that threshold (i.e. who would have ‘triggered’ the ITIS during screening); the area under the ROC curve; the actual sensitivity of that threshold; and its specificity. The prevalence of fever at each threshold in holders of marked questionnaires (as an estimate of the prevalence in this population of arriving travellers) is also shown, as well as the estimated PPV of ITIS for fever in this population. Temperature as a predictor of influenza infection Of the 1275 respiratory samples obtained from participating travellers, 30 were positive for influenza (3 Type A and 27 Type B), while 7 samples were invalid as they contained no human nucleic acid. The prevalence of influenza infection in holders of marked questionnaires with valid samples was 1.9% (20/1057). Most (90%; 27/30) influenza-positive participants were symptomatic, but none (0%) had a measured tympanic temperature of ≥37.8°C (99%CI 0% to 18%), and only two (7%) had a measured tympanic temperature of ≥37.5°C(99%CI 0.3% to 31%). Table 2 shows the ability of tympanic and ITIS temperatures to predict influenza infection in a population where the prevalence is 2% (the estimate of the prevalence of infection in this population of arriving travellers). With high sensitivity, specificity is very low. Combined with the low prevalence of influenza infection in this population, PPV is also very low. Influenza-positive participants reported that the first of their symptoms started between 12 hours and 24 days prior to answering the questionnaire, with symptom duration of 2 days or less in 11 participants, more than 2 and up to 5 days in 7 participants, and more than 5 days in 8 participants (3 were asymptomatic and 1 did not respond to this question).

Discussion The greatest potential for the use of ITIS to screen incoming or departing travellers for infectious diseases such as a pandemic strain of influenza would be as the first stage of screening; that is, to identify and select out a high risk group for further assessment, for example by questionnaire, body core temperature measurement, and/or respiratory sample collection. This would require very high sensitivity for raised body temperature, as any travellers who ‘slipped through’ the screening process would enter the community and potentially spread infection. In addition, core temperature would need to be a good predictor of infection. Can thermal scanning predict core temperature? This study shows that, among a group comprising both asymptomatic and symptomatic arriving international airline travellers, ITIS can have moderately high sensitivity and specificity for a high body core temperature of ≥37.8°C. However, the low prevalence of fever in arriving travellers means that the PPV is very low. Does temperature predict infection? Measurement of the sensitivity of fever for influenza infection requires that afebrile as well as febrile people, from the same population, are tested for influenza infection. There are few studies that have done this, as symptoms of ‘influenza-like illness’, which include fever, are usually criteria for entry to studies of influenza [17] , [18] , [19] . Such studies, with selected participants with a high prevalence of influenza infection, overestimate the sensitivity and dramatically overestimate the PPV of fever for influenza infection in unselected populations, such as airline travellers. A review of volunteer challenge studies [20] showed that not only were approximately 30% of influenza infections asymptomatic, but only 35% of those with symptoms had a measured fever >37.8°C. This study found a lower prevalence of fever among the participants infected with Influenza B (7/101) than with Influenza A H1N1(88/285; 31%) [20] . In this study, none of the 30 travellers subsequently identified as infected with influenza (most of whom had influenza B) had a temperature ≥37.8°C, and only two had a temperature ≥37.5°C. A tympanic temperature threshold of 36.2°C would be required to identify a high proportion (87%; table 2 ) of influenza-infected travellers. The ITIS temperature measures have better specificity than this (non-febrile) level of tympanic temperature for identifying influenza-infected travellers, but PPVs are all low at less than 3%. The ROC result for influenza infection shows that ITIS on their own are not much better than chance at identifying influenza-infected travellers. These results emphasise what is already known about fever as a symptom of influenza – while it clearly is one of the symptoms that can be experienced by people with influenza infection, it does not occur in all infected people [20] . The prevalence of fever is high in case series of patients with confirmed influenza infection [11] , [21] , since often one of the criteria that is often used to determine whether testing takes place is the presence of fever. However, where fever is not used as a criterion for influenza testing, the prevalence of fever is by no means 100%, even among people with severe symptoms. For example, among 106 patients hospitalised with respiratory disease [22] , 39% of those with confirmed pandemic (H1N1) 2009 infection did not have a temperature of ≥37.8°C at any time during admission. In this study, the predominance of Influenza B infection may partly explain the low prevalence of fever among infected participants (although the three with Influenza A all had tympanic temperatures <37.2°C ).Even with more pyrexigenic strains, among travellers, who by definition are not severely unwell and in fact who are mostly not unwell at all, the proportion of influenza infected people who are afebrile can be expected to be much higher than among hospitalised patients [22] (because the sicker infected people don't travel), as shown in this study. Limitations of this study It was a condition of conducting this study that we did not delay the transit of passengers through the airport by more than a few minutes and, therefore, measurements had to be made efficiently. We used a single measurement by an infrared tympanic thermometer as our ‘gold standard’ measure of core temperature. This approach may have introduced some random error into our results, but is unlikely to have caused systematic bias and is likely to be similar to the way that temperature would be confirmed in practice. In addition, our participants sat still at approximately 1m from the scanner for the ITIS measure and those who were wearing glasses were asked to remove them, steps likely to have provided greater accuracy than ITIS measures that are taken as numerous people walk past a fixed scanner in an arrivals hall. Therefore our study provides an assessment of the best results that could be expected from the use of ITIS in border screening for influenza. In this study, no influenza-infected travellers had a measured tympanic temperature ≥37.8°C. We do not believe that this was because of systematic errors in tympanic temperature measurements, as these were measured by trained nurses using standard thermometers. We acknowledge that the number of infected travellers was relatively small at 30 but the probability is only 0.005 (0.5%) that the prevalence of fever among the population of infected travellers arriving from Australia into Christchurch at this time was greater than 18%; in other words the vast majority of infected travellers in this population were afebrile. Among travellers, the proportion of influenza cases who are febrile may be low because those infected with influenza that is causing fever may feel too unwell to travel; 25% of travel-associated cases of pandemic (H1N1) 2009 infection with onset in Singapore were symptomatic on embarkation but the proportion who were febrile was not reported [11] . In addition, it is possible that unwell infected travellers had used anti-pyretics prior to or during the flight, but this is a limitation of ITIS rather than of our study. The study assessed the performance of ITIS in the real world, which includes the fact that some unwell people take anti-pyretics. Also, the flights that were part of this study were relatively short – 3 to 4 hours – and it is possible that on longer flights some of the infected travellers might have become febrile. However, it remains unlikely that fever would occur in all, or even most, infected travellers arriving at any international airport [23] . Good evidence on influenza virus transmissibility during the various phases of viral infection, (including afebrile infection and asymptomatic infection) is not available, but detection of viral RNA on a respiratory sample does not necessarily mean that the infected person is, or will be, infectious. We were not able to perform culture for influenza virus in this study, so it is possible that some of the infected travellers were not shedding viable virus. Although the approximately one third of participants whose symptoms were of 2 days' duration or less were likely to be in the early stages of their infection, those with longer duration of symptoms may not have been. Unfortunately the symptoms of influenza are so non-specific that it is difficult to estimate the stage of influenza infection in a traveller with, for example, a cough that has been present for several weeks. Nonetheless, it seems reasonable to conclude that at least a third, and probably more, of the infected (and afebrile) participants in this study were infectious on or after arrival into New Zealand. Implications Influenza-infected arriving travellers include those who are symptomatic (with or without fever), those who become symptomatic during the flight, those who will develop symptoms following arrival, and those who will never have symptoms. It is not known whether the latter group are infectious, but clearly only the first two categories could potentially be detected by entry screening. Most people who were infected but asymptomatic on boarding will still be asymptomatic on arrival at their destination [23] . However, in the absence of effective exit screening during the H1N1 2009 pandemic, some countries decided to use ITIS in entry screening with the hope that detecting travellers who were febrile on arrival would be worthwhile to reduce the probability of infected travellers entering the country, and that ITIS could detect them [1] . This study provides evidence to the contrary. The low PPV of ITIS measures for fever in this population means that the number of false positives who would require further investigation, presumably by taking a tympanic temperature, would be very high. In this study, using a front of face ITIS threshold of 35.4°C identifies 69% of travellers as requiring further investigation, of whom only 4.1% had a tympanic temperature ≥37.5°C. The PPV of any of the measures of temperature for influenza infection itself was lower, at less than 3%. However, the prevalence of disease is an important determinant of PPV, and the prevalence of influenza infection in this study, performed during the ‘influenza season’, was low at 1.9%. There are no other published estimates of the prevalence of influenza in arriving travellers, but it could be argued that the prevalence of infection would be higher during a pandemic, which typically infects a higher proportion of the population than seasonal influenza, than in this study. On the other hand, particularly if local containment strategies were in place in originating countries, the prevalence of infection in travellers might be lower during a pandemic. At the beginning of a pandemic, when effective entry screening would be most useful, the prevalence of infection among travellers and therefore the PPV will likely be much lower than the prevalence of seasonal influenza in this study. More importantly, raised temperature itself by any measurement technology is insufficiently sensitive for influenza infection for its measurement to be effective for mass screening in a pandemic situation. Use of ITIS to identify travellers at high risk of fever, measuring the core temperature of ITIS-positive travellers, and then taking specimens from those with high core temperatures would have failed to identify all the influenza-infected travellers in this study. Using a lower temperature threshold (however measured) for taking specimens could detect a high proportion of influenza-infected travellers only by taking specimens from what is likely to be an unfeasibly high proportion of travellers. Governments may decide to implement entry screening, including ITIS, for reasons other than to actually detect most influenza-infected arrivals, for example to deter unwell people from travelling, or to demonstrate to their citizens that they are doing everything they can to protect population health. The risks associated with this approach include the potentially very large opportunity cost of further investigating ITIS ‘positive’ travellers, including quarantine of those febrile on tympanic temperature measurement pending specimen processing, and the potential for the loss of public confidence in the pandemic response when it becomes clear that many infected travellers were not detected by the screening and entered the country. Conclusion In this study, during a seasonal epidemic of predominantly influenza type B, influenza-infected arriving travellers had a very low prevalence of fever. Consequently, ITIS would not have identified influenza-infected travellers even though it performed moderately well at detecting febrile travellers. Some aspects of this study may not generalise to a pandemic of Influenza A. Although febrile illness is more common in influenza A infections than influenza B infections, many influenza A infections are afebrile. Our findings therefore suggest that ITIS is unlikely to be effective for entry screening of travellers to detect influenza infection with the intention of preventing entry of the virus into a country.

Conclusion In this study, during a seasonal epidemic of predominantly influenza type B, influenza-infected arriving travellers had a very low prevalence of fever. Consequently, ITIS would not have identified influenza-infected travellers even though it performed moderately well at detecting febrile travellers. Some aspects of this study may not generalise to a pandemic of Influenza A. Although febrile illness is more common in influenza A infections than influenza B infections, many influenza A infections are afebrile. Our findings therefore suggest that ITIS is unlikely to be effective for entry screening of travellers to detect influenza infection with the intention of preventing entry of the virus into a country.

Conceived and designed the experiments: PCP ARD LCJ MB. Performed the experiments: ARD LCJ. Analyzed the data: PCP. Wrote the paper: PCP ARD LCJ MB. Background Infrared thermal image scanners (ITIS) appear an attractive option for the mass screening of travellers for influenza, but there are no published data on their performance in airports. Methods ITIS was used to measure cutaneous temperature in 1275 airline travellers who had agreed to tympanic temperature measurement and respiratory sampling. The prediction by ITIS of tympanic temperature (37.8°C and 37.5°C) and of influenza infection was assessed using Receiver Operating Characteristic (ROC) curves and estimated sensitivity, specificity and positive predictive value (PPV). Findings Using front of face ITIS for prediction of tympanic temperature ≥37.8°C, the area under the ROC curve was 0.86 (95%CI 0.75–0.97) and setting sensitivity at 86% gave specificity of 71%. The PPV in this population of travellers, of whom 0.5% were febrile using this definition, was 1.5%. We identified influenza virus infection in 30 travellers (3 Type A and 27 Type B). For ITIS prediction of influenza infection the area under the ROC curve was 0.66 (0.56–0.75), a sensitivity of 87% gave specificity of 39%, and PPV of 2.8%. None of the 30 influenza-positive travellers had tympanic temperature ≥37.8°C at screening (95%CI 0% to 12%); three had no influenza symptoms. Conclusion ITIS performed moderately well in detecting fever but in this study, during a seasonal epidemic of predominantly influenza type B, the proportion of influenza-infected travellers who were febrile was low and ITIS were not much better than chance at identifying travellers likely to be influenza-infected. Although febrile illness is more common in influenza A infections than influenza B infections, many influenza A infections are afebrile. Our findings therefore suggest that ITIS is unlikely to be effective for entry screening of travellers to detect influenza infection with the intention of preventing entry of the virus into a country.

We thank the Health Emergency Management Branch, Department of Health and Ageing, Australia for lending us the scanner; Christchurch International Airport Limited, New Zealand Customs Service, and the participating airlines for their cooperation and assistance; and Andrew Strathdee for the laboratory testing. We are grateful to Elisabeth Wells, Heath Kelly, Jonathan Van-Tam, and Rebecca Psutka for their helpful comments on drafts of this article, and to the anonymous reviewers for their helpful comments on the originally submitted version.

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e14490

oa_package/47/75/PMC3016318.tar.gz

PMC3016319

21245929

Introduction The management of patients suffering from acute radiation syndromes (ARS) still remains a major challenge. Survival of radiation induced bone marrow failure depends on the dose of radiation received and the intensity of supportive care which can protect from otherwise lethal infection and give surviving stem cells a chance to expand. To provide the best possible care for radiation accident victims in acts of terrorism or catastrophic incidences, medical countermeasures need to be made within the first few days for optimal efficacy [1] . The “response category concept” proposed by Fliedner et al [2] evaluates the radiation induced tissue damage and rates the hematopoietic score 4-H4 (the highest score for hematopoietic damage) as case with little probability for autologous recovery. Combined approaches including presenting symptoms, biomarkers and physical dosimetry are employed to categorize affected individuals for best medical countermeasures [3] . Overall measures include supportive care, treatment with growth factors within the first two weeks after radiation exposure, or hematopoietic stem cell transplantation (HSCT). Since radiation effects on blood stem cells occur at doses generally lower than on other critical organs, the rapidly emerging changes in the peripheral blood cell lineages dictate the treatment options. Animal and human studies indicate that hematopoietic pluripotent stem cells have a D 0 of about 95cG as indicated by Fliedner et al [1] . D 0 is the dose increment that reduces the cell survival to 37%. In fact, total body exposure at doses more than 7–8 Gy total body irradiation (TBI) in human corresponds to medullar eradication. Under this threshold spontaneous recovery from residual hematopoietic stem and progenitor cells may be expected within 30–50 days but going through cytopenic phases of granulocytic, megakaryocytic and erythrocytic lineages. HSCT should be considered if the victim's HSC pool is essentially irreversibly damaged. Interestingly, even after TBI, intrinsically radioresistant stem cells have been detected in distinct bone marrow (BM) areas comprising a residual hematopoietic stem and progenitor cell pool [4] . ARS does not only imply damage to the bone marrow. In a dose-dependent matter, it can also emerge as gastrointestinal and cerebrovascular syndromes leading to development of multiple organ dysfunctions [4] . Damage to the whole organism is related to a systemic inflammatory response. Different target organs are affected due to activation of the innate immune system, resulting in a significant release of inflammatory cytokines [5] . The pathophysiology appears comparable to that of acute graft-versus-host disease (GvHD) following allogeneic stem cell transplantation where a similar “cytokine storm” has been observed [6] . Long-term effects of ionizing radiation have been well documented in atomic bomb survivors in whom persistent signs of inflammation, e.g. increased plasma levels of tumor necrosis factor-α (TNF−-α), interferon-β, interleukin-6, and C-reactive protein, have been reported [7] . Additionally, oxidative stress after high dose ionizing radiation has been involved in delayed morbidity [4] . Management of ARS therefore relies on tissue damage repair processes that might be supported by therapies directed at mitigation of inflammation [4] . Efforts to improve outcome for affected individuals focus on the stem cell niche. Therefore, visionary therapies should augment niche activity to accelerate hematopoietic recovery in vivo. Several studies have demonstrated that BM osteoblasts regulate the HSC pool size in vivo via the Jagged1-Notch signaling pathway [1] . For example, parathyroid hormone receptor activation can increase the number of osteoblastic cells, resulting in Notch1-mediated expansion of HSCs [8] . One integrative part of the BM stroma are the mesenchymal stromal cells (MSCs), also described as osteoblastic progenitors [9] . MSCs have been proven to intervene with acute organ impairments. Cotransplanted with HSCs, MSCs augment hematopoietic recovery after chemo- or radiotherapy significantly decreasing the time to full hematopoietic and particularly platelet reconstitution [10] . Additionally, there is a large body of evidence for MSCs effectiveness in the treatment of steroid resistant GvHD without any side effects even when obtained from BM of third-party donors [11] , [12] . No HLA-matching is needed between donor and recipient because MSCs have been shown to be hypoimmunogenic and are not recognized by the recipients immune system even after repeated injections [11] , [12] . Finally, MSCs secrete a plethora of bioactive molecules [13] , [14] . Among these, several essential hematopoietic growth factors including IL6, IL11, LIF, SCF, and Flt3 ligand are produced but also factors with immunomodulatory effects, e.g. TGF-β1, prostaglandin E2, indoleamine 2,3-dioxygenase, and others. [14] , [15] . Additionally, vascular endothelial growth factor (VEGF) secreted by MSCs in large amounts might interfere with an early apoptotic cell death after irradiation [4] . Therefore, MSCs might be a good candidate for the modulation of hematopoietic niche activity. Altogether, we assumed that MSCs with their comprehensive trophic potential could serve as a readily available treatment option after severe radiation exposure.

Materials and Methods Mouse MSC generation and characterization Ethics Statement: Animal experiments were approved by the local ethical committee (License Department Hamburg) under application No. 64/02 and 84/05 and performed according to their guidelines. Female C57BL/6J-CD45.1 mice (The Jackson Laboratory) represented the recipient population, male C57BL/6J mice with the wild-type CD45 (CD45.2) were used as donor animals. Mouse MSCs were isolated from male bone marrow and expanded for 9 passages in DMEM/Ham's F12 medium (Biochrom) supplemented with 20% preselected fetal calf serum and 2mM glutamine (both: Invitrogen). The cells were retrovirally labeled with SF91-eGFP [45] at MOI = 3 (bulk population) and expanded or seeded for cloning into ten 96-well plates at 0.3 cells/well. Expanded bulk and selected clonal mMSCs of P14–P18 were characterized according to their differentiation capability into adipo-, chondro- and osteogenic lineages and phenotype as described [46] according to ISCT criteria [17] . Characterization of integration site pattern of clonal mMSCs was carried out as described via LM-PCR [45] . Genomic DNA was digested with Tsp509I giving rise to integration specific fragments of unique lengths. Linker cassettes with known sequences were ligated to the restriction sites. Primer were designed to bind at the known sequences of the provirus and the linker cassette. Amplified PCR products were loaded onto the gel. After gel extraction the bands (red stars) were sequenced. Using the BLAST algorithm, the obtained sequences were aligned to the mouse genome to identify the integration sites. Chromosome preparation and spectral karyotyping of eGFP-transduced bulk-mMSCs, passage 13 and clone IXH8, passage 20 were performed as described [47] . Mouse MSC transplantation TBI was performed using a Cs-137 radiation source. Lethally (9.5 Gy) irradiated female C57BL/6J-CD45.1 mice were i.v. transplanted within 8 hours with 10 6 cells divided into following groups: (i) male bulk (n = 28) or cloned (n = 59; irradiation controls n = 15) mMSCs of P15–P20 for investigation of long-term survival; (ii) clone IXH8 mMSCs (n = 32) to investigate the in vivo distribution of donor mMSCs; (iii) BM (n = 4, named HSC) or bulk mMSCs (n = 6, named MSC) for microarray analysis; (iv) HSC (n = 10) or bulk mMSCs (n = 9) to validate the differential gene expressions obtained with the microarray. Sample acquisition and examination From experimental animals of group (i) blood samples were taken retroorbitally and cell counts analyzed using a Coulter Onyx. Seven months later, PB, BM, thymus, lymph node, liver, spleen, lung, intestine, aorta/vena cava and abdominal fat were removed and used for genomic detection of the Y-chromosome (Sry) and/or eGFP via quantitative PCR ( Table S2 ). Parts of lung tissues were fixed in formalin, paraffin embedded and serial cuts stained with HE, von-Kossa stain visualizing calcium precipitates and Collagen I. Genomic DNA from PB and BM was isolated the same day using innuPREP Blood DNA Mini kit (analytikjena), DNA from all other snap frozen organs with Invisorb Spin Tissue Mini Kit (Invitek). To yield a standard curve, male/GFP-positive DNA was diluted at decreasing concentrations in female/GFP-negative DNA. Quantitative PCR reactions were performed on a Mx3000P (Stratagene). Reaction mixture contained 50 ng of DNA, SYBR Premix Ex Taq (Takara Bio INC), and 200 nM primers (MWG-BIOTECH AG). The threshold cycle (Ct) was determined for each reaction by the MxPro Software. Amplification efficiency was calculated by the sliding window method (LinReg software) [48] . Normalization of expression values was done using Rps27a for eGFP and control chro11 for Y-chromosome tests. Additionally, parts of freshly isolated PB, BM and thymus were investigated for donor derived eGFP- and CD45.2-antigen-expression using flow cytometry as described [45] . Blood samples from animals of group (ii) were taken at different time points (20 min, 2, 4, 8, 12 and 24 hours; n = 8 for each time point) after transplantation and isolated DNA used for Sry/eGFP-chromosomal quantitative PCR. At day 1 and 10, PB, BM, lung, liver and spleen were obtained for DNA isolation. All samples were quantitatively tested for a) the Y-chromosome of recipient cells and b) the stably integrated eGFP. From experimental animals of group (iii) at d21, BM from HSC- and mMSC-transplanted groups was flushed combining the cells of 2 or 3 mice respectively and used for RNA isolation using Invisorb Spin Cell RNA Mini Kit (Invitek). Non-manipulated BM of age-matched mice (n = 4, named BM) was used as control. Differential gene expression was investigated using CodeLink UniSet Mouse 20K I Bioarrays as described previously [46] . For each group (BM, HSC, MSC), two arrays were hybridized. Gene expression profiles of all genes were grouped by hierarchical clustering (TIGR MeV v.4.5.1; Manhattan distance, average linkage). All data is MIAME compliant and the raw and processed data has been deposited in a MIAME compliant database at the gene expression omnibus (GEO) under accession GSE21867. Validation of differential gene expression of genes with a fold change of ≥2.5 was done on the independent mouse group (iv) using d21 BM of 10 animals transplanted with HSCs and 9 animals transplanted with clonal IXH8 mMSCs. Primers used are listed in Table S2 . Differences in gene expressions in BM of MSC and HSC transplanted animals were determined based on the ΔΔCt method normalized for Taf12 and Rps27a . Statistical analysis For statistical analysis, unpaired and two-tailed Student's t-test was applied. P-values <0.05 were considered statistically significant.

Results MSCs promote hematopoietic recovery after lethal irradiation Dynamic evaluation of peripheral blood counts of animals treated with bulk MSCs revealed similar leukocyte and thrombocyte recovery as observed in recipients transplanted with HSCs [16] reaching normalization of white blood cells after 4 weeks ( Figure 1 ). Seven months after transplantation, 2/3 of recipients still were alive ( Table 1 ) and hematologically well with a normal distribution of peripheral blood cell (PB) populations ( Table 2 ). Using ligation-mediated (LM-) PCR we identified specific integration site (IS) patterns for each single clone in vitro ( Figure 2 ). For mMSC clone IVH7 we identified 2 IS, for VF10 3 IS, for VIIIE7 1 IS, for IXC2 2 IS and for IXH8 1IS. Integration sites shown twice might be due to incomplete digestion of the genomic DNA. The thickness of the bands doesn't resemble minor or major integration sites but is inherent to the method. Further analysis of single integration sites is shown in Table S1 . Transplantation of clonal mMSCs resulted in a superior survival rate of recipients treated with clone IXH8 (88%) compared to survival rates of approx. 30% obtained with other clones ( Table 1 ). No control animals without cell transplantation survived the TBI longer than 3 weeks. Impressively, clone IXH8 morphologically was different from all other cultures ( Figure 3 ) without any flattened cells and additionally showed a distinct phenotype with increased CD34 and CD45 but no CD105 expression. All other characteristics of this clone corresponded to the ISCT criteria [17] ( Figure 3 , Table 1 ), questioning the relevance of CD105 expression for MSC characterization. Transplanted donor cells are detectable short- but not long-term Stably integrated eGFP-sequences were used for tracing donor cells in recipients after transplantation. Immediately after injection, 20.2%±15.7 of transfused cells could be detected in the peripheral blood ( Figure 4a ). Within 24 hours, eGFP-positive donor cells were diluted out from PB and trapped in lungs but not in BM, spleen or liver ( Figure 4a insert). At day 10 and later on, no donor cells remained in any investigated tissue (PB, thymus, lymph node, liver, spleen, lung, intestine, aorta/vena cava and abdominal fat; not shown). A standard curve for eGFP-BM and assessment of donor cells in long-term survivors (see Materials and Methods ) is shown in Figure 4b . This corresponds to results from repeated PB flow cytometry analysis of recipients within the 7-month period where no eGFP-positive or CD45.2 donor cells were found (not shown). Although we transplanted male donor mMSCs into female recipients, the Y-chromosome was not available for chimerism analysis. Among various structural chromosomal alterations and massive aneuploidy we surprisingly detected the loss of the Y-chromosome in clonal mMSCs already during the in vitro expansion period using spectral karyotyping ( Figure 5 ). MSCs salvage endogeneous hematopoiesis While donor mMSCs did not home to the BM, the gene expression profile in BM changed significantly, clustering as a separate group compared to HSC transplanted mice or age-matched BM ( Figure 6a ). Impressingly, the clustering of all genes was done without any preselection giving rise to highly stable clusters. A heat map using hierarchical clustering of up- and downregulated genes in the MSC compared to the HSC groups with p≤0.01 and fold changes of ≥2.5 or ≤2.5 illustrates the variations between the samples ( Figure 6b ). Successful validation of selected genes by quantitative PCR ( Table 3 ) emphasized the beneficial role of mMSCs in endogenous hematopoietic reconstitution ( Figure 6c ). Transplantation of mMSCs upregulated genes in the BM involved in cell cycle and protection from oxidative stress (Cdkn1a, BRPK) as well as in anti-inflammatory and detoxification processes (Thbs2, Gstm5). In contrast, genes for regulation of proinflammation (Klk6, Klk1b5), protein degradation (Uchl1), adhesion/matrix formation (Sykb, Emid1, Col5a3), lipid synthesis (Gpam), and lymphoid development (Vpreb1, Rag2) were downregulated. The downregulation of genes involved in adhesion and matrix formation particularly suggests lower retention of hematopoietic progenitor cells within the stroma and higher potency to egress into the peripheral circulation.

Discussion While lower dose radiation victims may profit from supportive care, the situation is more serious after irradiation dose higher than 6 Gy. The only curative treatment in these cases is the transplantation of HSCs. However, this alternative depends on whether a suitable donor is available and can be provided quickly. The normal time required to find and deliver a HSC transplant spans over weeks, a frame too long for seriously affected individuals. HSCs reside in close association with osteoblasts and sinusoidal blood vessels within the bone marrow and this association contributes to the maintenance of the HSC pool in vivo. Self-renewal, proliferation and differentiation of HSCs are regulated through intrinsic signals from the BM niche in which the MSCs are a regulatory component [18] , [19] . We show here that lethally irradiated mice fully reconstituted the blood system through transplantation of mMSCs with similar kinetics as HSCs. Systemic administration of non-clonal mMSCs resulted in long-term survival of the majority of animals with normal blood cell distribution. Generation and expansion of MSCs from C57BL/6J mice is particularly difficult and time-consuming. Since it is well known that rodent adherent BM cells might contain HSCs over long periods caused by emperiopoiesis [20] among other mechanisms, we formally cannot exclude a contamination of MSC preparations with remaining HSCs. Therefore, clonal mMSCs were evaluated for their reconstitution potential. Here, surprisingly, one population of cloned mMSCs showed an even better reconstituting potential than the bulk population. This clone IXH8 was different from all other MSC cultures: a) morphologically, the clone consisted exclusively of spindle shaped cells indicative for significantly accelerated proliferation of MSCs [21] ; that was not observed in bulk and other clonal cultures, and b) flow-cytometrically, we detected increased hematopoietic CD34 and CD45 but no CD105 expression. CD105 has been described as a marker of proliferation and adhesion [22] . We assume that lack of CD105 might increase survival of recipients by lowering lung embolization. All other characteristics corresponded to the ISCT criteria [17] , questioning the relevance of CD105 expression particularly for mMSC characterization. Likewise, over-expression of CD105 in cultured endothelial cells has been shown to induce a marked increase in protein levels of inflammatory eNOS [23] , suggesting an anti-inflammatory action of CD105-negative mMSCs in our model. Surprisingly, none of the recipients transplanted with clonal mMSCs developed osteosarcomas or fibrotic lesions in lungs as has been observed with non-clonal cultures ( Figure 7 ). Taking together the differences in morphology, antigen expression and survival, we hypothesize that the composition of the expanded mMSCs (e.g. significant amount of nonproliferating osteoprogenitors) rather than the source of the population (mMSC or hMSC as discussed in [24] ) causes different outcomes after transplantation. The survival of recipient animals suggested the homing of mMSC to the BM. We tested this hypothesis using either the Y-chromosome of male donor cells or the stably integrated eGFP as detection marker. Y-chromosome-based chimerism analysis in female recipients could not detect donor cells in any investigated tissue including PB and BM, although animals survived long-term. Spectral karyotyping of clonal mMSCs revealed various structural chromosomal alterations and massive aneuploidy (as did bulk mMSCs) with loss of the Y-chromosome whereas bulk cultures of passage13 were still Y-positive. Hence, mMSCs not only accumulate chromosomal abnormalities during few in vitro passages [24] but also might lose sex-specific chromosomes. Despite this, no tumors or osseous inclusions were formed as shown in Figure 7 and previously described by others [24] , [25] . We assume that cloning of mMSCs selected defined populations which do not contain replicative senescent cells as has been regularly detected in bulk populations. The clonal mMSCs might not be prone to stable lung embolization and thus do not lead to eventual tumorous degeneration. Quantitative PCR for stably integrated eGFP-sequences also failed to detect any donor cells and no eGFP-positive cells were found in PB, BM or thymus by flow-cytometry. These results were unexpected, since forced in vitro differentiation of human MSCs suggested a potential differentiation capability into hematopoietic and endothelial cells, albeit to a rather low degree ( Figure S1 ). Although we cannot completely rule out the presence of single donor cells in the investigated tissues below the detection limit, hematopoietic recovery in recipients due to replenishment with donor cells is unlikely in our setting. Additionally, transplantation of cells with the shown high incidence of chromosomal aberrations would in high probability result in tumor formation. That was not the case in our recipients pointing at limited survival of the cells in vivo. This conclusion contradicts the results of earlier studies analyzing hematopoietic recovery after myeloablative TBI with blood-derived mMSCs [26] , [27] and showing donor characteristics in blood and BM. One fundamental difference between both cell sources is the immortalization of cells with SV40 used in these studies, potentially altering BM homing capability of MSCs. Therefore, our results support the concept of impaired transplantability of expanded MSCs [28] – [30] but also challenge the hypothesis of high plasticity of MSCs [31] . Our findings are in contrast to results obtained in immunodeficient mice and monkeys after MSC transfusion [32] – [34] . Francois et al. [32] detected 0.08% to 0.94% of injected human MSCs in several tissues of NOD/SCID mice after TBI+/− local irradiation. In this study, quantitative distribution but no long-term effects of hMSCs were studied. The low percentage might question the physiological relevance of homed hMSCs for intervention in irradiation-induced tissue damage. Recently, the main mechanism through which MSCs counteract global inflammation in sepsis has been attributed to reprogramming of macrophages [35] . This mechanism could be active also in ARS where TNF-α activated MSCs release prostaglandin E2 that acts on the macrophages through the prostaglandin EP2 and EP4 receptors and induces IL-10 secretion. It well might be that, based on this mechanism, the direct injection of hMSCs into the humerus of non-human primates as a single treatment [33] failed to detect any beneficial long-term effect of stromal cells. In an additional approach of ARS in monkey treated with MSC transfusion the one shown animal did not survive 12 days [34] . However, not many of the HSC cotransplanted animals survived long-term either. Because of the low number of experimental animals it remains to be determined if the cells (mean of 14.23% of MSCs were positive for SH2) or the model were not optimal for appropriate assessment. Taken together, our and results from others [36] support the trophic effects of MSCs rather than transdifferentiation. In contrast to “conventional” MSCs described here, mesodermally derived multipotent adult progenitor cells (MAPCs) with their extraordinarily high plasticity are unable to radioprotect lethally irradiated recipients but possess long-term multilineage hematopoietic repopulating activity, thus preceding HSCs in ontogeny [37] , [38] . However, their in vivo equivalent and the true nature (e.g. in vitro artifact) are still unknown. MAPCs seem to differ fundamentally from MSCs in their in vitro and in vivo differentiation potential. Kinetic analysis of the distribution of eGFP+ donor cells after i.v. transplantation substantiated the fast disappearance from PB. Mouse MSCs trapped in lungs quickly, however without long-term residence and embolization as revealed by lack of donor signals 10 days post-transplant. Additionally, no homing of donor mMSCs to the BM was evident, pointing to salvage of endogenous irradiation-surviving HSCs but not to reconstitution of hematopoiesis by donor MSCs. This conclusion is corroborated by the gene expression profile in BM of MSC-transplanted animals. MSCs in a complex mechanism counteracted factors, e.g. oxidative stress, inflammation and toxification by degradation products which suppress the recovery of remaining HSCs. The MSC-mediated regulation of the niche environment likely is a redundant system that is mediated by several molecules as has been shown for immunoregulation by MSCs [18] . Irradiation produces excessive inflammatory responses [39] , contributing to HSC death if left untreated. Among other organs, the lung is especially sensitive towards irradiation damage and may retain MSCs. MSCs interfere with inflammation by changing the gene expression profile not only in the lungs where they dock [40] but also in BM. To do so, MSCs need not necessarily home to the BM but might globally change gene expression or activate the production of systemically counteracting substances. This mechanism has been described for MSCs influencing myocardial infarction [40] . Similarly to mMSCs, injected hMSCs trapped in lungs and affect the production of the anti-inflammatory protein TSG6 which enhances myocardial repair without significant engraftment. Differentiation-independent paracrine MSC-effects also ameliorate acute kidney injury via anti-inflammatory, mitogenic and angiogenic actions [41] , [42] . Impressing clinical benefits of MSCs combining surgery and local cellular therapy for the treatment of severe radiation burns has been demonstrated. Uncontrolled clinical symptoms of radiation inflammatory waves successfully could be limited during patient's 8-month follow-up as evidenced by the decrease in the C-reactive protein level observed after each MSC administration thus confirming modulating activity of MSCs in inflammatory processes [43] . Secretion of a broad range of bioactive molecules which alter the tissue microenvironment is now believed to be the main mechanism by which MSCs achieve their therapeutic effect. The transplanted MSCs might, as a principal mechanism, export their inherent trophic effect to unorthodox sites [44] . The outcome is an enhanced regeneration of injured cells, stimulation of proliferation and differentiation of endogenous tissue progenitors, but also a decrease in inflammatory and immune reactions [14] . Our results present the evidence for this highly effective trophic mechanism working also in BM after lethal irradiation in mice. MSCs, moreover, might be helpful in alleviating myelosuppression due to chemotherapy and toxic drug reaction. Whether the results can be translated to humans still has to be shown. Because BM-derived MSCs are easily accessible, can be massively expanded, and stored for prolonged time, they are easily distributed to places in need. We suggest MSC-infusion as an efficient and immediate treatment option after irradiation injuries.

Conceived and designed the experiments: CL UMG ARZ. Performed the experiments: CL BBS HCO KE UMG CR BS KC JZ ANS. Analyzed the data: CL BBS HCO KE ANS. Wrote the paper: CL. Background Successful treatment of acute radiation syndromes relies on immediate supportive care. In patients with limited hematopoietic recovery potential, hematopoietic stem cell (HSC) transplantation is the only curative treatment option. Because of time consuming donor search and uncertain outcome we propose MSC treatment as an alternative treatment for severely radiation-affected individuals. Methods and Findings Mouse mesenchymal stromal cells (mMSCs) were expanded from bone marrow, retrovirally labeled with eGFP (bulk cultures) and cloned. Bulk and five selected clonal mMSCs populations were characterized in vitro for their multilineage differentiation potential and phenotype showing no contamination with hematopoietic cells. Lethally irradiated recipients were i.v. transplanted with bulk or clonal mMSCs. We found a long-term survival of recipients with fast hematopoietic recovery after the transplantation of MSCs exclusively without support by HSCs. Quantitative PCR based chimerism analysis detected eGFP-positive donor cells in peripheral blood immediately after injection and in lungs within 24 hours. However, no donor cells in any investigated tissue remained long-term. Despite the rapidly disappearing donor cells, microarray and quantitative RT-PCR gene expression analysis in the bone marrow of MSC-transplanted animals displayed enhanced regenerative features characterized by (i) decreased proinflammatory, ECM formation and adhesion properties and (ii) boosted anti-inflammation, detoxification, cell cycle and anti-oxidative stress control as compared to HSC-transplanted animals. Conclusions Our data revealed that systemically administered MSCs provoke a protective mechanism counteracting the inflammatory events and also supporting detoxification and stress management after radiation exposure. Further our results suggest that MSCs, their release of trophic factors and their HSC-niche modulating activity rescue endogenous hematopoiesis thereby serving as fast and effective first-line treatment to combat radiation-induced hematopoietic failure.

Supporting Information

We thank K. Schlagner (MSC characterization and mouse tissue preparation), A. Schmidt (MSC characterization) and G. Arman-Kalcek (i.v. injection of MSCs) for excellent technical support.

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e14486

oa_package/3c/ae/PMC3016319.tar.gz

PMC3016320

21245930

Introduction Elucidating the mechanisms governing cohesion during group movement is a central issue to our understanding of the evolution of social behaviour [1] . In order to maintain the benefits of group living (such as reduced predation risk, better foraging efficiency and the exchange of social information), mobile animals often have to synchronize their activities, forage collectively and move together by coordinating both the timing and direction of their movement decisions. Collective movements typically begin with some individuals first departing to a new area. Thus, movement initiations within resting or foraging group are instances of transient group splitting. Decision-making regarding movement may be especially critical for those first individuals that leave the group since they disproportionately increase their risk of predation [2] and potentially lose territorial defense benefits [3] . If benefits are linked to group size, as is expected [4] , there must exist some conflict between staying with others and taking the risk of departing to forage on higher quality resources or to reduce competition. Importantly, this conflict between leaving and staying also concerns not only the first individual to initiate the movement (the “initiator” [5] ), but also those individuals which have not yet departed. When some of the group members decide to move, the remaining individuals have to choose whether to follow those that have departed. If they do not, the group will remain split. Although individual movement decisions are known to be influenced by the actions of conspecifics [6] – [8] , the precise mechanisms are largely unknown [9] . A way of addressing this is to identify the stimulus-response function at the individual scale, that is the individual following rule that can account for the observed collective outcomes. Many theoretical or experimental studies have suggested that collective decisions to move emerge either from a kind of a pre-departure consensus building based on a voting procedure [1] , [10] , or from a combination of more individualistic decisions based on a behavioural switch when a quorum has been reached [9] , [11] – [16] . In most cases, however, they postulate the decision-making process at the individual scale and then test the model predictions at the collective scale, without explicit reference to experimental data at the individual scale [17] . However, different models at the individual scale can lead to the same predictions at the collective scale, provided their parameters can be freely adjusted. As a consequence, conclusions drawn from such models remain hypothetical regarding the full details of the information used by individuals to come to their decisions. To gain deeper insight into collective motion in animals, and to highlight the individual decision-making process, we analyzed quantitatively the individual responses in the course of collective departures for different groups of sheep (Merino breed). For this, we trained individual sheep to move towards a panel raised at the periphery of an arena [18] . A single trained individual was then introduced into groups of naïve sheep and used to initiate a collective movement. To identify the nature of the stimuli that trigger individuals' decisions to follow we characterized the stimulus-response function at the individual scale for all naïve individuals. Under our experimental conditions, in which environmental factors are controlled, the stimulus is purely social, and was provided by the behaviour of other group members. A key feature was the use of groups of different sizes ( N = 2, 4, 6, 8) so that sheep were exposed to various combinations of two factors: the number of departed individuals (including those sheep departing in response to the trained leader) and the number of non-departed individuals. We quantified the individual stimulus-response function by the probability per unit time to depart (or departure rate, expressed in s −1 ) when exposed to such combinations. Both factors (the number of departed and non-departed neighbours) were shown to significantly affect the departure rate. The insight that individuals integrate information about their departed and non-departed neighbours has several important functional consequences. First, the collective dynamics remain the same in groups of any size, and it therefore supports scalability at least up to group sizes where each individual can see each other. Furthermore, the parameter values that fit experiments are precisely the ones that minimize the duration (and thus potential costs) of the temporal split of the group that is the time elapsed between the trained departure and the departure of the last follower.

Material and Methods Experimental set-up Fieldwork was carried out in the experimental farm of Domaine du Merle (5.74°E, 48.50°N) in the South of France from January 2008 to March 2008, with females of Merinos d'Arles (three years old). The training set and the naïve set comprised 25 and 150 ewes respectively, which were randomly selected from a flock of 1600 females avoiding relatedness. Each ewe was marked on its back using a special paint in order to be identified. All the ewes were released every morning into enclosed paddocks situated within homogeneous meadows of Crau hay. The naïve set was penned up each evening in the same sheepfold as the training set. To investigate the dynamics of decision making, we triggered movement using an informed individual [23] . This series of experiments was realized with the same training and experiments procedures as in our previous study [23] . Sheep were trained (in five groups of five sheep) to become movement initiators. After two weeks of training, we obtained four trained, one well trained individual per training group, which answer on 95% of the test. Then, one trained individual was combined with sheep familiar with the sound and the panel, but naïve for the food target, i.e. habituated to the stimulus, and we used different group sizes: groups consisted of two ( N = 11 replications), four ( N = 7), six ( N = 11) and eight individuals ( N = 11), to obtain different arrangements of the number of departed D and non-departed S individuals. Groups of sheep were introduced in circular arenas (25 m diameter), in a flat homogeneous pasture [23] . Arenas were enclosed with sheep fences and visually isolated from immediate surrounding by a green polypropylene net. In each group tested, one trained sheep initiated a move towards a coloured panel raised under experimenter's control. Under these controlled condition, individual decisions to move depended mainly on other group members' behaviour. For that purpose a food reward was delivered on the ground at the foot of one of five panels laid at the periphery of the arena. Before raising one panel, a sound stimulus was delivered to synchronize the attention state of all sheep (head-up) so they could concurrently perceive the departure of the initiator. This could be compared to a situation of heightened attention of all group members such as may occur under conditions of predation risk in which it is important to be vigilant and to flee if necessary. The behaviour of sheep was recorded with a digital camera (Canon EOS D50) fixed on the top of the tower with the frequency of 1 frame per second. We use several trained individuals in order to prevent collective movements in response to behaviour of one potentially peculiar initiator. All naïve individuals were tested only once. We triggered a departure of only one individual in each group. Animal care and experimental manipulations were in accordance with the rules of the French Ethical Committee for animal experimentation. Data analysis The behaviour of each individual was quantified using a probabilistic stimulus/response function. Latency of the follower i corresponds to the time elapsed (in seconds) since the previous departure of individual i −1. The distributions of experimental following latencies fitted exponential distributions, indicating that the probability per unit time to depart (the log gradient of the exponential distribution) is constant over time for the same herd configuration (number of departed and non-departed). The experimental probability per unit time to follow (the following rate expressed in s −1 ) is the inverse of the mean departure latency. The latencies were gathered as a function of the number of D (departed) and the number of S (non-departed) individuals (see also Appendix S1 ). Most departures were well-defined and discrete events in our time scale, but when we observed two or three individuals departing simultaneously (within the same second), they are ascribed to the same departure rank and thus submitted to the same combination of number of departed and non-departed individuals. To perform our analysis at the individual scale, we assumed that the individual response functions were the same for all naïve individuals, and were stable over time. This is reasonable since naïve individuals were used only once, so that any potential effects of learning, exploration, habituation or uncontrolled social experience were discarded. This precluded also any potential effect of inter-individual affinity [8] . Moreover, the trained sheep had the same motivation to depart towards the panel and exhibited the same movement away from the group, that is the trained sheep walked directly to it [23] . No naïve sheep departed towards the target before the trained sheep [23] , and when they departed, they move towards the trained sheep (and not the panel). In control groups, with no trained sheep, naïve individuals never walked towards the panel [23] . This strongly suggests that naïves were engaged in a pure following behaviour. Response function fitting A simple linear regression on the log-transformed data was used to fit the parameters of the response function.

Results In all experiments, a consensus decision was observed. The departure of the trained sheep towards the visual panel always triggered a collective movement and all naïve sheep followed within a relatively short time (95% followed in less than twelve seconds). Moreover, the time course of collective departures did not depend on the group sizes (Kruskal test on time course: χ 2 = 2.045, df = 3, P = 0.56, Fig. 1 ) whilst one may have expected that larger groups would take a longer time to depart, even over this range of group sizes. Individuals' behavioural responses were quantified by calculating the departure rate separately for each combination of departed/non-departed group members. To quantify this departure rate, we assumed a continuous time Markovian jump process, that is, the probability per unit time displaying the response (in this study, following) is constant over time as long as the stimulus remains the same, and this probability jumps to a new value when the stimulus changes. This assumption was validated (see Data analysis and Appendix S1 ). To identify the nature of the stimulus, we considered that the state of the group changed each time a further individual followed. For instance, in a group of four individuals (one trained and three naïves), each naïve individual is assumed to witness the same group state from the time the trained individual departed until the first follower's departure. Then, from this departure until the next, the two still non-departed naïves witness a new group state which consists of the two departed individuals (the trained and the first follower) and one non-departed individual, and so on. Each time the group state changes, the rate of following may or may not change, depending on what the sheep are reactive to. Accordingly, if the rate of following changes from one group state to another, the two states can be considered as different stimuli. For all group sizes, the departure rate increased sharply with the number of departed individuals: individuals were increasingly stimulated to depart as the number of departed animals increased ( Fig. 2A ). Moreover, for a given number of departed individuals, the departure rate decreased with the number of non-departed individuals ( Fig. 2B ). For instance, when three individuals were already departed, the departure rate decreased from 0.62 s −1 in groups of four (one non-departed) to 0.30 s −1 in groups of six (three non-departed) to 0.27 s −1 in groups of eight (five non-departed). This suggests that sheep were responsive both to the departed and non-departed individuals since the following rate changed each time either the number of departed or the number of non-departed individuals changed. Models of the individual decision To verify this hypothesis, we tested the relevance of this model against more parsimonious models: responding only to the trained sheep departure (initiation), or responding only to the departed individuals. For testing, we used one simple equation: where μ is the departure rate (the response), α is the probability to follow when there is only one follower and one departed individual and D and S the number of departed and non-departed individuals respectively (the stimulus). Note that μ is null before the trained departure ( D = 0), which is consistent with the experiments since the collective movement was triggered by the trained individual in all cases (no naïve individual's departure towards the panel observed before the trained individual's departure). The parameters β and γ modulate the influence of D and S , and allow testing the three alternative models according to their values, Model 1 : Just mimic the initiator Under model 1, the following decision is stimulated by the departure of the initiator only, and is independent of whether other group members have departed or not. Therefore the departure rate should be the same for all naïves at any time, following . Model 2 : Mimic all the departed individuals The following decision is stimulated by the initiator and also by the already departed group members. The departure rate should monotonically increase with the number of departed animals, following . Model 3 : Mimicking the departed or staying with non-departed The following decision is stimulated by the group members which have already departed, as in model 2, but is concurrently inhibited by the ones which have not. The departure rate should increase with the number of departed animals, but decrease with the number of non-departed individuals, following equation (1). To test the adequacy of each model, we first adjusted the corresponding free parameters to the entire set of experimental values (departure rates, Fig. 2A ). Note that fitting model 3 required experiments with different group sizes, so that D and S are not colinear. Both factors had a significant effect in the full model (regression in the log domain: , respectively P β <10 −6 and P γ <2.10 −4 , F 2,13 = 102.9, P <10 −7 , r 2 = 0.94). To test the likelihood that model 3 is a better explanation than model 2 and model 1, we derived their corresponding AIC (Akaike Information Criterion, Table 1 ) [19] . Model 3 with both factors (departed and non-departed) is orders of magnitude (1400 times) more likely to be the best explanation for following rates compared to model 2 with departed individuals only. Models' predictions at the collective scale We used the departure rates fitted under each model ( Fig. 3 , left column, cross symbols) as input to compute the corresponding dynamics of the followers' departures. The model predictions obtained at the collective scale were compared to the experimental values for ( a ) the mean latency of the first follower's departure ( Fig. 3 , middle column) and ( b ) the mean duration from the trained individual's departure to the last follower's ( Fig. 3 right column). These predictions were derived from: which is the mean latency of the i th follower's departure (see Appendix S2 ). < t 1 > corresponds to the mean latency of the first follower ( a ), and < t N > corresponds to the mean latency of the last follower, which is the same as the mean duration of the collective move ( b ). Under model 1 the departure rates are independent from the number of departed and non-departed conspecifics. This model enforces a distribution of the fitted rates greatly different to the data ( Fig. 3A , α = 0.33 s −1 ). Accordingly, it yields bad predictions at the collective scale: the predicted first follower's mean latency is far too low while the mean duration increases continuously with an increasing group size ( Figure 3B–C ). Model 2 allows a distribution of the fitted rates that is closer to the observed responses since they are allowed to reflect the stimulating effect of the number of departed individuals ( Fig. 3D , α = 0.09, β = 1.4, r 2 = 0.80). However, the model still yields incorrect predictions for small groups ( Fig. 3E–F ). Only model 3, which also includes an inhibiting effect of still non-departed conspecifics, provides an accurate distribution of the fitted rates ( Fig. 3G , α = 0.19, β = 1.16 and γ = 0.6, r 2 = 0.94, see above). Accordingly, it yields predictions that are consistent with the experimental data at both the individual and collective level ( Fig. 3H–I ). This supports the hypothesis that individual response depends both on the number of departed and non-departed animals. Therefore our analysis reveals that sheep decision-making is not based on a single mimetic effect but rather on a rule which balances two mimetic opposite effects: follow the departed individuals but remain with the non-departed individuals. Hence, not only do sheep respond to the sudden events of departing conspecifics, they also integrate information about the steady state of still non-departed conspecifics. Functional consequences The evolutionary advantages of the distribution of departure latencies for species subject to predation are now well known [4] , and survival typically increases with an increasing group size [20] . The individual benefit of being in a group is therefore often an increasing function of the number of individuals N , at least up to some maximal extent [4] . Considering two populations, staying ( S ) and departed ( D ), the individual benefit can be estimated as: The individual benefit (equation 3a) and I M (equation 3b) are assumed to be proportional to the number of individuals being in the same behavioural state as the focal individual. When the i th individual departs, i is the rank of its departure and the transition for the whole group is: so that its benefit Δ I of staying compared to moving is the difference between the benefits to join the departing group and the benefits to remain with staying individuals expressed as: Equation 4 a allows the calculation of Δ I , and model 3 allows us to predict departure rates in group sizes untested in our experiments. Following a generic anti-predator strategy, it is beneficial to remain with the largest population. For a group of fixed size N , it is advantageous to remain while the number of departed individuals D < N /2, whereas it becomes beneficial to depart when D > N /2, and the benefit becomes positive (Δ I >0) when the departure rank i > N /2. Figure 4 demonstrates that the departure rates follow the same pattern for any group sizes. In fact, the experimental departure rate also increases with an increasing number of departed individuals ( Fig. 2A ), and increases sharply when the number of departed individuals is greater than half of the group, whatever the group size ( Fig. 4 ). An important property is therefore that the sheep decision-making process can scale and function effectively, for any group size. How sensitive are the collective dynamics to the parameters we found? In our experiments we found that the balanced effects of departed and non-departed individuals follow β = 2*γ, with β = 1.2. Using equation 2, we tested the sensitivity of the mean duration of collective moves (the latency from the initiator departure to the last follower's departure) to deviations of β (0<β<2, γ = β/2, for different group sizes N = 2, 4, 8 and 32, Fig. 5 ). The mean duration was found to depend strongly on the interaction of β and group size. The values of β and γ for which the mean duration was minimized were similar to those found in our experiment, and furthermore they maintained this property of minimizing the duration of split events whatever the group size considered. In addition, the variation of mean duration for all group sizes is also minimized for the values that best fit our experiments. This is likely to have an important functional consequence to group decision making, because it facilitates consensus, functions independently of group size and minimizes the proportion of time the group is split during the act of decision-making.

Discussion By analyzing the dynamics of individuals' reactions within an experimentally-induced collective departure under controlled conditions, we have been able to demonstrate that individual decision-making in sheep is based both on departed and non-departed group members. This mechanism is scalable in four group sizes of eight or less individuals, and the experimental parameters' values β and γ proved to be the ones which minimize the duration of fission events, whatever the group size. The temporal organization of individuals' responses was used to reveal the underlying individual decision-making process. This direct measurement of the individual sheep response demonstrated that their probability per unit time to display a response is constant over time as long as the stimulus remains the same (thus it is Markovian), and that both departed and non-departed individuals were necessary to account for these responses. The fit of the data at the collective scale showed that they were also sufficient . In other words, this is a parsimonious model that still fully explains the experimental results. We advocate that such a precise measurement should be made at both scales to fully explain collective behaviours. Ward et al. [9] and Sumpter et al. [16] previously proposed a model to account for moving decision in a Y-maze in three-spine sticklebacks, Gasterosteus aculeatus . They suggested that individuals' probability of following leaders increases sharply with an increasing group size of departing individuals, and that following behaviour is inhibited by undecided (non-departed) individuals. However, they fit their model only to the collective response, and thus the individual parameters remained hypothetical. The present study gives further support to their hypotheses that both departed and non-departed individuals matter. Petit et al. [21] recently proposed a model for the collective decision in one group of white-faced capuchin monkeys Cebus capucinus . The movement initiator was highly prone to cancel the movement when too few group members followed in a too short time. In their model, the probability following was also modulated by the ratio of the departed to the non-departed group members, and it appears to be a quorum at the collective level. However, disentangling both effects still has to be validated using different group sizes. Using different group sizes and individual measures which largely derive from a quantification of the dynamics at the collective scale in Merino sheep, Gautrais et al . [22] developed a model explaining the synchronization of states (resting and grazing) among group members. The transition between the two activities was also influenced both by active and inactive neighbours. Unlike this study, here we quantify the behaviour at the individual scale, and moreover, we control the initiation of the phenomenon. This methodology allows to isolate and highlight the social interactions and to exclude the contribution of any external stimulus. Our experimental model suggests that individual decisions to follow those that have departed is not a quorum-like response because the probability monotonically increases with the ratio of already departed and non-departed individuals. As a consequence, the probability of response to the number of departed is dependent on group-size. Moreover, all individuals would instantaneously follow the trained individual which moved away. Finally, it engaged all potential followers every time, and so there is no apparent threshold under which no moving at all would occur, as was found in ants and fish [9] , [11] . It is also noteworthy that the mean departure latency of the whole group did not vary with the experimental group size. This result stems from two balancing effects: the mean latency of the first follower decreases with group size ( Fig. 3H ), while the mean duration (from the trained individuals' departure to the last follower's departures) increases with it ( Fig. 3I ). The mean latency of the first follower decreases proportionally with (≈ N −0,5 ). This results from coupling between a pure sampling effect proportional to the number of potential followers N (see Appendix S2 ), and the individual departure rate decreasing by √ N . One prediction of the inhibiting effect of the non-departed individuals is that large groups should be more stable because their members should be less sensitive to initiations. To explain the formation of large animal groups, the first effect may therefore be necessary to over-compensate the latter. Natural selection is likely to result in decision-making rules that allow individuals to vary their behaviour efficiently across a wide range of environmental and social conditions [23] , [24] . Our results in sheep are congruent with this concept. First, the decision-making rule is scalable. Secondly, the experimentally estimated parameters β and γminimized the time needed for all group members to depart. In fact, we found that the parameter values that minimize the duration of group splitting correspond to the experimental ones. Sheep responses were compatible with a strategy that minimizes putative risks of predation by choosing to stay with the larger group ( Fig. 5 ) whatever the size of the departed and non-departed groups ( Figs. 4 and 5 ): low departure rate for the ( N /2) first movers (long departure latency = 1/departure rate, see Data analysis and Appendix S1 ), and when N /2 individuals have moved, they move as a cohesive unit with an increasing departure rate ( Fig. 2 ). Following these arguments, the first mover may still pay disproportionate costs when departing from the group. If the first mover possesses knowledge of the environment with respect to its most profitable foraging areas, the benefit of moving to a new food resource could compensate the cost of the predation risk. Social foragers are known to have to make a compromise between food and safety [4] , [25] . In our experiment the first mover is a trained individual which possesses pertinent information (the location of food reward) giving it a foraging advantage. Following social foraging theory, we can assume that for such individuals the benefits likely outweigh the risks incurred from moving away from the group, when the new area is not far (range: 10–20 m). This assumption could be tested in future experiments, for example by varying the reward given to the trained individuals, or the distance over which they must move. Since we used relatively small group sizes, we assumed that each individual was able to monitor continuously the behavioural state of each member of the group, but this assumption would become irrelevant in large groups where crowding restricts perception of others that are beyond immediate neighbours. In any case, even in large groups, one should consider only the closest neighbours as stimuli ( sensu Voronoi neighbours [26] ). Hence the sensitivity both to departed and non-departed neighbours should stand robustly also in large groups, but perhaps with different parameter values (α, β, γ) [27] . Finally, to investigate the full dynamics of collective motion in large groups [26] , [28] , [29] , models of individual reactions to neighbours inspired by the principles we found in the present study are likely to be informative. One would additionally need to quantify spontaneous initiations and the propagation of its effect across the group in order to understand how the following by the nearest neighbours would in turn trigger or not the departure of the farthest individuals [30] . Considering the ubiquitous nature of gregariousness observed in grazing herbivores, we therefore expect that in a wide range of ungulate species and more generally in many vertebrates, individuals may follow rules that conform to the general principles revealed here. Moreover, we assume that similar individual rules could be at work in situations where group members are confronted with different alternatives like activities or directions, particularly in the presence of trained individuals [8] , [31] .

Conceived and designed the experiments: MHP JG RB. Performed the experiments: MHP. Analyzed the data: MHP JG RB JLD. Contributed reagents/materials/analysis tools: MHP JG PA RB JLD. Wrote the paper: MHP JG IDC RB JLD. Individuals of gregarious species that initiate collective movement require mechanisms of cohesion in order to maintain advantages of group living. One fundamental question in the study of collective movement is what individual rules are employed when making movement decisions. Previous studies have revealed that group movements often depend on social interactions among individual members and specifically that collective decisions to move often follow a quorum-like response. However, these studies either did not quantify the response function at the individual scale (but rather tested hypotheses based on group-level behaviours), or they used a single group size and did not demonstrate which social stimuli influence the individual decision-making process. One challenge in the study of collective movement has been to discriminate between a common response to an external stimulus and the synchronization of behaviours resulting from social interactions. Here we discriminate between these two mechanisms by triggering the departure of one trained Merino sheep ( Ovis aries ) from groups containing one, three, five and seven naïve individuals. Each individual was thus exposed to various combinations of already-departed and non-departed individuals, depending on its rank of departure. To investigate which individual mechanisms are involved in maintaining group cohesion under conditions of leadership, we quantified the temporal dynamic of response at the individual scale. We found that individuals' decisions to move do not follow a quorum response but rather follow a rule based on a double mimetic effect: attraction to already-departed individuals and attraction to non-departed individuals. This rule is shown to be in agreement with an adaptive strategy that is inherently scalable as a function of group size.

Supporting Information

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e14487

oa_package/a4/82/PMC3016320.tar.gz

PMC3016321

21129171

Background Streptococcus pneumoniae and Haemophilus influenzae are major causes of community-acquired pneumonia (CAP) [ 1 , 2 ] and as Neisseria meningitidis they are important agents of meningitis [ 3 - 5 ]. Identification of the microbiological cause of CAP and meningitis is important, as it enables pathogen-directed antibiotic therapy. Conventional detection of bacteria is based on culture and phenotypic characterization. However, culture methods are time-consuming and have relatively low sensitivity, especially when antibiotics have been given to the patient prior to sampling [ 6 ]. The use of nucleic acid amplification tests, such as quantitative real-time polymerase chain reaction (qPCR), have enabled more sensitive and rapid detection of pathogens in respiratory secretions and cerebrospinal fluid (CSF). Several qPCR assays for the detection of S. pneumoniae [ 7 - 9 ], H. influenzae [ 10 - 12 ] and N. meningitidis [ 13 ] have been developed and multiplex detection of several target DNAs in a single tube is achievable [ 14 - 16 ]. Still, the specificity of methods used is an underestimated problem and commonly used targets have been shown to be unspecific and causing misleading results. An illustrative example is the pneumolysin ( ply ) gene for the detection of S. pneumoniae [ 17 - 19 ]. For detection of H. influenzae , a species with frequent exchange of genetic elements, the problem is even worse and most target genes used are problematic. The bexA is not present in all strains of H. influenzae [ 20 ], while 16 S rRNA and rnpB do not provide specific detection [ 21 ]. We have recently developed qPCRs for specific detection of S. pneumoniae , based on the Spn9802 fragment [ 17 ], and for the detection of H. influenzae , based on the outer membrane protein P6 [ 21 ]. Real time PCR assays for detection of N. meningitidis have been based on genes as porA [ 22 ] and ctrA [ 14 , 16 ]. Here we present a new quantitative multiplex PCR (qmPCR) method for detection of S. pneumoniae , H. influenzae and N. meningitidis . The method was evaluated on a collection of bronchoalveolar lavage (BAL) and cerebrospinal fluid specimens for detection of lower respiratory tract infection (LRTI) and meningitis due to these three bacteria species.

Methods Patients and controls From 1997 through 2000, 159 consecutively identified immunocompetent adult patients who were hospitalised for LRTI at the Department of Internal Medicine, Silkeborg County Hospital, Silkeborg, Denmark, were enrolled in a prospective study [ 23 ]. The criteria for LRTI were fever and/or an increased leukocyte count (≥ 11 × 10 9 /L), together with increased focal symptoms from the lower airways with at least one of three newly developed symptoms of increased dyspnoea, increased coughing and/or increased sputum purulence. The enrolled patients underwent standardized fibre-optic bronchoscopy within 24 hours from admission. For the present study, BAL fluid was available in 156 patients, median age 63 years (range 26-90 years). A chronic lung disease was documented in 72 patients (46%), 31% were current and 40% were previous smokers. New X-ray infiltrates were identified in 87 patients (56%). Antibiotics had been taken within 7 days prior to bronchoscopy in 103 cases (66%). As controls, 31 adult patients, median age 64 years (range 30-77 years), who consecutively underwent fibre-optic bronchoscopy for suspected malignancy and who did not have pulmonary infection were included. Nineteen of them had lung malignancies and 12 had no pathology identified by bronchoscopy or radiological examinations. Twenty-seven controls (87%) were current or previous smokers. CSF samples sent for culture to the Bacteriological Laboratory, Sahlgrenska University Hospital, Gothenburg, Sweden during a four year period were used in the study. Specimens were eligible if the total CSF white blood cell (WBC) count was ≥10 × 10 6 /L indicating meningeal inflammation. Only one CSF sample from each patient was included. Medical records of all patients included in the study were reviewed retrospectively for a final diagnosis, predisposing factors, treatment and outcome by one doctor. All 87 specimens were included in a study previously published for 16 S rRNA gene PCR [ 24 ] and the relevance of the PCR findings and bacterial cultures to the final diagnosis was evaluated and compared with the clinical findings and other laboratory results. The median age of the patients were 34 years (range 1 day- 91 years). Fibre-optic bronchoscope In brief, the fibre-optic bronchoscope was introduced through the nose or through the mouth. The tip of the bronchoscope was wedged into the segment of bronchus affected by a pulmonary infiltrate, or, if no infiltrate was available, into the middle lobe. A sterile, thin tube was then introduced into the working channel of the bronchoscope, and lavage was then performed. One to three portions of 60 mL of isotonic NaCl were used for lavage, and the aspirated fluid was collected in one single portion for microbiological analyses. Conventional diagnostic methods BAL fluid from the LRTI patients and the controls were analysed with culture (but no Gram staining) at the Department of Clinical Microbiology, Aarhus University Hospital (Aalborg, Denmark), within a maximum of 6 h from the time of sampling. The specimens were cultured on 5% horse blood agar and chocolate agar with semi-quantitative determinations by dispersion of 1 and 10 μL on each half of the plate. The plates were incubated in 5% carbon dioxide at 35°C for 24-48 h. From 152 LRTI patients, blood samples were collected for culture with a Bactec blood-culturing system (BioMérieux, Marcy-Etoile, France) at the Department of Clinical Microbiology, Aarhus University Hospital. Non-frozen urine samples collected from 142 LRTI patients were sent to the Department of Bacteriology, Mycology and Parasitology, Statens Serum Institute, Copenhagen, Denmark, and were analyzed for pneumococcal capsular polysaccharides by countercurrent immunoelectrophoresis [ 25 ]. CSF samples were submitted for routine bacterial culture and chemistry [ 26 ]. DNA extraction DNA from 0.2-0.5 mL BAL was extracted by the automatic MagNa Pure LC DNA-Isolation system (Roche Diagnostics). Bacteria DNA used for determination of the analytical sensitivity of the Spn9802 and the P6 PCRs was purified from cultured isolates ( S. pneumoniae CCUG 28588 T and H. influenzae CCUG 23946 T ) by phenol-chloroform extraction of bacteria harvested in exponential growth phase after culturing on chocolate agar at 37°C in 5% carbon dioxide and the concentration of DNA was determined by a Nanodrop instrument (NanoDrop Technologies, Inc. Wilmington, DE, USA). The genome copy number was determined according to conventional calculations based on molecular weight and one gene copy per genome. CSF samples (50 μL-1.5 mL) were centrifuged at 12 000 g for 20 min, after which DNA was extracted from the pellet with a bacterial DNA preparation kit (Roche Diagnostics, Indianapolis, USA), used according to the manufacturer's instructions. qmPCR The quantitative Spn9802 PCR for the detection of S. pneumoniae [ 17 ] was combined with the P6 PCR for the detection of H. influenzae [ 21 ] and the ctrA PCR for the detection of Neisseria meningitidis [ 14 ]. All primers and probes are shown in Table 1 where positions with lower case letters indicate locked nucleic acid [ 27 ]. The PCR for detection of N. meningitidis was used as described previously, except that 3.5 mmol/L MgCl 2 was used instead of 5.5 mmol/L and that the elongation time was 40 s instead of 1 min. All primers and probes were obtained from Thermo Hybaid, Interactiva Division (Ulm, Germany) except the Spn9802 FAM probe which was obtained from Eurogentec, Seraing, Belgium. The real-time PCR assay was performed in a Rotor-Gene 3000 instrument (Qiagen, Hilden, Germany). The optimized real-time PCR amplifications were performed in 25-μL reactions containing 0.3 μmol/L of each primer, 0.2 μmol/L of the Spn9802 FAM probe, 0.1μmol/L of the P6 JOE probe and ctrA ROX probe, 3.5 mmol/L MgCl 2 , 0.2 mmol/L deoxynucleoside triphosphate, and 1 U HotStar Taq polymerase (Qiagen) in PCR buffer. A total of 5 μL of target DNA was used in the assay. The qmPCR was performed according to the following program: 15 min of enzyme activation at 95°C, followed by 45 cycles of 95°C for 15 s and 60°C for 40 s. Reproducibility of analytical sensitivity and quantification The analytical sensitivity of the Spn9802, P6 and ctrA PCRs was determined by serial dilutions of target DNA in carrier tRNA (1μg/mL). Two experiments were performed with 5 to 600 genome copies per reaction tube and 2 to 4 tubes of each dilution. The reproducibility of quantification was evaluated by testing DNA preparations with known concentrations (duplicates of 500, 2,000 and 10,000 genome copies per PCR reaction) in five consecutive runs and also in 73 BAL samples and in 8 CSF samples. PCRs with primer/probe reagents in both monoplex and multiplex were tested in parallel. In addition we tested the reproducibility of quantification with positive control DNA of S. pneumoniae , H. influenzae and N. meningitidis in separate tubes and combined in a single tube. fucK PCR The fucK PCR was used as previously described [ 28 ], to confirm the presence of H. influenzae in samples which proved negative by culture but positive by qmPCR. lytA PCR For respiratory samples the lytA PCR was tested in a gel based PCR for S. pneumoniae as previously described [ 29 ]. In short, extracted DNA (10 μL) was added to a PCR mixture, and after 40 cycles, PCR products were detected on ethidium bromide-stained agarose gels. By serial dilution of bacterial strains, the detection level of lytA PCR has been shown to be 10 2 colony forming units (CFU)/mL sample [ 29 ]. 16 S rRNA PCR for CSF samples The primers and other PCR conditions used to amplify the 5'-half of the 16 S rRNA gene were previously described [ 24 ]. The PCR product was visualized in an agarosegel and DNA bands of expected size were cut from the gel, purified with a Qiaquick Gel Extraction kit (Qiagen) and subjected to cycle sequencing using the ABI prism Big Dye Terminator Sequencing Ready Reaction kit, v.1.1 (Applied Biosystems). The sequencing reaction products were analyzed using an ABI PRISM 310 Genetic Analyser (Applied Biosystems). After DNA sequence editing, the GenBank BLAST program was used for sequence comparisons. Ethics The study was performed according to the Declaration of Helsinki II and approved by the local ethical committee and all participating patients gave written consent.

Results A sensitive and specific multiplex PCR for quantitative detection of S. pneumoniae, H. influenzae and N. meningitidis was developed and evaluated on BAL samples from adults with LRTI and a control group, and on CSF samples from patients with meningitis. To establish the detection capacity of the Spn9802, the P6 and the ctrA assays, serial dilutions of target DNA with known concentration were repeatedly tested and the analytical sensitivity was 10-60 copies per PCR reaction for the Spn9802 assay, 3-30 copies per PCR reaction for the P6 assay and 5-50 copies per PCR reaction for the ctrA assay. As shown in Table 2 the analytical sensitivity and quantification was not affected by using a combined mixture of reagents and a combined DNA standard ( S. pneumoniae, H. influenzae and N. meningitidis ) in single tubes. Table 3A shows results of tests for S. pneumoniae and H. influenzae in the patient group. Of 156 LRTI patients S. pneumoniae was identified by conventional tests in 21 (13%) cases, and by qmPCR in 54 (35%) cases, including 47 cases using a cut-off level of 10 5 copies/mL. From the 21 patients with conventional (blood culture, BAL culture, or urinary antigen test) tests positive for S. pneumoniae , 20 were positive by qmPCR. In addition 34 cases with no conventional test positive for S. pneumoniae were positive with Spn9802 PCR of which 26 were also positive by lytA PCR. Of the 6 patients with pneumococcal bacteraemia, S. pneumoniae was identified by BAL culture in one case, by urinary antigen test in one case, and by qmPCR and lytA PCR in all the 6 patients. Similarly, among the 9 patients with positive urinary antigen test, S. pneumoniae was identified in 8 by BAL qmPCR and in seven by lytA PCR, and none by BAL culture. H. influenzae was not found in any blood culture but was detected by BAL culture in 31 cases, of which 28 also were positive by qmPCR. Of 44 cases proved negative by culture but positive by qmPCR, 41 were confirmed by fucK PCR. Among the 31 control patients S. pneumoniae and H. influenzae were identified by BAL culture in 2 (6%) and 3 (10%) cases respectively, by qmPCR in 8 (26%) and 11 (35%) cases (Table 3B ). Of 7 and 8 cases proved negative by culture but positive with qmPCR for S. pneumoniae and H. influenzae respectively, 2 were positive by lytA PCR for S. pneumoniae and 7 were positive by fucK PCR for H. influenzae . Figure 1 shows the qmPCR copy number of the LRTI patients and controls compared to results by culture, urinary antigen test and lytA PCR. Among the qmPCR positive subjects, the LRTI patients and controls had a similar mean log 10 of copy number 5.69 (standard deviation [SD] 1.53) versus 5.65 (SD 1.63); p = 0.79, for H. influenzae and 6.31 (SD 1.12) versus 5.93 (SD 0.96); p = 0.36, for S. pneumoniae ). If the cut-off limit for a positive qmPCR result was risen to 10 5 DNA copies/mL, the positivity rate among the controls would drop from 26% (8/31) to 16% (5/31) for S. pneumoniae and from 35% (11/31) to 19% (6/31) for H. influenzae . Similarly in the patient group the positivity rate would drop from 35% (54/156) to 30% (47/156) for S. pneumoniae and from 46% (72/156) to 20% (31/156) for H. influenzae . Table 4 shows the sensitivities and specificities of the qmPCR, with the detection limit of the PCR assay itself and a detection limit of 10 5 copies/mL. Since one urinary antigen test positive patient had Spn9802 DNA determined at <10 5 copies/mL and three culture positive patients had P6 DNA determined at <10 5 copies/mL, the raise of the cut-off limit to 10 5 copies/mL, would drop the sensitivities, but the specificities would increase. Among 103 patients treated with antibiotic before sampling, S. pneumoniae and H. influenzae were identified by culture in 6% (6/103) and 20% (21/103) respectively, and by qmPCR in 36% (37/103) and 53% (55/103) respectively. Of 22 patients positive by Spn9802 PCR and lytA PCR alone 19 of them had antibiotics prior to sampling. Figure 2 shows the quantitative results of the qmPCR compared to semi-quantitative culture of BAL specimens for S. pneumoniae and H. influenzae . There was no correlation between the measured DNA copy number/mL and the bacterial growth. Table 5 shows results of tests for S. pneumoniae and N. meningitidis in patients with meningitis. Of 87 CSF samples, S. pneumoniae and N. meningitidis were detected by culture in 5 (6%) and 2 (2%) samples, by 16 S rRNA PCR in 14 (16%) and 10 (11%) and by qmPCR and in 14 (16%) and 10 (11%) samples respectively. Altogether, culture, 16 S rRNA PCR and qmPCR were positive for S. pneumoniae in 14 cases, N. meningitidis in 10 cases, and H. influenzae in no case. If culture and the 16 S rRNA PCR in combination were used as reference standard for aetiology of meningitis, the sensitivities and specificities would be 100% and 100% for both S. pneumoniae and N. meningitidis . Two samples positive by the ctrA PCR were positive in the unspecific 16 S rRNA PCR and sequence analysis of the PCR product determined them as Neisseria spp. They were considered as N. meningitidis in the specificity calculation. Of 17 samples that were negative by culture but positive by qmPCR 8 samples originated from patients that were under antibiotic treatment (Amoxyclav PO, Cefotaxime IV or Cefuroxime IV) before lumbar puncture.

Discussion In this study we established a sensitive detection system that enabled simultaneous quantification of S. pneumoniae, H. influenzae and N. meningitidis DNA using qmPCR. The multiplex assay was reproducible and no change in detection and quantification capacity was seen when a combined mixture of reagents and a combined DNA standard ( S. pneumoniae/H. influenzae/N. meningitidis ) in single tubes was used (Table 2 ). This multiplex PCR assay reduced the expense of reagents and the required time for analysis. Antibiotic treatment prior sampling has been found to reduce the positivity rate of BAL culture from 92% to 55% in patients with severe community acquired pneumonia [ 6 ]. In this study 66% (103/156) of the patients had antibiotic therapy prior the sampling, this high rate of antibiotic treatment is probably the reason for the suboptimal specificity of the qmPCR. Of 78 samples which were negative by culture and positive for S. pneumoniae or H. influenzae by the qmPCR, 64 were treated with antibiotic 0-7 days prior to sampling. The high rate of prior antibiotic treatment was probably also the reason for the lack of correlation between DNA concentration and bacterial concentration determined by semi-quantitative culture (Figure 2 ). This lack of correlation between quantification of target DNA and culture contrasts to our previous analysis of nasopharyngeal aspirates from community acquired pneumonia patients, where a significant correlation was seen, but only 25% of patients were on antibiotic treatment when samples were collected in the previous study [ 17 , 21 ]. The evaluation of nucleic acid amplification tests by comparison with less sensitive reference methods such as culture is problematic. Several imperfect tests may be used to define a composite reference standard [ 30 ]. An alternative way to resolve cases with different test results is to use discrepant analysis where an additional method is used to determine the specimen status. Such analyses have been criticized [ 31 ], but is often the most realistic procedure for the evaluation of new methods that are more sensitive than an established reference method. In our study the Spn9802 target was evaluated by a composite reference standard and for the P6 target discrepant analysis was used. This resulted in increased specificity and a higher number of pneumonia cases with defined etiology. As expected the positive predictive values increase with increased specificity, thus with a cut-off limit of 10 5 copies/mL a positive PCR result is very reliable using an extended reference standard. We have recently shown that Spn9802 PCR and P6 PCR are specific for S. pneumoniae and H. influenzae when bacterial strains have been tested. Nevertheless, colonization of S. pneumoniae and H. influenzae in the respiratory tract is problematic for both culture and PCR. To overcome this problem semi-quantitative culture is often used. In our study a detection limit of 10 5 DNA copies/mL for positive Spn9802 and P6 PCRs yielded a high specificity but somewhat reduced the sensitivity. Similar results have been seen in previous studies [ 6 , 32 , 33 ] based on BAL culture and demonstrated that a cut-off of 10 4 -10 5 CFU/mL allow differentiation between colonization and infection of the lower respiratory tract. However, CFU/mL does not automatically correspond to the number of DNA copies/mL since several bacteria may aggregate and generate one colony although they constitute several genome equivalents. Furthermore, as described above antibiotic treatment before sampling and smoking habits have an effect on the number of detected bacteria. Thus patient treatment and the patient group characteristics affect the possibility of using quantification to differentiate between colonization and infection. When the multiplex PCR was applied on CSF samples, our assay was able to detect all the cases of N. meningitidis and S. pneumoniae that were found by culture and/or 16 S PCR in a previous study [ 24 ]. The problem of choosing optimal targets for S. pneumonia and H. influenzae has been addressed above. The primer pair used for N. meningitidis in our assay has previously been used in a multiplex assay for detection of bacterial meningitis [ 14 ] and even been evaluated in a major interlaboratory comparison of PCR-based identification of meningococci [ 34 ] as well as in other studies with satisfying results [ 35 , 36 ].

Conclusions Although culture is still indispensable in bacteriological diagnostics multiplex PCR enables concurrent diagnostics of viruses and fungi and provides a powerful tool for analysis. We conclude that the multiplex format of the assay facilitates diagnostics of S. pneumoniae , H. influenzae and N. meningitidis and is suitable for analysis of both respiratory tract tract and CSF specimens. The assay also enable detection after antibiotic treatment has been installed. Quantification increases the specificity of etiology for pneumonia.

Background Streptococcus pneumoniae and Haemophilus influenzae cause pneumonia and as Neisseria meningitidis they are important agents of meningitis. Although several PCR methods have been described for these bacteria the specificity is an underestimated problem. Here we present a quantitative multiplex real-time PCR (qmPCR) for detection of S. pneumoniae (9802 gene fragment), H. influenzae ( omp P6 gene) and N. meningitidis (ctrA gene). The method was evaluated on bronchoalveolar lavage (BAL) samples from 156 adults with lower respiratory tract infection (LRTI) and 31 controls, and on 87 cerebrospinal fluid (CSF) samples from meningitis patients. Results The analytical sensitivity was not affected by using a combined mixture of reagents and a combined DNA standard ( S. pneumoniae/H. influenzae/N. meningitidis ) in single tubes. By blood- and BAL-culture and S. pneumoniae urinary antigen test, S. pneumoniae and H. influenzae were aetiological agents in 21 and 31 of the LTRI patients, respectively. These pathogens were identified by qmPCR in 52 and 72 of the cases, respectively, yielding sensitivities and specificities of 95% and 75% for S. pneumoniae , and 90% and 65% for H. influenzae , respectively. When using a cut-off of 10 5 genome copies/mL for clinical positivity the sensitivities and specificities were 90% and 80% for S. pneumoniae , and 81% and 85% for H. influenzae , respectively. Of 44 culture negative but qmPCR positive for H. influenzae , 41 were confirmed by fucK PCR as H. influenzae . Of the 103 patients who had taken antibiotics prior to sampling, S. pneumoniae and H. influenzae were identified by culture in 6% and 20% of the cases, respectively, and by the qmPCR in 36% and 53% of the cases, respectively. In 87 CSF samples S. pneumoniae and N. meningitidis were identified by culture and/or 16 S rRNA in 14 and 10 samples and by qmPCR in 14 and 10 samples, respectively, giving a sensitivity of 100% and a specificity of 100% for both bacteria. Conclusions The PCR provides increased sensitivity and the multiplex format facilitates diagnosis of S. pneumoniae , H. influenzae and N. meningitidis and the assay enable detection after antibiotic treatment has been installed. Quantification increases the specificity of the etiology for pneumonia.

Authors' contributions GA: BH, KS and JB have planned the study; GA has done the laboratory work and written the draft. KS, JK and CW have provided clinical materials. All authors have contributed intellectually during the writing process and have read and approved the final manuscript.

Acknowledgements The study was supported by funds from the Uppsala-Örebro Regional Research Council.

CC BY

no

2022-01-12 15:21:45

BMC Microbiol. 2010 Dec 3; 10:310

oa_package/3d/6f/PMC3016321.tar.gz

PMC3016322

21134263

Background Pseudorabies virus (PRV), an alpha-herpesvirus, and the causative agent of Aujeszky's diseases of swine [ 2 ], is a commonly used model organism for studies in pathogenesis and the molecular biology of herpesviruses. Furthermore, it is widely utilized as a neural circuit tracer [[ 3 , 4 ] and [ 5 ]] and has been reported to be suitable as a vector for gene delivery to various cells [ 6 , 7 ] and as an oncolytic agent [ 8 ]. The gene expressions of herpesviruses are currently undergoing intensive investigation in consequence of the development of new technologies allowing simultaneous analysis of the expressions of multiple genes. DNA microarray approaches have been applied for the overall analysis of herpesvirus gene expression in several studies [[ 9 , 10 ] and [ 11 ]]. Microchip techniques are powerful tools that permit simultaneous measurement of the relative changes in quantity of thousands of genes of an organism, and the comparison of gene expression profiles under various circumstances. Quantitative real-time RT-PCR is a much more sensitive and accurate method, but, at least at present, it is not well suited for the analysis of large numbers of samples. The herpesvirus genome however is, within the range that can be successfully analysed with this technique [ 1 ]. The program of herpesvirus gene expression is controlled at multiple levels by complex interactions between viral and cellular factors. The lytic gene expressions of herpesviruses are strictly coordinated in a sequential cascade manner and are traditionally subdivided into immediate-early (IE), early (E) and late (L) phases. IE proteins are involved in the control of the synthesis of E and L genes. The IE180 gene ( ie180 ; homologue of the HSV ICP4 gene) is the only IE gene of PRV, and the most important regulator of viral gene expression. The E genes of herpesviruses are involved in various aspects of DNA synthesis, while most L genes mainly encode the structural elements of the virus. The antisense transcripts LLT (long latency transcript) and LAT (latency-associated transcript) overlapping the ICP4 and ICP0 (a homologue of ep0 in PRV), respectively, are reported to play important roles in the establishment of latency in HSV [ 12 ]. It has not yet been unequivocally clarified whether the expression of antisense transcript produced by the complementary DNA strand of the ie180 gene is controlled solely by the LAP (LAT promoter) producing LLT or also by a putative promoter (antisense promoter, ASP) localized on the inverted repeat of the PRV genome, producing a shorter transcript. In this study, we use the term 'antisense transcript' (AST) for the RNA molecule transcribed from the complementary DNA strand of the ie180 gene. It is well known that both the host response and the success of a pathogen are dependent on the quantity of particles infecting an organism; and, specifically in herpesviruses, the infecting dose determines whether the virus enters a latent state or induces an acute infection [ 13 ]. A further important question is whether the global gene expression profile of the virus genome is dependent on the number of virus particles entering the cells. In both traditional and microarray studies, herpesvirus gene expression has been analysed by using a relatively high multiplicity of infection, typically MOI~10 plaque-forming unit (pfu)/cell [ 9 - 11 ]. Theoretically, it is possible that herpesviruses express their genomes in a different manner when only a single virus particle infects a cell as compared with the situation when multiple virions enter a cell. In the present study, we addressed this issue by using low (0.1 pfu/cell) and high (10 pfu/cell) MOIs for the infection of cultured porcine kidney epithelial cells with wild-type PRV, and subsequently analysed and compared the expressions of 37 PRV genes and two antisense transcripts (AST and LAT) using the SYBR Green-based real-time RT-PCR technique.

Methods Virus, cells and infection Strain Kaplan (Ka) of pseudorabies virus (PRV) was used in our analyses. Immortalized porcine kidney (PK)-15 epithelial cells were applied for propagation of the virus. PK-15 cells were cultivated in Dulbecco's modified Eagle medium supplemented with 5% foetal bovine serum (Gibco Invitrogen) and 80 μ g gentamycin per ml at 37°C in the presence of CO 2 . The virus stock used for the experiments was prepared as follows. Rapidly-growing semi-confluent PK-15 epithelial cells were infected at an MOI of 0.1 pfu/cell and were incubated until a complete cytopathic effect was observed. The cell debris was removed by low-speed centrifugation (10,000 g for 20 min). The supernatant was concentrated and further purified by ultracentrifugation through a 30% sugar cushion at 24,000 rpm for 1 h, using a Sorvall AH-628 rotor. The number of cells in a culture flask (Corning, 150 cm 2 ) was 5 × 10 6 . In high-MOI and in low-MOI experiments, 5 × 10 7 and 5 × 10 5 pfu viral particles, respectively, were applied for the infections. Thus, in the high-MOI experiment, practically all the cells were infected, while in the low-MOI experiment, approximately 5 × 10 5 cells (10% of the cells in a culture flask) were infected by the virus. We used the same data for the low-MOI experiment as in a previous publication [ 1 ]. The two experiments were run simultaneously. We ran four independent sets of measurements for each time point in both low and high-MOI studies, but occasionally we had to remove data because of low amplification efficiencies or the amplification of non-specific products in the reaction." Thus, in some genes, instead of four, we only used three independent data. Infected cells were incubated for 1 h, followed by removal of the virus suspension and washing with phosphate-buffered saline. After the addition of new medium to the cells, they were incubated for 0, 1, 2, 4 or 6 h. In this study, mock-infected cells were used as controls, which were otherwise treated in the same way as the infected cells. Isolation of RNAs RNA was extracted by using the NucleoSpin RNA II Kit (Macherey-Nagel GmbH and Co. KG), as described previously [ 1 ]. Briefly, after the cells had been collected by centrifugation and lysed by buffer containing chaotropic ions, the nucleic acids were docked to a silica column. The DNA was removed with RNase-free DNase solution (supplied with the NucleoSpin RNA II Kit). Finally, the RNAs were eluted from the column in RNase-free water (supplied with the kit). To eliminate the residual DNA contamination, all RNA samples were treated by an additional digestion with Turbo DNase (Ambion Inc.). The concentrations of the RNA samples were measured by spectrophotometric analysis with a BioPhotometer Plus instrument (Eppendorf). RNA samples were stored at -80°C until further use. Reverse transcription 0.07 μ g of total RNA was reverse transcribed in a 5 μ l reaction volume, using SuperScript III Reverse Transcriptase (Invitrogen), a gene-specific primer (1 μ l), dNTP mix (0.25 μ l; 10 μ M final concentration), and 5× First-Strand Buffer (1 μ l). The reaction mix was incubated at 55°C for 60 min and incubation was stopped by holding at 70°C for 15 min. A no-RT control reaction was run to ensure that the RNA samples were free of DNA contamination. For the quantitative RT-PCR reactions, only DNA-free RNA samples were used. First-strand cDNAs were diluted 10-fold with Nuclease-Free Water (Promega Corp.) and stored at -80°C until use. The same primers were used for the RT reaction as in our previous publication [ 1 ]. Real-time PCR A Rotor-Gene 6000 cycler (Corbett Life Science) was used for the real-time quantitative PCR analysis. Each reaction (20 μ l final volume) contained the following components: 7 μ l of cDNAs, 10 μ l of Absolute QPCR SYBR Green Mix (Thermo Fisher Scientific), 1.5 μ l of forward and reverse primer (10 μ M each; we used the same primer pairs as described earlier [ 1 ]). The PCR cycling parameters were as follows: 95°C for 15 min (pre-incubation), and then 30 cycles of 94°C for 25 sec (denaturation), 60°C for 25 sec (annealing), and 72°C for 6 sec (extension). The specific amplification products (with a single peak at the predicted temperature) were identified by melting-point curve analysis. An additional detection step was included in the cycle program to avoid primer dimer detection for those primer pairs that produce primer dimers. The reliability of the primers was verified in our earlier publication [ 1 ]. Porcine 28 S rRNA was used as a loading control throughout the experiment. H 2 O was included as a no-template control, and cDNA derived from the reverse-transcribed RNAs of non-infected cells was used as a negative mock-infected control. SYBR Green-based real-time PCR was applied in this study because of the low costs and simple protocol [ 51 ]. Data analysis The following formula was used for calculation of the relative expression ratio (R): where E is the efficiency of amplification, Ct is the cycle threshold value, 'sample' is the examined PRV gene, and 'ref' is the 28 S rRNA. The Comparative Quantitation module of the Rotor-Gene 6000 Software (Version 1.7.87., Corbett Research) was used to calculate the real-time PCR efficiency for each sample. Thresholds were set by the software. The R values of both low and high-titre infections were maximized to the 6 h E Ct values of the high-MOI experiment. To measure the net change in R between two consecutive time points, R Δ was calculated via the following formula: R Δ = R (t+1) -R t . The rate of change was calculated as follows: R a = R (t+1) /R t . Pearson's correlation was used for the analysis of the relationship between low- and high-titre infections using the following formula [ 52 ]: The correlation measures the linear relationship between two variables, X and Y. Pearson's coefficient (r) is a number ranging from -1 to +1 that measures the degree of association between X and Y. If X and Y are independent, Pearson's correlation coefficient is 0. A positive r value for the correlation implies a positive association (large values of X tend to be associated with large values of Y, and small values of X tend to be associated with small values of Y). A negative value for the correlation means an inverse association (large values of X tend to be associated with small values of Y, and vice versa). In the analysis of the relationship between the low and high-titre infections, is the average R value of the low-titre infection at a given time point, and is the average R value at the same time point in the high-titre infection. S X and S Y are the SEM (standard error of the mean) values and n is the sample number.

Results and Discussion Experimental design In this study, PK-15 cells were infected with pseudorabies virus at MOIs of 0.1 and 10. Albeit the difference in the infectious dose in the two parallel experiments was 100-fold, an individual cell was invaded by only 10 times more virus particles in the high-MOI than in the low-MOI experiment (5 × 10 6 versus 5 × 10 5 infected cells), the reason for this being that in the latter case approximately 90% of the cells remained uninfected. Cells were harvested at 0, 1, 2, 4 and 6 h post-infection (pi), as in our earlier report [ 1 ]. We used 6 h as the maximum infection period in order to exclude the possibility of the initiation of new infection cycles in the low-MOI experiment. In this study, we analysed the expression of 37 genes (53% of the total PRV genes) and two antisense transcripts (AST and LAT) (Figure 1 and 2 [ 14 - 45 ]). For the calculation of relative expression ratios (Rs) (Additional file 1a ), we used the average 6 h E Ct values of the high-MOI experiments of both the "samples" and the "references" as controls, as in our earlier publication [ 1 ]. We used a correction factor of 10 for the calculation of R for the low-MOI experiment (Additional file 2a ). With this calculation technique, approximately the same numbers of infected cells, and hence the relative amounts of transcripts in an average infected cell, were compared in the two experiments. However, in the high-MOI experiment, the proportion of the genome copy number in an infected cell was also 10-fold higher on average, at least before the start of viral DNA replication (the first 2 h pi), the reason for this being that in the high-MOI experiment 10 virus particles infected an average cell, while in the low-MOI infection 10 per cent of the cells were infected with a single virus particle. Thus, to compare the gene expressions from a single virus DNA per cell, two normalizations are necessary: multiplication of the R values of the low-MOI data by 10, and division of the R values of the high-MOI data by 10. In some calculations, the original data were handled accordingly (see the indications in the particular cases). The relationship between the infectious dose and the genome copy number of PRV becomes non-linear in later stages of viral infection; the DNA copy numbers in the two experimental situations are therefore not comparable on the basis of the infectious dose. The R values of LAT and AST were calculated by using the 6 h E Ct values of the corresponding genes, ep0 and ie180 , respectively, as the reference gene. R Δ values were used to monitor the net change in the quantity of viral transcripts within a given period of time (Additional file 2b ). R a shows the ratio of the changes in the amounts of transcripts between two adjacent time points (Additional file 2c ). We considered two principles for the selection of genes for expression analyses. (1) We analysed the upstream genes of each nested gene cluster, the reason for this being that these genes are not overlapped by other genes, and the amounts of these transcripts are therefore proportional to their protein products. This is in contrast with the downstream genes, which, if transcribed from the promoter of an upstream gene, are not translated, because they do not have cap sequences that are required for the recognition by the ribosomes. (2) Furthermore, we analysed genes that are of primary importance in the regulation of global viral gene expression, such as ie180 , ep0 , vhs and ul54 . Gene expressions in the early stage of PRV infection In the first 2 h of infection, the viral DNA replication has not yet been initiated, and the copy number of viral genomes in a cell therefore corresponds with the infectious dose. In this analysis, we found that the mRNA levels of most examined PRV genes were higher in the cells infected with the high MOI than in those infected with the low MOI (Additional file 2a ) at both 1 h and 2 h pi. This was not unexpected since in the former case viral DNAs were represented in an approximately 10-fold higher proportion in an average infected cell. Exceptions to this were the transcripts ul1 , ul33 , and ul51 mRNAs at 1 h pi, and ul36, ul38, ul43 , and ul48 mRNAs at 2 h pi, and at both 1 h and 2 h: ie180 and ul30 mRNAs, as well as, LAT and AST. However, the expression levels normalized to the genome copy number (i.e. using R/10 values in the high-MOI infection) showed an inverse pattern: only a few genes were expressed at higher abundance in the high-MOI than in low-MOI infection (Additional file 2a ). AST was expressed at a considerably higher quantity in the cells infected with the low MOI than in those infected with the high MOI (R low MOI /R high MOI = 111-fold at 1 h, and 298-fold at 2 h pi). The expression rate of a single genomic region encoding the AST was even 10 times higher (1 h: 1110-fold and 2 h: 2980-fold) in the low-dose infection experiment (Additional file 2a ). In the high-dose infection 6 of the 37 genes ( ie180 , ul36 , ul50 , ul54 , us1 , and ul24 ) exhibited higher expression levels at 1 h than at 2 h pi. It should be noted that 3 of them ( ie180 , us1 and ul54 ) are regulatory genes. The fourth regulatory PRV gene, ep0 , is expressed at a very high level during the first 2 h in the high-MOI infection (R 1 h = 1.87, R 2 h = 2.05). Apart from ep0 , ul5 (R 2 h = 1.2) was the only gene that was expressed at a higher extent in the early stages of infection than at 6 h pi in the high-MOI experiment. The ie180 gene is the only one that was expressed in a higher amount at 1 h than at 2 h pi under both experimental conditions (Additional file 2 ). Overall, it appears that the 4 regulatory genes were expressed at relatively high levels before the onset of DNA replication in the high-MOI infection, which was not the case in low-MOI infection, with the exception of the ie180 gene. We think that the reason for the higher expression of regulatory genes at the onset of viral DNA replication in the high-MOI infection is that more regulatory proteins are needed to carry out the multiplication of a higher copy number of the viral genome. The rate of change in gene expression within the 1 h to 2 h interval (R 2h /R 1h ) was higher in more than two-thirds of the PRV genes (25/37) in the low-MOI than in the high-MOI infection (Additional file 2c ). The proportion of AST to ie180 mRNA molecules (R AST /R ie180 ) was 0.47 at 1 h pi, and 4.72 at 2 h pi in cells infected with the low MOI, while this ratio was extremely low (~0.01) at both 1 h and 2 h pi in the high-MOI infection (these data are only semi-quantitative since the primer efficiencies in the RT reaction are not necessarily equal for the two transcripts). Thus, the proportion of AST to ie180 mRNA [(R AST-low MOI /R ie-low MOI )/(R AST-high MOI )/R ie-high MOI )] was 39-fold higher at 1 h pi and 293-fold higher at 2 h pi in the low-MOI than in the high-MOI infection. In the early stages of PRV infection, the amount of AST was very high; it even significantly exceeded the level of ie180 mRNAs at 2 h pi in the low-MOI infection, while the amount of AST and also its ratio to ie180 mRNA were extremely low in the high-MOI infection. Moreover, ie180 mRNA is expressed to a significantly higher extent in the low-MOI experiment despite the 10 times lower copy number of PRV DNA in an infected cell, which is especially important because IE180 is a DNA-binding protein. We think that this observation reveals an important regulatory mechanism of the herpesviruses, which is as follows: in a high-titre infection, the virus initiates a lytic infection in a cell, while in a low-titre infection, the virus has the choice of whether to establish a dormant state or enter a lytic cycle in a cell. The molecular mechanism of this phenomenon might be based on the interaction of ie180 and AST genes at both the transcription and translation levels. (1) The ie180 protein might exert a negative effect on the synthesis of AST, such as in LAT in HSV [ 46 ] by binding the promoter of the antisense transcript. (2) Furthermore, the complementary transcripts might mutually influence each other's expression transcript by RNA-RNA interaction. In a low-MOI infection, the two transcripts exhibit a complementary expression pattern, which indicates a competition between the two transcripts. In a high-MOI infection, however, the high initial amount of ie180 gene product inhibits the expression of AST. The significance of this infection strategy could be that, in the case of a low-amount infection, the virus has no chance to invade the host cells; therefore, it is better to hide against the immune surveillance. The ep0 gene is expressed in higher quantity at both 1 h pi (4.22-fold) and 2 h pi (2.43-fold) in the high-MOI infection than in low-MOI infection, which is in contrast with LAT, its antisense partner, whose expression level was lower in the high-MOI infection (1 h: 0,5-fold; 2 h: 0,18-fold). Thus, the ratios of LAT to ep0 mRNA molecules were 8.33-fold higher at 1 h pi and 13.05-fold higher at 2 h pi in the low-MOI than in the high-MOI experiment, although, unlike AST, LAT is abundantly expressed in the high-MOI infection. Accordingly, similarly to AST, LAT is expressed in a significantly higher proportion to ep0 mRNA in the low-MOI infection in the early stages of infection, which may also be important as concerns of the replication strategy of the virus. Our analyses additionally showed that AST and LAT are, at least partly, expressed independently from each other, which supports the existence of separate elements controlling the expressions of the two antisense transcripts. Indeed, AST was suggested to be controlled by an antisense promoter (ASP) localized in the outer regions of inverted repeats [ 47 ]. Gene expressions in later stages of PRV infection At 4 h pi the transcript levels of more than three-quarters of the PRV genes (28/37) were still higher in the cells infected with the high MOI than in those infected with the low MOI (Additional file 2c ). However, in about two-third of the viral genes the rate of change (R a values) in the expression level was higher in the low-MOI than in the high-MOI infection (24/37 within the 2 h to 4 h period, and 25/37 within the 1 h to 4 h period) (Additional file 2c ). In the low-MOI infection, the amounts of 5 transcripts ( ul5 , ul44 , us1 and us6 ) were less than 10% of those in the high-MOI infection at 4 h pi. All of the examined us genes are expressed at a significantly lower level in the low-than in the high-titre infection at 4 h pi. There were significant decreases in the quantities of both AST and LAT in the low-titre infection at 4 h pi relative to the 2 h values (AST: a 59-fold decrease, and LAT: a 7-fold decrease). We explain this phenomenon by the negative effect of the regulatory genes on their antisense partners. Regulatory genes are upregulated at the onset of DNA replication (in order to facilitate this process), which exerts an inhibitory effect on the expression of AST and LAT. In contrast, there were increases in the amounts of antisense transcripts in the high-MOI (AST: an 11-fold increase, and LAT: a 7-fold increase) in this time interval. However, while LAT was expressed at high level (R = 1.3) under the high-MOI conditions, the AST expression remained extremely low (R = 0.013) in this period of infection. The amount of the ie180 transcript was practically unchanged within the 2 h to 4 h infection period under either infection conditions. There was a 4.7-fold increase in the ep0 mRNA level within the 2 h to 4 h infection period (R 4h /R 2h ) in the low-MOI infection, as compared with only 1.4 in the high-MOI experiment. On average, the amounts of mRNAs in low titre infection became higher than those in the high-infection titre by 6 h pi in more than half of the PRV genes (22/37). We assume that the reason for this might be that the ie180 gene, the major coordinator of gene expression, is expressed at higher levels at 4 and 6 h pi at low-MOI than at high-MOI infection. Moreover, in the high-MOI infection the amount of AST reached almost 30% of the transcript level in the low-MOI infection, while LAT was expressed at approximately the same level under the two infection conditions at 6 h pi. The genes expressed at lower levels in the low-dose infection appeared to be clustered on adjacent genomic locations (Figure 1 ). Each gene and the two antisense transcripts were expressed at higher rates (R a values) within the 4 h to 6 h period in the low-MOI than in the high-MOI infection without exception (Additional file 1c ) . In the high-MOI infection, 11 genes and LAT peaked at 4 h within the 6-h examination period, while in the low-MOI infection only the us3 transcript had a slightly lower R value at 6 h than at 4 h pi. The us3 gene was the only one among the 70 PRV genes which was expressed at a higher level at 4 h than at 6 h pi in another study [ 1 ]. Intriguingly, the ep0 mRNAs reached a 3.5-fold higher level in the low-dose than in the high-dose infection in an average cell at 6 h pi. Furthermore, at 6 h pi the ul1 and ul51 genes were expressed at an approximately 10 times higher level under the low-MOI than under the high-MOI conditions. Gene expression kinetics within the 0 to 6-h infection period The expression of most PRV genes basically differed under the two infection conditions (Additional file 1c ), which is in contrast with the case of rhesus monkey rhadinovirus (a γ-herpesvirus), whose lytic gene expression commences at a fixed pace in infected cells, regardless of the MOI [ 48 ]. Most genes were expressed at a lower level in a cell in the low-MOI experiment in the first 4 h of infection, but more than half of these gene products surpassed the high-MOI values by 6 h pi. The R values of 3 PRV genes ( ie180, ul1 and ul30 ) were higher in the low-MOI than in the high-MOI infection at every examined time point, while the opposite was true (the R values of high-MOI were always higher) in 13 genes: ul5, ul15, ul17, ul19, ul23, ul24, ul44, ul49.5, ul54, us6, us9, us1 and us3 (Figure 3 ). These latter genes form clusters on the basis of their localization on the genome (genes in close vicinity are underlined), which suggests that the adjacent genomic sequences might be under common regulatory control. This observation is supported by the similarity of the R a curves of adjacent genes (Additional file 1c ). For example, the expression rates of the ul36 , ul37 and ul38 genes were similar to each other in both experiments, but each of them exhibited an inverse expression pattern in the two infection conditions. All genes were expressed at a higher rate (R a ) within the 1 h to 6 h period of infection in the low-titre experiment, except for ie180 and the two antisense transcripts. The quantities of ie180 mRNAs were similar in the two experiments, except at 1 h pi, where the level of the transcripts was 2.8-fold higher in the low-MOI infection. Thus, the amount of total ie180 transcript in an infected cell appears to be under strict control, independently of the initial infection conditions. In contrast, the expression of the ep0 gene differed basically in the two experiments. In the high-MOI experiment, the amount of ep0 mRNAs was high from the first hour of infection, and its expression even declined by 6 h in the high-MOI infection, while the amount of these transcripts rapidly increased throughout the 6-h infection period in the low-dose infection, and reached a 3.5-fold higher level compared to that of the high-dose infection by 6 h. (Figure 4 ) The ratio of sense and antisense transcripts during the 6-h infection period displayed intriguing patterns. First of all, in the high-MOI infection the amount of AST and its ratio to ie180 mRNA were very low throughout the 6-h infection period. We demonstrated an inverse relationship in the expression kinetics of ie180 mRNA and AST and also ep0 mRNA and LAT in the low-MOI infection; however, we did not observe this inverse relationship between the complementary transcripts under the high-MOI conditions (Figure 5 ). In an earlier report [ 1 ], we showed that treatment of infected cells with cycloheximide (a protein synthesis blocker) resulted in significant increases in the amounts of both ie180 mRNA and AST, while phosphonoacetic acid (a DNA synthesis inhibitor) treatment led to a decrease in ie180 mRNA and a significant increase in the AST level. These results suggest a negative effect of the IE180 transactivator on ASP synthesis. We explain the huge drop in ASP level in the infected cells in the early stage of the high-MOI infection by the presence of a 10-fold higher amount of inhibitory IE180 protein localized in the tegument of the infecting virions [ 49 ]. The same reason could account for the lower ie180 mRNA level in the high-MOI infection. The us1 gene was expressed in the late kinetics in our earlier low-MOI analysis in both phophonoacetic acid-treated and non-treated samples. These results are in concordance with those of the present high-dose infection experiment, i.e. us1 mRNA was expressed at a relatively low level at 1 h, which even dropped by 2 h pi. The highest rate of us1 mRNA expression was observed at 4 h, with a rate (R 4 h/2 h = 13.32) typical of L genes. The Pearson correlation coefficients of the R, R Δ , and R a values precisely show the degree of similarity (or differences) of the expression kinetics of the genes in the low- and high-MOI experiments (Additional file 3 ). Several genes exhibited high correlations for all three parameters. For example, the ie180 , ul19 , ul21 , ul22 , ul42 and ul43 genes gave high correlation coefficients for the R, R Δ and R a values. The us1 gene behaved in an irregular manner; it gave a relatively high correlation for the R values, no correlation of R Δ , and an inverse correlation for the R a values. AST yielded relatively high negative values for all three parameters, indicating a significant negative correlation. The expressions of LAT under the two experimental conditions did not correlate on the basis of the R values, whereas it gave a very high negative correlation for its R Δ and R a values. The effect of the MOI on the overall gene expression of HSV-1 has been investigated by Wagner and colleagues [ 50 ], who found that, following the infection of cultured cells by wild-type virus at MOIs ranging from 0.05 to 5 pfu/cell, the temporal profiles of transcript abundance were essentially the same. This is in sharp contrast with our results. We explain the differences by the low resolution of the microarray technique that Wagner et al. used for their analysis. An analysis of the global transcription of Rhesus monkey rhadinovirus, a γ-herpesvirus, has revealed differential gene expression at different MOIs [ 48 ], but these data cannot be compared because they related to later time points (12, 24, 48 72 and 96 h) than in our analysis.

Results and Discussion Experimental design In this study, PK-15 cells were infected with pseudorabies virus at MOIs of 0.1 and 10. Albeit the difference in the infectious dose in the two parallel experiments was 100-fold, an individual cell was invaded by only 10 times more virus particles in the high-MOI than in the low-MOI experiment (5 × 10 6 versus 5 × 10 5 infected cells), the reason for this being that in the latter case approximately 90% of the cells remained uninfected. Cells were harvested at 0, 1, 2, 4 and 6 h post-infection (pi), as in our earlier report [ 1 ]. We used 6 h as the maximum infection period in order to exclude the possibility of the initiation of new infection cycles in the low-MOI experiment. In this study, we analysed the expression of 37 genes (53% of the total PRV genes) and two antisense transcripts (AST and LAT) (Figure 1 and 2 [ 14 - 45 ]). For the calculation of relative expression ratios (Rs) (Additional file 1a ), we used the average 6 h E Ct values of the high-MOI experiments of both the "samples" and the "references" as controls, as in our earlier publication [ 1 ]. We used a correction factor of 10 for the calculation of R for the low-MOI experiment (Additional file 2a ). With this calculation technique, approximately the same numbers of infected cells, and hence the relative amounts of transcripts in an average infected cell, were compared in the two experiments. However, in the high-MOI experiment, the proportion of the genome copy number in an infected cell was also 10-fold higher on average, at least before the start of viral DNA replication (the first 2 h pi), the reason for this being that in the high-MOI experiment 10 virus particles infected an average cell, while in the low-MOI infection 10 per cent of the cells were infected with a single virus particle. Thus, to compare the gene expressions from a single virus DNA per cell, two normalizations are necessary: multiplication of the R values of the low-MOI data by 10, and division of the R values of the high-MOI data by 10. In some calculations, the original data were handled accordingly (see the indications in the particular cases). The relationship between the infectious dose and the genome copy number of PRV becomes non-linear in later stages of viral infection; the DNA copy numbers in the two experimental situations are therefore not comparable on the basis of the infectious dose. The R values of LAT and AST were calculated by using the 6 h E Ct values of the corresponding genes, ep0 and ie180 , respectively, as the reference gene. R Δ values were used to monitor the net change in the quantity of viral transcripts within a given period of time (Additional file 2b ). R a shows the ratio of the changes in the amounts of transcripts between two adjacent time points (Additional file 2c ). We considered two principles for the selection of genes for expression analyses. (1) We analysed the upstream genes of each nested gene cluster, the reason for this being that these genes are not overlapped by other genes, and the amounts of these transcripts are therefore proportional to their protein products. This is in contrast with the downstream genes, which, if transcribed from the promoter of an upstream gene, are not translated, because they do not have cap sequences that are required for the recognition by the ribosomes. (2) Furthermore, we analysed genes that are of primary importance in the regulation of global viral gene expression, such as ie180 , ep0 , vhs and ul54 . Gene expressions in the early stage of PRV infection In the first 2 h of infection, the viral DNA replication has not yet been initiated, and the copy number of viral genomes in a cell therefore corresponds with the infectious dose. In this analysis, we found that the mRNA levels of most examined PRV genes were higher in the cells infected with the high MOI than in those infected with the low MOI (Additional file 2a ) at both 1 h and 2 h pi. This was not unexpected since in the former case viral DNAs were represented in an approximately 10-fold higher proportion in an average infected cell. Exceptions to this were the transcripts ul1 , ul33 , and ul51 mRNAs at 1 h pi, and ul36, ul38, ul43 , and ul48 mRNAs at 2 h pi, and at both 1 h and 2 h: ie180 and ul30 mRNAs, as well as, LAT and AST. However, the expression levels normalized to the genome copy number (i.e. using R/10 values in the high-MOI infection) showed an inverse pattern: only a few genes were expressed at higher abundance in the high-MOI than in low-MOI infection (Additional file 2a ). AST was expressed at a considerably higher quantity in the cells infected with the low MOI than in those infected with the high MOI (R low MOI /R high MOI = 111-fold at 1 h, and 298-fold at 2 h pi). The expression rate of a single genomic region encoding the AST was even 10 times higher (1 h: 1110-fold and 2 h: 2980-fold) in the low-dose infection experiment (Additional file 2a ). In the high-dose infection 6 of the 37 genes ( ie180 , ul36 , ul50 , ul54 , us1 , and ul24 ) exhibited higher expression levels at 1 h than at 2 h pi. It should be noted that 3 of them ( ie180 , us1 and ul54 ) are regulatory genes. The fourth regulatory PRV gene, ep0 , is expressed at a very high level during the first 2 h in the high-MOI infection (R 1 h = 1.87, R 2 h = 2.05). Apart from ep0 , ul5 (R 2 h = 1.2) was the only gene that was expressed at a higher extent in the early stages of infection than at 6 h pi in the high-MOI experiment. The ie180 gene is the only one that was expressed in a higher amount at 1 h than at 2 h pi under both experimental conditions (Additional file 2 ). Overall, it appears that the 4 regulatory genes were expressed at relatively high levels before the onset of DNA replication in the high-MOI infection, which was not the case in low-MOI infection, with the exception of the ie180 gene. We think that the reason for the higher expression of regulatory genes at the onset of viral DNA replication in the high-MOI infection is that more regulatory proteins are needed to carry out the multiplication of a higher copy number of the viral genome. The rate of change in gene expression within the 1 h to 2 h interval (R 2h /R 1h ) was higher in more than two-thirds of the PRV genes (25/37) in the low-MOI than in the high-MOI infection (Additional file 2c ). The proportion of AST to ie180 mRNA molecules (R AST /R ie180 ) was 0.47 at 1 h pi, and 4.72 at 2 h pi in cells infected with the low MOI, while this ratio was extremely low (~0.01) at both 1 h and 2 h pi in the high-MOI infection (these data are only semi-quantitative since the primer efficiencies in the RT reaction are not necessarily equal for the two transcripts). Thus, the proportion of AST to ie180 mRNA [(R AST-low MOI /R ie-low MOI )/(R AST-high MOI )/R ie-high MOI )] was 39-fold higher at 1 h pi and 293-fold higher at 2 h pi in the low-MOI than in the high-MOI infection. In the early stages of PRV infection, the amount of AST was very high; it even significantly exceeded the level of ie180 mRNAs at 2 h pi in the low-MOI infection, while the amount of AST and also its ratio to ie180 mRNA were extremely low in the high-MOI infection. Moreover, ie180 mRNA is expressed to a significantly higher extent in the low-MOI experiment despite the 10 times lower copy number of PRV DNA in an infected cell, which is especially important because IE180 is a DNA-binding protein. We think that this observation reveals an important regulatory mechanism of the herpesviruses, which is as follows: in a high-titre infection, the virus initiates a lytic infection in a cell, while in a low-titre infection, the virus has the choice of whether to establish a dormant state or enter a lytic cycle in a cell. The molecular mechanism of this phenomenon might be based on the interaction of ie180 and AST genes at both the transcription and translation levels. (1) The ie180 protein might exert a negative effect on the synthesis of AST, such as in LAT in HSV [ 46 ] by binding the promoter of the antisense transcript. (2) Furthermore, the complementary transcripts might mutually influence each other's expression transcript by RNA-RNA interaction. In a low-MOI infection, the two transcripts exhibit a complementary expression pattern, which indicates a competition between the two transcripts. In a high-MOI infection, however, the high initial amount of ie180 gene product inhibits the expression of AST. The significance of this infection strategy could be that, in the case of a low-amount infection, the virus has no chance to invade the host cells; therefore, it is better to hide against the immune surveillance. The ep0 gene is expressed in higher quantity at both 1 h pi (4.22-fold) and 2 h pi (2.43-fold) in the high-MOI infection than in low-MOI infection, which is in contrast with LAT, its antisense partner, whose expression level was lower in the high-MOI infection (1 h: 0,5-fold; 2 h: 0,18-fold). Thus, the ratios of LAT to ep0 mRNA molecules were 8.33-fold higher at 1 h pi and 13.05-fold higher at 2 h pi in the low-MOI than in the high-MOI experiment, although, unlike AST, LAT is abundantly expressed in the high-MOI infection. Accordingly, similarly to AST, LAT is expressed in a significantly higher proportion to ep0 mRNA in the low-MOI infection in the early stages of infection, which may also be important as concerns of the replication strategy of the virus. Our analyses additionally showed that AST and LAT are, at least partly, expressed independently from each other, which supports the existence of separate elements controlling the expressions of the two antisense transcripts. Indeed, AST was suggested to be controlled by an antisense promoter (ASP) localized in the outer regions of inverted repeats [ 47 ]. Gene expressions in later stages of PRV infection At 4 h pi the transcript levels of more than three-quarters of the PRV genes (28/37) were still higher in the cells infected with the high MOI than in those infected with the low MOI (Additional file 2c ). However, in about two-third of the viral genes the rate of change (R a values) in the expression level was higher in the low-MOI than in the high-MOI infection (24/37 within the 2 h to 4 h period, and 25/37 within the 1 h to 4 h period) (Additional file 2c ). In the low-MOI infection, the amounts of 5 transcripts ( ul5 , ul44 , us1 and us6 ) were less than 10% of those in the high-MOI infection at 4 h pi. All of the examined us genes are expressed at a significantly lower level in the low-than in the high-titre infection at 4 h pi. There were significant decreases in the quantities of both AST and LAT in the low-titre infection at 4 h pi relative to the 2 h values (AST: a 59-fold decrease, and LAT: a 7-fold decrease). We explain this phenomenon by the negative effect of the regulatory genes on their antisense partners. Regulatory genes are upregulated at the onset of DNA replication (in order to facilitate this process), which exerts an inhibitory effect on the expression of AST and LAT. In contrast, there were increases in the amounts of antisense transcripts in the high-MOI (AST: an 11-fold increase, and LAT: a 7-fold increase) in this time interval. However, while LAT was expressed at high level (R = 1.3) under the high-MOI conditions, the AST expression remained extremely low (R = 0.013) in this period of infection. The amount of the ie180 transcript was practically unchanged within the 2 h to 4 h infection period under either infection conditions. There was a 4.7-fold increase in the ep0 mRNA level within the 2 h to 4 h infection period (R 4h /R 2h ) in the low-MOI infection, as compared with only 1.4 in the high-MOI experiment. On average, the amounts of mRNAs in low titre infection became higher than those in the high-infection titre by 6 h pi in more than half of the PRV genes (22/37). We assume that the reason for this might be that the ie180 gene, the major coordinator of gene expression, is expressed at higher levels at 4 and 6 h pi at low-MOI than at high-MOI infection. Moreover, in the high-MOI infection the amount of AST reached almost 30% of the transcript level in the low-MOI infection, while LAT was expressed at approximately the same level under the two infection conditions at 6 h pi. The genes expressed at lower levels in the low-dose infection appeared to be clustered on adjacent genomic locations (Figure 1 ). Each gene and the two antisense transcripts were expressed at higher rates (R a values) within the 4 h to 6 h period in the low-MOI than in the high-MOI infection without exception (Additional file 1c ) . In the high-MOI infection, 11 genes and LAT peaked at 4 h within the 6-h examination period, while in the low-MOI infection only the us3 transcript had a slightly lower R value at 6 h than at 4 h pi. The us3 gene was the only one among the 70 PRV genes which was expressed at a higher level at 4 h than at 6 h pi in another study [ 1 ]. Intriguingly, the ep0 mRNAs reached a 3.5-fold higher level in the low-dose than in the high-dose infection in an average cell at 6 h pi. Furthermore, at 6 h pi the ul1 and ul51 genes were expressed at an approximately 10 times higher level under the low-MOI than under the high-MOI conditions. Gene expression kinetics within the 0 to 6-h infection period The expression of most PRV genes basically differed under the two infection conditions (Additional file 1c ), which is in contrast with the case of rhesus monkey rhadinovirus (a γ-herpesvirus), whose lytic gene expression commences at a fixed pace in infected cells, regardless of the MOI [ 48 ]. Most genes were expressed at a lower level in a cell in the low-MOI experiment in the first 4 h of infection, but more than half of these gene products surpassed the high-MOI values by 6 h pi. The R values of 3 PRV genes ( ie180, ul1 and ul30 ) were higher in the low-MOI than in the high-MOI infection at every examined time point, while the opposite was true (the R values of high-MOI were always higher) in 13 genes: ul5, ul15, ul17, ul19, ul23, ul24, ul44, ul49.5, ul54, us6, us9, us1 and us3 (Figure 3 ). These latter genes form clusters on the basis of their localization on the genome (genes in close vicinity are underlined), which suggests that the adjacent genomic sequences might be under common regulatory control. This observation is supported by the similarity of the R a curves of adjacent genes (Additional file 1c ). For example, the expression rates of the ul36 , ul37 and ul38 genes were similar to each other in both experiments, but each of them exhibited an inverse expression pattern in the two infection conditions. All genes were expressed at a higher rate (R a ) within the 1 h to 6 h period of infection in the low-titre experiment, except for ie180 and the two antisense transcripts. The quantities of ie180 mRNAs were similar in the two experiments, except at 1 h pi, where the level of the transcripts was 2.8-fold higher in the low-MOI infection. Thus, the amount of total ie180 transcript in an infected cell appears to be under strict control, independently of the initial infection conditions. In contrast, the expression of the ep0 gene differed basically in the two experiments. In the high-MOI experiment, the amount of ep0 mRNAs was high from the first hour of infection, and its expression even declined by 6 h in the high-MOI infection, while the amount of these transcripts rapidly increased throughout the 6-h infection period in the low-dose infection, and reached a 3.5-fold higher level compared to that of the high-dose infection by 6 h. (Figure 4 ) The ratio of sense and antisense transcripts during the 6-h infection period displayed intriguing patterns. First of all, in the high-MOI infection the amount of AST and its ratio to ie180 mRNA were very low throughout the 6-h infection period. We demonstrated an inverse relationship in the expression kinetics of ie180 mRNA and AST and also ep0 mRNA and LAT in the low-MOI infection; however, we did not observe this inverse relationship between the complementary transcripts under the high-MOI conditions (Figure 5 ). In an earlier report [ 1 ], we showed that treatment of infected cells with cycloheximide (a protein synthesis blocker) resulted in significant increases in the amounts of both ie180 mRNA and AST, while phosphonoacetic acid (a DNA synthesis inhibitor) treatment led to a decrease in ie180 mRNA and a significant increase in the AST level. These results suggest a negative effect of the IE180 transactivator on ASP synthesis. We explain the huge drop in ASP level in the infected cells in the early stage of the high-MOI infection by the presence of a 10-fold higher amount of inhibitory IE180 protein localized in the tegument of the infecting virions [ 49 ]. The same reason could account for the lower ie180 mRNA level in the high-MOI infection. The us1 gene was expressed in the late kinetics in our earlier low-MOI analysis in both phophonoacetic acid-treated and non-treated samples. These results are in concordance with those of the present high-dose infection experiment, i.e. us1 mRNA was expressed at a relatively low level at 1 h, which even dropped by 2 h pi. The highest rate of us1 mRNA expression was observed at 4 h, with a rate (R 4 h/2 h = 13.32) typical of L genes. The Pearson correlation coefficients of the R, R Δ , and R a values precisely show the degree of similarity (or differences) of the expression kinetics of the genes in the low- and high-MOI experiments (Additional file 3 ). Several genes exhibited high correlations for all three parameters. For example, the ie180 , ul19 , ul21 , ul22 , ul42 and ul43 genes gave high correlation coefficients for the R, R Δ and R a values. The us1 gene behaved in an irregular manner; it gave a relatively high correlation for the R values, no correlation of R Δ , and an inverse correlation for the R a values. AST yielded relatively high negative values for all three parameters, indicating a significant negative correlation. The expressions of LAT under the two experimental conditions did not correlate on the basis of the R values, whereas it gave a very high negative correlation for its R Δ and R a values. The effect of the MOI on the overall gene expression of HSV-1 has been investigated by Wagner and colleagues [ 50 ], who found that, following the infection of cultured cells by wild-type virus at MOIs ranging from 0.05 to 5 pfu/cell, the temporal profiles of transcript abundance were essentially the same. This is in sharp contrast with our results. We explain the differences by the low resolution of the microarray technique that Wagner et al. used for their analysis. An analysis of the global transcription of Rhesus monkey rhadinovirus, a γ-herpesvirus, has revealed differential gene expression at different MOIs [ 48 ], but these data cannot be compared because they related to later time points (12, 24, 48 72 and 96 h) than in our analysis.

Conclusion Our analysis has revealed that almost all of the examined PRV genes exhibited different expression dynamics under the two experimental conditions. Most PRV genes were expressed at a lower level in the low-MOI than in the high-MOI experiment in the early stages of infection; however, the reverse was true when the transcript levels were normalized to the genome copy numbers. In the low-MOI infection, slightly more than half of the PRV transcripts outran the high-MOI values by 6 h pi. The lower ie180 transcript per genome in the high-titre infection experiment might account for the lower level of global PRV gene expression per genome in the high-MOI infection. However, the expression of viral genes per DNA did not uniformly decrease; some genes even became more active in the high-MOI infection, which indicates the selective effect of the reduced availability of the IE180 protein. The most dramatic difference between the two MOI infections was observed in AST, which was expressed at a more than two log higher level in an infected cell in the low-MOI infection, which is a 3 log higher activity of a single DNA region encoding the ASP. The ratio of LAT/EP0 was also significantly lower in the high-than in the low-MOI infection. The reasons for and the mechanisms of these phenomena remain to be clarified. Furthermore, genes localized in adjacent regions on the viral genome exhibited similar expression properties, indicating the existence of synchronizing mechanisms of gene expression.

Background Herpesvirus genes are classified into distinct kinetic groups on the basis of their expression dynamics during lytic growth of the virus in cultured cells at a high, typically 10 plaque-forming units/cell multiplicity of infection (MOI). It has been shown that both the host response and the success of a pathogen are dependent on the quantity of particles infecting an organism. This work is a continuation of an earlier study [ 1 ], in which we characterized the overall expression of PRV genes following low-MOI infection. In the present study, we have addressed the question of whether viral gene expressions are dependent on the multiplicity of infection by comparing gene expressions under low and high-MOI conditions. Results In the present study, using a real-time RT-PCR assay, we address the question of whether the expression properties of the pseudorabies virus (PRV) genes are dependent on the number of virion particles infecting a single cell in a culture. Our analysis revealed a significant dependence of the gene expression on the MOI in most of these genes. Specifically, we found that most of the examined viral genes were expressed at a lower level at a low MOI (0.1) than at a high MOI (10) experiment in the early stage of infection; however, this trend reversed by six hour post-infection in more than half of the genes. Furthermore, in the high-MOI infection, several PRV genes substantially declined within the 4 to 6-h infection period, which was not the case in the low-MOI infection. In the low-MOI infection, the level of antisense transcript (AST), transcribed from the antiparallel DNA strand of the immediate-early 180 ( ie180 ) gene, was comparable to that of ie180 mRNA, while in the high-MOI experiment (despite the 10 times higher copy number of the viral genome in the infected cells) the amount of AST dropped by more than two log values at the early phase of infection. Furthermore, our analysis suggests that adjacent PRV genes are under a common regulation. This is the first report on the effect of the multiplicity of infection on genome-wide gene expression of large DNA viruses, including herpesviruses. Conclusion Our results show a strong dependence of the global expression of PRV genes on the MOI. Furthermore, our data indicate a strong interrelation between the expressions of ie180 mRNA and AST, which determines the expression properties of the herpesvirus genome and possibly the replication strategy (lytic or latent infection) of the virus in certain cell types.

Competing interests The authors declare that they have no competing interests. Authors' contributions JT carried out the standard and real-time PCR, the agarose and polyacrilamide gel electrophoresis, and the DNA sequencing, and participated in the evaluation of the primary data. DT took part by performing the reverse transcription reactions, purified PRV RNA, and propagated PK-15 cells. IT participated in performing the reverse transcription reactions. ZB coordinated the study, propagated viruses and isolated viral DNAs. All authors have read and approved the final manuscript. Supplementary Material

Acknowledgements This study was supported by Hungarian National Fund for Human Frontiers Science Program Young Investigator grant (No. RGY0073/2006) to Z.B.

CC BY

no

2022-01-12 15:21:45

BMC Microbiol. 2010 Dec 6; 10:311

oa_package/56/5a/PMC3016322.tar.gz

PMC3016323

21138563

Background Zoosporic plant pathogens in the phylum Oomycota of the Stramenopila kingdom include hundreds of devastating species that attack a broad range of economically important agricultural and ornamental crops as well as forest tree species [ 1 , 2 ]. These oomycetes, including Phytophthora and Pythium species, use motile zoospores for dispersal and plant infection [ 3 - 5 ]. Plant infection by zoosporic pathogens is often effective in nature despite the fact that the population density in primary inoculum sources is relatively low [ 6 - 9 ]. This has led to differing theories with regard to density-dependent zoospore behaviors and plant infection [ 10 - 17 ]. A recent study with Phytophthora nicotianae showed that plant infection may be regulated through zoosporic extracellular products in zoospore-free fluid (ZFF) which can promote infection by a single zoospore [ 18 ]. This indicates that the physical presence of the threshold density of zoospores at an infection site is not strictly required, and plant infection can be initiated efficiently through chemical communication by the population. However, it is not clear how such a process is carried out by a pathogen at its naturally occurring low population density, which would be unlikely to produce adequate levels of functional signals unless these signals were also produced by other organisms and readily accessible in the environment. Ca 2+ and autoinducer 2 (AI-2), two widespread and non-specific signaling molecules, are known to be produced by zoosporic oomycetes [ 19 - 21 ]. Ca 2+ plays a central role in autonomous encystment, adhesion and germination of cysts in zoosporic oomycetes [ 3 , 10 , 14 , 22 - 24 ]. However, it is not considered to be an autoinducer because Ca 2+ does not directly trigger cooperative behaviors of zoospores and acts more like a secondary messenger [ 18 ]. AI-2 was first detected in bacteria and is utilized for metabolism and quorum sensing in bacteria [ 25 - 27 ]. In the latter process, bacteria respond to these released signaling molecules or autoinducers to coordinate their communal behavior. Eukaryotes including oomycetes can also produce AI-2 or AI-2-like activities [ 21 , 28 - 30 ] although they do not use the LuxS pathway that most bacteria use [ 31 , 32 ]. Instead, AI-2 is formed spontaneously from D-ribulose-5-phosphate that is synthesized in these eukaryotes from pentose-phosphates by ribose phosphate isomerase (RPI) in the pentose-phosphate pathway [ 28 ]. AI-2 has been proposed as a universal signaling molecule in bacteria based on its role in inter-species signaling and postulated cross-kingdom communication [ 33 - 40 ]. However, the function of AI-2 in eukaryotes has not been established. The aim of this study was to investigate the nature of signal molecules in ZFF. Specifically, we identified inter-specific signaling activities of ZFF from four Phytophthora species and one Pythium species. We also assessed the potential of AI-2 along with another known bacterial autoinducer as signal molecules for communication among zoosporic species.

Methods Zoosporic oomycetes and culture conditions Four Phytophthora species, P. nicotianae (1B11), P. sojae (28G4), P. capsici (24F4), P. hydropathica (37E6) and one Pythium species Py. aphanidermatum (18H7) were used in this study. These species are distinct in morphology and genetics [ 2 , 47 ]. Specifically, P. nicotianae , P. capsici and Py. aphanidermatum have broad host ranges while P. sojae has a restricted host range, generally infecting only soybeans and lupines. P. hydropathica (37E6) originated from irrigation water and is a pathogen of nursery plants [ 48 ]. The isolates were maintained on clarified vegetable juice agar (CV8A) medium [ 49 ] at 23°C. Preparation of zoospore-free fluid Zoospore-free fluid (ZFF) from a particular species is designated with an abbreviated species name. For example, ZFFnic represents ZFF from a P. nicotianae zoospore suspension. ZFF was prepared from nutrient-depleted zoospore suspensions starting with sporangium induction as described previously [ 18 , 21 ]. Specifically, prior to sporangium production, P. sojae and Py. aphanidermatum were cultured for 3-4 days and the other species were cultured for 1-2 wk in 10% CV8 broth. After nutrient depletion (medium removal and water rinses), the mycelial mats were further incubated for 16-18 h for P. sojae and Py. aphanidermatum , 2-3 days for P. capsici and one week for the other species under fluorescent light at 23°C to obtain a desired number of sporangia. To induce zoospore release, the mats with sporangia were flooded with chilled SDW and kept under lights until the desired zoospore density was reached. ZFF was obtained by passing a zoospore suspension through a 0.2 μm pore-size filter after vortexing for 2 min. ZFF was used fresh or stored at -20°C. Freezing destroyed the aggregation-promoting activity of ZFF, but not its infection-promoting activity. Phytopathosystems, plant growth conditions, inoculum preparation and inoculation Four phytopathosystems, P. nicotianae × annual vinca ( Catharanthus roseus cv. Little Bright Eye), P. sojae × lupine ( Lupinus polyphyllus ), P. sojae × soybean ( Glycine max cv. Williams) and P. capsici × pepper ( Capsicum annum cv. California Wonder) were used. Annual vinca plants were prepared in the greenhouse where 4-wk old seedlings were grown in pine bark with fertilizer for 4-6 wk. Soybean and pepper seedlings were prepared by growing 9 seeds per pot in sterilized Soilless Potting Mix (Schultz Professional) supplied with fertilizer and fungicide for 2 and 4 weeks, respectively, in the greenhouse. For lupine plants, 10 germinated seeds per styrofoam cup were grown in sterilized vermiculite (Whittemore Com) and fertilizer solution 20-20-20 (Scotts) for 2 wk in the growth chamber. Single-zoospore inocula with an average concentration of one zoospore per drop (10 μl) were prepared by dilution of a fresh zoospore suspension at 10 4 ml -1 with a test solution to 100 zoospore ml -1 . Test solutions included SDW, dilutions from 1 mM purified AI-2 (Omm Scientific Inc, Dallas, TX) and ZFF from different species. To test whether ZFF was heat or freezing labile, ZFFnic boiled for 5 min or freeze thawed was also included. For determination of the infection threshold of P. capsici , the zoospore suspension was diluted in SDW to prepare inocula at 10 2 , 10 3 or 10 4 ml -1 , containing an average of 1, 10, or 100 zoospores per 10-μl drop. For inoculation with P. nicotianae , detached annual vinca leaves were used as described previously [ 18 ]. Each leaf was inoculated at 10 sites unless stated otherwise with a 10-μl drop of single zoospore inocula. Each treatment included six replicate leaves and was done at least three times. In the P. sojae × lupine phytopathosystem, each cotyledon of lupine plants received one 10-μl drop of a single zoospore inoculum. Each treatment included 10 cups. Each cup contained 5-10 plants. Inoculated plants were kept in a moist chamber at 23°C in the dark overnight, then at a 10 h/14 h day/night cycle until symptoms appeared. Plants with damping-off symptoms were recorded as dead plants. Each assay was repeated twice. Similarly, for soybean and pepper plant inoculation, two 10-μl drops of an inoculum containing single or multiple zoospores were placed on the hypocotyls of each plant which was laid on its side in a moist chamber. Inoculated plants were kept in the dark overnight and then placed upright in a growth chamber at 26°C until symptoms appeared. For soybean, each treatment included at least 3 replicate pots containing 7-9 plants and was repeated twice. For pepper plants, each inoculation was performed in 6 replicate pots containing 3-8 plants. Microscopy of zoospore activity To determine zoospore responses to ZFF and other chemicals, 30 μl zoospore suspensions at 10 4 zoospores ml -1 were added to 120 μl of a test solution in a well on a depression slide to obtain a density of 2 × 10 3 zoospores ml -1 . Test solutions included fresh or treated (boiled or freeze/thawed) ZFF, a serial dilution from purified AI-2 at 1 mM, or SDW. Each test contained two replicate wells per treatment and was repeated once. The slides were placed on wet filter paper in 10-cm Petri dishes and incubated at 23°C. Zoospore behaviors including encystment, aggregation, germination and differentiation in three random fields in each well were examined with an IX71 inverted microscope (Olympus America Inc., Pennsylvania, USA) after overnight incubation. Images were captured with the Image-Pro ® Plus software version 5.1 (Media Cybernetics, Inc, Maryland, USA). Transient RPI gene silencing mediated by dsRNA and RT-PCR analysis The procedure was adopted from that for P. infestans [ 41 ]. To obtain a template for preparation of sense and antisense RNAs by transcription, two pairs of primers containing the T7 RNA polymerase promoter in their forward or reverse sequences were designed for amplification of a partial RPI sequence extracted from the P. capsici genome http://genome.jgi-psf.org/PhycaF7/PhycaF7.home.html . These primers were dsRPIPcapF: 5'-CAA GCT AAG CAG CTC ATC GCC CA-3'; dsRPIPcapRT7: 5'-GTA ATA CGA CTC ACT ATA GGG CAA CAG GCA CCC CCT GGG TCC A-3'; dsRPIPcapR: 5'-CAA CAG GCA CCC CCT GGG TCC A-3'(TGGACCCAGGGGGTGCCTGTTG); and dsRPIPcapFT7: 5'-GTA ATA CGA CTC ACT ATA GGG CAA GCT AAG CAG CTC ATC GCC CA-3'. Concentrated PCR amplicons were transcribed to produce sense and antisense RNAs using Megascrit RNAi kit (Ambion). Both sense and antisense RNA were mixed to obtain dsRNA at 168 ng μl -1 . To silence RPI , P. capsici protoplasts were transfected with the dsRNA. For each transfection, 24 μl of dsRNA (4 μg) was dried under vacuum (20-30 min) and then suspended in 10 μl PEG and 0.8 M mannitol solutions, respectively then incubated with 10 μl Lipofectin (Invitrogen) for 15 min prior to mixing with 20 μl P. capsici protoplasts. Protoplasts were prepared using a modified transformation protocol for P. sojae [ 50 ]. After further incubation for 24 h at 23°C, the mixture was transferred to 200 ml pea broth with ampicillin and vancomycin then 4 ml was transferred into each well of 12-well plates. To determine RPI expression in dsRNA-treated lines, mycelia from each well (line) were subcultured and extracted for RNA on day 7 using the Qiagen RNeasy plant kit. RNA was prepared from the lines before (T 0 ) and two weeks after transfer (T 1 ) as well as from the wild type culture. All the RNAs were treated with the RNase-Free DNAase Set (Qiagen), quantified and subjected to reverse transcription using the SuperScript III Reverse Transcriptase kit (Roche) followed by PCR using primers RPIPcapF: 5'- CAG ACG TCG CAG ATA CTA TTA ACC A-3'; and RPIPcapR: 5'-CTC CAG GAA GTA ATG CAT GAC ACA A-3' for RPI and actin housekeeping gene primers [ 50 ] for an endogenous control. The PCR products were then analyzed by electrophoresis. Detection of AHL activity Acyl-homoserine lactone (AHL) activity was determined with an Agrobacterium tumefaciens AHL reporter strain (KYC55/pJZ410/pJZ384/pJZ372) [ 46 ]. The reporter strain cannot produce AHLs but has plasmids containing a traI-lacZ reporter fusion and the regulator TraR driven by a T7 expression system. In the presence of exogenous AHLs, the over-expressed TraR activates the reporter fusion, resulting in production of β-galactosidase. The reporter can detect a broad range of AHLs ranging from 4- to 18-carbon acyl moieties at nanomolar levels [ 46 ]. We monitored LacZ activity by observing X-gal hydrolysis colorimetrically in the culture plates [ 51 ] and quantified the activity using the lactose analog ONPG (orthonitrophenyl-galactopyranoside) in a spectrophotometric assay [ 46 ]. ZFFnic and ZFFsoj from different zoospore suspensions, their ethyl acetate extracts, four positive controls (N-hexanoyl-, N-octanoyl-, N-decanoyl-, and dodecanoyl-DL-homoserine lactones (Sigma-Aldrich, Atlanta, Georgia, US) and a negative control (SDW) were included in the experiments. All AHLs were assessed at concentrations of 10 nM and 100 nM. In plate assays, 10 μl of ZFF, a synthetic AHL or SDW was injected at the center of the test plates with a pipette once the overlay was set. After incubation at 28°C for 2 days, LacZ activity was measured by the diameter of the blue area in test plates. The experiments were performed four times, and each experiment had two replicate plates. In spectrophotometric assays, the reporter was pre-induced in the AT medium containing antibiotics and stored at -80°C. The thawed cells were resuspended in AT medium (1:1000). A 200-μl aliquot of ZFF or SDW, or 50 μl of synthetic AHL was added to glass tubes containing 2 ml suspension. Cultures were grown on a shaker at 28°C until OD 600 = 1.0 (1.5 days). The bacterial cells in each tube were lysed by the addition of 800 μl of Z buffer, 20 μl of 0.05% SDS and 30 μl of chloroform followed by vortexing. LacZ activity was measured using the Miller Unit at OD 420 for the supernatant after the reaction with 100 μl of ONPG was ended by 1 M Na 2 CO 3 . The experiment was carried out in replicate and performed twice. Statistical analysis Data from independent experiments were processed and statistically analyzed using ANOVA in Excel. All P-values were determined based on one-way ANOVA unless otherwise stated.

Results and Discussion ZFF interspecific stimulation of zoosporic infection Zoospore-free fluids were prepared from suspensions at a density of 10 4 zoospores ml -1 or higher of Phytophthora nicotianae (ZFFnic), P. capsici (ZFFcap), P. hydropathica (ZFFhyd), P. sojae (ZFFsoj) and Pythium aphanidermatum (ZFFaph) and evaluated in three phytopathosystems. Inoculation of annual vinca ( Catharanthus roseus ) with suspensions containing an average of one zoospore of P. nicotianae in any of the four ZFFs resulted in significantly higher infection ( P < 0.001) compared to the control (SDW). Specifically, percentages of sites infected were 39%, 21%, 11%, and 15% for ZFFaph , ZFFhyd, ZFFnic, and ZFFsoj, respectively compared to 3% for SDW (Figure 1A ). Similarly, ZFFaph, ZFFhyd, ZFFnic and ZFFsoj stimulated infection of lupine ( Lupinus polyphyllus ) by P. sojae (Figure 1B ), while ZFFcap and ZFFsoj stimulated infection of soybean ( Glycine max ) by P. sojae (Figure 1C ). These results indicate that ZFF from the different Phytophthora species and Py. aphanidermatum contained one or more signals stimulating zoosporic infection by P. nicotianae and P. sojae that are active across species boundaries. Many plants are attacked by multiple oomycete species [ 1 ]. The ability of oomycete pathogens to benefit from the presence of related (or unrelated) species is presumably a selective advantage, especially if the diverse pathogens are competing for a limited resource (i.e. the host plant tissue) and/or the initial population density of each individual pathogen population is low. Such self-interested cooperation may have further advantages if the effector molecules released by each pathogen species have complementary or synergistic capabilities for suppressing plant defenses. ZFF inter-specific regulation of zoospore aggregation To determine whether ZFF may also have cross-species activity in regulating zoospore aggregation, fresh zoospores of P. nicotianae and P. sojae at a concentration (2 × 10 3 ml -1 ) below normal aggregation thresholds (approx. 10 6 ml -1 ) were cross incubated in multiwell plates with ZFFsoj or ZFFnic and compared with those in SDW. Zoospores of P. nicotianae in ZFFsoj and those of P. sojae in ZFFnic aggregated (Figure 2C and 2G ) as if they were in ZFF produced by their own species. As expected, zoospores of neither species aggregated in SDW (Figure 2D and 2H ). ZFFcap and ZFFaph did not stimulate zoospore aggregation by P. nicotianae or P. sojae zoospores. However, they did stimulate germination of cysts of both P. nicotianae and P. sojae (Figure 2A, B, E, F ), which may explain their activity in promoting plant infection (Figure 1 ). It was interesting that zoospores of P. capsici did not aggregate even at a density of 10 6 zoospores ml -1 . These results indicate that the signal(s) involved in aggregation are somewhat species-restricted and may be different from those mediating the infection process. AI-2 is not involved in zoospore communication and promotion of plant infection To test whether AI-2 may be involved in zoospore communication and promotion of plant infection, purified AI-2 was used in place of ZFF. AI-2 was tested at a wide concentration range of 0.01 μM -1 mM for its effects on P. nicotianae zoospore behaviors and plant infection; the concentration of AI-2 in ZFF was estimated to be less than 2 μM [ 21 ]. Under the microscope, an increased number of zoospores treated with AI-2 lysed before encystment and failed to germinate as the AI-2 concentration was increased (Table 1 ). Zoospore aggregation was not observed at any concentration tested. In infection experiments with annual vinca, AI-2 did not promote single zoospore infection at any concentration. Interestingly, AI-2 induced hypersensitive response (HR)-like micro-lesions on the inoculated sites at 100 μM and higher. These results indicated that AI-2 was not responsible for any of the zoospore signals found in ZFF. As a complementary test for the ability of AI-2-like molecules to mediate zoospore communication and promote plant infection, we cloned and silenced the ribose phosphate isomerase (RPI) gene of P. capsici . RPI converts ribose-5-phosphate to ribulose-5-phosphate, which can spontaneously convert to AI-2-like molecules under physiological conditions [ 28 ]. RPI was proposed to be responsible for production of AI-2-like molecules in zoosporic pathogens [ 21 ]. To silence the RPI gene of P. capsici , protoplasts of P. capsici were treated with RPI dsRNA. If RPI had a role in production of zoospore signaling molecules, RPI -silenced lines would be expected to require much higher zoospore concentrations to infect plants than the wild type due to reduced or blocked AI-2 production by the inocula. One third of the 48 T 0 lines regenerated 7 days after dsRNA exposure showed no or decreased expression with RPI compared to the endogenous control actin detected using RT-PCR. Half of these silenced or down regulated RPI lines retained the same reduced transcript levels two weeks after being transferred to fresh media (T 1 ) (Figure 3E ). Five T 1 lines were simultaneously tested for zoospore threshold for infection. The resulting disease incidences were very similar to those produced by wild type P. capsici at zoospore inoculum concentrations ranging from 10 2 to 10 4 ml -1 (Figure 3A-D ) ( P = 0.705; P = 0.065; P = 0.598, respectively). These results indicate that RPI silencing had no significant impact on zoospore communication during infection. The ZFF activity of the silenced lines was not evaluated due to the transient nature of dsRNA-mediated silencing [ 41 ] and insufficient numbers of T 1 zoospores for ZFF production. Nevertheless, these findings are consistent with the conclusion that AI-2-like molecules that might be produced via the action of RPI are not required for infection at low inoculum densities. The function of AI-2-like activities produced by zoosporic oomycetes remains unclear although it regulates bacteria quorum sensing [ 21 ]. Two-way communication has been observed between eukaryotes and bacteria such as Leguminosae and bacterial rhizobia [ 42 ] and between mycorrhiza and Streptomyces [ 43 ]. In the former case, plants release flavonoids that bind LysR-family transcriptional regulators in the bacteria, leading to the production of Nod factor that facilitates nitrogen fixation. In the latter case, fungal metabolites stimulate the bacteria to produce auxofuran which promotes growth of both the fungus and the host plants. Perhaps zoosporic oomycetes utilize AI-2 to attract quorum sensing bacteria which subsequently release factors that facilitate plant infection. Indeed, bacteria have been shown to benefit sporangium production by zoosporic oomycetes [ 44 ]. Involvement of other molecules in ZFF activity Acyl-homoserine lactones (AHLs), or bacterial autoinducer 1, are utilized by zoospores of the green seaweed Enteromorpha ( Ulva ) for communication in the search for settlement surfaces [ 45 ]. A bioassay was performed using the Agrobacterium tumefaciens reporter strain KYC55/pJZ410/pJZ384/pJZ372 [ 46 ] in plate and spectrophotometric tests to determine whether this molecule is present in ZFF. LacZ activity was detected in all four positive control plates at nM concentrations of AHL but not in ZFFnic or ZFFsoj prepared from zoospore suspensions at > 10 4 spores ml -1 nor in concentrated extracts from them obtained with ethyl acetate. These results indicate that zoospores from these oomycete species do not produce AHLs which therefore cannot be responsible for any ZFF activity. Temperature sensitivity of ZFF activities To begin to characterize the signal molecules in ZFF we tested their temperature sensitivity. ZFFnic did not stimulate zoospore aggregation after a freeze-thaw or heat treatment, suggesting that the molecule promoting this behavior may be a protein or lipoprotein that is sensitive to heat and freezing. On the other hand, freeze-thaw did not affect the activity of ZFFnic in promoting plant infection by zoospores (data not shown). In addition, ZFFnic boiled for 5 minutes remained as active as the untreated in promoting infection (Figure 4 ), indicating that the molecule which stimulates plant infection is temperature insensitive and different from that involved in aggregation.

Results and Discussion ZFF interspecific stimulation of zoosporic infection Zoospore-free fluids were prepared from suspensions at a density of 10 4 zoospores ml -1 or higher of Phytophthora nicotianae (ZFFnic), P. capsici (ZFFcap), P. hydropathica (ZFFhyd), P. sojae (ZFFsoj) and Pythium aphanidermatum (ZFFaph) and evaluated in three phytopathosystems. Inoculation of annual vinca ( Catharanthus roseus ) with suspensions containing an average of one zoospore of P. nicotianae in any of the four ZFFs resulted in significantly higher infection ( P < 0.001) compared to the control (SDW). Specifically, percentages of sites infected were 39%, 21%, 11%, and 15% for ZFFaph , ZFFhyd, ZFFnic, and ZFFsoj, respectively compared to 3% for SDW (Figure 1A ). Similarly, ZFFaph, ZFFhyd, ZFFnic and ZFFsoj stimulated infection of lupine ( Lupinus polyphyllus ) by P. sojae (Figure 1B ), while ZFFcap and ZFFsoj stimulated infection of soybean ( Glycine max ) by P. sojae (Figure 1C ). These results indicate that ZFF from the different Phytophthora species and Py. aphanidermatum contained one or more signals stimulating zoosporic infection by P. nicotianae and P. sojae that are active across species boundaries. Many plants are attacked by multiple oomycete species [ 1 ]. The ability of oomycete pathogens to benefit from the presence of related (or unrelated) species is presumably a selective advantage, especially if the diverse pathogens are competing for a limited resource (i.e. the host plant tissue) and/or the initial population density of each individual pathogen population is low. Such self-interested cooperation may have further advantages if the effector molecules released by each pathogen species have complementary or synergistic capabilities for suppressing plant defenses. ZFF inter-specific regulation of zoospore aggregation To determine whether ZFF may also have cross-species activity in regulating zoospore aggregation, fresh zoospores of P. nicotianae and P. sojae at a concentration (2 × 10 3 ml -1 ) below normal aggregation thresholds (approx. 10 6 ml -1 ) were cross incubated in multiwell plates with ZFFsoj or ZFFnic and compared with those in SDW. Zoospores of P. nicotianae in ZFFsoj and those of P. sojae in ZFFnic aggregated (Figure 2C and 2G ) as if they were in ZFF produced by their own species. As expected, zoospores of neither species aggregated in SDW (Figure 2D and 2H ). ZFFcap and ZFFaph did not stimulate zoospore aggregation by P. nicotianae or P. sojae zoospores. However, they did stimulate germination of cysts of both P. nicotianae and P. sojae (Figure 2A, B, E, F ), which may explain their activity in promoting plant infection (Figure 1 ). It was interesting that zoospores of P. capsici did not aggregate even at a density of 10 6 zoospores ml -1 . These results indicate that the signal(s) involved in aggregation are somewhat species-restricted and may be different from those mediating the infection process. AI-2 is not involved in zoospore communication and promotion of plant infection To test whether AI-2 may be involved in zoospore communication and promotion of plant infection, purified AI-2 was used in place of ZFF. AI-2 was tested at a wide concentration range of 0.01 μM -1 mM for its effects on P. nicotianae zoospore behaviors and plant infection; the concentration of AI-2 in ZFF was estimated to be less than 2 μM [ 21 ]. Under the microscope, an increased number of zoospores treated with AI-2 lysed before encystment and failed to germinate as the AI-2 concentration was increased (Table 1 ). Zoospore aggregation was not observed at any concentration tested. In infection experiments with annual vinca, AI-2 did not promote single zoospore infection at any concentration. Interestingly, AI-2 induced hypersensitive response (HR)-like micro-lesions on the inoculated sites at 100 μM and higher. These results indicated that AI-2 was not responsible for any of the zoospore signals found in ZFF. As a complementary test for the ability of AI-2-like molecules to mediate zoospore communication and promote plant infection, we cloned and silenced the ribose phosphate isomerase (RPI) gene of P. capsici . RPI converts ribose-5-phosphate to ribulose-5-phosphate, which can spontaneously convert to AI-2-like molecules under physiological conditions [ 28 ]. RPI was proposed to be responsible for production of AI-2-like molecules in zoosporic pathogens [ 21 ]. To silence the RPI gene of P. capsici , protoplasts of P. capsici were treated with RPI dsRNA. If RPI had a role in production of zoospore signaling molecules, RPI -silenced lines would be expected to require much higher zoospore concentrations to infect plants than the wild type due to reduced or blocked AI-2 production by the inocula. One third of the 48 T 0 lines regenerated 7 days after dsRNA exposure showed no or decreased expression with RPI compared to the endogenous control actin detected using RT-PCR. Half of these silenced or down regulated RPI lines retained the same reduced transcript levels two weeks after being transferred to fresh media (T 1 ) (Figure 3E ). Five T 1 lines were simultaneously tested for zoospore threshold for infection. The resulting disease incidences were very similar to those produced by wild type P. capsici at zoospore inoculum concentrations ranging from 10 2 to 10 4 ml -1 (Figure 3A-D ) ( P = 0.705; P = 0.065; P = 0.598, respectively). These results indicate that RPI silencing had no significant impact on zoospore communication during infection. The ZFF activity of the silenced lines was not evaluated due to the transient nature of dsRNA-mediated silencing [ 41 ] and insufficient numbers of T 1 zoospores for ZFF production. Nevertheless, these findings are consistent with the conclusion that AI-2-like molecules that might be produced via the action of RPI are not required for infection at low inoculum densities. The function of AI-2-like activities produced by zoosporic oomycetes remains unclear although it regulates bacteria quorum sensing [ 21 ]. Two-way communication has been observed between eukaryotes and bacteria such as Leguminosae and bacterial rhizobia [ 42 ] and between mycorrhiza and Streptomyces [ 43 ]. In the former case, plants release flavonoids that bind LysR-family transcriptional regulators in the bacteria, leading to the production of Nod factor that facilitates nitrogen fixation. In the latter case, fungal metabolites stimulate the bacteria to produce auxofuran which promotes growth of both the fungus and the host plants. Perhaps zoosporic oomycetes utilize AI-2 to attract quorum sensing bacteria which subsequently release factors that facilitate plant infection. Indeed, bacteria have been shown to benefit sporangium production by zoosporic oomycetes [ 44 ]. Involvement of other molecules in ZFF activity Acyl-homoserine lactones (AHLs), or bacterial autoinducer 1, are utilized by zoospores of the green seaweed Enteromorpha ( Ulva ) for communication in the search for settlement surfaces [ 45 ]. A bioassay was performed using the Agrobacterium tumefaciens reporter strain KYC55/pJZ410/pJZ384/pJZ372 [ 46 ] in plate and spectrophotometric tests to determine whether this molecule is present in ZFF. LacZ activity was detected in all four positive control plates at nM concentrations of AHL but not in ZFFnic or ZFFsoj prepared from zoospore suspensions at > 10 4 spores ml -1 nor in concentrated extracts from them obtained with ethyl acetate. These results indicate that zoospores from these oomycete species do not produce AHLs which therefore cannot be responsible for any ZFF activity. Temperature sensitivity of ZFF activities To begin to characterize the signal molecules in ZFF we tested their temperature sensitivity. ZFFnic did not stimulate zoospore aggregation after a freeze-thaw or heat treatment, suggesting that the molecule promoting this behavior may be a protein or lipoprotein that is sensitive to heat and freezing. On the other hand, freeze-thaw did not affect the activity of ZFFnic in promoting plant infection by zoospores (data not shown). In addition, ZFFnic boiled for 5 minutes remained as active as the untreated in promoting infection (Figure 4 ), indicating that the molecule which stimulates plant infection is temperature insensitive and different from that involved in aggregation.

Conclusion This study demonstrated inter-specific activities of zoospore extracellular products in promoting zoospore aggregation and plant infection by Phytophthora . Zoosporic oomycetes contain hundreds of species and are widespread in irrigation water and plant production fields. However, specific populations detected in primary inoculum sources are not in sufficient numbers to produce signals that could promote plant infection. Inter-specific chemical communication (probably self-interested) as a strategy used by zoosporic pathogens for effective plant infection provides insights into the destructiveness of these pathogens and the importance of the microbial community and the environment in the infection process. AI-2 was excluded as a signal for communal behavior in zoosporic oomycetes, despite its detection in ZFF and widespread presence in the environment. AI-2 synthase RPI and purified AI-2 both were not required for regulation of zoospore aggregation and infection. AHLs also were excluded because of their absence in ZFF. Thus, zoosporic oomycetes may use completely different chemicals from bacteria for quorum sensing. Analysis of ZFF revealed that functional signals controlling zoospore aggregation and plant infection differ in molecular composition. The former is not temperature labile and acts upon a restricted number of species while the latter is heat labile and non-species-specific. Identifying these molecules will facilitate our understanding of the mechanisms underlying natural plant infection by these pathogens and may lead to innovative control strategies.

Background Oomycetes attack a huge variety of economically and ecologically important plants. These pathogens release, detect and respond to signal molecules to coordinate their communal behaviors including the infection process. When signal molecules are present at or above threshold level, single zoospores can infect plants. However, at the beginning of a growing season population densities of individual species are likely below those required to reach a quorum and produce threshold levels of signal molecules to trigger infection. It is unclear whether these molecules are shared among related species and what their chemistries are. Results Zoospore-free fluids (ZFF) from Phytophthora capsici , P. hydropathica , P. nicotianae (ZFFnic), P. sojae (ZFFsoj) and Pythium aphanidermatum were cross tested for stimulating plant infection in three pathosystems. All ZFFs tested significantly increased infection of Catharanthus roseus by P. nicotianae . Similar cross activities were observed in infection of Lupinus polyphyllus and Glycine max by P. sojae . Only ZFFnic and ZFFsoj cross induced zoospore aggregation at a density of 2 × 10 3 ml -1 . Pure autoinducer-2 (AI-2), a component in ZFF, caused zoospore lysis of P. nicotianae before encystment and did not stimulate plant infection at concentrations from 0.01 to 1000 μM. P. capsici transformants with a transiently silenced AI-2 synthase gene, ribose phosphate isomerase ( RPI ), infected Capsicum annuum seedlings at the same inoculum concentration as the wild type. Acyl-homoserine lactones (AHLs) were not detected in any ZFFs. After freeze-thaw treatments, ZFF remained active in promoting plant infection but not zoospore aggregation. Heat treatment by boiling for 5 min also did not affect the infection-stimulating property of ZFFnic. Conclusion Oomycetes produce and use different molecules to regulate zoospore aggregation and plant infection. We found that some of these signal molecules could act in an inter-specific manner, though signals for zoospore aggregation were somewhat restricted. This self-interested cooperation among related species gives individual pathogens of the same group a competitive advantage over pathogens and microbes from other groups for limited resources. These findings help to understand why these pathogens often are individually undetectable until severe disease epidemics have developed. The signal molecules for both zoospore aggregation and plant infection are distinct from AI-2 and AHL.

Authors' contributions PK conceived of the study, carried out the experiments and drafted the manuscript. BMT identified the RPI gene sequence, participated in designing experiments for RPI cloning, silencing and expression, and helped interpret the data and write the paper. PAR maintained cultures of isolates used in all experiments and participated in drafting and editing the manuscript. BWKL conducted chemical analysis of AI-2 in ZFFs and participated in drafting and editing the manuscript. ZSZ has been involved in design and coordination of this study as well as editing of the manuscript. CH participated in conceiving of the study, drafting and editing the manuscript. All authors read and approved the final manuscript.

Acknowledgements The authors are indebted to Dr. Jun Zhu at the University of Pennsylvania School of Medicine for providing AHL-reporter strain KYC55 and the assay protocol. This work was supported in part by grants to CH from USDA-NIFA (2005-51101-02337 and 2010-51181-21140) and to ZSZ from NIAID/NIH (1R01AI058146) as well as an oomycete genomics and bioinformatics training fellowship to PK, supported by the NSF Research Collaboration Networks grant to BMT for the oomycete community.

CC BY

no

2022-01-12 15:21:45

BMC Microbiol. 2010 Dec 7; 10:313

oa_package/8a/43/PMC3016323.tar.gz

PMC3016324

21143837

Background A group of diverse pathogens has the potential to cause high morbidity and mortality in humans -especially if carried by aerosols- even though they do not pose a major threat to public health under normal circumstances. The most menacing bacterial pathogens of this group are Bacillus anthracis , Francisella tularensis and Yersinia pestis , and these organisms are listed as category A biothreat agents (classification of the CDC, USA, http://www.bt.cdc.gov/agent/agentlist-category.asp ) because of the potential danger of their deliberate release. Exposure to aerosolized B . anthracis spores and F . tularensis can lead to inhalational anthrax and tularemia. Y. pestis may cause pneumonic plague, which, unlike the other two diseases, may also spread from person to person. To reduce the public health impact of such highly pathogenic micro-organisms, rapid and accurate diagnostic tools for their detection are needed. Timely recognition of disease agents will enable appropriate treatment of exposed individuals which will be critical to their survival, and the spread of disease can be reduced by taking appropriate public health measures. Classical identification involves culturing suspect pathogens, but although culturing can be very sensitive, these methods are time consuming, not very specific, involve extensive biosafety measures and some organisms simply resist cultivation. Real-time qPCR methods for the detection of pathogens can be equally or more sensitive, and can also provide higher speed and specificity. Also, molecular methods require only preparatory handling of samples under biosafety conditions and can be easily scaled-up, which is important for speeding up investigations and control of disease progression in outbreak situations. Despite these manifold advantages, detection of DNA does not yield information about the presence of viable organisms. Multiplexing qPCR detection offers several advantages, including reduction of sample volume and handling time (reducing the analysis time, cost and opportunities for lab contamination). Also, false-negative results can be reduced through co-amplification of internal controls in each sample, and using multiple redundant genetic markers for each organism reduces the chance that strain variants are missed. Amplification of multiple signature sequences per organism will also reduce false-positive results in complex samples. False positives can be an issue if detection relies on single targets when analyzing environmental samples, due to the presence of homologous sequences in related organisms or unknown sources [ 1 , 2 ]. Therefore, it is essential to validate the qPCR using multiple strains, including of closely related organisms. The selection of suitable signature sequences is an essential requirement for reliable PCR assays. The suitability of signature sequences may be based on their function, e.g. detection of virulence factors supplies important information. But also the stability of their association with the pathogen is of importance. For instance, virulent B . anthracis can be recognized by its virulence plasmids pXO1 and pXO2 [ 3 ] which contain genes that confer toxin production and capsule synthesis activities, respectively. However, there are also chromosomally encoded factors that are important for the full virulence of B . anthracis [ 4 ]. Also, recent studies have shown the occurrence of a plasmid homologous to pXO1 in a pathogenic B . cereus strain [ 5 ] as well as genes homologous to genes on pXO2 in environmental Bacillus isolates [ 2 ]. This underscores the importance of inclusion of a chromosomal signature for B. anthracis in addition to the detection of plasmid genes. Similarly, virulent Y. pestis possesses 3 plasmids involved in virulence, but these plasmids are not stable and pathogenic Y. pestis lacking any of these plasmids exists [ 6 ]. Several reports have described real-time PCR assays for the detection of B. anthracis [ 7 - 10 ], Y. pestis [ 6 , 11 , 12 ] and F. tularensis [ 13 - 15 ]. Some assays were designed in multiplex format, but only few of these included internal controls for DNA amplification [ 10 , 16 ] and none included an internal control for successful DNA extraction. Here, we report the highly reliable and sensitive detection of these three pathogens that we achieved by developing multiplex qPCRs for 3 organism-specific markers and 1 internal control. By using a B. thuringiensis gene as internal control, it is possible to use the highly refractory spores of this near relative of B. anthracis as a control for both DNA extraction and qPCR amplification. The assays were extensively validated and were used on different real-time PCR platforms. The multiplex qPCRs are being applied in screening protocols and our setup allows straightforward expansion of the detection capabilities by inclusion of additional pathogens.

Methods Selection signature sequences An initial selection of potential signature sequences for specific detection of B . anthracis , F . tularensis and Y. pestis was based on previous reports and on the availability of sequences through public databases (NCBI/EMBL). The selection was based on functional and on technical criteria. Since 4 reporter dyes can be reliably differentiated by using qPCR instruments, and 1 channel was reserved for the internal control, we selected 3 signature sequences per organism. If possible, signature sequences included virulence genes since these are significant diagnostic markers. Such virulence genes are often located on plasmids. Besides plasmid-encoded targets, at least one chromosomal target was included to account for plasmid transfer and loss. Plasmids may be transferred between closely related species of Bacillus or Yersinia [ 8 ]. Plasmids can be cured from B . anthracis [ 31 ] and Y. pestis [ 6 ], and virulent plasmid-deficient Y. pestis strains occur in nature [ 6 ]. Also, near-neighbor species carrying closely-related plasmids [ 5 ] should be distinguished from B . anthracis . Finally, although B . anthracis has two plasmids that are required for virulence, there are also chromosomally encoded factors that are important for the full virulence [ 4 ]. If available, a multicopy sequence was included to enhance sensitivity. Unique targets present only in the organism of interest were preferred over targets differentiating homologues in related species only by sequence differences. Finally, an important consideration for the selection of targets was the quality of sequence information available from the public databases. This sequence quality concerned the number of sequences, their length and their coverage of strain diversity. For each potential target sequence, representative sequences were retrieved from NCBI/EMBL. BLAST searches were then performed to retrieve all homologous sequences from nucleotide and bacterial genome databases. All available sequences were aligned and consensus sequences were created using an accept level of 100% (to make sure the consensus sequence displayed all sequence variation). For B . anthracis , genes were selected on the multicopy virulence plasmids pXO1 and pXO2, and on the chromosome. The consensus alignment from the toxin gene cya included this gene from the homologous pBCXO1 plasmid which is present in a virulent B . cereus strain [ 5 ]. The chromosomal target for B . anthracis , the spore structural gene sspE , is not a unique gene as it is present in all Bacillus . Nevertheless, this sequence was selected since the sequence differences between B . anthracis and other species within the closely related B . cereus group were sufficient for designing highly selective oligonucleotides. Also, the presence of a substantial number of sequence entries in the databases (> 200) enabled a reliable consideration of the sequence diversity of B . cereus group isolates. For F . tularensis , the multicopy insertion sequence IS Ftu2 was selected for the detection of F. tularensis . Cross reaction with other Francisella species such as F. philomiragia could not be ruled out based on the available sequences, and a region of the outer membrane protein gene fopA was selected for the specific detection of all subspecies from the species F . tularensis . A specific location in the pdpD gene, which is absent from F. tularensis subspecies holarctica , was selected for the design of a probe for the detection of F . tularensis subspecies tularensis (type A) [ 14 ]. For Y. pestis , genes were selected on Y. pestis specific virulence plasmids pPCP1 and pMT1. The plasmid pCD1 was not used as it is shared by other pathogenic Yersinia species. A chromosomal sequence of unknown function that had been identified using comparative genome hybridization [ 17 ] was selected as Y. pestis specific chromosomal target. Spores of B . thuringiensis were used as internal control, not only for DNA amplification but also for successful DNA extraction. This member of the B . cereus group is closely related to B . anthracis and forms similar spores, while it contains species-specific plasmids. The B . thuringiensis plasmid gene encoding insecticidal crystal proteins ( cry genes) was used as the signature sequence for the detection of DNA released from this organism's spores. Sequence analysis tools, bioinformatics software Sequences retrieved from NCBI/EMBL were organized and aligned using the software package Kodon (Applied Maths, Ghent, Belgium). Comprehensive sequence alignments were made by performing BLAST searches from the selected targets to make sure all available sequence homologues were included in the alignments. Oligonucleotides for multiplex qPCR assays and for conventional PCR assays were designed using the software package Visual Oligonucleotide Modeling Platform version 6 (DNA software Inc. Ann Arbor, USA). The design strategy for multiplex qPCR assays was as follows. First, a hydrolysis probe and primer set were designed for the B . thuringiensis internal control. Then, for each selected signature sequence a hydrolysis probe was designed, followed by the design of the corresponding primer set. A different strategy was chosen for the B . anthracis assay, because its chromosomal target sspE has homologues in other Bacillus , notably the internal control B . thuringiensis . To make sure that detection of B . anthracis sspE was highly selective, the exact positions of probe and primers were guided based on visual inspection of the alignment. Probe and primers were located in regions with mismatches between Bacillus species (notably between B . thuringiensis and B . anthracis ), and the primers were designed such that mismatch positions were located at their highly discriminating 3'-ends. Oligonucleotides that were calculated by the design software were first checked against the consensus alignment to exclude designs not covering all sequence variants, and were then evaluated using the simulation module of Visual OMP. All oligonucleotides designed were validated in silico by using BLAST searches in general and microbial genomes databases (NCBI/EMBL). Sequencing Sequences were obtained from the cry1 gene from B. thuringiensis strain ATCC 29730 and from the sspE gene from all B. anthracis strains in our culture collection, B. thuringiensis ATCC 29730 and B. cereus strains WSBC 10583, 10619, 10766, 10483, 10572, 10705, 10770 and 10865 (Additional file 1 Table S1). In addition, the pla gene was sequenced from DNA extracted from muscle tissue derived from a dissected specimen of Rattus rattus . Primers used for sequencing are displayed in Additional file 2 Table S2. PCR products were purified by using ExoSAP-IT (USB, Cleveland, USA) and DNA sequencing reactions were performed in both directions using BigDyeTerminator v3.1 (Applied Biosystems, Nieuwerkerk a/d IJssel, the Netherlands) on a 48-capillary 3730 DNA Analyzer sequencer (Applied Biosystems, Nieuwerkerk a/d IJssel, the Netherlands). Accession numbers: HQ222846 to HQ222861 and HQ606074. PCR and real-time qPCR Oligonucleotides were synthesized by Biolegio (Biolegio, Nijmegen, the Netherlands). Conventional PCR was used to produce amplicons from signature sequences. Amplification was carried out using the HotStar Taq Master Mix Kit (Qiagen, Westburg, the Netherlands) and 400 nM primers in a total reaction volume of 50 μl. Primer sets were designed using Visual OMP software (Additional file 2 Table S2). Thermocycling conditions were as follows: 95°C for 15 min, 40 cycles at 95°C for 30 sec, 55°C for 30 sec and 72°C for 30 sec, followed by a final step at 72°C for 7 min. Thermocycling reactions were carried out in a Px2 thermal cycler (Thermo Electron Corporation, Breda, the Netherlands). All qPCR reactions were carried out in a final volume of 20 μl containing iQ Multiplex Powermix (Bio-Rad, Veenendaal, the Netherlands), 200 nM of each primer and 100-300 nM hydrolysis probes and 3 μl of DNA template. Probes concentrations had been optimized to yield minimal spectral overlap between fluorescence level of the reporter dyes for each target in a multiplex assay and were 100, 200, 300 and 300 nM for FAM, JOE, CFR590 and Cy5 labeled probes respectively. The multiplex real-time qPCR assays had been designed for an optimal annealing temperature of 60°C and the thermal cycling conditions were as follows: First enzyme activation at 95°C for 5 min, followed by amplification and detection by 45 cycles at 95°C for 5 sec and 60°C for 35 sec. Each real-time qPCR experiment included a negative (no template) control. Measurements were carried out on a Lightcycler 480 (Roche, Almere, the Netherlands). An iQ5 (Bio-Rad) instrument was used for routine screening purposes. Analyses were performed on the instruments software: LightCycler 480 Software release 1.5.0. SP3 and iQ5 Optical Systems Software version 2.0. Cq values were calculated using the second derivative method on the LightCycler and the Base Line Subtracted Curve Fit method on the iQ5. Color compensations were carried out on both instruments as follows. PCR amplifications were performed using single primer-probe sets for each reporter dye and under identical reaction conditions as during multiplex amplification. The PCR reactions thus produced contained single dyes in relevant concentrations and these were used for color compensation runs according to the manufacturers' guidelines. Verification of PCR product sizes were carried out on the 2100 Bioanalyzer instrument (Agilent Technologies, Eindhoven, the Netherlands) using the DNA 1000 kit. Bacterial isolates and genomic DNA preparation The detection limits and specificities of the assays were evaluated using genomic materials from the bacterial strains and other sources displayed in Additional file 1 Table S1. The pathogen panel included (besides a variety of Eukaryal organisms): 8 B. anthracis strains and 31 near relatives (22 B. cereus , 5 B. thuringiensis and 4 B. mycoides ), 21 F. tularensis strains (16 subspecies holarctica , 4 tularensis and 1 novicida ) and 4 of the closest related species F. philomiragia , 23 Y. pestis (including Antiqua , Mediaevalis and Orientalis biovars) and 3 strains from the closest relative Y. pseudotuberculosis and 7 strains from Y. enterocolitica . From most of the B. anthracis , F. tularensis and Y. pestis strains we only had genomic DNA (lysates) available to verify specificity of our assays. Several strains were available as live cultures in our laboratory and these were used as resource for the production of larger quantities of genomic DNA. B. anthracis and Y. pestis strains were acquired from the NCTC (National Culture Type Collection, UK) and the Pasteur Institute (France). The Francisella holarctica strain was a patient isolate. Other genomic materials were lysates from bacterial cultures provided by other researchers as mentioned in the acknowledgements. Cultivation of these strains was carried out in a BSL3 glove-box. Colonies from B. anthracis , F. tularensis and Y. pestis were grown on Columbian sheep blood agar plates and chocolate agar plates. Single colonies were transferred to liquid BHI (Brain Heart Infusion, 27 g/L) medium. After cultures had grown to visible turbidity, 1.4 ml cell culture was centrifuged and the pellet was resuspended in 250 μl TE pH 8. Cells were incubated for 30 minutes at 100°C. Lysed cultures were filtered through a 0.22 μm sterile Ultrafree-MC spinfilter (Millipore, Amsterdam, the Netherlands) and the filtrate was subsequently transported from the BSL3 facility for handling under normal laboratory conditions. Cultures from non-target bacteria that were used in the specificity panel were obtained from the culture collection at the RIVM. These cultures were cultivated under BSL2 conditions and lysates of these cultures were used for specificity testing. DNA extraction and purification was carried out by using NucliSens Magnetic Extraction Reagents (bioMérieux, Boxtel, the Netherlands) following the manufacturers instructions. This method performed best with regard to efficiency and ease-of-use when compared to other kits. This comparison was carried out as follows. Dilution series of a mixture of genomic DNA from B. anthracis , Y. pestis and F. tularensis , and spores from B. thuringiensis were added to various powders including milk powder, soy powder, silica and maize powder, and DNA was extracted by using the powersoil DNA isolation kit (MO BIO Laboratories, Carlsbad, USA), the ultraclean microbial DNA isolation kit (MO BIO) and the NucliSens Magnetic Extraction Reagents (bioMérieux) according to the manufacturers instructions. The extracts were measured by using the developed qPCR. DNA concentrations were measured using the NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Wilmington, USA). DNA samples were stored at 4°C for use within 1 week and at -20°C for longer storage. Spore suspension for use as internal control Spore suspensions of B. thuringiensis strain ATCC 29730 (var. galleriae Heimpel) were obtained from Raven Biological Laboratories (Omaha, Nebraska, USA). These washed spores were counted by microscopy and then aliquotted and stored at 4°C. The amount of spores that needs to be added to samples to obtain suitable Cq values for this internal control must be determined empirically for each stock spore suspension. Ten-fold serial dilutions were made from the spore stock and DNA was extracted from 50 μl portions of each dilution by using the Nuclisens Magnetic Extraction Reagents (bioMérieux). The developed real-time qPCR assays were used to determine the amount of spores required for a Cq value between 32 and 35. Limit of detection, efficiency and repeatability Characterization of qPCR performance was guided by the MIQE guidelines [ 32 ]. The validation was carried out by using genomic DNA as well as purified PCR amplicons including > 100 bp upstream and downstream from the qPCR amplification sites. The latter were used to compose template mixes of desired composition and quantities, while maintaining secondary structures in the primer binding regions. Detection limits (LOD) for genomic DNA were determined by using purified DNA from cultures of B. anthracis strain Vollum, F. tularensis strain tularensis ATCC 6223 and Y. pestis strain Harbin. DNA was purified from lysates of these strains. The concentration of purified genomic DNA was measured by using the NanoDrop 1000 spectrophotometer. Serial dilutions of genomic DNA were used to calculate LODs from the proportion of positive qPCRs at each dilution. Four replicates of eight serial dilutions of genomic DNA were measured by qPCR. Based on the results, an additional measurement was performed on 4 replicates of 8 novel serial dilutions. The measurements included at least one dilution with all replicates positive and one with all replicates negative. A probit analysis was performed using SPSS Statistics 18.0.0 to calculate the DNA concentration that could be measured with 95% probability. DNA templates for measuring the detection limits from the different signature sequences were amplified from the bacterial strains mentioned above. In addition, the pdpD signature sequence from F. tularensis tularensis was amplified from ATCC 6223. To generate suitable amplicons for testing the different real-time qPCR targets, primers were designed for amplification of a signature sequence with a size of 400-800 bp, extending beyond both ends of the region amplified by the real-time qPCR. Primer sequences are displayed in Additional file 2 Table S2. After amplification, PCR products were purified and the number of DNA copies in amplicon solutions was calculated from their sizes and concentrations. Amplicon dilutions were used to calculate the LOD from the proportions of positive qPCRs at each dilution. First, 5 replicates of 8 dilutions around the estimated detection limit were measured using a mixture of equal amounts of target amplicons. Based on the results, an additional measurement was performed on 10 replicates of 8 novel dilutions. After scoring positive results, a probit analysis was performed to calculate the DNA concentration that could be measured with 95% probability. Efficiency and repeatability were calculated from the log-linear portion of the calibration curve, covering 6 orders of magnitude. The calibration curve was made using amplicon mixtures as templates containing the signature sequences (as described before). Four replicate measurements were obtained from each dilution. For calculation of the repeatability, the lowest template concentration was not used as the standard deviation (SD) near the detection limit was not consistent with those obtained for the other concentrations. Dynamic range internal control To establish a concentration range for the applicability of the internal control, serial dilutions were made of internal control cry1 target amplicon (0, 2·10 1 , 2·10 2 , 2·10 3 , 2·10 4 , 4·10 4 copies per reaction) in the presence of a mixture of the 3 organism specific target amplicons, each at a concentration of 20 copies per reaction. These target amplicon mixtures were amplified in triplicate by using the developed qPCR assays and Cq values were used to infer possible inhibition of PCR amplification. To investigate inhibitory effects on the amplification of organism-specific targets, triplicate measurements were performed on amplicons of the multicopy targets ( cya , pla and IS Ftu2 ) diluted as above in the presence of the 2 other organism-specific target amplicons, each at a concentration of 20 copies per reaction.

Results Design of multiplex hydrolysis probe assays A selection of signature sequences for the specific detection and partial characterization of B . anthracis , F . tularensis and Y. pestis was based on previous reports [ 4 - 6 , 8 , 11 - 14 , 17 ], and sequence data accessible via public databases (NCBI/EMBL). Additional sequences were obtained from sspE genes from a number of strains from the Bacillus cereus group in our culture collection and from the cry1 gene from B . thuringiensis strain ATCC 29730. Based on signature sequence alignments, regions were identified that were shared by all strain variants and sufficiently different from homologous sequences to design selective oligonucleotides for multiplex real-time qPCR assays (see Table 1 ). In order to achieve a reliable as well as rapid method for the detection of B . anthracis , Y. pestis and F . tularensis , the cry1 gene from B. thuringiensis was included in the multiplex qPCR assays. Inclusion of this gene permitted the development of B . thuringiensis spores as internal control for DNA extraction as well as amplification. The amount of spores that must be added to the samples before DNA extraction to obtain the desired Cq value was determined from serial dilutions of the spores. Specificity and coverage of strain diversity A DNA panel from the Bacterial and Eukaryal species listed in Additional file 1 Table S1 was used to validate the specificity of the developed real-time qPCR assays. The pathogen-specific targets showed no cross-reactivity and very near relatives could be differentiated as evidenced by the absence of amplification from various members of the Bacillus cereus group, Yersinia pseudotuberculosis, Y. enterocolitica and Francisella philomiragia . From the latter species, one out of the four strains that were tested showed very weak amplification of the multicopy sequence IS Ftu2 , but none of the strains showed amplification of the F. tularensis signature sequence fopA . The assays detected all available strains from the targeted organisms. Nevertheless, the genomic marker ypo 393 was amplified from only one strain (NCTC 10329) out of four from a Y. pestis cluster from Nairobi. Additional information about the pathogens could be derived from the detection of particular plasmid combinations in the B . anthracis and Y. pestis assays, and from the detection of the pdpD gene [ 14 ] in the F. tularensis assay. This was confirmed by the anticipated absence of the pdpD gene in the 16 F. tularensis holarctica strains we tested. However, the probe designed for pdpD detection could not discriminate between subspecies tularensis and novicida . Based on the available sequences from F. tularensis mediasiatica , amplification of pdpD from this subspecies will occur as well, however, we did not have genomic materials to verify this. Amplification of the pla target from Rattus rattus DNA was unexpected and seemed to indicate cross-reactivity. To confirm pla amplification we investigated DNA from 10 rats, including 3 from the related species Rattus norvegicus (Additional file 1 Table S1). Sequencing of the amplification product from these samples revealed the presence of a pla gene highly similar to that of Y. pestis (99% identity), while no sequences with any homology to these sequences were encountered in the published rat genome. Therefore, the amplification does not invalidate our assay but highlights the fact that the pla gene alone is not a sufficient diagnostic marker for the presence of Y. pestis . The internal control gene cry1 was amplified from several Bacillus cultures in addition to B. thuringiensis . Efficiency, dynamic range, precision and detection limit Ten-fold independent serial dilutions from purified target amplicons (PCR products containing target sequences) were used to generate calibration curves and calculate PCR amplification efficiencies. As shown in Table 2 efficiencies for the different targets ranged between 94.5% and 94.8% for B. anthracis , between 95.9% and 98.2% for F. tularensis and between 93.1% and 93.2% for Y. pestis . The efficiency for amplification of the internal control target cry1 varied slightly between the assays and was near that of the organism-specific targets. The reaction was linear over 6 orders of magnitude, from 1.5·10 2 to 1.5·10 7 target copies per reaction (data not shown). The precision of the different qPCR assays was calculated from 4 replicates of 5 independent dilutions. Mean Cq values and standard deviations (SD) were calculated from each dilution. As shown in Table 2 there is a high repeatability for the different targets, with SDs around 0.05 Cq. Only at very low concentrations (high Cq values) near the limit of detection, the SD exceeded 1 Cq (data not shown). To determine the analytical sensitivity for each single target, dilutions of target amplicons near the detection limit were measured by using the developed assays. The analytical sensitivity for genomic DNA was calculated from dilutions of purified genomic DNA from selected pathogens. The fraction of positive reactions in replicate dilutions were scored and a probit analysis was used to calculate the limit of detection (LOD), which is the lowest concentration at which 95% of positive samples are detected. The LOD for single targets could be expressed as copy numbers as the target amplicons were of known size. Table 2 shows LODs of below 10 copies for the various targets. For genomic DNA, LODs based on the most sensitive target were for B. anthracis 15.7 fg, for F. tularensis 0.6 fg and for Y. pestis 29.6 fg. Co-amplification targets in multiplex assay Large concentration differences between DNA templates in a multiplex PCR may lead to competition for reaction components and impaired amplification of the rarer templates. Divergence of target concentrations could originate from different copy numbers of the targets within the pathogen genome, or from differences between the numbers of organisms that are detected simultaneously. Although there is limited copy number variation for the selected targets, multicopy sequences such as insertion sequences and plasmid genes could outnumber single-copy targets by a factor of more than 200 [ 3 , 18 ]. To exclude an inhibitory effect of the dominant amplification product in the multiplex reaction, dilution series of the high copy number targets ( cya , pla and IS Ftu2 ) were made in the presence of a constant and low concentration of the other targets from that organism, and measured by the multiplex qPCRs (Figure 1A-C ). No inhibitory effect (increasing Cq) was observed, even if the excess target considerably exceeded the maximum ratio that could be anticipated. Significant concentration differences are possible between the pathogen specific targets and the internal control target, as these organisms could be mixed in very different quantities. Inhibition of the internal control (IC) by excess pathogen DNA is not a problem as the function of the IC is to exclude false negative results (a positive pathogen signal makes an additional IC signal irrelevant). In contrast, it is essential that inhibition of pathogen targets by the internal control is prevented. To determine the boundaries within which IC B . thuringiensis DNA could be added to pathogen DNA without interfering with the detection of low pathogen concentrations, a dilution series of the IC target amplicon ( cry1 gene) was made in the presence of a constant and low concentration of pathogen targets and measured by the multiplex qPCRs. As shown in Figure 1D-F , the amplification of 20 copies of pathogen targets was inhibited (increasing Cq) if more than 200 copies of the internal control target were present for B. anthracis or more than 2000 copies for Y. pestis and F. tularensis . Moreover, rare targets were still detectable at much higher excess ratios of internal control, even though at higher Cq values.

Discussion Multiplexing and the reduction of false negative and false positive results In this report, we describe the development of multiplex qPCRs for the rapid and reliable detection of B . anthracis , F . tularensis and Y. pestis . The assays include a signature sequence from B. thuringiensis which allows the use of its spores as combined internal control for both DNA extraction and subsequent DNA amplification. As Bacillus spores are among the most resistant of microbial structures, DNA extraction from such spores can be considered to be a reliable indicator for successful DNA extraction from other microbes. Application of internal controls is especially important when measuring environmental samples because these tend to contain various sorts of PCR inhibitors. The internal control helps preventing false negative results, which are further reduced by the sensitivity of the methods and by the recognition of multiple signatures per organism. Multiplexing reduces the chance that the pathogen escapes detection due to modification or loss of plasmids or genes (natural or by manipulation). Multiple diagnostic signatures per pathogen will also help reducing false positive detection, which is particularly important in complex (environmental) samples which may contain homologous genes of yet uncharacterized origin[ 1 , 2 ]. The genera Bacillus , Francisella , and Yersinia each include species ranging from nonpathogenic environmental species, through symbionts and facultative pathogens, to highly virulent human and animal pathogens. Comparative genomic sequencing and typing studies have indicated that the sequence similarity and gene composition of species having very different lifestyles can be very high [ 1 , 19 - 21 ] Also, bacterial genomes are dynamic and non-target organisms could acquire diagnostic sequences by lateral gene transfer, especially if present on plasmids [ 22 ]. An additional reason for including multiple targets is that for B. anthracis and Y. pestis , a full picture of virulence requires the detection of several markers. Although virulent Y. pestis usually contains three plasmids, strains deficient in one or more plasmids may cause fatal infections [ 6 ]. Assays relying on one signature sequence for the detection of a pathogen [ 10 , 23 , 24 ], suffer from the constraints mentioned above, especially when analyzing environmental samples [ 1 ]. For instance, Y. pestis subgroup Pestoides lacks the plasminogen coagulase ( pla ) gene [ 25 ] that is used as the major and sometimes only target for the detection of Y. pestis [ 23 , 26 ]. On the other hand, we found that the pla gene may yield false positive results in certain matrices (unpublished). In addition to relying on multiple targets, false positives are further reduced by the high specificity of the developed assays for the selected targets, which was confirmed by in silico and in vitro validations. Selected targets Inclusion of chromosomal markers in addition to virulence plasmids is important due to the occurrence of B. anthracis and Y. pestis strains lacking virulence plasmids. These strains, as well as yet uncharacterized closely related environmental species, share genomic traits that could lead to misidentification. Fully virulent B. anthracis strains possess plasmids pXO1 and pXO2. However, the detection of plasmids only, as for instance commercial kits do, cannot detect plasmid-deficient B. anthracis strains such as Sterne and CDC 1014. Moreover, B. cereus strains carrying plasmid highly similar to those of B. anthracis ( B. cereus G9241) are not correctly identified. Several chromosomal markers have been used for the detection of B. anthracis (e.g. BA813 , rpoB , gyrA , gyrB , saspB , plcR , BA5345 , BA5510 ), but only recently a locus was described for qPCR that did not yield any false positive results from closely related Bacillus [ 27 ]. We have developed an alternative chromosomal signature sequence ( sspE ) for use in real-time PCR. This marker has previously been used for specific detection of B. anthracis , but differentiation required melting curve analysis [ 8 ]. By selecting highly discriminating positions for primers and hydrolysis probe, we achieved specific detection without post-PCR analysis. For Y. pestis , it is equally important to detect chromosomal sequences in addition to its plasmids, as plasmid-deficient virulent Y. pestis has been described [ 6 ]. Most of the chromosomal targets that have been described previously did not differentiate Y. pestis from closely related Y. pseudotuberculosis or Y. enterocolitica [ 12 ]. The chromosomal signature sequence we developed for Y. pestis detection was based on a previous study employing comparative genome hybridization to identify chromosomal regions specific for Y. pestis [ 17 ]. We selected a different region than the ypo 2088 target which was used by these authors and later by Matero et al. [ 16 ], because examination of published genomes revealed that strain Y. pestis antiqua (accession # CP000308) does not possess this region. Although ypo 339 was present in all 20 Y. pestis sequences currently publicly available, 3 out of 4 isolates from the Nairobi cluster appeared to lack this signature sequence. Hence, although ypo 393 is a reliable signature sequence for most Y. pestis , strains lacking this sequence do exist. Our results illustrate that even if signature sequences selected for diagnostic purposes are based on a considerable amount of sequences available from genomes and sequence databases, uncharacterized strain variants may exist or new variants may arise that do not posses a particular target sequence. Conversely, amplification of the cry1 gene from some Bacillus strains other than B. thuringiensis was not anticipated as these strains were not known to contain the plasmids carrying cry genes or homologues. Since it concerned related, spore-forming Bacillus strains, these could also be used as internal controls. The primary focus of our assays was the sensitive and specific detection of the selected pathogens, minimizing false negative and false positive results. Strain differentiation was considered to be of only secondary interest. For F. tularensis , sensitive detection requires detection of the multicopy sequence IS Ftu2 . The targeted tranposase can also be present in F. philomiragia , but strain ATCC 225017 for instance, has only one copy with mismatches in the probe and reverse primer. This explains the very low cross-reactivity with the four strains we investigated. Nevertheless, specific detection of the species F. tularensis was confirmed by additional detection of the fopA gene [ 13 , 15 ]. Further subspecies information could be obtained from the pdpD target, which is known to be absent in subspecies holarctica (type B) [ 14 ] and was indeed not detected in the 16 strains we tested. With all targets positive, subsequent research is warranted however, as presence of this gene could also imply presence of the subspecies novicida and mediasiatica [ 28 ]. Subspecies mediasiatica is, similar to subspecies holarctica , a considerable public health threat although both species are less pathogenic compared to subspecies tularensis . Subspecies novicida is less pathogenic than the other subspecies and has been involved in only a limited number of human cases. Sensitivity The analytical sensitivity for detection of the different signature sequences is very high (Table 2 ). Hence, the presence of only a few genomes should enable detection of the organisms of interest at 95% probability, especially when based on multicopy signature sequences. For F. tularensis this means that only 0.3 genomic equivalents (GE) were sufficient for the detection, considering a genome size of 1.9 megabases. For B. anthracis and Y. pestis , reliable estimates of GE could not be made due to the variable and sometimes significant contribution of plasmids to the total amount of DNA measured [ 3 , 18 ]. But, using approximate plasmid copy numbers, a detection limit of 4 GE for B. anthracis and 6 GE for Y. pestis can be calculated. The LODs were similar or lower than those reported previously [ 13 , 14 ] and lower than those of other multiplex assays for these pathogens [ 12 ]. A correlation between the copy numbers of the targeted genes and the LOD for genomic DNA can be expected. For F. tularensis gDNA, the LOD was indeed highest based on the detection of the single-copy fopA target, lower when based on the 2-copy pdpD and lowest when based on the approximately 20-copy IS Ftu2 (Table 2 ). Also for Y. pestis , an inverse correlation between gDNA LOD and expected target copy number was observed (Table 2 ). Nevertheless, a more pronounced difference would be expected based on the high relative abundance of pla carrying plasmids that has been reported [ 18 ]. Probably, the gDNA we used contained fewer plasmids, as was supported by a Cq difference between the chromosomal target and pla of only approximately 2 (data not shown). For B. anthracis , the LOD of gDNA was highest when based on the detection of the pXO1 plasmid marker cya , while high copy numbers for the pXO1 plasmid carrying this gene have been reported [ 3 ]. This discrepancy could be due to the gDNA preparation we used for calculating LODs. Although Coker et al. reported relative amounts of pXO1 and pXO2 of respectively 11.5 and 1.6, for the same strain we used ( B. anthracis Vollum), variation in pXO plasmid copy numbers could also result from the growth phase at which DNA was harvested [ 3 ]. Our data correspond better to the lower plasmid copy numbers reported by other authors [ 29 , 30 ]. Nevertheless, all reports agree that pXO1 is present in multiple copies. The relatively high LOD for gDNA detection based on cya can probably partly be explained by a low amplification efficiency near the detection limit as the LOD for the detection of cya target amplicons is also relatively high (Table 2 ). Internal control As was shown in Figure 1 the cry1 gene from B. thuringiensis spores can be used as internal control without affecting sensitive detection of the pathogens of interest. However, addition of more than 200 copies of cry1 per reaction lead to a Cq increase for the detection of the B. anthracis plasmid targets. For diagnostic purposes, we use a spore suspension that yields a Cq value between 32 and 35 for the detection of cry1 to prevent any interference with the detection of pathogen DNA. The amount of spores that needs to be added to yield this Cq should be determined for each new batch as it will vary with each new spore stock, and the DNA extraction protocol used. The observed inhibition highlights that multiplex qPCR can be problematic if it is used for the detection of mixed pathogens present in different quantities as amplification of targets from a dominant organism could inhibit the detection of an uncommon pathogen. Assays for the detection of single targets from multiple pathogens simultaneously, such as that described for B. anthracis , F. tularensis and Y. pestis detection [ 23 ], should therefore be carefully evaluated for this inhibition effect. Environmental testing Application of the multiplex qPCR assays directly on human specimens or environmental samples could save time and prevent loss of DNA during extraction. However, we use the assays only after a DNA extraction protocol, in order to prevent unanticipated inhibition by diverse matrices. Our laboratory has compared several commercially available DNA extraction kits for use in a BSL3 facility, and selected one that combined efficient DNA extraction with ease-of-use and applicability in the restricted BSL3 environment. We have been using the developed qPCRs for the analysis of samples suspected for the presence of these pathogens with B. thuringiensis spores added before DNA extraction under BSL3 biosafety conditions. Hundreds of samples containing all sorts of solid materials and liquids have been analyzed without yielding false positive readings.

Conclusion The multiplex qPCR assays that were developed for B. anthracis , F. tularensis and Y. pestis allow the rapid detection of 3 pathogen-specific targets simultaneously without compromising sensitivity. Together with the application of an internal control for both DNA extraction and DNA amplification, this assures highly reliable detection, while template consumption and laboratory effort is kept at a minimum. These considerations are particularly advantageous in the context of biothreat samples which may be used for additional tests and for surge capacity during an outbreak. The detection of multiple targets decreases the chance of false-positive and false-negative results and provides additional information about virulence.

Background Several pathogens could seriously affect public health if not recognized timely. To reduce the impact of such highly pathogenic micro-organisms, rapid and accurate diagnostic tools are needed for their detection in various samples, including environmental samples. Results Multiplex real-time PCRs were designed for rapid and reliable detection of three major pathogens that have the potential to cause high morbidity and mortality in humans: B. anthracis , F. tularensis and Y. pestis . The developed assays detect three pathogen-specific targets, including at least one chromosomal target, and one target from B. thuringiensis which is used as an internal control for nucleic acid extraction from refractory spores as well as successful DNA amplification. Validation of the PCRs showed a high analytical sensitivity, specificity and coverage of diverse pathogen strains. Conclusions The multiplex qPCR assays that were developed allow the rapid detection of 3 pathogen-specific targets simultaneously, without compromising sensitivity. The application of B. thuringiensis spores as internal controls further reduces false negative results. This ensures highly reliable detection, while template consumption and laboratory effort are kept at a minimum

Authors' contributions IJ: conceived the study and designed the experiments, performed oligonucleotide designs and statistical analyses, interpreted experimental results and wrote the manuscript; RAH: participated in the design of the experiments, carried out and interpreted the experimental work, and helped to draft the manuscript; JMB: helped carrying out experiments; BvR: coordinated the work. All authors read and approved the final manuscript. Supplementary Material

Acknowledgements We gratefully acknowledge Horacio Gill from the Centro Nacional de Microbiologia, Instituto de Salud Carlos III, Majadahonda, Spain, Rickart Knuttson and Joakim Ågren from the National Veterinary Institute (SVA), Uppsala, Sweden, the Swedish Defense Research Agency (FOI), Umea, Sweden, Karen Kempsell from the Health Protection Agency (HPA), Porton Down, UK, and Jasper Kieboom from TNO Defense and Safety, Rijswijk, the Netherlands, for providing genomic materials. Frans Reubsaet, Maaike de Vries, Marieke Opsteegh and Chantal Reusken from CIB, RIVM are acknowledged for sharing bacterial cultures and other genomic materials. This work was funded by a SOR strategic research grant from the RIVM.

CC BY

no

2022-01-12 16:12:49

BMC Microbiol. 2010 Dec 8; 10:314

oa_package/a0/a4/PMC3016324.tar.gz

PMC3016325

21144051

Background Internalin A (InlA) is a sortase achored, cell wall protein and a critical factor in the pathogenesis of the foodborne Gram-positive pathogen Listeria monocytogenes . InlA stimulates L. monocytogenes entry into normally non-phagocytic intestinal enterocytes [ 1 ]. The protein is 800 amino acids (aa) in length and composed of seven distinct domains (Figure 1a ); (i), 1-35 aa: a consensus N-terminal signal sequence (SS); (ii), 35-78 aa: forms a cap at the N-terminus tip of mature protein (C); (iii), 79-407 aa: 15 Leucine rich repeats (LRR) with 14 containing 22 aa (repeat 6 contains 21 aa) (shaded grey); (iv), 415-495 aa: an inter-repeat domain (IR); (v), 518-706 aa: three β-repeat domains, which may serve as a stalk to project the sickle shaped LRR out from the cell surface (β1, β2 and β3); and (vi), 767-771 aa: a sortase cleavage site (LPPTG) for peptidoglycan cross linking and (vii) 771-800 aa: a membrane targeting sequence (TM)[ 2 ]. Domain (iii) containing the LRR's of InlA is sufficient to stimulate enterocyte uptake [ 3 , 4 ]. The enterocyte ligand for InlA was identified as E-cadherin (CDH1) [ 5 ], which is required by host cells for the formation of tight junctions and to promote cellular polarization, communication and differentiation [ 6 ]. The localization of CDH1 on the basolateral face of differentiated cells suggested that invasion was a secondary event, occurring after non-specific uptake by M cells [ 5 ]. Oral infection studies using rats [ 7 ] and mice [ 8 ] provided support for this hypothesis. However, oral infections resulted in the invasion of enterocytes in a guinea pig model [ 9 ]. Human colonic Caco-2 enterocyte cells are also directly permissive to infection in vitro [ 9 , 10 ]. These seemingly anomalous results are due to the reduced affinity of murine CDH1 (mCDH1) for InlA. The reduced affinity was localized to amino acid 16 which is a proline in guinea pig and human CDH1 (hCDH1) but in rats and mice a glutamic acid is present [ 11 ]. This discovery led to the development and application of a transgenic mouse model expressing both human and murine CDH1 within intestinal enterocytes, which conclusively demonstrated the role of InlA in the pathogenesis of orally acquired L. monocytogenes [ 12 ]. In an elegant study, the site of enterocyte cell extrusion at the tips of intestinal villi was identified as a mechanism for exposing CDH1 on the apical surface at multicellular junctions [ 13 ]. More recently, a transgenic mouse strain that ubiquitously expresses human E-cadherin has been developed to demonstrate a role for InlA (and InlB) in fetoplacental listeriosis [ 14 ]. The crystal structure of InlA in complex with hCDH1 demonstrated the structural importance of proline 16 for the interaction [ 15 ]. In silico analysis confirmed that the reduced affinity of InlA for mCDH1was essentially due to the steric hindrance imposed by the bulky glutamic acid at aa 16, which therefore could not interact with the hydrophobic pocket (between LRR's 5, 6 and 7 of InlA) created by the removal of one amino acid from LRR 6 [ 15 ]. Overall the crystal structure identified 28 residues of hCDH1 that interact with the residues across the LRR region. Structural data and the invasion results from previous research [ 3 , 4 ] have confirmed the essential nature of the LRR's in the InlA::CDH1 interaction. Small animal model of listeriosis have a number of significant limitations. Even though rabbits and guinea pigs possess a permissive CDH1, they have recently been shown to be resistant to systemic infection due to a species specificity observed in the InlB/host interaction [ 16 ]. InlB is required for efficient hepatocyte/endothelial cell invasion in the mouse model and in certain human cell lines. A novel approach to address the lack of appropriate animal models focused on the 'murinization' of L. monocytogenes rather than the 'humanization' of mice [ 17 ]. Rational protein design based on the structural data of the InlA/hCDH1 complex, identified two mutations in InlA (Ser192Asn and Tyr369Ser) that dramatically increased the affinity for both hCDH1 and mCDH1. This allowed the development of a variant of L. monocytogenes EGD-e (EGD-InlA m ) capable of establishing systemic infections in C57BL/6J mice after oral inoculation [ 17 ]. However, the strain also exhibited a 2-fold increase in adhesion and consequently invasion into human cells, suggesting that the alteration in tropism towards mice also could enhance the virulence towards humans. To address any remaining concerns regarding human virulence of murinized L. monocytogenes , we conducted random mutagenesis of InlA combined with surface display on a non-invasive, Gram-positive, Lactococcus lactis to identify mutations that improve the entry into a colonic murine cell line. Using the CT-26 cells as a selection tool, multiple positive mutations in independent clones were identified with an enrichment in the InlA/hCDH1 interacting residues. The inlA genes from 4 L. lactis clones were separately recombined into the inlA chromosomal locus in EGD-eΔ inlA generating EGD-e A to D. Also, a version of EGD-InlA m [ 17 ] was created in order to permit comparison with our newly generated InlA mutant strains. In contrast to the strategy employed by Wollert et al. [ 17 ] we utilised preferred Listeria codons for the mutated 192Asn and 369Ser and designated the strain; EGD-e InlA m * . Strains were competed against EGD-e InlA m * in oral murine competitive index assays [ 18 ]. A novel aa mutation was identified which enhanced InlA/mCHD1 interaction compared to EGD-e. In agreement with earlier studies [ 17 ] the adherence/invasion into Caco-2 cells and virulence by murine intravenous infection of the codon-optimized EGD-e InlA m * strain was indistinguishable from EGD-e, while EGD-e InlA m * alone exhibited highly reproducible murine oral infections.

Methods Bacterial and Cell Culture Bacterial strains, plasmids and oligonucleotides are described in Table 1 . For the routine propagation of L. lactis MG1363 derivative NZ9000, cells were grown statically at 30°C in M17 (Oxiod) broth containing 0.5% w/v filter sterilized glucose (GM17). L. monocytogenes were cultivated in BHI (Oxiod) and Escherichia coli grown in LB at 37°C with shaking at 200 rpm. For growth on agar, respective broths were solidified with 1.5% (w/v) agar (Merck). For blue/white screening in L. monocytogenes , X-gal (Merck) was incorporated into BHI agar at 100 μg/ml. Antibiotics were added when required: erythromycin E. coli - 250 μg/ml, L. monocytogenes - 5 μg/ml and chloramphenicol L. lactis - 5 μg/ml. Plasmids were isolated from NZ9000 after overnight growth in 10 ml of GM17. To lyse, the pellet was resuspended in 500 μl of P1 buffer (see Qiagen manual) containing 30 μg of lysozyme and incubated for 30 min at 37°C. The lysate was processed as described in the Qiaprep spin miniprep kit (Qiagen). A nisin filtrate for P nisA induction was isolated from the supernatant of an overnight L. lactis culture of NZ9700 (filter sterilized through 0.22μM low protein binding filters - Millipore), aliquots frozen at -20°C. For all InlA induction experiments, overnight L. lactis NZ9000 cultures (containing pNZ8048 plasmids) were diluted 1:20 in 10 ml of fresh GM17 and grown to an OD 600 nm of 0.5 (approximately 2 h). The expression of inlA was induced with 10 μl of nisin and grown for a further hour to an OD = 1.0 (5×10 8 cfu/ml). The murine (CT-26) and human (Caco-2) colonic epithelial cell lines were routinely cultured at 37°C in 5% CO 2 . Media was composed of DMEM glutamax, 10% FBS, Pen/Strep and 1% non essential amino acids with all cell culture media purchased from Gibco. Oligonucelotides were purchased from Eurofins MWG Operon. Production of electrocompetent Lactococcus lactis The protocol of Holo and Nes [ 19 ] was adapted for the transformation of L. lactis MG1363 derivative NZ9000. A GM17 overnight culture of NZ9000 was diluted 1:100 into 5 ml of GM17 containing 500 mM sucrose and 2.5% glycine (GS-GM17). This culture was inoculated into 50 ml of fresh GS-GM17 and grown overnight. The 50 ml culture was inoculated into 400 ml of fresh GS-GM17, grown to OD600 of 0.3 and cells were subsequently harvested by centrifugation at 4,000 × g for 20 min at 4°C. The pellet was resuspended in 200 ml of ice cold SGB (500 mM sucrose and 10% (w/v) glucose - filter sterilized), centrifuged, resuspended in 100 ml SGB and left on ice for 15 min. The cells were centrifuged, resuspended in 50 ml SGB and left on ice for 15 min before a final centrifugation and re-suspension with 2 ml SGB. Cells were frozen at -80°C in 40 μl aliquots. To electroporate, cells were thawed on ice, mixed with 4 ul of pellet paint (Novagen) precipitated DNA and transferred to a 1 mm electroporation cuvette (Biorad). Cells were pulsed at 20 kV/cm, 200 Ω and 25 μF, regenerated in 1 ml GM17 containing 2 mM CaCl2/20 mM MgCl2 for 1.5 h and then plated onto GM17 agar containing 5 μg/ml chloramphenicol. An efficiency of 1 × 10 7 cfu/μg was routinely obtained with pNZ8048. Cloning of InlA into pNZB The unique Bgl II site up stream of the nisA promoter in pNZ8048 was removed by linearization of the vector with Bgl II and ends blunted with T4 DNA polymerase. The vector was religated to generate pNZB. The inlA gene was PCR amplified (primers IM194 and IM188) as described previously [ 20 ], digested with Nco I/ Pst I and ligated into the complementary digested pNZB. Ligations were directly electroporated into NZ9000 as described above and the sequence of the inlA gene was verified by DNA sequencing. QuikChange mutagenesis in L. lactis Primers for site directed mutagenesis (SDM) (Table 1 ) were designed according to the Quikchange SDM manual (Stratagene). All plasmid template isolated from NZ9000 strains was methylated with Dam methylase following manufacturer recommendations (New England Biolabs). The PCR thermocycling conditions were conducted as described previously [ 21 ]. Separate 50 μl KOD hotstart high fidelity polymerase PCR reactions were preformed with each primer for 10 cycles and an extension time of 5 min 30 sec. After 10 cycles the reactions were combined and continued for an additional 18 cycles. Amplimers were column purified (Qiaquick PCR purification kit, Qiagen) and digested overnight with Dpn I (Roche). Digests were pellet paint precipitated and the half of the digest directly electroporated into NZ9000. Between 200 and 1000 colonies were obtained per transformation. The protocol was repeated to combine SDM changes. From the final mutagenized plasmids, Bgl II/ Bst XI fragments containing the LRR region of InlA were excised and ligated into complementary digested pNZB inlA WT . Isolation of cell wall proteins Cell wall proteins were isolated from nisin induced 10 ml NZ9000+pNZB inlA WT culture as described by previously [ 22 ], except cells were rendered as protoplasts for 1 hr at 30°C without mutanolysin. Blotted proteins were probed with the InlA specific monoclonal antibody described by Hearty et al [ 23 ]. Random Bank of inlA mutants in NZ9000 The generation of a randomly mutated inlA bank between amino acids 74 and 512 (containing the LRR) of InlA was accomplished by error prone PCR with Mutazyme II (Stratagene). Plasmid DNA (pNZB inlA WT ) was used as template in the reaction (primers IM317 and IM318) and a 1.3 kb fragment amplified between two naturally occurring restriction sites ( Bgl II and Bst XI). From the mutagenesis reactions, four different mutation rates by varying the amount of template used ((iii) 1000 ng (iv) 250 ng (v) 10 ng and (vi) 0.1 ng). This equates to a sliding scale of increasing mutation frequency. Each amplimer pool was digested with Bgl II and Bst XI and ligated into complementary digested pNZ8048b inlA . The ligations (100 ng of pNZB with 240 ng of inlA ) were pellet paint precipitated and electroporated into electrocompetent NZ9000 (repeated twice). For each pool a total of 40,000 colonies were obtained with plasmid religations accounting for 0.125% of the total (about 50 colonies per 40,000). The colonies from each mutation frequency were pooled and frozen at -80°C. From each mutation frequency, 10 individual colonies were subjected to plasmid isolation as described above and the mutated region sequenced to access the level of mutagenesis. CT-26 and Caco-2 invasion assays Overnight cultures of NZ9000 containing pNZB only or pNZB inlA derivatives (Figure 1a ) were induced as described above. A one ml aliquot was then pelleted at 4,000 × g for 5 min and resuspended in 1 ml of DMEM. Cells were centrifuged, resuspended in fresh DMEM and then diluted to a multiplicity of infection of 25:1. L. monocytogenes cells were grown as described previously prior to invasion [ 20 ]. CT-26 [ 24 ] and Caco-2 cells were seeded at 2 × 10 4 and 1 × 10 5 cells, respectively and grown for 4 days until confluency in 24 well plates (Falcon). On the day prior to use, antibiotics were removed from the media. On the day of use, cells were washed twice with DMEM to remove FBS. Both cell types were invaded for 1 h at 37°C in 5% CO 2 , washed once with Dulbecco's PBS (Sigma) and then overlayed with DMEM containing 10 (Caco-2) or 100 μg/ml (CT-26) gentamicin for 1 h. Monolayers were washed a further three times with PBS to remove residual antibiotic and then lysed with 1 ml of ice cold sterile water. Bacterial cells were enumerated by serial dilution in PBS and plated on GM17 agar containing 5 μg/ml chloramphenicol. The remaining lysate from error prone PCR pools were inoculated into GM17 containing 5 μg/ml chloramphenicol, grown overnight, stocked at -80°C with the protocol repeated for seven passages through CT-26 cells. EGD-e derivatives were plated onto BHI agar. Internalin A chromosomal mutagenesis in L. monocytogenes A 2 kb fragment was PCR amplified (primers IM467 and IM490) from the appropriate mutated pNZ8048b inlA plasmid, with primer design incorporating the first 16 nt upstream of the inlA GTG start codon. The amplimers were digested with Nco I/ Pst I, ligated into complementary digested pORI280 and transformed into E. coli strain EC10B (Table 1 ). The plasmids pORI280 and pVE6007 we co-transformed into EGD-eΔ inlA and mutagenesis preformed as described by previously [ 20 ]. The reconstruction of the inlA locus was identified by colony PCR (primers IM317 and IM318) with the integrity of the gene confirmed by DNA sequencing. Intragastric versus intravenous infections of Balb/c mice For all murine experiments, 6-8 week old female Balb/c mice (Harlan) were used. All experiments were approved by the institutional ethics committee. Tail vein intravenous infections were conducted as described previously [ 18 ] with an inoculum comprised of equal numbers of EGD-e::pIMC3 kan and EGD-e InlA m * ::pIMC3 ery (2 × 10 4 total in 100 μl). For oral inoculation, overnight cultures were centrifuged (7,000 × g for 5 min), washed twice with PBS and resuspended at 5 × 10 10 cfu/ml in PBS containing 100 mg/ml of CaCO 3 . A 200 μl inoculum was comprised of either a single strain (5 × 10 9 cfu) or a two strain mixture (5 × 10 9 of each strain). Mice were intragrastrically gavaged and the progression of infection followed over a three day time course. For bioluminescent imaging, mice were anesthetized on day 1 through to day 3 with isoflurane gas and imaged in a Xenogen IVIS 100 (Xenogen) at a binning of 16 for 5 min. Mice were euthanized with spleen and livers aseptically removed, imaged (binning of 8 for 5 min) and enumerated as previously described [ 18 ].

Results A L. monocytogenes gentamicin protection assay for murine cells Invasion into Caco-2 cells by L. monocytogenes is dependent on the expression of functional InlA [ 10 ]. We confirmed that a L. monocytogenes mutant producing InlA without the LRR and IR domain (Δ inlA ) is severely compromised in invasion, while an over expressing InlA strain exhibits dramatically enhanced invasion (Figure 2 ). To establish an equivalent murine assay for L. monocytogenes we used monolayers of CT-26 cells (murine colonic carcinoma cell line) originally isolated from Balb/c mice chemically treated to induce tumor formation [ 24 ]. While CT-26 cells are not enterocyte like (they exhibit an undifferentiated-fibroblast appearance [ 25 ]), the results from invasion assays showed that they provide characteristics suitable for use as an invasion model (Figure 2 ). The Δ inlA strain displayed a slight reduction (not statistically significant) in invasion compared to EGD-e, while over expression of InlA resulted in a modest increase in invasion. We speculate that this is due to a reduced affinity of InlA for mCDH1, however we have not assayed for mCDH1 production by CT-26 cells. Heterologous expression was then employed to distinguish InlA from additional virulence determinants on the surface of the L. monocytogenes . We chose to use the well characterized nisin inducible expression system [ 26 ] (Figure 1 ) to produce full length InlA on the surface of L. lactis . The system was chosen because production of functional InlA on the cell surface of L. lactis had previously been documented [ 27 ]. We compared the entry of L. lactis containing vector only ( L. lactis -pNZB), producing wild type InlA ( L. lactis InlA WT ) or producing InlA containing the Ser192Asn and Tyr369Ser, but with different codon usage to the previously described murinized InlA m [ 17 ] ( L. lactis InlA m * ) into Caco-2 and CT-26 cells. The presence of InlA on the cell surface was confirmed by Western blot analysis (Figure 1b ). The level of invasion for L. lactis -pNZB into Caco-2 cells is similar to that observed for EGD-eΔ inlA (Figure 2 and 3 ). As L. lactis is non invasive, the surviving bacterial cells probably represent bacteria not killed by the gentamicin treatment rather than internalized cells, as documented previously [ 1 ]. A similar level of entry into Caco-2 cells was observed for L. lactis InlA WT and L. lactis InlA m * , while entry into CT-26 cells was 27-30 fold greater for L. lactis InlA m * compared to L. lactis InlA WT (Figure 2 ). In contrast to a previous report [ 11 ], we observed an increased invasion into a murine cell line by the L. monocytogenes strain over-expressing InlA WT in contrast to the plasmid only control (Figure 2 ). A similar trend was observed when the L. monocytogenes InlA over-expressing strain and Δ inlA strain were compared (Figure 2 ) and was also seen in experiments in the L. lactis background (Figure 3 ). These results could be due to the high level of inlA expression from the P nis and P help promoters, amplifying the differences in InlA on the surface of L. lactis and L. monocytogenes cells (Figure 2 and 3 ). We interpret these results as evidence of a specific interaction between InlA and a cell surface receptor on CT-26 cells which stimulates bacterial cell entry. To summarise, we have established a gentamicin protection assay, capable of discriminating InlA mediated invasion into a murine cell line. Generation and screening of a random bank of InlA LRR mutants To generate diversity within the inlA gene we applied error prone PCR to the LRR region (between naturally occurring Bgl II/ Bst XI sites - Figure 1a ). Four separate banks were created containing different levels of mutation frequency, each containing about 40,000 L. lactis clones. Initial assessment by DNA sequencing of ten clones from each bank identified mutations throughout the LRR region with the level of mutation correlating with the concentration of input template DNA for the error prone PCR (data not shown). To identify positive mutations, pools were invaded through CT-26 cells en masse as detailed in Figure 4 . Sequential passages through CT-26 cells were required to remove the background functional InlA from the pools (Figure 5 ). Of the four banks only the highest mutation frequency resulted in an initial recovery below that of wild type InlA, which suggested that a significant number of clones contained inactivating mutations. From passage two through six a significant enrichment in positive mutations was observed, with a leveling off at passage seven (Figure 5 ). From passage six, eight clones from each bank were sequenced (Table 2 ) and assayed individually using both CT-26 and Caco-2 cells (Figure 6 ). All clones exhibited enhanced entry into CT-26 cells while no apparent differences for cell entry into Caco-2 cells were observed (compared to L. lactis InlA WT ). However, no clones were identified which were capable of matching the level of L. lactis InlA m * mediated entry into the murine cells. Sequence analysis revealed that 23 of the 32 clones contained amino acid changes in residues involved in direct interaction with CDH1. Of the four banks, only the lowest mutation frequency contained multiple clones with the same mutation (Gln190Leu), with this single amino acid change also found in one clone from an additional bank (Table 2 ). Characterization of murinized L. monocytogenes : competitive index assays Four inlA sequences conferring enhanced invasion into CT-26 cells were selected to be re-created in the chromosome of L. monocytogenes EGD-e. The mutations constituted two single aa changes for EGD-e A (Asn259Tyr) and EGD-e B (Gln190Leu). While three aa changes were introduced into EGD-e C (T164A, K301I, Q303E) and EGD-e D (S173I, L185F, L188I). These mutations were chosen based on the frequency of isolation in L. lactis (EGD-e B and C), the ability to attribute the phenotype to an aa change (EGD-e A) and the isolation of mutations all confined within one LRR (EGD-e D). A fifth strain was also created based on the Lmo-InlA m mutation [ 18 ], except with Listeria optimized codons for 192Asn and 369Ser, and was used as a positive control (EGD-e InlA m * ). Sequencing confirmed the integrity of the newly introduced mutations, with equivalent levels of InlA expressed on the surface of the strains as compared to EGD-e (assessed by western blot - data not shown). InlA m strain (termed EGD-e InlA m * ) was compared to the parental EGD-e strain for invasion into Caco-2 and CT-26 monolayers. No differences in invasion (Figure 7a ) or adherence (data not shown) were observed to Caco-2 cells, while the invasion of EGD-e InlA m * was significantly higher than EGD-e into CT-26 cells. We then compared the virulence of EGD-e and EGD-e InlA m * by competitive index (CI) assays via the intravenous (i.v.) (Figure 7b ) or intragastric (i.g.) (Figure 7c ) route in Balb/c mice. For i.v. inoculated mice, no differences in the kinetics of infection were observed for either strain (Figure 7b ). This confirms that the two amino acid changes in InlA m do not impact on the virulence of EGD-e InlA m * once the gastrointestinal tract is bypassed. However, EGD-e InlA m * was significantly more virulent when infected by the i.g. route, with higher counts obtained from livers and spleens and a significantly higher CI value (p < 0.001) for both day two (Liver 28.9, Spleen 10.6) and day three (Liver 24.9, Spleen 7.7 - Figure 7c ). Neither strain was recovered form the liver nor spleen at day one post infection. Subsequent competitive index experiments were conducted by the i.g. route comparing EGD-e InlA m * against the strains expressing the InlA mutations identified by the CT-26 cell screen (Figure 7d ). Of the four recreated strains, only EGD-e A (N259Y) gave a higher CI than EGD-e in the liver (0.19 vs 0.05) whereas identical values (0.12) were obtained for the spleens. Further experimentation will be required to access the contribution of the N259Y mutation, and it would be intriguing to see if the recombination of this mutation into EGD-e InlA m * would further enhance murine pathogenicity. It is interesting to note that the strain in which InlA m (with Listeria optimized codons for 192Asn and 369Ser) was recreated (EGD-e InlA m * ) did not exhibit enhanced invasion or adhesion to Caco-2 cells, which is a marker for human virulence, in contrast to the previously published results [ 17 ]. To further explore the progression of i.g. infection, we repeated the Balb/c inoculations with either EGD-e or EGD-e InlA m * tagged with a constitutive bioluminescent lux marker and mice were imaged for bioluminescence on each subsequent day [ 18 ]. The EGD-e InlA m * strain exhibited uniform clinical signs of L. monocytogenes infection by day 2 [ 28 ], while these characteristics were absent from the EGD-e group even prior to sacrifice at day 3. Consistent with the clinical scores very little light was observed from the EGD-e group, while increasing light levels were obtained from the EGD-e InlA m * group on days 1 and 2, with a distinct foci evident in the abdomen in all 5 mice by day 3 (Figure 8a ). Upon ex vivo imaging of the livers, a low signal was present in the gall bladder in 3 of the 5 EGD-e infected mice, whereas a much stronger signal was found from the gall bladders of all EGD-e InlA m * (5 out of 5) infected mice, with infection across the liver also observed (Figure 8a ). The EGD-e InlA m * infected gall bladders were also found to be to twice the size of the EGD-e group. Further work is necessary to determine the exact extent of gall bladder colonization in these animals relative to hepatocyte infection. Enumeration of the livers and spleens confirmed that the EGD-e InlA m * strain produced highly reproducible i.g. infections, with the levels recovered comparable to day three i.v. infections in the liver (Figure 8b ). A much larger degree of variation was observed in the EGD-e group, with statistically significant differences in bacterial counts observed between the two strains (Figure 8b ). The mechanism of gall bladder colonization is currently unknown [ 29 , 30 ] and warrants further investigation. The EGD-e InlA m * strain is capable of establishing highly reproducible colonization of the gall bladder upon i.g. inoculation. This strain will be extremely useful in examining factors required for gastrointestinal transit and gall bladder colonization.

Discussion It is now well established that the murine model of listeriosis is limited by a poor interaction between the bacterial invasion protein InlA and its host ligand mCDH1. This is in direct contrast, to the efficient interaction between InlA and hCDH1. The discrepancy is due to a glutamate at residue 16 in mouse (and rat) E-cadherin rendering these host species relatively resistant to infection by the oral route and limiting their use as laboratory models for certain L. monocytogenes- mediated disease processes [ 11 ]. Recent studies have developed an engineered mouse strain expressing 'humanized' E-cadherin for studies of oral and fetoplacental listeriosis [ 14 ]. An alternative approach has utilized structure-based engineering to 'murinize' the bacterial InlA protein in order to increase affinity for murine E-cadherin [ 17 ]. This approach has provided key insights into the interaction between InlA and CDH1. While murinization was highly successful, we reasoned that additional points of contact may also improve the interaction with mCDH1. We therefore developed a system to select random mutations in InlA that enhance invasion of murine cells in order to identify novel amino acid interactions and to determine if 'murinization' of the strain can be improved. L. lactis was used as a surrogate host for this process in order to prevent generation of Listeria mutants with increased affinity for human cells. Previous research had shown that heterologous expression of InlA from the native P inlA promoter in L. lactis could stimulate invasion into cultured human colonic enterocytes and guinea pig enterocytes in an oral infection model [ 27 ]. Additional properties of L. lactis such as high transformation efficiency (4 × 10 4 cfu for ligations) allowed us to generate multiple random libraries of substantial size and enabled the direct transformation of SDM constructs. Also the nisin inducible system enabled a high level of InlA expression on the surface of L. lactis in a background with relatively few sortase A anchored proteins. The ability of L. lactis InlA m * to facilitate uptake into murine cells encouraged us to use multiple rounds of en masse enrichment of InlA mutant libraries through CT-26 cells. The cumulative results from each passage showed a continued improvement in the invasion efficiency, suggestive of an enrichment of positive clones. A surprising level of diversity in InlA clones was apparent (across the 4 banks) with 25 of the 32 clones analyzed exhibiting unique sequences. Only bank iii with the lowest frequency of mutations exhibited a degree of clonality (4/8 were Q190L). This suggests that we have not yet uncovered the full complement of mutations within the banks which confer enhanced invasion capabilities. Directed evolution of the inlA gene has the potential to uncover mutations not predicted by a structure-based approach (Table 2 ). With respect to the Q190L mutation the glutamine at residue 190 found on LRR 6 within the hydrophobic pocket, and forms a hydrogen bond to proline 16 in hCDH1. The change to leucine may affect the pocket and improve access of glutamic acid 16 in mCDH1. Of all the single amino acid changes, the N259Y mutation exhibited the single greatest invasion increase into CT-26 cells. Combining this mutation with either T399I or L149 M was shown to reduce or enhance invasion, respectively, with the negative effect of the T399I confirmed by the reduction in invasion efficiency observed when combined with additional positive mutations (bank IV, clone 8 versus bank IV, clone 1-Table 2 ). Further biochemical studies will be required to identify the role these mutations play to enhance the interaction with mCDH1. The previously identified single aa changes at residues 192 and 369 [ 17 ] each increased invasion ~20 fold, whereas the combined 192 + 369 mutations increased invasion ~30 fold. The identical aa change at residue 369 was also isolated from our error prone PCR bank. However, this clone contained additional mutations that resulted in a reduced level of invasion compared to the 369 single mutant. The CDH1 interacting amino acids appear to be highly conserved and recalcitrant to change [ 31 ]. From a collection of 101 inlA gene sequences mapped onto the InlA crystal structure [ 32 ], three naturally occurring InlA variants were identified which could potentially mediate an interaction with hCDH1, with one (Lys301Glu) also identified through the random mutagenesis approach in our study. However, while all mutants containing this residue had a positive effect on invasion into CT-26 cells, the exact contribution of this residue could not be assessed as additional mutations were present in all clones. Further analysis of individual clones from each bank or the application of additional selection is required due to the diversity uncovered (25 of the 32 clones analyzed were different). This diversity and the enhanced invasion of all the clones examined confirms that amino acids additional to the ones previously examined [ 17 ] can modulate the affinity for CDH1. Despite the analysis of 32 clones from our enriched bank of InlA variants, we failed to detect mutations that yielded invasion rates comparable to the murinized InlA described by Wollert and coworkers [ 17 ]. In terms of developing usable models of murine listeriosis the approach of 'murinizing' the bacterial strain arguably has a number of benefits over the development of humanized mouse lines. Development of the modified bacterium will permit utilization of this strain in existing mouse lines (including existing knock-out murine models) and distribution of the murinized strain is relatively straightforward, as is the creation of new mutations in the EGD-e InlA m * background. However, the 2-fold enhanced adherence and invasion to human (Caco-2) cells of the L. monocytogenes Lmo-InlA m [ 17 ] could be a potential cause for concern as it is suggestive of a slight enhancement of virulence towards humans. The procedure used to create that strain required multiple prolonged incubations at 42°C [ 17 , 33 ]. It has been recently shown that high temperature growth of L. monocytogenes can induce spontaneous mutation, suggesting that high temperature growth should be minimized to avoid the acquisition of secondary mutations [ 34 ]. We re-created the InlA mutations described by Wollert et al ., [ 17 ] to create EGD-e InlA m * using only two temperature shifts to 37°C and six passages under non-selective conditions [ 20 ]. Another difference between the Lmo-InlA m and EGD-e InlA m * strain were the nucleotide changes made to create the mutated amino acids. In the EGD-e InlA m * strain the two codons were chosen based on the codon usage from genome analysis, with the most commonly used triplets applied. In each case usage was 50% higher than the one used in Lmo-InlA m . For the asparagine 192, AAT compared to the AAC codon was chosen (31.8 vs 14.4 per 1000 codons). While for serine 369 TCT compared to TCG codon was chosen (12.8 vs 6.2 per 1000 codons). The invasion data for Lmo-InlA m agreed with the biophysical characterization which showed an enhanced interaction for InlA with CDH1 [ 35 ] however as recently shown, synonymous mutations leading to mRNA sequence changes can also affect substrate specificity or protein activity [ 36 ]. To access the role of codon usage or strain background, competitive index experiments will need to be conducted to directly compare Lmo-InlA m with EGD-e InlA m * .

Conclusions The research presented here generated random InlA variants with enhanced invasion into the CT-26 cell line most likely through an increased affinity for mCDH1. Novel mutations in InlA were readily identified from the random mutagenesis approach and a number (including the N259Y mutation) are worthy of further study. The approach used here indicates that other random or targeted mutagenesis strategies may uncover mutations that further enhance protein-ligand binding. In particular we suggest that screening approaches such as biopanning [ 37 ] using the first extra cellular domain of mCDH1 as bait or a site-saturation mutagenesis approach (the analysis of all amino acid combinations at a single residue) [ 38 ] may uncover further potential interactions. We have demonstrated that the newly created strain, EGD-e InlA m * does not have an enhanced affinity for human cells (unlike the predecessor EGD-InlA m ) while displaying highly reproducible oral infections in the mouse model. The use of this murinized L. monocytogenes strain will prove a useful tool in analysing the gastrointestinal phase of listeriosis. The additional residues identified here as playing a role in InlA::CDH1 interactions will inform our ongoing efforts to create safer 'murinised' versions of L. monocytogenes which will help us to combat this often fatal pathogen.

Background Internalin A (InlA) is a critical virulence factor which mediates the initiation of Listeria monocytogenes infection by the oral route in permissive hosts. The interaction of InlA with the host cell ligand E-cadherin efficiently stimulates L. monocytogenes entry into human enterocytes, but has only a limited interaction with murine cells. Results We have created a surface display library of randomly mutated InlA in a non-invasive heterologous host Lactococcus lactis in order to create and screen novel variants of this invasion factor. After sequential passage through a murine cell line (CT-26), multiple clones with enhanced invasion characteristics were identified. Competitive index experiments were conducted in mice using selected mutations introduced into L. monocytogenes EGD-e background. A novel single amino acid change was identified which enhanced virulence by the oral route in the murine model and will form the basis of further engineering approaches. As a control a previously described EGD-InlA m murinized strain was also re-created as part of this study with minor modifications and designated EGD-e InlA m * . The strain was created using a procedure that minimizes the likelihood of secondary mutations and incorporates Listeria -optimized codons encoding the altered amino acids. L. monocytogenes EGD-e InlA m * yielded consistently higher level murine infections by the oral route when compared to EGD-e, but did not display the two-fold increased invasion into a human cell line that was previously described for the EGD-InlA m strain. Conclusions We have used both site-directed mutagenesis and directed evolution to create variants of InlA which may inform future structure-function analyses of this protein. During the course of the study we engineered a murinized strain of L. monocytogenes EGD-e which shows reproducibly higher infectivity in the intragastric murine infection model than the wild type, but does not display enhanced entry into human cells as previously observed. This murinized L. monocytogenes strain will provide a useful tool for the analysis of the gastrointestinal phase of listeriosis.

Authors' contributions All authors read and approved the final manuscript. IRM devised the study, carried out the experimental work and wrote the manuscript; PGC carried out murine infection work; CH and CGMG devised and guided the study and helped to draft the manuscript.

Acknowledgements The authors would like to thank Richard O'Kennedy and Stephen Harty for generously supplying the InlA monoclonal antibody. We would like to acknowledge the funding received from the Irish Government under the National Development Plan 2000-2006 and the funding of the Alimentary Pharmabiotic Centre by the Science Foundation of Ireland Centres for Science Engineering and Technology (CSET) programme.

CC BY

no

2022-01-12 15:21:45

BMC Microbiol. 2010 Dec 13; 10:318

oa_package/17/1c/PMC3016325.tar.gz

PMC3016326

21156049

Background Deinococcus radiodurans is an extreme bacterium known for its resistance to ionizing radiation (IR), ultraviolet (UV) radiation, oxidative stress, and desiccation [ 1 , 2 ]. It has been reported that D. radiodurans can recover from exposure to γ-radiation at 15 kGy, a dose lethal to most life forms. IR can directly damage biomacromolecules and can also produce reactive oxygen species (ROS) that can indirectly attack both proteins and DNA [ 3 , 4 ]. Therefore, cellular defense against ROS-induced protein and DNA damage is proposed to be important to the radiation resistance of D. radiodurans [ 5 ]. Manganese plays an important role in the antioxidant systems of bacteria and can relieve the phenotypic deficit of sod -null Escherichia coli [ 6 ]. Interestingly, Daly and coworkers found that the Mn/Fe ratio of most IR-resistant bacteria is higher than that of IR-sensitive bacteria. The group also found that D. radiodurans grown in manganese-deficient medium was relatively more sensitive to IR than the bacteria grown in manganese-containing medium, suggesting that the accumulation of intracellular manganese ions can protect proteins from ROS-induced damage and can help in the survival of D. radiodurans in extreme environments [ 5 , 7 , 8 ]. Although manganese can improve cellular ROS resistance, excess manganese is toxic to cells. Thus, maintenance of the intracellular Mn concentration homoeostasis is a challenge. In bacteria, two main classes of manganese transporters have been identified--Nramp H+-Mn 2+ transporters and the ATP-binding cassette (ABC) Mn 2+ permeases [ 9 ]. Recently, a manganese efflux system was identified in Streptococcus pneumoniae , and this was found to play important roles in host pathogenesis and H 2 O 2 resistance [ 10 ]. Many genes involved in the maintenance of manganese ion homeostasis have been reported in D. radiodurans , such as dr1709 , dr2523 [ 11 ], dr2539 [ 12 ], and dr0615 [ 13 ]. Therefore, it would be very interesting to determine whether D. radiodurans possesses a similar manganese efflux system. In this study, we identified a manganese efflux gene ( dr1236 ) in D. radiodurans and demonstrated that it plays an important role in maintaining the homeostasis of intracellular Mn. The null mutant mntE - was highly sensitive to manganese ions. When the intracellular level of manganese ions was increased by mutating dr1236 , the mutant showed clearly enhanced resistance to oxidative stress. Our results also demonstrated that increased intracellular Mn levels could substantially suppress protein oxidation (carbonylation) in D. radiodurans exposed to H 2 O 2 , indicating that manganese transport and regulation may be involved in the cellular resistance of D. radiodurans to oxidative stress.

Methods Strains and media All the strains and plasmids used in this study are listed in the supporting information (Table 1 ). The D. radiodurans strains were cultured at 30°C in TGY (0.5% Bacto tryptone, 0.1% glucose, and 0.3% Bacto yeast extract) medium with aeration or on TGY plates supplemented with 1.2% Bacto agar. Disruption and complementation of dr1236 The mutant dr1236 gene was constructed as described previously [ 23 ]. Briefly, ~600-bp DNA fragments immediately upstream and downstream from dr1236 were amplified from the genome of the R1 strain using the primer pairs ME1/ME2 and ME3/ME4, respectively (Table 2 ). These two fragments were digested with Bam HI and Hin dIII, respectively, and cloned to the streptomycin-resistance DNA fragment from pKat-aadA [ 24 ] pretreated with the same enzymes. The ligation product was transformed into D. radiodurans R1, and mutant colonies were selected on TGY plates containing 8 μg/mL streptomycin. Null mutants were confirmed by PCR and sequencing, and the resulting mutant was designated mntE - . A complementary plasmid was constructed and transformed into the mntE - mutant as described previously [ 25 ]. Briefly, the dr1236 gene with the Nde I and Bam HI sites was amplified with primers ME5/ME6. The PCR product was ligated to the pMD18-T simple vector (Takara, JP), and the product was designated pMDmntE . After digestion with Nde I and Bam HI, the target gene MntE was ligated to Nde I- and Bam HI-predigested pRADK [ 23 ]. The complementation plasmid was confirmed by PCR and DNA sequence analyses and transformed into the mntE - strain. Cation sensitivity assay Cation sensitivity assays were carried out as described previously [ 18 ]. Solutions (1 M) of manganese chloride, manganese sulfate, calcium chloride, magnesium chloride, zinc chloride, cobalt (II) chloride, copper chloride, ferric chloride, and ferrous sulfate (Sigma) were prepared in milli-Q water and filter-sterilized by passing through 0.22-μm filters. Cells grown to the early stationary phase in TGY broth were plated on TGY plates and overlaid with 5-mm sterile filter discs containing 10 μL of various cation solutions. The plates were incubated for three days, and the inhibition zone of each disc was measured. To measure the growth of mntE - and R1, 1 × 10 5 cfu mL -1 were grown in TGY supplemented with increasing concentrations of MnCl 2 . The OD 600 value was measured 12 h post incubation (mean ± SD of three experiments). Inductively coupled plasma-mass spectrometry (ICP-MS) assay For the ICP-MS assays [ 26 ], the cells were cultured in TGY broth that had been pretreated with Chelex (Sigma) to remove any cations and supplemented with 50 μM manganese, 10 μM ferric chloride, 100 mM magnesium, or 100 mM calcium chloride. Cells (OD 600 = 0.6-0.8) were harvested by centrifugation, washed three times with phosphate-buffered saline (PBS) containing 10 mM EDTA, and rinsed three times with PBS without EDTA. Cells (1/10 of the total volume) were withdrawn to measure the dry weight, and the remaining cells were treated with nitric acid and used for the ICP-MS assay. Survival curves of the mntE - mutant and R1 R1 and mntE - cells were cultured in TGY broth with or without 50 μM manganese to OD 600 = 1.0, centrifuged, and then resuspended in phosphate buffer. For the γ-irradiation treatment, the suspension was irradiated with different doses of 60 Co γ-radiation for 1 h on ice. After the irradiation treatment, the cells were plated on TGY plates and incubated at 30°C for three days. The colonies were then counted. For the UV treatment, the cells were plated on TGY plates and exposed to different doses of UV radiation at 254 nm. For the H 2 O 2 treatment, the cultures were treated with different concentrations of H 2 O 2 for 30 min and then plated on TGY plates. Protein carbonylation assay Cells grown to OD 600 = 0.5 were treated with H 2 O 2 (30 mM), harvested, and resuspended in PBS containing 1% (by volume) β-mercaptoethanol and 1 mM phenylmethanesulfonyl fluoride. The cells were disrupted by sonication, and the cell-free extracts were used for the protein carbonylation assay. The protein concentrations were determined by the Bradford method. The cell-free extracts were incubated with 400 μL of 10 mM 2, 4-dinitrophenyl hydrazine (DNPH) in 2 M HCl for 2 h in the dark. After precipitation with ice-chilled 10% trichloroacetic acid (TCA), the precipitated proteins were washed three times with 50% ethyl acetate in ethanol. The decolorized precipitates were evaporated and dissolved in 1 mL of 6 M guanidine hydrochloride. The solution was centrifuged, and the absorbance of the supernatant was determined at 370 nm against a protein control that had been processed in parallel but with 2 M HCl instead of DNPH. The protein carbonyl content is defined as mM/mg protein. Statistical analysis Student's t -test was used to assess the significance between results, and p < 0.05 was considered as significant.

Results and discussion D. radiodurans encodes a putative manganese efflux protein By searching the D. radiodurans genome http://www.ncbi.nlm.nih.gov/ , we identified a manganese efflux protein homologue that was annotated as the conserved hypothetical protein DR1236 based on its extensive sequence similarity (25% identity, 49% similarity) to the manganese efflux protein (sp1552) of Streptococcus pneumoniae [ 10 ]. Similar to most cation diffusion facilitator (CDF) proteins, DR1236 has six putative transmembrane domains (TMDs) http://www.ch.embnet.org/software/TMPRED_form.html . The most conserved region of the CDF protein is the TMD region, which is probably involved in metal transfer [ 14 ]. Sequence alignment was performed with the CLUSTAL W program available on the EMBL web page http://www.ebi.ac.uk . The alignment Sp1552 and DR1236 revealed the presence of highly conserved sequences in metal transfer regions III and VI (Figure 1 ). Moreover, the DXXXD motif, which is conserved in the manganese efflux protein, was also present in DR1236 (224 DAGVD 230). mntE is essential for the manganese resistance of D. radiodurans To confirm the specific substrate and roles of DR1236 in D. radiodurans , the null mutant of dr1236 ( mntE - ) and wild-type revertant mntE strains were constructed (Figure 2 ). Metals including manganese are essential yet potentially toxic to bacteria [ 15 ]. Supplementation with certain metal ions can inhibit the growth of an exporter system mutant [ 16 , 17 ]; therefore, this phenotype is used to verify certain mutants. In this study, wild-type R1 and dr1236 ( mntE - ) were grown on TGY plates overlaid with discs saturated with 10 μL of different metal ion solutions (1 M) containing manganese, magnesium, cobalt, calcium, copper, zinc, nickel, or iron ions. As shown in Figure 3A/B , the growth of the mntE - mutant was strongly inhibited by the manganese ions, but the mutant grew normally in the presence of other cations. Moreover, the wild-type revertant showed a growth phenotype similar to that of R1, indicating that growth inhibition of the mntE - mutant was due to the interruption of dr1236 . To further investigate the influence of manganese ions on the mntE - mutant, different concentrations of manganese ions were added to TGY medium, and the growth of the mntE - mutant was measured (Figure 3C ). The results showed that in comparison with R1, the growth of the mntE - mutant was clearly delayed in the presence of low concentrations of manganese ions. When the manganese concentration increased, the growth defect phenotype became more pronounced. This phenotype is similar to that observed in Rosch's study in which the growth of S. pneumoniae having a disrupted calcium efflux system was more severely inhibited at higher calcium concentrations [ 18 ]. The mntE - mutant shows high intracellular manganese concentrations To confirm that the mntE - mutant had lost its ability to export manganese ions, the intracellular manganese ion levels of wild-type R1 and the mntE - mutant were measured by inductively coupled plasma-mass spectrometry (ICP-MS). As expected, when grown on TGY medium supplemented with manganese ions, the manganese ion level in the mntE - mutant was almost four-fold higher than that in wild-type R1. However, there was no significant difference in the intracellular Fe ion concentrations of R1 and the mutant (Figure 4A ). Similar results were obtained when the mntE - mutant and wild-type R1 were grown on TGY medium (Figure 4B ). This result indicates that Dr1236 is a manganese ion exporter. The mntE - mutant shows higher resistance to γ-radiation, UV, and oxidative Recently, there has been a debate on whether the high intracellular Mn/Fe ratio of D. radiodurans contributes to the extreme oxidative resistance of this microorganism. Daly et al proposed that the high Mn/Fe ratio can effectively suppress protein carbonylation and increase radiation resistance [ 7 , 8 ]. In contrast, Sukhi et al and Shashidhar et al argued that D. radiodurans exhibits the same radiation resistance even when the intracellular Mn/Fe ratio changed substantially [ 19 , 20 ]. Since the intracellular manganese ion level was significantly increased, a cell survival experiment was carried out to confirm whether the high intracellular manganese ion level could contribute to cellular resistance. The D 10 value represents the irradiating dose required to reduce the population by 90%. Here, the D 10 value was proposed to assess the resistant ability of R1 and mntE - mutant to different stresses. As shown in Figure 5 the resistance of the mntE - mutant under different stresses was higher than that of R1, and the D 10 values of the mntE - mutant were 14000 Gy γ-radiation, 700 J/m 2 UV, and 50 mM H 2 O 2 , whereas that for R1 was 11000 Gy γ-radiation, 600 J/m 2 UV, and 40 mM H 2 O 2 . Moreover, when R1 and mntE - mutant were cultured in TGY supplemented with 50 μM manganese, their resistance to different stresses also increased remarkably, and it is consistent with their intracellular manganese level (Figure 5 ). The results suggest that there is a correlation between the intracellular manganese level and cellular oxidative resistance, which is consistent with the data from Daly's studies [ 8 ]. Although the role of manganese in the oxidative resistance of D. radiodurans remains unclear, our study implies that an increase in the intracellular manganese level may be one of the responses to oxidative stress. Moreover, it is notable that the UV resistance of the mntE - mutant also increased. Generally, UV light results in DNA damage, and only high doses of UV cause oxidative damage. Therefore, it is interesting to speculate that the UV resistance of the mntE - mutant may be indirectly enhanced by manganese ions. In fact, many important DNA repair enzymes use Mn 2+ as the cofactor [ 21 ], and manganese accumulation may have a positive effect on gene function. Furthermore, a high intracellular manganese level is also known to have an important effect on the expression of many genes including stress response genes [ 10 ]. The mntE - mutant shows a lower protein oxidation level under oxidative stress The protein carbonylation level is an important index of intracellular oxidative damage to proteins [ 8 ]. Previous reports have shown that the proteins of IR-sensitive bacteria are more vulnerable than those of D. radiodurans to ROS-induced protein oxidative damage [ 7 ]. Therefore, we measured and compared the levels of protein carbonylation in the mntE - mutant and wild-type R1. Notably, the level of protein carbonylation in the mntE - mutant decreased to nearly 50% of that in R1 after H 2 O 2 treatment (Figure 6 ), indicating that the mutation of mntE resulted in a lower level of protein oxidation than that observed in the wild type. This suggests that the high level of intracellular manganese ions could enhance cellular resistance by protecting proteins against ROS damage.

Results and discussion D. radiodurans encodes a putative manganese efflux protein By searching the D. radiodurans genome http://www.ncbi.nlm.nih.gov/ , we identified a manganese efflux protein homologue that was annotated as the conserved hypothetical protein DR1236 based on its extensive sequence similarity (25% identity, 49% similarity) to the manganese efflux protein (sp1552) of Streptococcus pneumoniae [ 10 ]. Similar to most cation diffusion facilitator (CDF) proteins, DR1236 has six putative transmembrane domains (TMDs) http://www.ch.embnet.org/software/TMPRED_form.html . The most conserved region of the CDF protein is the TMD region, which is probably involved in metal transfer [ 14 ]. Sequence alignment was performed with the CLUSTAL W program available on the EMBL web page http://www.ebi.ac.uk . The alignment Sp1552 and DR1236 revealed the presence of highly conserved sequences in metal transfer regions III and VI (Figure 1 ). Moreover, the DXXXD motif, which is conserved in the manganese efflux protein, was also present in DR1236 (224 DAGVD 230). mntE is essential for the manganese resistance of D. radiodurans To confirm the specific substrate and roles of DR1236 in D. radiodurans , the null mutant of dr1236 ( mntE - ) and wild-type revertant mntE strains were constructed (Figure 2 ). Metals including manganese are essential yet potentially toxic to bacteria [ 15 ]. Supplementation with certain metal ions can inhibit the growth of an exporter system mutant [ 16 , 17 ]; therefore, this phenotype is used to verify certain mutants. In this study, wild-type R1 and dr1236 ( mntE - ) were grown on TGY plates overlaid with discs saturated with 10 μL of different metal ion solutions (1 M) containing manganese, magnesium, cobalt, calcium, copper, zinc, nickel, or iron ions. As shown in Figure 3A/B , the growth of the mntE - mutant was strongly inhibited by the manganese ions, but the mutant grew normally in the presence of other cations. Moreover, the wild-type revertant showed a growth phenotype similar to that of R1, indicating that growth inhibition of the mntE - mutant was due to the interruption of dr1236 . To further investigate the influence of manganese ions on the mntE - mutant, different concentrations of manganese ions were added to TGY medium, and the growth of the mntE - mutant was measured (Figure 3C ). The results showed that in comparison with R1, the growth of the mntE - mutant was clearly delayed in the presence of low concentrations of manganese ions. When the manganese concentration increased, the growth defect phenotype became more pronounced. This phenotype is similar to that observed in Rosch's study in which the growth of S. pneumoniae having a disrupted calcium efflux system was more severely inhibited at higher calcium concentrations [ 18 ]. The mntE - mutant shows high intracellular manganese concentrations To confirm that the mntE - mutant had lost its ability to export manganese ions, the intracellular manganese ion levels of wild-type R1 and the mntE - mutant were measured by inductively coupled plasma-mass spectrometry (ICP-MS). As expected, when grown on TGY medium supplemented with manganese ions, the manganese ion level in the mntE - mutant was almost four-fold higher than that in wild-type R1. However, there was no significant difference in the intracellular Fe ion concentrations of R1 and the mutant (Figure 4A ). Similar results were obtained when the mntE - mutant and wild-type R1 were grown on TGY medium (Figure 4B ). This result indicates that Dr1236 is a manganese ion exporter. The mntE - mutant shows higher resistance to γ-radiation, UV, and oxidative Recently, there has been a debate on whether the high intracellular Mn/Fe ratio of D. radiodurans contributes to the extreme oxidative resistance of this microorganism. Daly et al proposed that the high Mn/Fe ratio can effectively suppress protein carbonylation and increase radiation resistance [ 7 , 8 ]. In contrast, Sukhi et al and Shashidhar et al argued that D. radiodurans exhibits the same radiation resistance even when the intracellular Mn/Fe ratio changed substantially [ 19 , 20 ]. Since the intracellular manganese ion level was significantly increased, a cell survival experiment was carried out to confirm whether the high intracellular manganese ion level could contribute to cellular resistance. The D 10 value represents the irradiating dose required to reduce the population by 90%. Here, the D 10 value was proposed to assess the resistant ability of R1 and mntE - mutant to different stresses. As shown in Figure 5 the resistance of the mntE - mutant under different stresses was higher than that of R1, and the D 10 values of the mntE - mutant were 14000 Gy γ-radiation, 700 J/m 2 UV, and 50 mM H 2 O 2 , whereas that for R1 was 11000 Gy γ-radiation, 600 J/m 2 UV, and 40 mM H 2 O 2 . Moreover, when R1 and mntE - mutant were cultured in TGY supplemented with 50 μM manganese, their resistance to different stresses also increased remarkably, and it is consistent with their intracellular manganese level (Figure 5 ). The results suggest that there is a correlation between the intracellular manganese level and cellular oxidative resistance, which is consistent with the data from Daly's studies [ 8 ]. Although the role of manganese in the oxidative resistance of D. radiodurans remains unclear, our study implies that an increase in the intracellular manganese level may be one of the responses to oxidative stress. Moreover, it is notable that the UV resistance of the mntE - mutant also increased. Generally, UV light results in DNA damage, and only high doses of UV cause oxidative damage. Therefore, it is interesting to speculate that the UV resistance of the mntE - mutant may be indirectly enhanced by manganese ions. In fact, many important DNA repair enzymes use Mn 2+ as the cofactor [ 21 ], and manganese accumulation may have a positive effect on gene function. Furthermore, a high intracellular manganese level is also known to have an important effect on the expression of many genes including stress response genes [ 10 ]. The mntE - mutant shows a lower protein oxidation level under oxidative stress The protein carbonylation level is an important index of intracellular oxidative damage to proteins [ 8 ]. Previous reports have shown that the proteins of IR-sensitive bacteria are more vulnerable than those of D. radiodurans to ROS-induced protein oxidative damage [ 7 ]. Therefore, we measured and compared the levels of protein carbonylation in the mntE - mutant and wild-type R1. Notably, the level of protein carbonylation in the mntE - mutant decreased to nearly 50% of that in R1 after H 2 O 2 treatment (Figure 6 ), indicating that the mutation of mntE resulted in a lower level of protein oxidation than that observed in the wild type. This suggests that the high level of intracellular manganese ions could enhance cellular resistance by protecting proteins against ROS damage.

Conclusions Although it is known that the Mn/Fe ratio of D. radiodurans is higher than that of other bacteria, little is known regarding the maintenance of the intracellular manganese ion level in this bacterium. So far, only one manganese efflux system has been identified in bacteria [ 10 ], and it is still unknown whether this system exists in D. radiodurans [ 22 ]. In this study, we identified a MntE homolog in D. radiodurans . As expected, our results showed that the intracellular manganese ion level was almost four-fold higher in the mutant than in R1. Furthermore, we also found that the oxidative level of mntE - proteins decreased to almost one half that of R1. On the other hand, the data also revealed that manganese accumulation is dangerous to the mntE - mutant. Based on these data, we conclude that dr1236 is indeed a mntE homologue and is indispensable for maintaining manganese homeostasis in D. radiodurans . The results provide additional evidence that intracellular manganese ions are involved in the radiation resistance of D. radiodurans . However, because the intracellular Mn/Fe ratio and the Mn concentration of mntE - both increased in this study, we could not clarify whether the Mn/Fe ratio or the Mn concentration is more important for stress tolerance. Therefore, global analysis of the regulation of the intracellular manganese ion level is necessary in further studies.

Background Deinococcus radiodurans accumulates high levels of manganese ions, and this is believed to be correlated with the radiation resistance ability of this microorganism. However, the maintenance of manganese ion homeostasis in D. radiodurans remains to be investigated. Results In this study, we identified the manganese efflux protein (MntE) in D. radiodurans . The null mutant of mntE was more sensitive than the wild-type strain to manganese ions, and the growth of the mntE mutant was delayed in manganese-supplemented media. Furthermore, there was a substantial increase in the in vivo concentration of manganese ions. Consistent with these characteristics, the mntE mutant was more resistant to H 2 O 2 , ultraviolet rays, and γ-radiation. The intracellular protein oxidation (carbonylation) level of the mutant strain was remarkably lower than that of the wild-type strain. Conclusions Our results indicated that dr1236 is indeed a mntE homologue and is indispensable for maintaining manganese homeostasis in D. radiodurans . The data also provide additional evidence for the involvement of intracellular manganese ions in the radiation resistance of D. radiodurans .

Authors' contributions HXS and YJH conceived and designed the study. HXS performed the experiments and wrote the manuscript. GZX, BT and HC participated in the discussion of the experimental results. HDZ and ZTS carry out the protein carbonylation analysis. All authors read and approved the final manuscript.

Acknowledgements This work was supported by a grant from the National Basic Research Program of China (2007CB707804), a grant from the National Hi-Tech Development Program (2007AA021305), a key project of the National Natural Science Foundation of China (30830006), a major scientific and technological project for significant new drugs creation (2009ZXJ09001-034), a major project for genetically modified organisms breeding (2009ZX08009-075B), a grant from the National Natural Science Foundation of China (30870035), the project "Application of Nuclear Techniques in Agriculture" from the Chinese Ministry of Agriculture (200803034), and a grant from Zhejiang Provincial Natural Science Foundation (Y3090032).

CC BY

no

2022-01-12 16:12:49

BMC Microbiol. 2010 Dec 14; 10:319

oa_package/9f/a0/PMC3016326.tar.gz

PMC3016327

21138591

Background Many studies have shown that ambient pollution is consistently associated with adverse health outcomes [ 1 - 6 ], but mechanisms accountable for these associations have not been fully elucidated. Suggested biological mechanisms linking air pollution and cardiovascular diseases include direct effect on the myocardium, disturbance of the cardiac autonomic nervous system, pulmonary and systematic oxidative stress and inflammatory response that triggers endothelial dysfunction, atherosclerosis and coagulation/thrombosis [ 7 ]. Understanding relative roles of such potential is a priority of recent air pollution epidemiology. Some studies have demonstrated that exposures to particulate matter (aerodynamic diameter ≤2.5 μm, PM 2.5 ) and ozone are associated with global oxidative stress [ 7 - 11 ]. Others reported that the exposures were associated with heart rate variability (HRV), plasma homocysteine and C-reactive protein and such effects were modified by genetic polymorphisms related to oxidative defenses [ 12 - 16 ]. In living cells, reactive oxygen species (ROS) are continuously generated as a consequence of metabolic reactions, which may cause oxidative damage to nucleic acids. DNA damage may be repaired by the base excision repair pathway. The resulting repair product, 8-Hydroxy-2'-deoxyguanosine (8-OHdG), is the most common DNA lesion [ 17 ] and is not affected directly by either diet or cell turnover [ 18 ]. Therefore, 8-OHdG is a good biomarker for ROS or oxidative stress. A limited number of epidemiological studies reported that 8-OHdG was associated with exposures to indoor and ambient pollution or smoking, but they were conducted among a small number of children or occupationally exposed employees [ 9 , 10 , 19 ]. Oxidative stress caused by air pollution may be implicated in the development of respiratory disease, cardiovascular disease, lung cancer and other diseases [ 20 - 22 ]. Our recent study found that the elevated urinary 8-OHdG was associated with pollutants often thought of as secondary or formed through photochemical reactions after emission (PM 2.5 , nitrogen dioxide, NO 2 , maximal one-hour ozone, O 3 , sulfate, SO 4 2- or organic carbon, OC), but not with directly emitted primary pollutants (black carbon, BC, carbon monoxide, CO or elemental carbon, EC), suggesting that secondary pollution plays a stronger role in oxidative stress [ 23 ]. Several studies have demonstrated that certain genetic polymorphisms related to oxidative stress modified effects of PM on cardiovascular responses [ 6 , 13 , 14 ], but a set of examined single nucleotide polymorphisms (SNPs) was very limited. Further, these studies only indirectly implicated oxidative stress as none of these outcomes was a direct measure of oxidative stress. For example, some studies reported that associations between exposure to PM 2.5 and heart rate variability (HRV) were modified by polymorphisms of the glutathione-S-transferase M1 (GSTM1) gene [ 14 ] or heme oxygenase-1 (HMOX) [ 15 ], enzymes that reduce impacts of ROS. Our previous studies examined a set of genotypes related to oxidative stress and found that polymorphisms of hemochromatosis (HFE) and glutathione S-transferase T1 (GSTT1) significantly modified associations of PM 2.5 with plasma homocysteine [ 12 ]. Anh et al. [ 24 ] reported that vitamin D-related genes (group-specific component, GC) were significantly associated with the serum D-vitamin concentrations that related to prostate cancer. However, the selection of certain genes is somewhat arbitrary and the use of an array of genes is vulnerable to false positives from multiple comparisons, a major issue in genetic association studies. In this study, we aimed to examine whether daily ambient OC, SO 4 2- and maximal one-hour O 3 were associated with urinary 8-OHdG based on our previous findings [ 23 ] and such associations were modified by genotypes related to oxidative stress in the Normative Aging Study population (NAS). Because of multiple comparisons, we used the Multiple Testing Procedures (MTP) modified by our team, multtest in the R project http://www.r-project.org to identify significant SNPs from a set of candidate genes [ 25 - 28 ].

Methods Study population Data were obtained from a longitudinal NAS [ 29 ]. Briefly, the NAS is a longitudinal aging population initiated by the Veterans Administration (VA) in 1963. A total of 2,280 men from the greater Boston area free of known chronic medical conditions were enrolled. Subjects were asked to return for examinations every three to five years in the study center, including routine physical examinations, laboratory tests, collection of medical history, social status information, and administration of questionnaires on smoking history, food intake and other factors that may influence health. All participants provided written informed consents and the study protocol was approved by the institutions. By 2006, only did a small proportion of participants remain in the cohort, as many participants had died or were lost to follow up. A total of 320 participants, who still remained in this cohort, were included in our analyses, visiting the clinic between January 2006 and December 2008 for measurement of urinary 8-OHdG and other covariates (no repeated measurements). 8-hydroxy-2'-deoxyguanosine and plasma analysis of B vitamins Urinary 8-OHdG analysis was conducted by Genox Corp (Baltimore, MD). A competitive enzyme-linked immunosorbent assay was used to analyze urinary 8-OHdG [ 30 , 31 ]. The measurement methods have been described elsewhere [ 23 ]. Folate, vitamin B6 and B12 in fasting plasma were analyzed at the USDA Human Nutrition Research Center on Aging at Tufts University. Folate and vitamin B12 were examined by radioassay using a commercially available kit from Bio-Rad (Hercules, CA); vitamin B6 (as pyridoxal-5-phosphate) by an enzymatic method using tyrosine decarboxylase. Further details are described elsewhere [ 32 , 33 ]. Plasma creatinine was measured with urine 8-OHdG using spectrophotometric assay. The method has been described elsewhere in details [ 34 ]. Air pollution and Weather Data Averages of daily OC, SO 4 2- and maximal one-hour O 3 were used in this study. O 3 and OC were provided by the Massachusetts Department of Environmental Protection and SO 4 2- was measured at Harvard School Public Health monitoring station. For each day, SO 4 2- , OC and O 3 values were averaged for periods for up to four weeks before the visit as these averaging periods were shown to be most relevant in our previous analyses. Findings from our previous study showed that 8-OHdG were only associated with the secondary pollutants [ 23 ]. To adjust for weather condition, we used apparent temperature as an index, defined as a person's perceived air temperature, given the humidity [ 35 ]. Genotypes In order to avoid the arbitrary selection of genes, we selected all 20 oxidative stress-related SNPs available in the NAS database. We examined effect modification using the array of candidate SNPs, including catalase (CAT, rs480575, rs1001179, rs2284367 and rs2300181), HFE H63 D (rs1799945), HFE C282Y (rs1800562), GSTM1, GSTT1, GSTP1 I105V (rs1695), GSTP1 A114V (rs1799811), HMOX (rs2071746, rs2071747, rs2071749, rs5995098), HMOX-1 VNTR, GC (rs2282679, rs1155563), glutamate cysteine ligase catalytic subunit (GCLC, rs17883901) and glutamate cysteine ligase modifier (GCLM, rs2301022 and rs3170633). HFE is related to cellular uptake of metals that are related to ROS generation and inflammation [ 8 , 36 ]. Glutathione pathways play a vital role in cellular defenses against ROS [ 14 , 37 - 39 ]. Similarly, GC, GCLC and GCLM are related to glutathione-related metabolism [ 40 , 41 ]. CAT helps catalyze hydrogen peroxide, a powerful ROS into water and molecular oxygen to maintain oxidative balance [ 39 , 42 ]. HMOX-1 was categorized into two levels (any short and both long) based on repeated number of microsatellite (GTn) because previous studies have shown that a high GT repeats at 5'-flanking region may reduce HMOX-1 inducibility by ROS and has been associated with increased risk of cardiovascular diseases [ 15 , 43 , 44 ]. Previous studies have shown that variations of HFE C282Y, HFE H63 D, HMOX-1, GSTs genes modify associations of PM 2.5 or BC with HRV or homocysteine [ 12 - 15 ]. Multiplex polymerase chain reaction assays were designed using Sequenom SpectroDESIGNER software (Sequenom Inc, San Diego, Calif) by inputting sequence containing the SNP site and 100 bp of flanking sequence on either side of the SNP. Assays were genotyped using the Sequenom MassArray MALDI-TOF mass spectrometer (Sequonom, CA, USA) with semiautomated primer design (SpectroDESIGNER, Sequenom) and implementation of the very short extension method [ 45 ]. Assays failing to multiplex were genotyped using the TaqMan 5' exonuclease [Applied Biosystems (ABI), Foster City, CA, USA] with primers from ABI using radioactive labeled probes detected with ABI PRISM 7900 Sequence Detector System [ 46 ]. Statistical analyses Statistical analyses were performed with R version 2.9.1. First, we fitted linear regression models to separately examine the association of a single pollutant with urinary 8-OHdG at different day moving averages up to four weeks during the study period to decide which day moving averages for each pollutant were strongly associated with 8-OHdG for effect modification assessment. We used the log-transformation of 8-OHdG to minimize residuals and to stabilize the variance. We identified a priori the following variables as important potential confounders based on our previous NAS studies and other studies [ 9 , 12 , 14 ]: age, body mass index (BMI), alcohol consumption (≥2 drinks/day; yes/no), smoking status (never, former, current), pack-years of cigarettes smoked, plasma folate, vitamin B6, B12, use of statin medication (yes/no) and season and chronic disease status (cardiovascular disease, diabetes and chronic cough). We controlled plasma folate, vitamin B6, B12, age, BMI and pack-years of cigarettes smoked as continuous variables and adjusted for alcohol consumption, smoking status, use of statin medication and season as categorical variables. We adjusted for temperature using three-day moving average of apparent temperature with linear and quadratic terms due to the potential nonlinear relationship between temperature and 8-OHdG. In addition, we adjusted for creatinine clearance rate using the Cockcroft-Gault formula ([140 - age(year)]*weight(kg)]/[72* serum creatinine(mg/dL)]) [ 47 ]. We also adjusted for chronic disease status (cardiovascular disease or chronic respiratory diseases) as a dummy variable [ 23 ]. We examined effect modification by each of candidate SNP via adding an interaction term of the SNP and the pollutant simultaneously with both the main effect terms adjusting for the same covariates as the above [ 12 , 23 ]. Because two dozens of candidate SNPs were involved in the analyses, results were vulnerable to the multiple comparison problem. To decrease type I errors, we used MTP model to identify the significance of interaction terms of individual SNP and pollutant. The current version of MTP allows one to identify the significance of a group of candidate variables to reduce the false discovery rate meanwhile adjusting for a group of fixed covariates. We used MTP to identify the significance of the group of interaction terms. Because the current version of MTP in R can only include one term that varied across models, our team modified it to include two terms, i.e., the main effect term of genes and the interaction term of one pollutant and genes. We used the family-wise error rate (fwer) for type I error adjustment, step-down max T (sd.maxT) for method and default values for others in MTP. We briefly described the rationale here. More details about the rationale are described elsewhere [ 25 - 27 ]. MTP is based on Bootstrap estimation of the null distribution samples and the data generating distribution P. Samples are drawn at random with replacement from the observed data. The program generates B bootstrap samples from hypotheses M and obtains M × B samples or M × B matrix of test statistics. Then, based on the M × B matrix of test statistics, the bootstrap estimates or test statistics are induced. There are several methods to define type I error and calculate adjusted p-values in MTP. We selected family-wise error rate and step-down maxT methods in this study. In step-down procedures, the hypotheses corresponding to the most significant test statistics are considered successively, with further tests depending on the outcomes of earlier ones. Therefore, it is more powerful than a single-step. The adjusted p-values for the step-down maxT procedures are given by [ 26 ] where Pr refers to p-value, H denotes hypothesis, and T means test statistic. MTP directly reported adjusted p-values. An advantage of this method as opposed to only rejection or not of hypotheses, is that it is not needed to determine the level of the test in advance. This study reported adjusted p-values. Then, we quantitatively estimated associations between the pollutants and 8-OHdG across those carrying variants of the significant genes identified by MTP with significant interactions using the bootstrap method with the combination of coefficients of the main effect and the interaction [ 6 ].

Results Table 1 shows the descriptive statistics of the demographic characteristics, health and environmental variables among the NAS population during 2006-2008 at visit (n = 320). There were no repeated measurements in this study. Table 2 shows distributions of polymorphisms of candidate genes. Among 320 participants, wild types were dominant for CATs, HFEs, GSTP1 (rs1799811), HMOX (rs2071749) and GCLC, but the situation varied for other candidate genes. There were no obvious differences for the distributions of wild and heterozygous types in GCLM, GC and GSTP1 (rs1695). Heterozygous types for HMOX (rs2071746 and rs2071749) consisted of large components. 80.9% and 48.8% of subjects were classified as non-deletions for GSTT1 and GSTM1, respectively. Mean of the HMOX-1 GC repeated number was 25.8 (SD: 3.3) with median 24. We first fit the linear regression model to estimate associations of OC, SO 4 2- and maximal one-hour O 3 with 8-OHdG using moving averages of pollutants up to four weeks. Results show that main effects varied across different day moving averages and 24-, 20- and 18-day moving averages were strongest associated with SO 4 2- , OC and maximal one-hour O 3 , respectively, which were used to assess effect modifications. The detailed information has been reported elsewhere [ 23 ]. For an IQR increases in 24-, 20- and 18-day moving averages of daily SO 4 2- , OC and maximal one-hour O 3 , urinary 8-OHdG increased by 29.0% (95% CI: 5.9%, 52.1%), 27.6% (95% CI: 3.6%, 51.6%) and 54.3% (95% CI: 7.6%, 100.9%), respectively. Before examining effect modification, we categorized each candidate gene into a dummy variable so that the gene and the pollutant of interest only have one interaction term. We combined the homozygous and heterozygous types for appropriate genes known as the non-wild type (dominant model) due to small number of the homozygous type. We also combined the homozygous and heterozygous short repeat for HMOX-1, referred to as any short (Table 2 ). Then, we identified candidate genes that executed significant effect modification as aforementioned. Adjusted p-values in MTP model show that GSTP1 A114V (rs1799811) marginally significantly modified the effect of SO 4 2- on 8-OHdG (adjusted p = 0.091). CAT (rs2286367) (adjusted p = 0.037), GSTM1 (adjusted p = 0.037), GC (rs2282679) (adjusted p = 0.025) and GC (rs1155563) (adjusted p = 0.027) significantly modified effects of OC on 8-OHdG. There was no significant effect modification for O 3 (Table 3 ). As sensitive analyses, we used different options in MTP for typeone (type I error) (tail probabilities for error rate, TPPER; false discovery rate, FDR) and methods (single-step maximum T, ss.maxT; single-step minimum P ss.minP; step-down minimum P, ss.minP). Similar trends were found in spite of some variations. We also categorized pack-years of cigarettes smoked using tertiles as cut-off and re-ran MTP model. Results were similar to those using continuous variable for pack-years of cigarettes smoked. Figure 1 shows the estimated effects of OC or SO 4 2- on 8-OHdG across subpopulations carrying different genotypes, for those SNPs where an interaction with p < 0.10 was found.

Discussion We found that associations of the secondary pollutants, specifically OC and SO 4 2-, with 8-OHdG, a direct oxidative stress-related biomarker, were modified by polymorphisms in genes related to oxidative defenses. This is significant for several reasons. First, the finding that genetic polymorphisms in the oxidative defense pathway modified the association suggests that it is not due to chance or confounding, since neither should be associated with the genotypes of the individuals. Second, while considerable focus has been placed recently on freshly generated traffic particles, such as BC or ultrafine particle number, this study confirms that particles, including particles from coal burning power plants, play a role in increasing systemic oxidative stress. The specific polymorphisms that modified the associations were GSTP1 (rs1799811), GSTM1, CAT (rs1799811) and GC (rs22826799, rs1155563). We found 8-OHdG was more strongly associated with SO 4 2- among those carrying the wild type of the GSPT1, and more strongly associated with OC among those carrying the wild type of CAT (rs2284367), the non-deletion of GSTM1 and the non-wild type of the GCs (rs2282679 and rs1155563) comparing with other types of the corresponding genes (Figure 1 ). Based on our knowledge, it is the first time that MTP has been used to identify significant gene-environment interactions. MTP has advantages over some other approaches to controlling for false discovery rates in which a group of fixed covariates are adjusted for while a set of variables were compared. Several studies have examined effect modification and found that people carrying variants of oxidative stress-related genes are differentially susceptible to air [ 12 - 14 , 16 , 48 ]. Human GSTs are subdivided into several classes, among which GSTT1, GSTM1 and GSTP1 have been extensively investigated [ 12 , 14 , 49 , 50 ]. GSTM1 or GSTT1 catalyzes the conjugation of glutathione to numerous potentially genotoxic compounds [ 50 ]. Individuals with the deletion of GSTM1 or GSTT1 have been shown to reduce GST activity and thus may be unable to eliminate toxins as efficiently when they expose to oxidative pollutants [ 50 ]. Schwartz et al. [ 14 ] found that PM 2.5 was significantly associated with high frequency of HRV among those without the GSTM1 allele, but not for those with the allele. Gilliland et al. [ 48 ] reported that exposure to in utero maternal smoking was associated with increased prevalence of early onset asthma among those without GSTM1 allele, but not for those with GTSM1 allele. Similarly, Romieu et al. [ 51 ] found that GSTM1 null children were more sensitive to ozone exposure. However, all the aforementioned studies did not report whether there were significant effect modifications. Differential results from these stratification analyses might also be attributed to statistical powers across subpopulations or differential distributions of other controlled or uncontrolled covariates across subpopulations. This study observed that GSTM1 significantly modified associations of OC with 8-OHdG, but paradoxically that the GSTM1 null allele provided protection against exposure. Our recent study examined whether variations of a set of genes altered effects of black carbon and PM 2.5 on plasma homocysteine in this population and found that GSTT1 (but not GSTM1) significantly modified associations between pollutants and homocysteine. PM 2.5 and black carbon were more strongly associated with homocysteine among those carrying GSTM1 allele comparing those without the allele although no significant interactive effects were found [ 12 ]. Different findings of effect modification by GSTM1 variation across studies may reflect differences of exposure, outcome and population, measurement errors in exposure or phenotype, and by chance. Similar situations also appeared in other studies [ 52 , 53 ]. Therefore, statistical effect modification may be inconsistent with biological interaction. Further research or meta-analysis is needed for GSTM1. In contrast, few studies have examined the function of GSTP1 A114V (rs1799811) on diseases with inconsistent results [ 54 - 57 ]. None of these studies found the GSTP1 is significantly associated with the outcomes of interest although some studies found positive trends. Therefore, the functions of the polymorphisms have not been determined. Several studies examined effect modifications of GSTT1 on various endpoints but no significant effect modification was found [ 58 - 60 ]. For example, Melén et al. [ 59 ] examined whether GST modified traffic-related pollution effect on childhood allergic disease and found that carriers with variants of GSTP1 (rs1799811) were higher susceptible to NO x . Our study found the variation of GSTP1 showed a protective effect of SO 4 2- on 8-OHdG. However, other two studies did not find any evidence that the GSTP1 modified effects of black carbon or smoking on blood pressure or Parkinson's disease occurrence [ 58 , 60 ]. Inconsistent observed findings may be attributable to various sources as aforementioned. In this study, it may also related to the small number of variants in this population, which probably lead to unstable estimates. Therefore, its functions remain to be clarified by others (Table 2 ). GC, vitamin D-related genes, is related to the vitamin D metabolism [ 61 ]. Vitamin D is activated to form 1, 25-dihydroxyvitamin D in the liver and kidney and then transported in serum to different tissues by the vitamin D-binding protein, which is encoded by GC [ 61 ]. Studies show that polymorphisms of vitamin D-related genes are associated with various cancers, cardiovascular diseases and respiratory diseases [ 62 - 64 ]. Ahn et al. [ 61 ] examined variations of 212 SNPs related to vitamin D metabolism and found that all four SNPs of GC (rs1212631, rs2282679, rs7041, rs1155563) are significantly associated with the concentration of serum vitamin D. When these four SNPs were simultaneously included in the multivariate model, only two SNPs (rs22679, rs1155563) were significantly associated with vitamin D. In this study, we found that the two SNPs of GC (rs22679, rs1155563) were associated with 8-OHdG in this study. The mechanisms remain to be clarified yet. Catalase is a protein of 526 amino acids, encoded by the catalase gene with 34 kb pairs of nuclear acids [ 65 ]. Catalase is the main regulator of hydrogen peroxide metabolism [ 66 ]. Catalase enzyme mutations may reduce its activity and probably results in the increase of the hydrogen peroxide concentrations in the tissues [ 62 ]. Inherited catalase deficiency results in acatalasemia (homozygous state) and hypocatalasemia (heterozygous) and is related to increased plasma homocysteine concentrations [ 42 , 67 , 68 ]. Our previous study reported that the variation of CAT modified associations between particle matter and plasma homocysteine concentrations [ 12 ]. Experimental toxicology studies have shown that air pollutants act via the oxidative stress pathway [ 8 , 36 , 69 ]. Ghio et al. [ 36 ] found that homozygous Belgrade rats functionally deficient in divalent metal transporter-1 display decreased metal transport from the lower respiratory tract and have stronger lung injury than control littermates, when exposed to oil fly ash containing iron. Belgrade rats cannot transport iron and other divalent metals across membranes via HFE gene regulated processes. They also reported that healthy volunteers exposed to concentrated ambient air particles had increased concentrations of blood fibrinogen and induced mild pulmonary inflammation [ 8 ]. Tamagawa et al. [ 69 ] reported that five-day and four-week exposures to PM 10 caused acute and chronic lung and systematic inflammation of New Zealand rabbits. There are several strengths in this study. First, we used MTP model to identify the significance of a group of candidate genes while we examined effect modification by genes on air pollution effects. This method overcame some problems in this kind of studies, such as arbitrary selection of a few significant genes or high false discovery rate when individually examining a set of genes. Secondly, this study was conducted in a relatively large population. Information of participants was well measured and collected. However, several limitations also exist with this study. First, we used air pollution data collected from a single monitoring site for personal pollution exposure and therefore, some extent misclassification might happen. A recent study compared ambient concentrations with personal exposures with monitoring measurement and results show that ambient measures were good surrogates for PM 2.5 and SO 4 2- in both winter and summer, but O 3 was only good in summer, not well in winter [ 70 ]. Nevertheless, with non-differential misclassification, any potential bias would be expected toward the null. Second, MTP has several options to select type I error and several methods to calculate adjusted p-values. Using bootstrap re-sampling methods will result in different estimates when a MTP model is rerun. These will introduce the uncertainties in model selections [ 25 - 28 ]. In addition, the NAS consists of an aged population and non-Hispanic white men were dominant. Thus, the findings are not well generalizable to other populations.

Conclusions This study found that variations of oxidative stress-related genes modified effects of OC or SO 4 2- on 8-OHdG. This suggests that effects of OC or SO 4 2- on 8-OHdG and other endpoints may be through the oxidative stress pathway.

Background Air pollution is associated with adverse human health, but mechanisms through which pollution exerts effects remain to be clarified. One suggested pathway is that pollution causes oxidative stress. If so, oxidative stress-related genotypes may modify the oxidative response defenses to pollution exposure. Methods We explored the potential pathway by examining whether an array of oxidative stress-related genes (twenty single nucleotide polymorphisms, SNPs in nine genes) modified associations of pollutants (organic carbon (OC), ozone and sulfate) with urinary 8-hydroxy-2-deoxygunosine (8-OHdG), a biomarker of oxidative stress among the 320 aging men. We used a Multiple Testing Procedure in R modified by our team to identify the significance of the candidate genes adjusting for a priori covariates. Results We found that glutathione S-tranferase P1 (GSTP1, rs1799811), M1 and catalase (rs2284367) and group-specific component (GC, rs2282679, rs1155563) significantly or marginally significantly modified effects of OC and/or sulfate with larger effects among those carrying the wild type of GSTP1, catalase, non-wild type of GC and the non-null of GSTM1. Conclusions Polymorphisms of oxidative stress-related genes modified effects of OC and/or sulfate on 8-OHdG, suggesting that effects of OC or sulfate on 8-OHdG and other endpoints may be through the oxidative stress pathway.

Abbreviations BC: black carbon; OC: organic carbon; EC: element of carbon; SNP: single nucleotide polymorphism; NO2: nitrogen dioxide; CO: carbon monoxide; O3: ozone; 8-OHdG: 8'-hydroxy-2'-deoxyguanosine; PM 2.5 : particulate matter ≤2.5 μm in aerodynamic diameter; GST: glutathione S-tranferase; CAT: catalase; GC: group-specific component; HFE: hemochromatosis; HOMX: heme oxygenase-1; GCLC: glutamate cysteine ligase catalytic subunit; GCLM: glutamate cysteine ligase modifier; Competing interests The authors declare that they have no competing interests. Authors' contributions CR was responsible for study design, data analyses, result interpretation and manuscript writing. JS was responsible for study design, data collection and result interpretation. Other coauthors participated in the study design, data collection and result interpretation. All authors read and approved the final manuscripts.

Acknowledgements This work was supported by the National Institute of Environmental Health Sciences grants ES014663, ES 15172, and ES-00002, by U.S. Environmental Protection Agency grant R832416 and USDA Contract 58-1950-7-707. The Normative Aging Study is supported by the Cooperative Studies Program/Epidemiology Research and Information Center of the U.S. Department of Veterans Affairs, and is a component of the Massachusetts Veterans Epidemiology Research and Information Center. It is partially supported by Harvard-NIOSH ERC Pilot (T42 OH008416).

CC BY

no

2022-01-12 15:21:45

Environ Health. 2010 Dec 7; 9:78

oa_package/4b/42/PMC3016327.tar.gz

PMC3016328

21162750

Background Lipopolysaccharide (LPS), which is expressed on the outer membrane of essentially all Gram-negative bacteria, is a potent inducer of pro-inflammatory cytokines, including tumor necrosis factor-α (TNF-α interleukin-1β (IL-1β), IL-6, IL-8, arachidonic acid metabolites and nitric oxide [ 1 ]. LPS can also induce corticosteroid production by the host, which tends to suppress further production of pro-inflammatory cytokines. Some conditions leading to dysregulated production of inflammatory cytokines by the host can produce profound alterations in metabolic, cardiovascular, immunological, haemostatic, and endocrine functions, which may ultimately lead to septic shock [ 1 - 3 ]. Less profound inflammatory responses have also been implicated in the pathogenesis of atherosclerosis, cancer, stroke and diabetes in human subjects [ 4 - 7 ]. Proteasomes are essential for numerous physiological processes, including signal transduction, transcriptional activation, cell cycle progression, and certain immune cell functions [ 8 ]. We have reported a potentially important central role for proteasomes in inflammation and other macrophage functions [ 8 ]. Proteasomes often exist as 26 S multi-subunit complexes containing a 20 S proteolytic proteasome and a 19 S regulatory complex. Correspondingly, the 20 S proteasome is comprised of a variety of distinct protein subunits that account for the different proteolytic activities of the 20 S proteasome. Several different exogenous inhibitors or activators of proteasome function have been described, and these inhibitors act by blocking, or activating, the proteolytic activity of the individual protein subunits of the 20 S proteasome. We, and others, have reported that tocotrienols interfere with the formation of atherosclerotic plaque, and possess hypocholesterolemic, antioxidant, anti-inflammatory, antithrombotic, and anti-proliferative (anticancer) properties [ 9 - 22 ]. Tocotrienols are naturally occurring compounds containing a chroman ring and a farnesylated unsaturated side-chain with analogs of α-, β-, γ- and δ-type. These tocotrienols are minor constituents of natural vitamin E (predominantly α-tocopherol) which has a saturated side-chain attached to a chroman ring (Figure 1 ). Tocotrienols lower serum total- and LDL-cholesterol levels by inhibiting hepatic β-hydroxy-β-methylglutaryl coenzyme A (HMG-CoA) reductase activity through a post-transcriptional mechanism, which induces degradation of the reductase enzyme [ 19 ]. An unsaturated side-chain is essential for inhibition of hepatic HMG-CoA reductase activity. On the other hand, tocopherols (vitamin E) are well known for their characteristic antioxidant activity, but they do not increase reductase degradation or lower serum total or LDL-cholesterol levels [ 10 , 16 ]. The positive effects of tocotrienols as hypocholesterolemic, antioxidant, and anticancer agents have been confirmed in animal systems and various cell lines by many investigators [ 15 - 22 ]. Moreover, the far superior efficacy of tocotrienols versus tocopherols (vitamin E) as antioxidants has been established, and δ-tocotrienol is found to be the most potent among the known tocotrienols [ 10 , 17 , 18 , 22 ]. Tocotrienols also show non-antioxidant properties in various in vitro and in vivo models. Perhaps most importantly, tocotrienols interact with the mevalonate pathway leading to the lowering of cholesterol levels, the prevention of cell adhesion to endothelial cells, the suppression of tumor cell growth, and glutamate-induced neurotoxicity [ 10 - 12 , 23 , 24 ]. The present investigation was carried out to evaluate the mechanisms by which tocotrienols inhibit inflammation. Murine systems were chosen in the present study, because they are relatively resistant to the lethal effects of endotoxin compared to rabbit and sheep [ 1 , 3 , 25 ]. Recently, rodent models have been developed which are very sensitive to bacterial endotoxin and thus, female mice were chosen due to their increased sensitivity to LPS-induced inflammation. These experiments were designed to test our hypothesis that tocotrienol treatment modulates proteasomal activity, induces corticosteroids synthesis, and blocks LPS-induced signaling pathways that contribute to the inflammatory process.

Materials and methods Reagents Highly purified, deep rough chemotype LPS (Re LPS) from E. coli D31m4 was prepared as described by Qureshi et al. [ 26 ]. For tissue culture studies, Dulbecco's Modified Eagle Medium (DMEM), heat-inactivated low-endotoxin fetal bovine serum (FBS), and gentamicin were all purchased from Cambrex (Walkersville, MD). Thioglycollate was purchased from Sigma, Aldrich (St. Louis, MO) and the RNeasy mini kit from QIAGEN sciences (Germantown, MD). Substrates for the chymotrypsin-like activity of the proteasome were purchased from Calbiochem (La Jolla, CA). Tocotrienol rich fractions (TRF) of palm oil were provided by Malaysian Palm Oil Board, Kuala Lumpur, Malaysia (previously known as "Palm Oil Research Institute of Malaysia" [PORIM], Kuala Lumpur, Malaysia). "Proteasome-Glo" assays kits for chymotrypsin-like activity (substrate: Suc LLVY-Glo, Succinyl-leucine-leucine-valine-tyrosine-aminoluciferin), trypsin-like activity (Z-LRR aminoluciferin, Z-leucine-arginine-arginine-aminoluciferin), and post-glutamase activity (post-acidic, substrate, ZnLPnLD-Glo, Z-norleucine-proline-norleucine-aspartate-aminoluciferin) of the proteasome were purchased from Promega (Madison, WI). RAW 264.7 cells (TIB 71) were purchased from American Type Culture Collection (Manassas, VA), and 4-week-old BALB/c female mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Purification of α-tocopherol, α-, γ-, and δ-tocotrienols from TRF of palm oil The individual components of tocotrienol rich fraction (TRF) of palm oil were purified as described recently [ 10 ]. The purity of α-tocopherol and individual tocotrienols were established by high pressure liquid chromatography (HPLC) against their respective pure standards [ 10 ]. Effects of various tocols (tocopherols + tocotrienols) on the chymotrypsin-like activity of 20 S rabbit muscle proteasomes Proteasomal activities of the 20 S rabbit muscle proteasomes (0.4 μg/mL) were assayed with synthetic peptide substrates in 0.02 M Tris-HCl buffer (pH 7.2). The substrate used for the chymotrypsin-like activity was 100 μM of succinyl-Leu-Leu-Val-Tyr-amino-methyl-coumarin. Fluorescence was measured (absorption at 360 nm and emission at 460 nm) using an FLX 800 microplate fluorescence reader (Bio-Tek Instruments, Winooski, VT). Cell culture and maintenance The RAW 264.7 cells or mouse peritoneal macrophages were maintained in DMEM supplemented with 10% heat inactivated fetal bovine serum (FBS) and 10 mg gentamicin (in 500 mL) at 37°C in a humidified atmosphere with 5% CO 2 as described previously [ 27 ]. Cells were cultured in 6-well plates as described in the legends to the figures. Effects of δ-tocotrienol (concentrations of 10 - 640 μM) after 60 min treatment on different proteasomal activities (chymotrypsin-like, trypsin-like, and postglutamase) in RAW 264.7 whole cells In order to check the comparative inhibitory effect of tocols, only the most potent δ-tocotrienol was tested on the chymotrypsin-like, trypsin-like, and post-glutamase activities of the proteasome using RAW 264.7 whole cells; the following experiment was carried out. RAW 264.7 cells (10 × 10 4 cells/100 μl/well) were added in white plates 96-well, Fisher, 0877126), followed by the addition of various concentrations of δ-tocotrienol (10, 20, 40, 80, 160, 320, or 640 μM in 100 μL; dissolved in 0.2% dimethyl sulfoxide (DMSO). The mixtures were incubated at 37°C in an incubator at 5% CO 2 for 60 min. After incubation period, the cells in the 96-well plates were taken out 20 min prior to the addition of Caspase-Glo reagent (brought to room temperature before addition to the wells). Caspase-Glo reagent (100 μL) was added to each well to a total volume of 200 μL/well (tris buffer, pH 7.5; 0.02 M). The plate were covered with a plate sealer, removed from light and incubated at room temp for 10 min. The relative luminescence units (RLU) of assays were read with a Promega Plate Luminometer. The chymotrypsin-like, trypsin-like, or post-glutamase activities were quantitated by measuring luminescence after stimulation of RAW 264.7 whole cells with various doses of δ-tocotrienol in a Luminometer (Promega), according to the directions of manufacturer. Effects of various tocols (α-tocopherol, α-tocotrienol, γ-tocotrienol or δ-tocotrienol) on the secretion of TNF-α in LPS-stimulated RAW 264.7 cells The levels of TNF-α in RAW 264.7 cell culture supernatants were determined after treatment with LPS (1 ng/mL) and various doses (4, 8, 16 μM) of α-tocopherol, α-tocotrienol, γ-tocotrienol or δ-tocotrienol by Quantikine M ELISA kit (R&D System, Minneapolis, MN) according to manufacturer's instructions. The lower limit of detection for TNF-α in this method is approximately, 5.0 pg/mL [ 28 , 29 ]. The TNF-α levels in mouse's serum or thioglycollate-elicited peritoneal macrophages were also quantified by using the same method [ 28 , 29 ]. Effects of α-, γ-, and δ-tocotrienols on release of LPS-induced TNF-α in serum of 6-wk-old female BALB/c mice All mice used in this study received humane care in compliance with the principles of laboratory animal care formulated by the National Society of Health Guide for the Care and Use of Laboratory Animals (DHHS Publication No. [NIH] 85 - 23, revised 1985). The experimental procedures involving animals were reviewed and approved by the Institutional Animal Care and Use Committee of UMKC, Medical School, MO. Female BALB/c mice were acclimatized to the new environment for fourteen days and were fedad libitum regular commercial mice diet and had free access to water throughout the experiment. A 12 h light and 12 h dark cycle was maintained during feeding period. All samples were dissolved in 5% triethylamine solution. Seven groups of 4-wk-old female BALB/c mice (3/group) were acclimatized (for two wks), and then injected either with saline (two control groups), dexamethazone or various doses (2.5, 5.0, and 10.0 μg/kg body weight) of α-tocopherol, α-tocotrienol, γ-tocotrienol, or δ-tocotrienol. One h later, all mice were injected intraperitoneally with pure LPS ( E. coli D31m4; 10 ng/mouse). After two h, all mice were sacrificed, serum was collected, and the levels of TNF-α were quantified using a radioimmunoassay kit according to the manufacturer's directions [ 28 , 29 ]. Effects of α-, γ-, and δ-tocotrienols on the induction of corticosterone and adrenocorticotropic hormone (ACTH) in serum of LPS-stimulated 6-wk-old female BALB/c mice To further probe the anti-inflammatory effects of various tocotrienols in mice, we queried whether tocotrienols would also induce the serum levels of corticosterone and adrenocorticotropic hormone (ACTH) after LPS treatment of mice as reported recently [ 3 ]. The levels of serum (used from the above experiment) corticosterone and ACTH were estimated according to published method [ 1 , 3 ]. Isolation of thioglycollate-elicited peritoneal macrophages from BALB/c mice, total cellular RNA isolation and RT-PCR Ten 4-wk-old female BALB/c mice were acclimatized to the new environment for fourteen days, and thioglycollate-elicited peritoneal macrophages were prepared as described previously [ 29 ]. Macrophages (2 × 10 5 ) were treated with either pure LPS (10 ng/treatment), LPS + α-tocopherol (25, 50, or 100 μM), LPS + δ-tocotrienol (10, 20, or 40 μM). All the samples were dissolved in 0.2% dimethyl sulfoxide (DMSO). The assay mixtures were incubated at room temperature for 4 h and then centrifuged at 2,000 rpm for 20 min. The supernatants were removed and concentrations of TNF-α were determined using Quantikine M ELISA kit (R&D System, Minneapolis, MN) according to the manufacturer's directions [ 29 ]. The total RNA was isolated from each pellet with RNeasy mini kit according to the manufacturer's instructions. To check the purity of the total RNA, Reverse transcriptase polymerase chain reaction (RT-PCR) was conducted using a 1-step kit (Qiagen, Chatsworth, CA) according to the manufacturer's instructions. Detection of cell viability Viability of peritoneal macrophages treated with and without LPS plus dexamethasone (positive control), α-tocopherol or various doses of δ-tocotrienol was determined by trypan blue dye exclusion or a quantitative colorimetric assay with 3-(4,5)-dimethylthiozol-2,5-diphenyltetrazolium bromide (MTT) as described previously [ 30 ]. Statistical analyses The analyses demonstrated the affect of various isomers of tocols (α-tocopherol or α-, γ-, δ-tocotrienols) within the groups. Stat View software (version 4.01, Abacus Concepts, Berkeley, CA) was used for the analyses of treatment-mediated effects as compared to the control group. Treatment-mediated differences in various inflammatory markers variables were evaluated using a two-way ANOVA, and when F test indicated a significant effect, the differences between the means were analyzed by a Fisher's protected least significant and least difference test. Data were reported as means ± SD in text and Tables. The level for statistical significance level was established at 5% ( P < 0.05).

Results The results of the present study on the inhibition of inflammation by tocotrienols are presented in the same in two different formats. For each figure, 'A' shows the raw values for each of the treatments and controls, and 'B' shows the percent change compared to controls. Inhibition of chymotrypsin-like activity of 20 S rabbit muscle proteasomes by various tocols (tocopherols + tocotrienols) We have previously shown that the proteasome is a central regulator of inflammation; we therefore queried whether tocotrienols also affect proteasome activity, thus affecting the induction of cytokines. The results of this study reveals that inhibition of chymotrypsin-like activity of the 20 S rabbit muscle proteasomes was dose-dependent between 4 μM and 16 μM for α-tocotrienol (6% - 36%), γ-tocotrienol (12% - 32%), and δ-tocotrienol (51% - 55%) as compared to control groups (Figure 2A & 2B ). An insignificant reduction (5%) in 20 S proteasomes activity was observed with α-tocopherol (Figure 2A & 2B ). These studies further suggest that δ-tocotrienol is the most effective isomer of the tocotrienols for inhibiting the chymotrypsin-like activity of rabbit muscle proteasomes. Impact of δ-tocotrienol (concentrations of 10 - 640 μM) on different proteasomal active sites (chymotrypsin-like, trypsin-like, and post-glutamase) in RAW 264.7 whole cells after 60 min treatment The effects of δ-tocotrienol on the chymotrypsin-like, trypsin-like, and post-glutamase activities of proteasomes were also carried out in RAW 264.7 murine macrophages. The chymotrypsin-like activities decreased to 46% of controls with 40 μM δ-tocotrienol treatment of cells and then increased gradually as shown in Figure 3 . A similar trend was observed with the trypsin-like activity and post-glutamase activities in response to δ-tocotrienol (Figure 3 ). Inhibition of the secretion of TNF-α in LPS-stimulated RAW 264.7 cells by various tocols (α-tocopherol, α-tocotrienol, γ-tocotrienol or δ-tocotrienol) The effects of tocotrienols on the secretion of TNF-α with, and without, LPS-stimulation in RAW 264.7 cells was investigated. A significant reduction of secretion of TNF-α, 9% - 33% ( P < 0.05) was observed with tocotrienols in LPS-stimulated RAW 264.7 cells in a dose-dependent manner (Figure 4A & 4B ). Tocotrienols did not inhibit TNF-α secretion in the absence of LPS stimulation (data not shown in the figure). Cell viability for all above experiments was 90% - 95%. α-, γ-, and δ-Tocotrienols reduced serum TNF-α levels in LPS-stimulated 6-wk-old BALB/c mice To further probe the anti-inflammatory effects of various tocols in mice, we queried whether dietary supplementation with tocotrienols would also block LPS-induced TNF-α secretion in mice. A dose-dependent inhibition (20% to 48%; P < 0.05) of LPS-induced serum TNF-α levels was observed with α-, γ-, and δ-tocotrienols as compared to control diets (Figure 5A & 5B ). The reduction with α-tocopherol (5% - 9%) was insignificant as compared to control or tocotrienol treatments (Figure 5A & 5B ). These results also suggest that δ-tocotrienol is the most effective isomer in reducing LPS-induced TNF-α secretion in mice. α-, γ-, and δ-Tocotrienols induce corticosterone in serum of 6-wk-old BALB/c mice We have previously shown that LPS treatment of mice induces high endogenous corticosterone levels, and that corticosterone levels are inversely correlated with TNF-α serum concentrations [ 9 ]. Therefore, to further explore mechanisms responsible for the anti-inflammatory effects of tocotrienols, we queried whether tocols increase corticosterone levels produced in response to LPS. For this experiment BALB/c mice were injected i.p. with tocotrienols. A significant ( P < 0.02) rise in LPS-induced serum corticosterone (19%, 31%, and 41%) levels was induced by treatment with α-, γ-, or δ-tocotrienols as compared to the control group (Figure 6A & 6B ). These results suggest that the decrease in the synthesis of TNF-α could be due to the rise in the endogenous corticosteroids, which may modulate the synthesis of inflammatory cytokines [ 3 ]. α-, γ-, and δ-Tocotrienols induce adrenocorticotropic hormone in serum of 6-wk-old BALB/c mice The impact of α-tocopherol and various tocotrienols on the levels of adrenocorticotropic hormone was also determined in the serum obtained from the experiment described above. A corresponding significant ( P < 0.02) dose-dependent rise in LPS-induced serum adrenocorticotropic hormone levels was observed (81%, 118%, and 145%) by treatment with α-, γ-, or δ-tocotrienols, respectively, as compared with the control group (Figure 7A & 7B ). An insignificant rise in the levels of corticosterone (7%) and adrenocorticotropic hormone (14%) with α-tocopherol was also observed (Figure 6B & 7B ). In summary, these results clearly indicate that δ-tocotrienol is most effective in inhibiting the serum levels of TNF-α and also in inducing the serum levels of corticosterone and adrenocorticotropic hormone as compared to α-, γ-tocotrienols and α-tocopherol, which were the least effective compounds as compared to their respective control groups. δ-Tocotrienol inhibits LPS-induced TNF-α, IL-1β, IL-6, and iNOS gene expression in thioglycollate-elicited peritoneal macrophages of 6-wk-old BALB/c mice In order to define the mechanism of action of tocotrienols with respect to lowering inflammation, thioglycollate-elicited peritoneal macrophages were prepared from BALB/c female mice, and treated with α-tocopherol, or δ-tocotrienol 1 h before, or at the same time as LPS. After 4 h of the treatment, total cellular RNA was extracted and reverse-transcribed, and gene analysis was performed by RT-PCR and southern blot analyses. The results of gene expression studies clearly demonstrate that δ-tocotrienol is most effective in blocking LPS-induced gene expression of TNF-α Figure 8A & 8B , IL-1β, IL-6 and iNOS at low doses of 10 μM and 20 μM (Figure 9A & 9B ). There were significant ( P < 0.02) reductions of TNF-α (45%), IL-1β (63%), IL-6 (46%) and iNOS (32%) mRNA with a dose of 20 μM compared to their respective control values, and significant increases (40%, 17%, 26%, 13%) in the levels of these mRNA's, with a dose of 40 μM of δ-tocotrienol, as compared to their respective 20 μM control values (Figure 8B & 9B ). These results are in agreement with earlier results of tocotrienols on the activities of chymotrypsin-like, trypsin-like and post-glutamase in RAW 264.7 whole cells (Figure 3 ). The α-tocopherol (doses of 25, 50 or 100 μM) did not have any impact on the gene expression of TNF-α, IL-1β, IL-6 and iNOS compared to their respective control values (Figure 8A , Figure 9B ).

Discussion It is well established that LPS-induced TNF-α production requires activation of NF-κB and activation of proteasomal activity. The results presented here demonstrate that tocotrienols block production of key pro-inflammatory cytokines induced by LPS. We hypothesized that this blockade of LPS-induced inflammation by tocotrienols could be due to modulation of proteasomal activity [ 31 ], since tocotrienols have been shown to be effective inhibitors of β-hydroxy-β-methylglutaryl coenzyme A (HMG-CoA) reductase, (the rate-limiting enzyme in cholesterol biosynthesis), and are very potent antioxidant and hypocholesterolemic agents [ 12 ]. It has also been demonstrated that HMG-CoA reductase is degraded via the ubiquitin-proteasome pathway [ 32 , 33 ]. We have shown that tocotrienols can inhibit chymotrypsin-like activity of 20 S rabbit muscle proteasome, and that δ-tocotrienol specifically inhibits the chymotrypsin-like, trypsin-like and post-glutamase activities of the proteasome in a dose-dependent manner in RAW 264.7 whole cells at concentrations below 80 μM; at higher concentrations (80 - 640 μM) tocotrienole enhance these activities as has been previously reported for lactacystin [ 34 , 35 ]. Tocotrienols also inhibit the LPS-stimulated secretion of TNF-α in RAW 264.7 cells and in serum of BALB/c mice, and a corresponding significant ( P < 0.02) rise was observed in the serum levels of corticosterone and adrenocorticotropic hormone compared to controls. Moreover, δ-Tocotrienol is effective in reducing LPS-induced gene expression of TNF-α, IL-1β IL-6 and iNOS at low concentrations. Interestingly, at high concentrations δ-Tocotrienol activates chymotrypsin-like, trypsin-like, and post-glutamase activities, and upregulates LPS-induced gene expression of TNF-α, IL-1β, IL-6 and iNOS. It seems that tocotrienols actually upregulates the proteasomal activity, and thus can upregulate LPS-stimulated transcription of several pro-inflammatory genes at higher concentrations. Therefore, the capacity of δ-tocotrienol to modulate inflammation may be attributable, in part, to inhibition and activation of the chymotrypsin-like, trypsin-like and post-glutamase' activities of mouse macrophages. The present results also demonstrated that δ-tocotrienol is the most effective tocol among the known, α- and γ-tocotrienols for inhibition or induction of various cytokines involved in inflammation, whereas α-tocopherol does not play any role in inhibiting LPS-induced inflammation. This conclusion has been further confirmed in a recent study [ 32 ] which demonstrated that δ-tocotrienol is very effective in inducing ubiquitination and degradation of HMG-CoA reductase. δ-tocotrienol also blocks the processing of sterol regulatory element-binding proteins-2 (SREBPs-2) which, when activated, can enhance transcription of genes encoding cholesterol biosynthetic enzymes, including HMG-CoA reductase [ 32 ]. On the other hand, tocopherols neither increase degradation of reductase, nor decrease SREBP-2 processing [ 32 ]. The unsaturated isoprenoid side-chain in tocotrienol molecules is important for its biological activity and it was suggested that this could stimulate binding of HMG-CoA reductase to Insig (endoplasmic reticulum membrane protein). Alternatively, the side-chain might allow improved penetration and distribution in cell membranes [ 32 , 33 ]. This may contribute to the differences observed in the potency of tocotrienols versus tocopherols in stimulating HMG-CoA reductase in vitro ubiquitination [ 32 , 33 ] thus confirming our previous and present results showing that δ-tocotrienol (one methyl group) is more potent than γ-tocotrienol (two methyl groups) due to the number of methyl groups which abolishes regulatory activity with respect to HMG-CoA reductase degradation and SREBP-2 processing [ 10 , 11 , 32 ].

Conclusions Although LPS is a potent inducer of inflammation via the proteasome, it also activates an anti-inflammatory response by inducing corticosteroid production. In summary, δ-tocotrienol treatment has several anti-inflammatory effects that could be mechanistically analogous to the modulation of proteasomal activity by lactacystin, a well established proteasome inhibitor that can either increase or decrease proteasomal activity under different conditions [ 34 , 35 ]. The results of our current study also demonstrate that tocotrienols increase levels of adrenocorticotropic hormone (ACTH) and corticosteroids produced in response to LPS. The mechanisms by which LPS plus tocotrienols stimulate the hypothalamus-pituitary-adrenal (HPA) axis and the sites of action are currently unclear [ 3 ]. Regardless of the mechanism, however, the capacity of tocotrienols to increase the host corticosteroid response to LPS is likely to modulate inflammation. Other possible mechanisms for the anti-inflammatory properties of tocotrienols also include its superior antioxidant activity compared to α-tocopherol (vitamin E, [ 10 , 17 , 22 ]). All these latter mechanisms such as corticosteroid induction and antioxidant activities may be dependent on the proteasomal modulation by tocotrienols.

Background Inflammation has been implicated in cardiovascular disease, and the important role of proteasomes in the development of inflammation and other macrophage functions has been demonstrated. Tocotrienols are potent hypocholesterolemic agents that inhibit β-hydroxy-β-methylglutaryl coenzyme A reductase activity, which is degraded via the ubiquitin-proteasome pathway. Our objective was to evaluate the effect of tocotrienols in reducing inflammation. Lipopolysaccharide (LPS) was used as a prototype for inflammation in murine RAW 264.7 cells and BALB/c female mice. Results The present results clearly demonstrate that α-, γ-, or δ-tocotrienol treatments inhibit the chymotrypsin-like activity of 20 S rabbit muscle proteasomes (> 50%; P < 0.05). Chymotrypsin, trypsin, and post-glutamase activities were decreased > 40% ( P < 0.05) with low concentrations (< 80 μM), and then increased gradually with concentrations of (80 - 640 μM) in RAW 264.7 whole cells. Tocotrienols showed 9 - 33% ( P < 0.05) inhibitions in TNF-α secretion in LPS-stimulated RAW 264.7 cells. Results of experiments carried out in BALB/c mice demonstrated that serum levels of TNF-α after LPS treatment were also reduced (20 - 48%; P < 0.05) by tocotrienols with doses of 1 and 10 μg/kg, and a corresponding rise in serum levels of corticosterone (19 - 41%; P < 0.05) and adrenocorticotropic hormone (81 - 145%; P < 0.02) was observed at higher concentrations (40 μM). Maximal inhibition of LPS-induced TNF-α was obtained with δ-tocotrienol (10 μg/kg). Low concentrations of δ-Tocotrienols (< 20 μM) blocked LPS-induced gene expression of TNF-α, IL-1β, IL-6 and iNOS (> 40%), while higher concentrations (40 μM) increased gene expression of the latter in peritoneal macrophages (prepared from BALB/c mice) as compared to control group. Conclusions These results represent a novel approach by using natural products, such as tocotrienols as proteasome modulators, which may lead to the development of new dietary supplements of tocotrienols for cardiovascular diseases, as well as others that are based on inflammation.

Abbreviations ACTH: adrenocorticotropic hormone; DMSO: dimethyl sulfoxide; FBS: fetal bovine serum; IL-1β: interleukin-1β; IL-6: interleukin-6; IL-8: interleukin-8; HMG-CoA: β-hydroxy- β-methylglutaryl coenzyme A; HPLC: high pressure liquid chromatography; iNOS: inducible nitric oxide synthase; LDL: low density lipoprotein; LPS: lipopolysaccharide; MTT: 3-(4,5)-dimethyl-2,5-diphenyltetrazolium bromide; RNA: ribonucleic acid; RT-PCR: reverse-phase transcriptase polymerase chain reaction; SREBP-2: sterol regulatory element-binding protein-2.; TCL: total cholesterol; Tocols: mixtures of α-, β-, γ-, δ-tocopherols + α-, β-, γ-, δ-tocotrienols; TNF-α: tumor necrosis factor-α; TRF: tocotrienol rich fraction. Competing interests The authors declare that they have no competing interests. Authors' contributions JCR (graduate student) has carried out most of the experiments. All the authors were involved in the design of the study. CJP edited the manuscript. All the authors have read and approved the final version.

Acknowledgements This study was supported in part by Advanced Medical Research (AMR) and a grant from the Malaysian Palm Oil Board, Kuala Lumpur, Malaysia (previously known as Palm Oil Research Institute of Malaysia [PORIM]) and NIH grant 50870 (NQ). The study was carried out under a FDA approved IND number 36906.

CC BY

no

2022-01-12 15:21:45

Lipids Health Dis. 2010 Dec 16; 9:143

oa_package/c4/cd/PMC3016328.tar.gz

PMC3016329

21129219

Background The most important cause of mortality in diabetic patients is cardiovascular disease (CVD) [ 1 ] and observational studies have shown that diabetes mellitus (DM) increases risk of CVD nearly two to three folds [ 1 , 2 ]. In addition, some cohort studies conducted in American and European populations have suggested that rates for coronary heart disease (CHD) events are equivalent to both individuals with prior CHD and diabetic subjects without prior CHD [ 1 - 4 ]. Therefore, the American Diabetes Association recommended that, in secondary prevention, the diabetic patients should be treated for the same lipid and blood pressure targets as subjects with previous myocardial infarction [ 5 ]. Some other studies have found that cardiovascular risk are lower in subjects with diabetes but without CHD, than in persons with CHD and without diabetes [ 6 - 8 ]; however a few studies have also shown contradictory results [ 2 , 3 ]. Importantly, it is not clear whether these findings can be generalized to other populations because relative mortality risks may differ with ethnicity [ 9 - 11 ]. It is estimated that developing countries in Asia and in the Middle East, particularly in Persian Gulf states, will have the largest increases in the prevalence of diabetes by 2030, which is related to major shift in life style and nutrition transition in these countries [ 12 - 17 ]. Nevertheless, to our best knowledge, there is no study to show the equivalency of diabetes and prior CHD for risk of CHD in this region. The prevalence of Type 2 diabetes is reported to be more than 14% in Tehran, Iran, with an estimated incidence of new cases in about 1% of the population per year [ 14 , 15 ]. There is a high prevalence of CHD in this area too [ 18 ] and it has been shown that the Iranian population with diabetes has a high risk for CVD, independent of traditional risk factors [ 19 ]. Hence, we compared the risk of CHD events among diabetic participants without prior CHD to that of participants diagnosed as having CHD but not diabetes in the framework of the Tehran Lipid and Glucose Study (TLGS) which is a population based study conducted on a representative population of the capital city, Tehran [ 20 ]. Furthermore, we analyzed the results separately for newly diagnosed DM and known DM, and in addition to a prior history of CHD, we applied electrocardiographic criteria for the presence of CHD at baseline.

Methods Study population The TLGS is a prospective population-based study to determine the risk factors for non-communicable diseases among a representative urban population of Tehran [ 20 ]. The sampling method has been described elsewhere [ 20 ]. Briefly, a total of 15005 individuals aged 3 years and over who were residences of district No.13 of Tehran were selected using multistage cluster random sampling method. Subjects were categorized into the cohort and intervention groups, the latter to be educated for implementation of life style changes [ 20 ]. From among 8071 individuals, aged ≥ 30 years, who participated in the first phase of TLGS (February 1999 to August 2001), 5981 subjects had complete data of electrocardiogram (ECG) and history of CHD. Excluding participants with other missing data (n = 155) resulted in 5826 subjects at baseline; of those we followed 5198 (64.4% of total) until 20 March 2008 with a median follow up of 7.6 years (Figure 1 ). Written informed consent was obtained from all subjects and the ethical committee of Research Institute for Endocrine Sciences approved this study. Clinical and laboratory measurements at enrolment The information was collected during a personal interview, and completion of a questionnaire for demographic factors, medical history, medication use and smoking. Physical examination for blood pressure and anthropometrical measurements was performed. Blood pressure was measured twice in a seated position after 15 min resting using a standard mercury sphygmomanometer and the mean of the two measurements was considered as the subject's blood pressure. Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m 2 ). Two trained technicians recorded a 12-lead ECG according to the standard recording protocol developed by the School of Public Health, the University of Minnesota using a PC-ECG 1200 machine [ 21 ]. Two qualified physicians coded the ECGs in parallel according to the Minnesota codes using a measuring loupe specially manufactured by University of Minnesota [ 21 ]. Then, for assurance of quality a third qualified physician recoded 10% of ECGs. All the data were double-checked. Fasting plasma glucose (FPG) and lipid profiles were tested after 12-14 hours overnight fasting. Standard oral glucose tolerance test (OGTT) was performed in participants without treated diabetes. All analyses were done at the TLGS research laboratory on the day of collection. Plasma glucose, serum total cholesterol, high-density lipoprotein cholesterol (HDL-C) and triglyceride (TG) levels were measured by using previously reported methods [ 20 ]. Definition of terms Family history of premature CVD reflected prior diagnosis of CVD in female first-degree relatives, aged less than 65 years or male first-degree relatives under the age of 55 years old. Smoking was defined as history of current smoking (including daily or occasional). Newly diagnosed DM (NDM) was defined as individuals with FPG ≥ 7.0 mmol/L (126 mg/dl) or 2-h postchallenge glycemia (2hPG) ≥ 11.1 mmol/L (200 mg/dl) without any history of glucose-lowering medications and Known DM (KDM) was defined as subjects who have been treated with any kind of glucose-lowering medications. Prevalent CHD at enrolment Prevalent (or prior) CHD was defined as self-reported CHD or ECG positive CHD. Self-reported CHD was defined as a positive answer to the question as to "whether the subject has ever had a prior diagnosis of CHD by a physician". Electrocardiogram positive CHD was defined according to the Whitehall criteria [ 22 ] which categorize subjects into three groups, non CHD, probable CHD (any Minnesota codes of 1.1.1 through 1.1.7and 1.2.1 through 1.2.8) and possible CHD (codes of 1.3.1 through 1.3.6, 4.1.1 through 4.4, 5.1 through 5.3 or 7.1.1 through 7.1.2). In our analysis, we considered both probable and possible CHD as a single definition of ECG-positive CHD. We divided our cohort into six groups based on their clinical status at the baseline examination: - Subjects without newly diagnosed or known DM and without prior CHD (DM-/CHD-) - Subjects without newly diagnosed or known DM but with prior CHD (DM-/CHD+) - Subjects with NDM but without prior CHD (NDM+/CHD-) - Subjects with NDM and with prior CHD (NDM+/CHD+) - Subjects with KDM but without prior CHD (KDM+/CHD-) - Subjects with KDM and with prior CHD (KDM+/CHD+) Definition of CHD outcome Each TLGS participant is under continuous surveillance for any medical event leading to hospitalization during the previous year by telephone call and he/she is questioned by a trained nurse regarding any medical conditions. If a related event has occurred, a trained physician collects complementary data during a home visit, and when deemed necessary, a visit to the respective hospital for collecting data from the participant's medical files was done. In the case of mortality, data are collected from the hospital or death certificate by an authenticated local physician. Collected data are evaluated by an outcome committee consisting of a principal investigator, an internist, an endocrinologist, a cardiologist, an epidemiologist and the physician who collected the outcome data; other experts are invited for evaluation of non-communicable disorders as needed. Specific outcome for every event is assigned according to ICD-10 criteria and American Heart Association classification for cardiovascular events [ 20 ]. In the current study, CHD events as outcomes included cases of definite myocardial infarction (MI) diagnosed by electrocardiogram (ECG) and biomarkers, probable MI (positive ECG findings plus cardiac symptoms or signs but biomarkers showing negative or equivocal results), unstable angina pectoris (new cardiac symptoms or changing symptom patterns and positive ECG findings with normal biomarkers), angiographic proven CHD and CHD death. Statistics The results for continues variables are given as mean (± SD) and for categorical variables as percentages. Comparisons of the baseline characteristics among DM/CHD groups were done using ANOVA or χ2 test and we used t-test or χ2 test to compare any group with the DM-/CHD+ group considering Bonferroni correction for multiple comparisons. Cox proportional hazard regression models were used to estimate the hazard ratio (HR) of CHD events for DM/CHD groups (DM-/CHD+, NDM+/CHD-, NDM+/CHD+, KDM+/CHD-, KDM+/CHD+) in both genders, given DM-/CHD-as the reference group. Follow up duration was defined as the period between the entrance to study and the end point; end point was considered as the first CHD event and participants were censored at non-CHD death or end of follow-up. Analyses were done with adjustment for age alone and for age with other possible risk factors including systolic blood pressure, body mass index (BMI), total cholesterol, TG/HDL-C [ 23 ], family history of premature CVD, smoking and being in the intervention group. The proportional hazards assumption in the Cox model was assessed using log minus log plot of survival and Schoenfeld residual test. All proportional hazards assumptions were generally appropriate (p-value for global test of proportional hazards assumption was > 0.1 in men and women). A paired homogeneity test, which is a Wald test of the linear hypothesis of the Cox model regression coefficients, was performed to test the null hypothesis that the hazard ratio for NDM+/CHD-and KDM+/CHD-was equal to that for DM-/CHD+. Statistical analyses were performed using SPSS for windows version 15 and STATA version 10, and p-values less than 0.05 were considered statistically significant. All analyses were separated by gender regarding significant interaction between gender and some of the DM/CHD categories to predict CVD events.

Results In comparison to men who were not included in the study, those who were had less history of smoking (27% vs. 31.5%). Women included were younger (46.9 vs. 47.8 years), had lower systolic blood pressure (122 vs. 124 mmHg) and higher BMI (28.6 vs. 28.1 kg/m2) than those not included (all p < 0.05). Baseline characteristics for all groups are shown in Table 1 and 2 . Compared to the DM-/CHD+ group, NDM+/CHD-had higher TG/HDL-C and BMI values in men and higher TG/HDL-C, BMI and systolic blood pressure values in women; also KDM+/CHD-had higher systolic blood pressure and TG/HDL-C in women. Overall, 212 (9.4%) men and 146 (5%) women had CHD events during 16033 and 21433 person-years of follow up; the corresponding incidence density of CHD event were 13.2 and 6.8 per 1000 person-years in men and women respectively. In men, 46 (20.2%) of DM-/CHD+, 26 (17.1%) of NDM+/CHD-and 11(15.9%) of KDM+/CHD-individuals experienced CHD events; these values in women were 39 (13.7%), 19 (9.0%) and 25 (18.1%) respectively. Table 3 and 4 and Figure 2 highlight the HR of coronary heart disease in relation with DM and prior CHD in men and women. Age adjusted analysis revealed notably higher risks in DM-/CHD+, NDM+/CHD-and KDM+/CHD-groups compared with DM-/CHD-in both genders. However, in multivariate analysis, DM-/CHD+ and NDM+/CHD-in both genders and KDM+/CHD-only in women remained as significant predictors of CHD events. The risks of patients with DM or CHD were higher in women than men (in all of the DM/CHD categories) and p- value for the effect modification by gender reached the significant level of 0.05 in DM-/CHD+ and KDM+/CHD-groups ( p = 0.002 for both). Hazard ratios of CHD events were 2.1 (95%CI: 1.4-3.1), 1.7 (1.1-2.7) and 1.7 (0.9-3.3) for DM-/CHD+, NDM+/CHD-and KDM+/CHD-respectively in men, values which were 5.2 (3.2-8.3), 3.1 (1.8-5.6) and 6.2 (3.6-10.6) in women. Considering both NDM and KDM in a DM+/CHD-group, the HR of this group was 1.73 (95%CI: 1.15-2.60) and 4.37 (2.74-6.97) compared to DM-/CHD-group in men and women, respectively. Individuals with known or newly diagnosed DM and prior CHD demonstrated additive risks, i.e. HRs of 3.2 (95% CI: 1.9-5.6) and 4.2 (2.2-7.8) in men and 6.4 (3.2-12.9) and 8.0 (4.3-14.8) in women for NDM+/CHD+ and KDM+/CHD+ respectively. Finally, as shown in Table 3 and 4 , the results of the Wald test, used to compare the hazard ratios, highlighted that the risk of CHD events did not differ between KDM+/CHD-and DM-/CHD+ in both genders, however the risk of NDM+/CHD-was marginally lower than the HR of DM-/CHD+ in women ( p = 0.085).

Discussion When history of CHD or ischemic ECG changes was used to define prior CHD, this study showed that Type 2 diabetes is a "CHD equivalent" in a 7.6-year follow-up of Iranian subjects. The risk of CHD associated with diabetes was relatively equal to that associated with prior CHD, each conferring a ≈2 and ≈ 4-fold increased risk of CHD in men and women, respectively. Our findings are consistent with the highly cited Haffner et al' paper, in which Finnish known Type 2 diabetic patients without prior myocardial infarction have as high a risk of myocardial infarction as nondiabetic patients with prior myocardial infarction [ 24 ]. Because the Haffner et al. study included only 69 DM-/CHD+ individuals, the power of the study to detect differences between the 2 groups was limited. Similarly, in a population based study in Denmark, diabetic patients requiring glucose-lowering agents exhibited a cardiovascular risk comparable to nondiabetic with a prior myocardial infarction, regardless of sex and diabetes type [ 25 ]. However, Our findings are in contrast with those of Lee et al, in their Atherosclerosis Risk in Communities study (ARIC), in which after adjustment for multiple baseline risk factors, patients who had a history of myocardial infarction without diabetes at baseline had a 1.9 times the risk of fatal CHD or nonfatal myocardial infarction, compared with diabetic patients without a prior history of myocardial infarction [ 26 ]. Furthermore, in the Nurses' Health Study [ 27 ] and Health professional Follow-up study [ 8 ], Type 2 diabetic patients without myocardial infarction had a lower risk of CHD compared with myocardial infarction patients without diabetes. Recently, data from the REACH registry do not fully support the concept that diabetes is a cardiovascular risk equivalent, and the rate of pooled CVD mortality and nonfatal myocardial infarction in nondiabetic subjects with previous CAD was more than 50% higher than that seen in diabetic patients with risk factors only [ 28 ]. It is important to note that, the studies addressing these issues have great differences in age distribution, designs, and definitions of the study populations that could explain the discrepancies with our findings. Recently we found that diabetes, in particular the known cases, had higher risk more so in women than in men for CVD events [ 19 ]. In the present study, according to the diabetes status i.e. new vs. known cases, in women both the KDM+/CHD-and NDM+/CHD-groups had significant risk for incident CHD in comparison with the DM-/CHD-group (Risk factor adjusted HR ≈ 6 and ≈ 3 respectively, p < 0.001); however, in men only NDM+/CHD-, resulted in a significant risk, in comparison with DM-/CHD-(HR: 1.7, 95% CI: 1.07-2.69). The lack of association of KDM+/CHD-with incident CHD in men might be attributed to the limited statistical power. As acknowledged by Gonzalez-Clemente et al, it would probably have been better to directly compare persons who were DM+/CHD-with those who were DM-/CHD+ [ 29 ], hence comparing directly the hazard ratios of the these groups using the paired homogeneity test highlighted no difference between the KDM+/CHD-and DM-/CHD+ groups in risk factor adjusted analysis in prediction of incident CHD. However, an important finding of the current study was that regarding NDM in women, the risk of incident CHD in multivariate analysis for participants with DM -/CHD+ was marginally higher than those with NDM+/CHD-, although this did not reach a significant level. Consistent with our findings, a similar impact of diabetes in those receiving glucose-lowering agents (i.e. KDM) vs. those with prior CHD in prediction of incident CHD was demonstrated in the Multiple Risk Factor Intervention Trial (MRFIT) of men, 35 to 57 years old [ 30 ]. Similarly, Carnethon et al in a longitudinal study of men and women aged ≥ 65 years highlighted that, CHD mortality risk was similar between participants with CHD alone vs. diabetes alone [ 31 ]. Recently Andrersson el, in a multivariate regression analyses showed that patients with diabetes and absence of significant coronary artery disease at angiography have impaired systolic longitudinal left ventricular function and a global diastolic dysfunction, which is likely to be associated with adverse prognosis [ 32 ]. In the present study, diabetic subjects with prior CHD had the worst prognosis, by far more harmful in women than men (multivariate adjusted HR 7.19 vs. 3.58, respectively) similar to the reports of other studies [ 2 , 27 , 29 ] highlighting the importance of secondary prevention in patients with both coexisting disorders, especially in women. Our data concerning higher risk of CVD in women than men with DM +/CHD+ is comparable with female sex predominance in acute coronary syndrome patients with diabetes from a hospital based study from Iran [ 33 ]. Our study has several strengths. This is the first population-based study in Caucasians of Middle East region, conducted to determine the equivalency of diabetes and prior CHD for risk of CHD event. We included newly diagnosed DM in our study based on the OGTT result; furthermore, the ischemic change of ECG was added for defining prevalent CHD. However, most of the studies on this issue, included diagnosis of CHD and diabetes based on data provided by the patients themselves, which may have led to misclassification of subjects in the various groups. Finally, our study also adjusted for major CVD risk factors in the statistical model.

Conclusions The current study highlighted the finding that known diabetic patients in both genders and newly diagnosed diabetes especially in men, exhibited a CHD risk comparable to nondiabetics with a prior CHD, regardless of risk factors, furthermore diabetic subjects with prior CHD had the worst prognosis, by far more harmful in women than men; reinforcing thus the urgent need for intensive care and prophylactic treatment for cardiovascular diseases.

Background To investigate whether the known diabetes mellitus (KDM) or newly diagnosed diabetes mellitus (NDM) could be regarded as a coronary heart disease (CHD) risk equivalent among a relatively young Middle East population with high prevalence of diabetes mellitus (DM). Methods A population based cohort study of 2267 men and 2931 women, aged ≥ 30 years. Prior CHD was defined as self-reported or ECG positive CHD at baseline, KDM as subjects using any kind of glucose-lowering medications and NDM according to fasting plasma glucose and 2-h postchallenge glycemia. Participants were categorized to six groups according to the presence of known or newly diagnosed DM and CHD at baseline (DM-/CHD-, DM-/CHD+, NDM+/CHD-, NDM+/CHD+, KDM+/CHD-, KDM+/CHD+) and Cox regression analysis were used to estimate the hazard ratio (HR) of CHD events for these DM/CHD groups, given DM-/CHD-as the reference. Results During 7.6-year follow up, 358 CHD events occurred. After controlling traditional risk factors, HRs of CHD events for DM-/CHD+ group were 2.1 (95% CI: 1.4-3.1) and 5.2 (3.2-8.3) in men and women respectively. Corresponding HRs for NDM+/CHD-were 1.7 (1.1-2.7) and 3.1 (1.8-5.6) and for KDM+/CHD-were 1.7 (0.9-3.3) and 6.2 (3.6-10.6) in men and women respectively. The HRs for NDM+/CHD+ and KDM+/CHD+ groups (i.e. participants with history of both diabetes and CHD) were 6.4 (3.2-12.9) and 8.0 (4.3-14.8) in women and 3.2 (1.9-5.6) and 4.2 (2.2-7.8) in men, respectively. The hazard of CHD events did not differ between KDM+/CHD-and DM-/CHD+ in both genders using paired homogeneity test, however the HR for NDM+/CHD-was marginally lower than the HR for DM-/CHD+ in women ( p = 0.085). Conclusions KDM patients in both genders and NDM especially in men exhibited a CHD risk comparable to nondiabetics with a prior CHD, furthermore diabetic subjects with prior CHD had the worst prognosis, by far more harmful in women than men; reinforcing the urgent need for intensive care and prophylactic treatment for cardiovascular diseases.

Limitations of the study This study has important limitations that should be considered. First, we used positive history of CHD and the ECG-defined CHD as criteria to define prevalent CHD at baseline. The principal difference between these criteria might be caused by over reporting, non-Q-wave myocardial infarctions, or the interventions implemented to alleviate blockage of the coronary arteries [ 22 ]; unfortunately, at baseline the questionnaire of TLGS did not include questions about use of thrombolytic therapy, coronary artery bypass surgery, or percutaneous transluminal coronary angioplasty. Furthermore, population-based studies have found self-reported MI, CHD and stroke to be moderately or highly accurate in determining disease status [ 34 ]. Second, we defined newly diagnosed DM with a single OGTT at baseline as suggested for epidemiological studies; however, this might lead to attenuation of the association between newly diagnosed DM and cardiovascular events, resulting in an underestimation of the risk in this group. Third, the duration of diabetes could be a factor that explains the conflicting results in studies aiming to determine diabetes as a CHD risk equivalent, since longer duration of diabetes is associated with an increased risk of CVD [ 29 ]. Recently, Dagenais et al in the Quebec Cardiovascular Study emphasized finding that compared to men with incident CVD, men with incident diabetes had a lower risk for CVD mortality during the first 5 years after the diagnosis of diabetes, but subsequently there was no difference between the 2 groups for cardiovascular and total mortality [ 35 ]. Finally, our population was selected from middle-aged Middle East Caucasians and therefore we cannot make inferences beyond a similar group. Competing interests The authors declare that they have no competing interests. Authors' contributions FH (MD, associate professor of internal medicine and endocrinology) participated in the design of the study and drafting the manuscript. NF (MD, MPH) performed the statistical analysis and participated in the drafting the manuscript. DK (MD, MPH, PhD Candidate in Epidemiology) participated in the statistical analysis and revised the manuscript critically for important intellectual content. FS (MD, assistant professor of cardiology) and FA (MD, professor of internal medicine and endocrinology) participated in the design of the study and all authors have given final approval of the version to be published.

Acknowledgements This study was supported by grant No. 121 from the National Research Council of the I.R. Iran. We express appreciation to the participants of district 13, Tehran and outcome committee particularly Dr. A. Ghanbarian and Dr. F. Eskandari. The linguistic help given by Ms N. Shiva is much appreciated.

CC BY

no

2022-01-12 15:21:45

Cardiovasc Diabetol. 2010 Dec 4; 9:84

oa_package/b3/85/PMC3016329.tar.gz

PMC3016330

21143859

Introduction Low grade inflammation and activation of the innate immune system play a role in the common pathogenesis of both insulin resistance and endothelial dysfunction and subsequently the development of type 2 diabetes (T2D) and atherosclerosis. Several proinflammatory cytokines, acute phase-reactants and cell adhesion molecules seem to play a role in low grade inflammation, and today there is substantial evidence supporting the role of C-reactive protein (CRP), interleukin-6 (IL-6) and tumor necrosis factor-α (TNFα) among others as cardiovascular risk markers and participants in the pathogenesis of obesity, T2D and cardiovascular disease (CVD) [ 1 - 4 ]. CRP still remains the most validated biomarker in this context, although the substantial knowledge about CRP as a predictor of cardiovascular events is being supplemented by studies on emerging markers [ 5 ]. Matrix γ-carboxyglutamate (Gla) protein (MGP) is an extracellular matrix protein with wide tissue distribution, not only detectable in the normal blood vessels but also in calcified atherosclerotic plaques [ 6 ]. The main function of MGP is to inhibit medial calcification of arteries and thereby to protect the normal environment in the vascular wall [ 7 - 9 ]. Its main sources are cartilage and the vessel wall where it is synthesized by chondrocytes and vascular smooth muscle cells (VSMCs), respectively [ 8 , 10 , 11 ]. The importance of MGP to prevent calcification in soft tissues in vivo has been documented in the MGP knockout mouse model, where calcification of the elastic lamellae in the tunica media resulted in rupture of the large arteries within 8 weeks [ 8 ]. The function of MGP as an inhibitor of vascular calcification is exercised in the local tissues and no biological function of circulating MGP has been demonstrated [ 12 ]. However, serum MGP levels have been found to be significantly increased in patients with severe atherosclerosis and circulating MGP can be regarded as a marker of atherosclerosis reflecting the degree of inhibition of ongoing calcification processes in the vascular wall. This is consistent with the high MGP mRNA expression observed in atherosclerotic vessels and plaques in patients with type 1 diabetes [ 9 , 13 ]. YKL-40 is a 40 kDa heparin- and chitin-binding glycoprotein which is secreted in vitro by a variety of cells. In vivo YKL-40 is found in subpopulations of macrophages and VSMCs in different tissues with inflammation and extracellular matrix remodeling as in atherosclerotic plaques [ 14 , 15 ]. YKL-40 induces maturation of monocytes to macrophages and is secreted by macrophages during late stages of differentiation and by activated macrophages [ 14 , 15 ]. YKL-40 has also been shown to be an adhesion and migration factor for vascular cells and it is secernated by differentiated VSMCs [ 14 , 15 ]. The aim of the present study was to evaluate levels of markers of calcification (MGP) and inflammation (YKL-40, hsCRP) in patients with T2D and/or ischemic heart disease (IHD) and in healthy control subjects.

Materials and methods Study population The study population consisted of 1) patients with T2D (n = 45); 2) patients with IHD (n = 37); patients with both T2D and IHD (n = 20) and 4) healthy glucose tolerant controls (n = 20). Patients with T2D, but without IHD, were all without clinical or biochemical signs of diabetic complications. The patients with T2D were all on oral anti-hyperglycemic treatment, none received insulin. All healthy control subjects were Caucasian and had a sedentary lifestyle, and did not take any medication. Individuals with inflammatory conditions such as ongoing infectious disease/use of antibiotics, rheumatic or connective tissue disorders were excluded from the study. Furthermore, individuals with known chronic obstructive pulmonary disease, kidney disease or cancer were also excluded, because levels of YKL-40 are known to be increased in these conditions [ 14 ]. The study was approved by the local ethics committee (H-B-2007-058) and investigations conformed to the principles of The Helsinki Declaration. Clinical data All participants underwent a clinical examination including an ECG. A medical history was obtained and medications were recorded. The following baseline characteristics were also registered: known prior myocardial infarction (MI) and/or previous coronary revascularization procedures (percutaneous coronary intervention and coronary artery bypass grafting). MI was defined according to international guidelines http://www.escardio.org . IHD was defined as a previous episode with increased plasma coronary markers, ECG verified MI or previous coronary revascularization. All patients with IHD in the study had been referred to a myocardial perfusion imaging if they were considered to have an intermediate risk of having coronary artery disease (CAD) (symptoms of transient chest pain and/or worsening of chest pain when exercising and/or transient referred pain to the upper limbs or neck) or had a history of CAD with renewed suspicion of ischemia. All patients had an abnormal myocardial perfusion. All other patients in the study were without symptoms of cardiovascular disease and had a normal resting ECG. Measurements Analyses of the following markers of inflammations were performed: 1) Human MGP (mAb 3-15 ) was measured using an ELISA (Biomedica Medizinprodukte, Vienna, Austria). Standard range was 0-90 nmol/l, lower detection limit was 0.3 nmol/l and intra- and interassay coefficients of variation were 5.5% and 8.0%, respectively. 2) Plasma YKL-40 was measured using an ELISA method (Quidel, USA). Measuring range was 20-300 ng/ml, with intra- and interassay coefficients of variation of 5.8% and 6.0%, respectively. 3) CRP was measured with a highly sensitive, latex-particle-enhanced immunoturbidimetric assay (DAKO, Glostrup, Denmark) with a measuring range of 0.2-80 mg/l and a lower detection limit of 0.03 mg/l. Statistical analyses Following a test of statistical (log-) normality, data are presented as mean ± SD or as median and interquartile range (IQR). Comparisons between the groups were performed by a One-Way ANOVA. A non-parametric test was used if a variable exhibited a clear non-Gaussian distribution. The χ2-test was used for categorical variables. Analyses of associations were performed in the total study population using linear regression models with MGP, YKL-40 or hsCRP as dependent variable. Univariate analyses of correlations of either one of the biomarkers YKL-40, hsCRP and MGP were performed prior to multivariate analyses. Multivariate analyses including age, gender, patient category, cholesterol levels, blood pressure, and presence of IHD or diabetes were performed for biomarkers with significant outcome in the univariate analyses. All p-values were calculated as two-sided, and a p-value < 0.05 was considered significant. Analyses were made with the statistical software package SPSS15.0 (SPSS inc., Chicago, Il, USA).

Results Clinical data Clinical characteristics of the different patient categories of the study population are presented in Table 1 . There was an equal distribution between genders. Patients with T2D without IHD and the control subjects were younger that patients with IHD (p < 0.001). There was a significant difference in systolic but not in diastolic blood pressure between groups and more patients with both T2D and IHD had hypertension. Significantly lower levels of LDL-cholesterol and higher levels of HDL-cholesterol were found in patients with IHD when compared to patients without IHD (p < 0.0001). Patients with IHD were more frequently treated with ACE inhibitors, beta blockers (BB), statins and diuretics as compared to T2D patients without IHD (p < 0.001 for ACE, BB and statins; p = 0.002 for diuretics). Markers of calcification and inflammation MGP MGP levels were significantly higher in patients with both T2D and IHD, when compared to patients with either IHD or T2D (p < 0.001 for all comparisons)(Table 2 ). Moreover, MGP levels in non-diabetic patients with IHD were higher than in T2D patients without IHD (p = 0.007) (Figure 1a ). Univariate analyses revealed significant correlations between MGP and age (r = 0.43), HDL- and LDL-cholesterol levels (r = 0.59 and r = -0.47, respectively), LDL:HDL-cholesterol ratio (r = -0.62), statin (r = 0.55) and ACE treatment (r = 0.47) and patient category (r = 0.73) (all p ≤ 0.001) (Table 3 ). A similar pattern was found in subgroup analyses of patients with either IHD only or T2D only. No correlation was found between MGP levels and serum creatinine levels (r = -0.27, p = 0.45). In the total group of control subjects and IHD patients without T2D, MGP correlated positively with hsCRP (r = 0.50, p < 0.0001) and YKL-40 (r = 0.30; p = 0.03), whereas it correlated negatively to hsCRP in the total group of T2D patients without and with IHD (r = -0.28, p = 0.03). In multiple linear regression analyses MGP was associated with patient category (r = 0.36, p < 0.001), and HDL-cholesterol levels (r = 0.29, p < 0.001) adjusting for the significant covariates. Levels of MGP were significantly higher in patients treated with statins (2.32 ± 0.37 vs. 1.83 ± 0.41, p < 0.001) and ACE-inhibitors (2.36 ± 0.43 vs. 1.95 ± 0.43, p < 0.001) compared to non-treated. YKL-40 Levels of YKL-40 were significantly higher in patients with IHD and T2D (p < 0.05) (Figure 1b )). However, levels of YKL-40 did not differ between patients with both T2D and IHD and the control group (p = 0.1). No differences were seen in levels of YKL-40 between patient treated with statins (p = 0.13) and ACE-inhibitors (p = 0.20) (data not shown) when compared to non-treated patients. Only patient category (r = 0.22, p = 0.04) and LDL: HDL-cholesterol ratio (r = 0.31, p = 0.02) was significantly associated with levels of YKL-40 in a multiple regression model adjusting for the significant covariates. CRP Levels of hsCRP were significantly higher in any patient group when compared to the control group (p ≤ 0.001 for all comparisons) (Figure 1c ). Patients with T2D but no IHD had significantly higher levels of hsCRP compared to patients with IHD (p = 0.008), but not compared to patients with T2D and IHD (p = 0.16). No differences were seen in levels of hsCRP between patient treated with statins (p = 0.14) and ACE-inhibitors (p = 0.86) (data not shown) when compared to non-treated patients. In a multiple regression analysis hsCRP was only associated with BMI (r = 0.42, p < 0.001) after adjusting for the significant covariates. None of the biomarker levels differed between genders (p > 0.4).

Discussion The present study is to our knowledge the first to compare MGP expression in patients with either T2D or IHD or both. We describe increased serum MGP levels in patients with either T2D or IHD as well as in patients with both T2D and IHD, when compared to healthy control subjects. Significantly higher levels of MGP were found in patients with IHD when compared to patients without IHD, while no difference was seen when comparing patients with and without T2D. A stepwise increase was seen in serum MGP, with the highest levels in patients with both T2D and IHD. Based on our results it seems as if increased MGP not only is a marker of IHD characterized by intima calcification and subsequent atherosclerosis, but also of media sclerosis as typical seen in diabetes since an apparent additive concentration of MGP are documented in patients with both T2D and IHD. As in a previous study, we also found associations of MGP concentrations with levels of HDL- and LDL- cholesterol, consistent with the evidence that traditional lipid risk factors are significantly associated with circulating MGP [ 13 ]. Atherosclerosis and vascular calcification are extremely complex mechanism, with a long list of potential derangements on different pathways, and biomarkers have been extensively evaluated and become an important tool, helping to improve patient care [ 16 ]. Novel biomarkers of vascular inflammation, atherosclerosis and calcification and advanced glycation have been associated with both macrovascular late complications in high-risk T2D patients and with the severity of atherosclerosis in patients with carotid artery disease and lower limp artery disease [ 16 - 20 ]. In individuals without compromised kidney function as in the present study, levels of circulating MGP depend primarily on the synthesis and secretion and clearance of MGP from VSMC and subsequent binding of MGP to calcified areas within the vascular wall. We found significantly increased levels of serum MGP in patients with T2D and in patients with IHD, when compared to healthy control subjects, which is in agreement with previous studies demonstrating increased serum MGP levels in patients with severe atherosclerosis and in patients with type 1 diabetes [ 13 ]. We found an additive effect of T2D and IHD on MGP levels, which has never been demonstrated previously. This effect is probably due to a more advanced and general cardiovascular disease in patients with both T2D and IHD, when compared to patients with either T2D or IHD. Furthermore, MGP levels were significantly higher in non-diabetic patients with IHD compared to patients with T2D without clinical IHD, implying that the calcification process seems more active in conditions with clinical atherosclerosis, but obviously also active in conditions with increased risk of media and intima calcification as in T2D when compared to healthy subjects. It is also well known that the risk of myocardial infarction (MI) is similar in T2D patients without prior MI and in non-diabetic patients with prior MI, and that the risk is even higher in diabetic patients with previous MI [ 21 ]. Vascular calcification occurs at two anatomic sites, in the intima where it is associated with atherosclerosis and in the tunica media as Mönckeberg's sclerosis [ 22 ]. Media sclerosis is most commonly seen in patients with diabetes where it develops independent of atherosclerosis, implying different etiological mechanisms. Media calcifications are associated with VSMCs, whereas intima calcification in atherosclerosis occurs in macrophage and lipid rich atherosclerotic lesions. VSMCs grown in culture contain high levels of MGP, which is thought to limit the rate of calcification. MGP is a strong local inhibitor of vascular calcification, and although circulating MGP has no known biological function [ 12 ], it may reflect inhibition of calcification processes in the vascular wall. In healthy vessels MGP is synthesized at a low rate, probably because the need for calcification inhibition is low [ 9 , 23 ]. In vessels from patients with diabetes, MGP levels are lower than in normal vessels, which suggest that reduced MGP in diabetes may predispose to calcification [ 24 ]. However, levels of MGP has been demonstrated to be high in calcified vessels, where MGP most probably is expressed as an inhibitory counteraction of the calcification process [ 6 , 23 ]. Inflammation is probably an initial event very early on in plaque formation leading to calcification and IHD [ 25 ]. It has been demonstrated that statins, through their capacity to inhibit inflammation, can reduce the increased osteogenic and inflammatory activities seen with plaque progression, and this is correlated with the reduction in calcification [ 26 ]. In the present study we found that significantly more patients with IHD were treated with statins, so we cannot exclude that this has influenced the levels of MGP. However, since statin treatment reduces inflammation and can ameliorate the calcification response in the vessel wall, higher MGP levels would be expected had the patients been without statin treatment. Taken together, our and previous studies suggest that arterial calcification may lead to increased MGP expression, probably in a feedback attempt to reduce bone-like formation of calcium deposits in the vessel walls. However, it is still unclear whether serum MGP levels reflect activation, tissue production and deposition. Several studies have documented, that low-grade inflammation and activation of the innate immune system is involved in the common pathogenesis to atherosclerosis, endothelial dysfunction, insulin resistance and the development of T2D. We found significantly increased YKL-40 concentrations in patients with either T2D and/or IHD, thus confirming previous studies [ 27 - 30 ]. We also found a correlation between YKL-40 and patient category. As in a previous study of patients with T2D [ 27 , 28 ], no correlation was found between hsCRP and YKL-40 implying, that YKL-40 and CRP are produced and secernated independently of each other. Studies have shown, that increased levels of YKL-40 are independently associated with the presence and extent of CAD [ 29 - 32 ]. In patients with myocardial infarction even higher YKL-levels are documented [ 31 ]. YKL-40 has also been found to be associated with all-cause as well as cardiovascular mortality in both patients with stable IHD [ 31 ] and in the general population above 50 years of age without known diabetes or IHD [ 28 ]. In patients with type 1 diabetes, increasing levels of YKL-40 are seen with increasing levels of albuminuria, suggesting that YKL-40 might be able to be used as an early marker of CVD [ 33 ]. The present finding seems to support this hypothesis. It is well-known that serum CRP is increased in T2D patients [ 1 ]. We found a 7-fold higher concentration of serum hsCRP among T2D patients without IHD and a 3-fold higher in T2D patients with IHD, when compared to healthy control subjects. Furthermore, we found a strong correlation between serum hsCRP and BMI and patient category. It is also well known that CRP is a cardiovascular marker even within ranges considered normal [ 34 - 36 ]. In the present study we found a nearly 4-fold higher concentration of serum hsCRP in non-diabetic patients with IHD together with a strong correlation between serum hsCRP and BMI and HDL-cholesterol levels, in patients with and without IHD. Levels of hsCRP were significantly lower in non-diabetic patients with IHD compared to patients with T2D without IHD. Based on our and previous studies it can be speculated, that the low grade inflammation seen in both IHD and T2D declines in diabetes, when diabetic patients develop symptomatic atherosclerosis, and therefore make hsCRP as a biomarker of IHD in diabetic patients questionable. In the present study, more patients with IHD received statin treatment. Statins inhibit inflammation, and early treatment of inflammation seems to ameliorate the calcification response in the vessel wall [ 26 ]. We found increased levels of the inflammatory markers YKL-40 and hsCRP in patients with T2D and IHD, when compared to control subjects, however no differences were found in levels of these inflammatory markers between statin treated and non-treated patients. Statin treatment in especially IHD patients could also influence the levels of both YKL-40 and hsCRP, but without this treatment we would expect even higher levels of both markers in the IHD groups. In a recent study of patients with stable coronary artery disease (CAD), it was shown that YKL-40 levels were significantly reduced in statin treated patients compared to non-treated patients [ 37 ]. The median YKL-40 levels were 77% higher in the non-treated CAD group compared to the non-treated patients in our study, and median YKL-40 levels in statin treated CAD group were comparable to our statin treated patients, so we do not think that the participants in the two studies are comparable. We found that patients treated with statins had higher levels of MGP, and since we also found a positive correlation between HDL levels and MGP and a negative correlation between LDL levels and MGP, it could be speculated that higher MGP levels were an indication of better clinical and prognostic regulation of lipid levels due to statin treatment. Higher MGP levels could therefore be a positive indicator for the prognosis due to its function as an inhibitor of medial calcification of arteries. In a multiple regression analyses there were however no association between MGP levels and statin use or LDL-levels, adjusting for significant covariates, but we do not now if MGP can be used as a biomarker or positive indicator for prognosis of IHD. In conclusion, in patients with T2D and IHD we found increased levels of plasma MGP which may indicate progressing media and intima calcification processes. Moreover, these processes were paralleled by increased inflammation, which may reflect an initial event very early on in the processes leading to calcification

Objective and design Low grade inflammation is of pathogenic importance in atherosclerosis and in the development of cardiovascular disease (CVD) and type 2 diabetes (T2D). Matrix GLA protein (MGP), an inhibitor of medial calcification of arteries, is increased in patients with atherosclerosis. In the present study levels of markers of calcification (MGP) and inflammation (YKL-40, hsCRP) were evaluated in patients with T2 D and/or ischemic heart disease (IHD). Materials and methods The study population consisted of 1) patients with T2D (n = 45); 2) patients with IHD (n = 37); patients with both T2D and IHD (n = 20) and 4) healthy controls (n = 20). Biochemical parameters were measured in venous blood samples. Results Levels of MGP, YKL-40 and hsCRP were increased in patients with IHD and/or T2D (p < 0.0001) and patients with T2D and IHD had higher MGP levels (p < 0.001). In multiple linear regression analyses MGP was associated with patient category (r = 0.36, p < 0.001), and HDL-cholesterol levels (r = 0.29, p < 0.001) adjusting for the significant covariates. Conclusions In patients with T2D and/or IHD we found increased levels of plasma MGP indicative of a progressing calcification process. This process is paralleled by increased levels of YKL-40 and hsCRP, which most likely reflect the concomitant low grade inflammatory state in these patients

Conflicts of interest The authors declare that they have no financial or non-financial competing interests. Authors' contributions SBT participated in the design, coordination of the study, and interpretation of data and drafted the manuscript. CNR participated in analysis and interpretation of data, and helped to draft the manuscript. BZ participated in acquisition of data and conducts of the study and helped to draft the manuscript. HV conceived of the study and participated in the design, coordination, conduct of the study, performed the statistical analysis and helped to draft the manuscript. All authors have read and approved the final version of the manuscript.

Acknowledgements We wish to thank Ulla Kjaerulff-Hansen and Tonni Loeve Hansen, Endocrine Research Lab., Copenhagen University Hospital Herlev, Denmark, for skilful laboratory assistance. Funding The study was supported by grants from Aase and Ejnar Danielsens Foundation, Timber Merchant Vilhelm Bangs Foundation, The Hede-Nielsen Foundation and The Danish Council for Independent Research, Medical Sciences (FSS)

CC BY

no

2022-01-12 15:21:45

Cardiovasc Diabetol. 2010 Dec 8; 9:86

oa_package/0b/60/PMC3016330.tar.gz

PMC3016331

21245931

Introduction Myxococcus xanthus is a Gram-negative soil bacterium with a complex life cycle including diverse social behaviors such as “group hunting” and fruiting body formation [1] , [2] . A crucial feature of M. xanthus behavior is its ability to move in the direction of cell's long axis on solid surfaces by a flagella-independent mechanism called "gliding" [3] . Gliding motility is widespread in nature and has been shown to be essential in biofilm formation and the pathogenesis of motile species [4] , [5] , [6] , [7] , [8] . Genetic and phenotypic analyses have shown that M. xanthus gliding motility is regulated by the A (adventurous) and the S (social) motility systems [9] , [10] . The A motility system allows movement of isolated cells and does not require cell-cell contact, while the S motility system is typically employed for coordinated group movement of cells. The two systems appear to operate independently [11] but in a coordinated fashion [12] , and it has been suggested that each motility system provides different selective advantages to cells on surfaces containing different concentrations of agar, while enabling M. xanthus to adapt to a variety of physiological and ecological environments [13] . S motility in M. xanthus is mechanistically equivalent to twitching motility, the flagella-independent form of bacterial translocation over moist surfaces employed by Pseudomonas aeruginosa and Neisseria gonorrhoeae [7] . Mutations that abolish S motility (for a recent review, see [3] ) normally affect type IV pili (TFP) biogenesis [14] , the exopolysaccharide (EPS) component of the extracellular matrix (ECM) [15] , [16] , [17] or the lipopolysaccharide O-antigen [18] , [19] , [20] . Further functional studies have shown that S motility is powered by cycles of TFP extension and retraction [21] , [22] and depend on the interaction of TFP with EPS [23] . The highly organized and coordinated process of movement with cell-cell contact in a community of cells appears to be the normal state of affairs in S motility [9] , [23] , [24] , [25] , [26] as well as twitching motility [7] , [27] . However, it has been demonstrated that individual isolated cells can show TFP-dependent movement in M. xanthus [22] , P. aeruginosa [28] and N. gonorrhoeae [29] when in contact with certain substrates (for a review, see [7] ). M. xanthus strains deficient in A motility are capable of moving as single cells on a polystyrene surface in 1% methylcellulose [22] , even though on 1.5% agar gliding of individual cells requires A motility [9] , [10] . It is not fully understood how the methylcellulose-containing system compensates for the absence of the cell-cell contact normally necessary for S motility in M. xanthus [30] . M. xanthus cells have been observed to conduct single-cell S motility only on some surfaces with high water content [22] , while the biological relevance of this unique behavior has not been established. In this study, we conducted a detailed investigation of S motility of M. xanthus and a variety of mutant derivatives in methylcellulose-containing medium to analyze these apparent discrepancies, which led to further information about the roles of TFP and EPS in S motility.

Materials and Methods Bacterial growth conditions and construction of strains M. xanthus cells were grown in CYE medium [41] at 32°C on a rotary shaker at 300 rpm for 24 hr and harvested by 5,000×g centrifugation for 10 min. The M. xanthus strains used in this study are listed in Table 2 . Mx4-mediated generalized transduction [41] was used to construct strains SW2019 (A::Tn5- lac Ω1215, ΔdifA ) and SW2020 (A::Tn5- lac Ω1215, ΔdifA , ΔpilA ). SW2019 and SW2020 were constructed by transducing an Mx4 lysate of MXH1216 (A::Tn5- lac Ω1215) into SW504 ( ΔdifA ) and SW2017 ( ΔdifA , ΔpilA ), respectively, and selecting for kanamycin (Km) resistance. An in-frame deletion plasmid of aglZ was generated using MXH2265 ( ΔaglZ ) chromosomal DNA as the PCR template with the primers AGLZ-up ( CCGGAATTCAACCGCCCGATAGGATGTTC ) and AGLZ-down ( CGCGGATCCTGGCGTACTCCCACTCGTAGAG ). The amplified regions were digested with EcoRI and BamHI and ligated into similarly digested pBJ113 [42] to create the plasmids pBJ113-aglZ. After being verified by sequencing, the deletion plasmid pBJ113-aglZ was electroporated into SW504 ( ΔdifA ), DK10410 ( ΔpilA ) and SW2017 ( ΔdifA, ΔpilA ) to construct SW2021 ( ΔaglZ , ΔdifA ), SW2022 ( ΔaglZ , ΔpilA ) and SW2023 ( ΔaglZ , ΔdifA, ΔpilA ), respectively. Deletion mutants were selected by their galactose resistant and kanamycin sensitive phenotypes and confirmed with PCR and sequence analyses. 0.3% agar assay for S motility Swarm plates (MOPS medium solidified with 0.3% Bacto agar) were inoculated and S motility on agar surfaces was recorded as described previously [13] . Plates were incubated at 32°C for 1 hr before imaging. The movements of individual cells in groups were tracked employing the tetrazolium staining assay [43] . Methylcellulose assay for S motility This assay was based on a previously published protocol [22] with the following modifications. The harvested M. xanthus cells were diluted in MOPS buffer (10 mM MOPS, 8 mM MgSO 4 , pH 7.6) to 5×10 6 cell/ml. One microliter of cells was transferred onto different surfaces (described below) and allowed to settle for 10 min. The cells were then carefully overlaid with 200 μl of 1% methylcellulose in MOPS buffer and placed at 32°C for 1 hr before recording. For the methylcellulose assay with insoluble polysaccharides, 5 μl of a 1 mg/ml chitosan suspension (molecular weights from 100K to 300K, Acros Organics, USA) or cellulose suspension (Acros Organics) was added to the cells prior to the methylcellulose overlay. The motility of M. xanthus cells in 1% methylcellulose was tested on different surfaces. Polystyrene plates (CostarTM cell culture plates, Fisher Scientific, USA) were used as a polystyrene surface material (called Ps surface in this paper). The SuperBlock pre-treated polystyrene surface (called blocked-Ps surface in this paper) was prepared by coating 24-well polystyrene plates with 600 μl SuperBlockTM T20 (PBS) blocking buffer (Thermo Scientific, USA) three times according to the manufacturer's instructions. Microscopic imaging and analysis of motility in methylcellulose Cell movements were monitored with a Nikon Eclipse TE200 inverted microscope through a 60× objective, captured with a SPOT RT740 CCD camera (Diagnostic Instrument, USA) and recorded with SPOT advanced software (Version 4.6, Diagnostic Instrument). A heat plate system (Brook Instrument Corp, USA) was mounted onto the stage of the microscope to maintain the culture temperature at 32°C. Continuous images were taken at 10 s intervals and stored as TIFF image sequence files. For all the Supporting Information Videos, Microsoft videos (5 frames/s, wmv file) were exported with SPOT software resulting in a 50× faster than real-time replay speed. For velocity measurements and trajectory tracking, the TIFF image sequences were imported to Manual Tracking [44] , a plug-in for ImageJ software [45] . For each condition, 20 different cells moving along their long axes were tracked at their leading poles for 2 min (13 frames). Velocity was calculated with the Manual Tracking plug-in by measuring the distance from a starting to an ending point within two frames. A static synthetic view of cell motility tracks was generated and the recorded xy coordinates were exported to Microsoft Excel software to present the data as plots. One color was applied for each trajectory. To calculate the percentage of tethered cells (with one end of the cell attached to the solid surface such that the cell body was lifted up), 20 random frames were selected and the tethered cells were manually counted as previously described [36] . Motile percentages were calculated by estimating the number of motile versus total non-tethered cells as previously described [46] . Cells were defined as isolated single cells if the individual cells were separated from other cells by a distance greater than approximately one cell length [24] . TFP retraction assay The effects of different complex carbohydrates on TFP retraction were evaluated with a previously described mixing assay [23] . The purified EPS was isolated from M. xanthus DK1622 and quantified as previously described [23] , [47] . To test the rescue of hyperpiliated EPS − mutants by different complex carbohydrates, about 5×10 8 EPS − SW504 cells were mixed with purified EPS (1 mg/ml carbohydrate), insoluble chitosan (1 mg/ml, 100-300K), methylcellulose (1 mg/ml, Fisher) or granular agar (1 mg/ml, Fisher) prior to incubation in cohesion buffer at 32°C for 30 min [23] . Cell-surface pili were then analyzed with Western blotting using anti-pilA antibody [36] as previously described [48] . Generation and purification of anti-EPS antiserum Purified EPS from DK1622 cells [23] , [47] was used to challenge two rabbits to raise polyclonal anti-EPS antibodies. Immunizations were performed by GenScript (USA) based on established protocols [49] . The antiserum was preabsorbed with acetone powder prepared from the EPS deficient mutant strain SW504 ( ΔdifA ) as described by Harlow & Lane [49] . Immuno dot blots Dot blots of whole-cell lysates were performed according to standard protocols [49] . Cell pellets of DK1622, SW504 and SW810 were collected and suspended in 1% SDS solution at 5×10 9 cell/ml. Cells were then lysed by boiling for 10 min. Ten μl of each cell lysate was applied to a single spot on a nitrocellulose membrane and probed with the anti-EPS antiserum (1∶200 diluted). ELISA for Pilin and EPS adsorbed on different surfaces Purified truncated pilin (PilA (29–220) ) protein and anti-PilA antibody were prepared as previously described [36] . For ELISA, truncated PilA protein was dissolved in deionized water to a concentration of 10 μg/ml. Then 50 μl of PilA solution was added to the wells of CostarTM polystyrene 96-well plates w/wo SuperBlock pre-treatment, and 50 μl water was used as control. The plates were incubated at 37°C for 1 hr. Next, the PilA protein adsorbed on the surface of the well was detected with a standard ELISA [49] . The anti-PilA antibody used in this assay was 2000-fold diluted and peroxidase-conjugated goat anti-rabbit IgG was 5000-fold diluted. The optical density at 450 nm was determined and data was calculated as the average of five repeats. To detect the EPS shed from M. xanthus cells onto different surfaces, 50 μl DK1622 or SW504 cells at 1×10 8 cell/ml in MOPS buffer was added to the wells of CostarTM polystyrene 96-well plates w/wo SuperBlock pre-treatment while 50 μl of MOPS buffer and 50 μl purified EPS (10 μg/ml carbohydrate in MOPS buffer) were used as controls. ELISA for EPS was performed as described above, except that the anti-EPS antiserum was 1000-fold diluted. EPS staining and confocal microscopy MOPS agar layers in cover slide bottom chambers were prepared following previously described methods [50] . M. xanthus DK10547, a gfp -expressing derivative of DK1622 [31] , was spotted on the agar surface and kept in a humidity chamber for a 12 hr incubation periods. Carbohydrates present in the EPS portion of the cell swarm were stained with 10 μg/ml of Alexa633-conjugated derivatives of wheat germ agglutinin lectin (WGA, Molecular Probes, USA) in MOPS buffer [50] . The specimens were viewed using a PASCAL 5 confocal laser scanning microscope (Zeiss, Jena, Germany) after a 30 min incubation period at 32°C in the dark. The scanning module of the system was mounted onto an inverted microscope (Axiovert 200 M). Excitation at 488 nm with an argon laser in combination with a 505–530 nm bandpass emission filter were used for imaging of GFP-expressing cells. 633 nm excitation with a helium-neon laser and a 650 nm longpass emission filter were used to reveal Alexa633-conjugated lectin. Images were acquired through a 63x/NA1.4 oil lens.

Results M. xanthus cells lacking EPS displayed TFP-dependent motility on polystyrene surfaces in methylcellulose In M. xanthus, A motility is advantageous on relatively firm and dry surfaces (e.g. 1.5% agar), while S motility is dominant on wet surfaces (e.g. on 0.3% agar or in 1% methylcellulose) [13] , [22] . M. xanthus and different mutants derivatives lacking distinct motility features were systematically compared regarding their motility in methylcellulose relative to their S motility on 0.3% agar surface ( Table 1 ). Consistent with previous findings [13] , A − S + cells, like strain MXH1216 (A:: Tn5 ) and MXH2265 ( ΔaglZ ), showed S motility mainly in cell groups, while A − S − cells (including EPS − and pilus − cells) were not motile by S motility on 0.3% agar. In contrast, on polystyrene surfaces in 1% methylcellulose, S motility was frequently detected in both isolated and grouped S + M. xanthus cells, which was readily distinguished from the A motility of wild-type cells by their relative differences in velocities [22] . Interestingly, the EPS − cells, including strain SW2019 (A::Tn5, ΔdifA ) and SW2021 ( ΔaglZ , ΔdifA ), also showed a rapid motility phenotype in methylcellulose, while cells lacking TFP, like strains SW538 (A::Tn5, ΔpilA ) and SW2022 ( ΔaglZ , ΔpilA ), did not. To further study this phenomenon, different cells carrying mutations affecting A or S motility were placed on polystyrene surfaces in 1% methylcellulose-containing medium. The velocities of EPS − cells, including strains SW504 ( ΔdifA , Video S1 ) and SW810 ( ΔepsA ), were generally higher than the speed of wild-type cells ( Fig. 1B ). The movements of EPS − cells in this system were not due to A motility, since strains SW2019 (A::Tn5, ΔdifA ) and SW2021 ( ΔaglZ , ΔdifA ), derivatives of SW504 ( ΔdifA ) lacking A motility exhibited motility similar to their ΔdifA parent ( Fig. 1 ). These rapid movements were characteristic of TPF-dependent motility. In contrast, in the respective mutant strains defective in surface pilus biogenesis, ie. DK10410 ( ΔpilA ), SW2022 ( ΔaglZ , ΔpilA ) and SW2023 ( ΔaglZ , ΔdifA , ΔpilA ), rapid motility was totally eliminated ( Fig. 1A and B ). Although EPS has been identified as the trigger for TFP retraction [23] and is considered a key component for S motility on agar [15] , [16] , [17] , M. xanthus S motility on polystyrene surfaces in methylcellulose appeared to be independent of EPS. S motility of EPS − M. xanthus cells in methylcellulose was dependent on the interaction of TFP with the polystyrene surface The retraction of TFP in EPS deficient mutants, such as strain SW504 ( ΔdifA ), is stimulated by mixing the cells with isolated EPS, which suggests the specific recognition of M. xanthus EPS by its TFP [23] . Using the retraction assay [23] , the interactions of TFP with methylcellulose and agar were investigated. Methylcellulose was unable to trigger pilus retraction in EPS − SW504 cells (lane 2, Fig. 2A ), while insoluble chitosan (partially deacetylated chitin) and purified EPS were able to do so (lane 3 and 4, Fig. 2A ), which excluded the possibility that methylcellulose replaced the role of EPS in M. xanthus S motility. Consistent with the S motility deficient phenotype of SW504 on agar, granular agar was also unable trigger the TFP retraction of SW504 cells (Lane 5, Fig. 2A ). Next, an ELISA assay was employed to evaluate the possible interaction between pilin and polystyrene surface. The purified PilA protein was shown to adhere onto the polystyrene surface and the protein's binding was remarkably reduced after coating the surface with SuperBlock buffer ( Fig. 2B , grey columns). Consistent with the observed lack of PilA binding, the S motility of SW504 cells on the blocked surface was greatly inhibited ( Fig. 1 , Video S2 ) and their tethering abilities, motile percentages and velocities decreased significantly ( Fig. 1 and Fig. 2B ). These results suggested that the interaction between TFP and the polystyrene surface enable EPS − M. xanthus cells to move by S motility in methylcellulose. To dispel the concern that the SuperBlock treatment of the surface might interfere with TFP retraction, a chitosan suspension, which is known to be able to trigger the retraction of M. xanthus TFP ( Fig. 2A and [23] ), was mixed with SW504 ( ΔdifA ) cells that were placed on the surfaces before adding the methylcellulose overlay. As shown in the serial images in Fig. 3 , on both unblocked and blocked surfaces, SW504 cells located within a certain distance to chitosan granules (corresponding to the length of TFP) performed jerky motions towards chitosan granules ( Video S3 and S4 ), which were most likely caused by the retraction of TFP. At the same time, similar motions were not observed when mixing SW504 cells with cellulose granules. These controls confirmed that the blocking treatment prevented binding of TFP on polystyrene surfaces but did not interfere with the general ability of TFP to retract. S motility of EPS + M. xanthus cells on blocked polystyrene surfaces in methylcellulose Further experiments were conducted to investigate the role of the interaction between TFP and the polystyrene surface in S motility of EPS + cells. Such interactions were likely due to nonspecific binding compared to the recognition of M. xanthus EPS by TPF [23] . The blocking of the polystyrene surfaces had complicated influences on S motility of MXH2265 ( ΔaglZ , A − EPS + ) cells. On the blocked polystyrene surface at 1 hr, the S motility of isolated MXH2265 cells was inhibited and the motile percentage of cells was reduced to 33.2±8.3% from 81.5±5.2% on the unblocked polystyrene surface. On blocked surfaces, the motile percentage of MXH2265 ( ΔaglZ , A − EPS + ) isolated cells increased to 56.7±12.1% after 6 hrs incubation ( Video S5 ), whereas the motile percentage of SW2021 ( ΔaglZ , ΔdifA, A − EPS − ) cells at 6 hr (13.5±5.3%) remained similar to the ratio observed at 1 hr (9.6±4.7%). By tracking the motility trajectories ( Fig. 4 , Video S6 and S7 ), on the unblocked surface, MXH2265 cells were observed to move and separate themselves from the cell groups (like the light blue trajectory shown in Fig. 4A ). On the blocked surface, motility of isolated MXH2265 cells was inhibited at 1 hr, while the cells in groups remained motile ( Fig. 4B ). These results suggested that S motility of the isolated EPS + cells was partially dependent on the nonspecific interaction of TFP with polystyrene surfaces in methylcellulose. EPS + M. xanthus cells deposit EPS on both blocked and unblocked surfaces to enable surface-independent movement We noticed that about 30% of isolated EPS + cells were still motile on the blocked surface, while only about 9% of EPS − cells were able to move under the same condition ( Fig. 2B ). One possible explanation for this difference is that EPS + cells would eventually deposit sufficient EPS on the blocked surface to support single-cell S motility. In order to detect the EPS attached on the surface, a polyclonal anti-EPS antiserum was prepared using purified wild-type EPS. Dot blots showed that this antibody was able to recognize the EPS of wild-type cells, and had a low background for EPS − cells ( Fig. 5A ). Consistent with our hypothesis, purified EPS was shown to bind to the polystyrene surface even after treatment with blocking buffer ( Fig. 5B ). Considerable amounts of EPS were detected on both blocked and unblocked surfaces, after washing off the wild-type cells, while neither surface retained any detectable amounts of EPS after exposure to SW504 ( ΔdifA ) cells ( Fig. 5B ). Since the ELISA used to detect EPS on polystyrene surfaces could not be applied to agar surfaces, confocal laser scanning microscope (CLSM) with specific lectin staining was employed to reveal the EPS shed on agar surfaces. The results showed that DK10547 cells ( Fig. 5C -green), a gfp -expressing derivative of DK1622 [31] , followed the EPS trails ( Fig. 5C -pink) and aggregated on spots with high EPS concentration, which also suggested the presence of an agar surface covered by EPS under M. xanthus swarms. S motility of EPS − M. xanthus cells in groups is not coordinated on polystyrene surface in methylcellulose The S motility of EPS + cells in groups is most likely dependent on EPS regardless of the surfaces involved and EPS has been suggested to be important for coordinated movements in M. xanthus [32] , [33] . Our results showed that the S motility of EPS − M. xanthus cells in groups was not coordinated on polystyrene surface in methylcellulose ( Fig. 6 ). With the EPS and cell-cell contact, EPS + cells in groups formed initial multi-cellular aggregates after being incubated in methylcellulose at 32°C for 6 hr, and the cellular movement was apparently coordinated ( Fig. 6A ). However, the EPS − cells only exhibited random movements in groups and did not form aggregates even after 6 hr incubation in methylcellulose ( Fig. 6B ). By examining the cellular distribution in the spreading zones, some specific swarming structures, like peninsulas and trails ( Fig. 6C ), were indentified in EPS + cells, and such structures are usually observed in the swarms resulting from S motility of M. xanthus on agar [16] , [24] . However, the spreading zones of EPS − cells were most composed of randomly distributed single cells ( Fig. 6D ). These results confirmed that, in methylcellulose, EPS was likely a key component in coordinating movements in groups of cells.

Discussion The force-generating mechanism involved in M. xanthus S motility was shown to involve the extension and retraction of pili [21] , [22] , [34] , which is triggered by the specific interaction between TFP and EPS [23] . The requirement for EPS in M. xanthus S motility on agar surfaces has been previously demonstrated [15] , [16] , [17] . However, in the present study we found that in methylcellulose, nonspecific interactions of TFP with the polystyrene surface compensated for the absence of EPS as an anchoring substratum and allowed single-cell TFP-dependent motillity in M. xanthus . Consistent with our findings, the TFP of different bacteria can bind to a variety of surfaces, including inert surfaces, bacterial cells, and eukaryotic cells, and can mediate both colonization of the surfaces and intimate contact through pili retraction [7] , [35] . Similar to S motility in M. xanthus , the twitching motility in P. aeruginosa and N. gonorrhoeae is primarily a social activity. However, it has also been shown that isolated cells move by twitching motility when in contact with different surfaces [28] , [29] . The nonspecific interaction of TFP with the surface might also play an important role in the aforementioned twitching motility of isolated cells. We also observed that the presence of a general anchoring surface in a viscous environment could not substitute for the role of cell surface EPS in cell group movements of M. xanthus . Coordinated movements are required for the swarming and aggregation of A − S + cell groups on agar surfaces [24] . On polystyrene surfaces in methylcellulose, the TFP-dependent motility of EPS − cells was not coordinated and cells failed to aggregate or perform group swarming. For EPS + cells, TFP specifically recognized and bound onto the EPS of an adjacent cell or in the slime trail [23] , which coordinated cellular group formation into peninsulas, trails and swarm aggregates. The asymmetry of EPS distribution might regulate the movements in M. xanthus swarming, while the uniformity of nonspecific anchoring surface distribution might not do so. Our study provided the direct evidence for a definitive role for EPS in coordinating S motility in cellular groups, which is another important biological function of EPS in M. xanthus . Our results for S motility of M. xanthus isolated cells in methylcellulose medium are summarized in Fig. 7 . A − EPS − M. xanthus cells were deficient in S motility on agar because of the absence of EPS to bind with TFP and trigger retraction ( Fig. 7A , adapted from [23] ). Inert surfaces, like polystyrene and glass, are normally employed in the methylcellulose motility assays [22] , [30] . On these surfaces, the A − EPS − cells display uncoordinated S motility. It has been observed in previous studies that polystyrene surfaces allowed tethering of TFP [22] and the BSA-coated polystyrene surface reduced the amount of tethering by M. xanthus wild-type cells [36] . Consistent with these results, our findings suggest that polystyrene provided an anchoring surface for S motility of EPS − cells in methylcellulose ( Fig. 7B ). Results similar to those obtained with the cells on the polystyrene surface were obtained with cells on a glass surface (data not shown), which indicated that TFP also interacted nonspecifically with glass. S motility of isolated M. xanthus EPS + cells has been observed on polystyrene surface in methylcellulose [22] but rarely on 1.5% agar [9] . The need for adjacent EPS might explain why M. xanthus cells normally move by S motility in groups ( Fig. 7C , adapted from [23] ). In our model ( Fig. 7D ), single-EPS + cell S motility in methylcellulose system can be divided into two categories: one involves activation by the nonspecific interaction of TFP with the polystyrene surface and the other is triggered by the binding of TFP to the EPS deposited on the surfaces. Motile individual cells initially attached to virgin surfaces could modify the surface by shedding their EPS during movement, and subsequently provide a specific anchor for other isolated cells to display S motility. Consistent with this assumption, it was observed that inhibition of the S motility of isolated EPS + cells could be overcome following incubation, which possibly may result from the development of a blocked surface due to the shedding of EPS from the cells. In addition, single-cell S motility of M. xanthus was only observed under certain conditions invovling high water content (e.g. 1% methylcellulose solution), which might have a bio-ecological relevance for M. xanthus . Cellular motility provides bacteria with the capacity to actively seek out favorable environments and avoid hazardous situations, thereby facilitating growth and survival in nature [3] . Myxococci have been commonly found in terrestrial habitats [37] , [38] and they are also suggested to be adapted to environments with periodic dry spells as well as living permanently in fresh water habitats [39] . At the same time, some Myxococcus strains have been isolated from aqueous samples [40] , which indicates that these bacteria might be even more widely spread. In the environments with high water content, M. xanthus cells mainly depend on their S motility to travel and colonize [13] . Thus, the EPS independent single-cell S motility described in this report could possibly promote rapid exploratory movement of colonies over newly hydrated surfaces. However, the role of cell floating in this S motility is still unknown. In the methylcellulose motility assay, the 1% methylcellulose most likely provides a highly viscous medium rather than replacing the function of EPS. In methylcellulose, the Brownian motion of the cells is greatly reduced and gliding motility could be readily examined [22] . The buoyancy provided by the liquid medium could reduce the cell body weight acting on the surface and subsequently the resistance to motility caused by friction between a cell body and the surface. Under these conditions, S motile cells might not need to bind as tightly to the EPS and the nonspecific binding may be sufficient to activate S motility.

Conceived and designed the experiments: WH RL WS. Performed the experiments: WH MH RL. Analyzed the data: WH MH. Contributed reagents/materials/analysis tools: JW ZY YL. Wrote the paper: WH MH RL WS. Social motility (S motility), the coordinated movement of large cell groups on agar surfaces, of Myxococcus xanthus requires type IV pili (TFP) and exopolysaccharides (EPS). Previous models proposed that this behavior, which only occurred within cell groups, requires cycles of TFP extension and retraction triggered by the close interaction of TFP with EPS. However, the curious observation that M. xanthus can perform TFP-dependent motility at a single-cell level when placed onto polystyrene surfaces in a highly viscous medium containing 1% methylcellulose indicated that “S motility” is not limited to group movements. In an apparent further challenge of the previous findings for S motility, mutants defective in EPS production were found to perform TFP-dependent motility on polystyrene surface in methylcellulose-containing medium. By exploring the interactions between pilin and surface materials, we found that the binding of TFP onto polystyrene surfaces eliminated the requirement for EPS in EPS - cells and thus enabled TFP-dependent motility on a single cell level. However, the presence of a general anchoring surface in a viscous environment could not substitute for the role of cell surface EPS in group movement. Furthermore, EPS was found to serve as a self-produced anchoring substrate that can be shed onto surfaces to enable cells to conduct TFP-dependent motility regardless of surface properties. These results suggested that in certain environments, such as in methylcellulose solution, the cells could bypass the need for EPS to anchor their TPF and conduct single-cell S motility to promote exploratory movement of colonies over new specific surfaces.

Supporting Information

We thank Drs. Patricia Hartzell, Mitch Singer, Martin Dworkin and Dale Kaiser for providing bacterial strains, Sissi Chen for technical help in material preparation, Dr. Xuesong He for helpful discussion, and Dr. Howard Kuramitsu for careful review of the manuscript.

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e16102

oa_package/2b/04/PMC3016331.tar.gz

PMC3016332

21245932

Introduction The ocean and the land are ‘sinks’ for the excess atmospheric CO 2 produced by burning of fossil fuels, deforestation and land use changes. Twenty five to thirty percent of the total anthropogenic CO 2 emissions produced since the industrial revolution have been absorbed by the oceans [1] . This excess CO 2 dissolves in the surface ocean, causing increased hydrogen ion (H + ) concentrations and decreased carbonate ion (CO 3 2− ) concentrations in seawater [2] . The result is a decline in ocean pH and calcium carbonate (CaCO 3 ) saturation states. Through this process, known as ‘ocean acidification’, pH has already dropped by 0.1 pH units (an increase in H+ concentration of almost 30%) since the 1800s, and the rate of change is predicted to increase considerably, with a further decline of 0.2–0.5 pH units expected by 2100 [3] , [4] . While the implication of ocean acidification to the chemistry of the open ocean is reasonably well understood, the ecological implications for marine fauna and flora are harder to predict [5] , [6] . The taxonomic groups considered most susceptible to ocean acidification are calcifying organisms with CaCO 3 skeletons and shells, such as corals, coralline algae, coccolithophores and molluscs. This is due to the predicted reduced availability of the CO 3 2− ions they require for precipitation of CaCO 3 [5] , [7] . The degree of susceptibility of these calcifiers to ocean acidification is thought to be influenced by their mineralogy (but see [8] , [9] ). For example, skeletons made of the CaCO 3 polymorph aragonite are generally more soluble than those comprised of calcite, but calcite skeletons containing high proportions of magnesium may be even more soluble than aragonite [10] . Organisms without CaCO 3 skeletons or shells have also been shown to be affected through disruption of their acid-base balance and respiration (e.g., [11] – [13] ). Experiments to date have shown that species’ responses are variable, due in part to differences in functional responses and environmental conditions (e.g., [14] , [15] ). Importantly too, not all species, nor all life stages of a particular species, have responded negatively to pCO 2 conditions predicted for the future (e.g., [16] , [17] ). Especially fast rates of change are expected in the Southern Ocean due to its cold water temperatures (and thus higher solubility of CaCO 3 , [18] ) and CO 3 2− saturation levels that already are lower than temperate regions. As early as 2030 during winter months, the Southern Ocean is predicted to become undersaturated in aragonite [19] , [20] ; a situation that has probably not occurred in at least the last 400,000 years [4] . Studies of potential effects of future levels of pCO 2 on high latitude calcifying organisms are at present limited. Those conducted to date have shown reduced calcification and higher dissolution rates of shells of live pteropods [4] , [21] and dissolution of bivalve, gastropod and brachiopod shells [9] , and have documented a 30% decline in shell weights of foraminifera since the late 1800s [22] . In contrast, fertilisation, early embryogenesis and larval development of high Antarctic sea urchins were unaffected except at pH below 7.3 (i.e., pCO 2 predicted for 2300; [23] , [24] ), and similar results were noted for fertilisation and early development of nemerteans [24] . Laternula elliptica is a large, infaunal, suspension-feeding bivalve with a circum-Antarctic distribution. It has a wide depth range (1–460 m), but is particularly common in shallow waters where it is routinely found at densities of 10's of individuals m −2 [25] and often much higher (>170 ind. m −2 ; [26] , authors' unpubl. obs). In McMurdo Sound, Ross Sea, L. elliptica occurs in a variety of habitat types, ranging from homogeneous soft sediments to coarse pebbly sand deposits between rocks (authors' unpubl. obs). L. elliptica is a key species in Antarctic coastal benthic ecosystems through its influence on benthic-pelagic coupling [25] and has a shell comprised purely of aragonite [27] . In a laboratory study, [9] demonstrated significant dissolution of adult L. elliptica shells after 28 days exposure to pH 7.4. However, as their study was conducted on empty valves only, the biological response by the L. elliptica and their potential to compensate for this dissolution remains unknown. Here we describe a laboratory experiment designed to assess the effect of a change in ocean pH on the functioning of adult Laternula elliptica from the Ross Sea, Antarctica. We compared L. elliptica ' s biological response at current Antarctic pH (7.99) to that predicted for the high Antarctic in the following decades (7.78; [4] , [19] ),and to a considerably elevated, ‘glacial’ pH (8.32). A suite of response variables, chosen to incorporate a range of molecular level (gene function, of stress and growth proteins) and whole organism responses (respiration and physiological condition) important in the functioning of this bivalve, were assessed. Specifically, we monitored effects on L. elliptica (i) shell, via measurement of chitin synthase ( CHS ) gene expression, a key enzyme involved in the shell formation process [28] , and changes in shell weight; (ii) metabolism and growth, via measurement of oxygen consumption [29] , physiological condition [30] and protein synthesis (i.e., total RNA, [31] ); and (iii) stress, via measurement of heat shock proteins ( HSP , specifically HSP70 gene expression, [32] ). By examining this range of interlinked variables we were able to build a more comprehensive picture of the likely effect of ocean acidification on the functioning of this key Antarctic bivalve.

Methods Laternula elliptica were collected from 20 m depth at Granite Harbour, Ross Sea, Antarctica by SCUBA divers on 15 November 2008. Their habitat consisted of loosely compacted coarse, sandy sediments, interspersed with pebbles and cobbles, and with very low CaCO 3 content (<1%). Live L. elliptica (average 72.8 mm shell length (SL), range 54.2–87.9 mm) were transported to Wellington, New Zealand on 20 November 2008, where they were immediately transferred to a laboratory facility and held in running seawater. Seawater was sourced from the adjacent Wellington Harbour (WH) and was used in a single pass flow through system, filtered on a 0.1 μm filter and chilled to the experimental temperature of −1.76°C (± 0.0008 SE; range −1.93 to −1.60°C). WH seawater was of similar salinity to that at the Granite Harbour collection site (34.1 and 34.6, respectively). pH/pCO 2 manipulation Three pH treatment levels were chosen for the experiment. These included an ‘Antarctic control’ (pH 7.99), a lowered pH (pH 7.78) which represents that predicted for 2100 by [4] , and an elevated pH (pH 8.32) which last occurred in the Antarctic 20,000 years ago [33] . The Antarctic control pH treatment level was chosen based on measurements of water samples collected at the study site. The calculated pCO 2 equivalents for the elevated, Antarctic control, and lowered pH treatments are 187, 430, and 735 μatm, respectively (see Table 1 ). The pH of the chilled seawater was manipulated by bubbling food-grade 100% CO 2 gas into a 60 litre tank of WH seawater to reduce the pH to 7.6 pH units. This low pH water was then mixed with WH seawater in additional 60 litre ‘header’ tanks to produce the Antarctic control and lowered pH treatments, respectively. The quantity of low pH water pumped to the header tanks was controlled via a Proportional Integrative Derivative feedback loop from industrial pH controllers (ATI Q 45P, Analytical Technology Inc. USA). The controllers measured the header tank pH through high quality glass pH electrodes (Hamilton, Switzerland) (Liq-Glass rated to -10°C to +100°C) and were temperature compensated using PT100 resistance thermometers (Servotech, New Zealand). The pH electrodes were calibrated regularly. The more accurate pH measurements derived from spectrophotometric readings (described below) were used to correct for any drift in the pH control system between calibrations. The elevated pH treatment was generated by chilling WH seawater to experimental temperature. Water from the header tanks was gravity-fed to replicate tanks (each 20 l) at a rate of 140 ml min −1 . There were six replicates of each of the three experimental treatments. Experimental setup Laternula elliptica were individually numbered using permanent marker pen, their SL measured using electronic callipers, and 8 individuals were added to each replicate tank. Each L. elliptica was placed into tanks in a small mesh basket, which ensured they were held in a lifelike orientation (siphons uppermost). The tanks did not contain sediment. Seawater pH in the replicate tanks was initially set at chilled WH levels (8.32). The pH was then gradually lowered to the target levels over the following 24 h, reaching these on 27 November 2008 (Day 0). L. elliptica were fed twice per week with a commercial algae mix (Shellfish diet, Reed Mariculture, US). Chemical characteristics of seawater Water samples for alkalinity (A T ) were collected from the header tanks throughout the experiment, on December 4 th and 18 th 2008, and March 25 th 2009 (n = 13, 14 and 13 for pH 8.32, 7.99 and 7.78, respectively), and preserved using HgCl 2 . A T was determined using a closed cell potentiometric titration method [34] . The accuracy of this method is estimated to be 1.5 μmol kg −1 , based on the analyses of Certified Reference Material supplied by Andrew Dickson. pCO 2 , aragonite saturation (Ω Ar ), carbonate ion concentrations ([CO 3 2− ]) and dissolved inorganic carbon (C T ) were calculated from the measured A T and pH at the average experimental water temperature and salinity using the refitted [35] equilibrium constants [36] . Salinity of WH seawater was measured daily as conductivity with Endress Hausser CLS30 - DIC4A probes. For the first 38 days of the experiment (from 26/11/08 to 4/1/09), pH was measured manually twice daily using a pH electrode (WTW Multi 340i with sensorex glass probe). From 20/12/08 onwards, pH was measured spectrophotometrically using an automated system (see [37] for details) which sampled the header tanks hourly and the replicate tanks every 12 h. The dual measurements made during the 16 day overlapping time period allowed us to correct the electrode-based measurements using the more accurate spectrophotometric measurements, and thus attain a continuous record for the experiment. For spectrophotometric pH measurements, seawater from the tank being measured was drawn into a syringe pump (Kloehn model 55022, fitted with a 50 ml syringe), then mixed with an aliquot of thymol blue dye (mixing ratio 49:1). The dye-seawater mixture was pumped into a spectrometer (Ocean Optics USB4000) fitted with a 1 cm path length quartz cuvette. The spectral data were combined with reference data (no dye) to determine the absorbance at 435, 596 and 735 nm. The syringe pump and spectrophotometer assembly were located in a box which was temperature controlled at 4.2°C, and the temperature of the cuvette was recorded along with the absorbance data. The water sampling, pH measurement and data logging was automated, and controlled by a LabView programme. pH on the total hydrogen scale was calculated from the measured absorbances and the thymol blue dye parameters (pK2, e1, e2, and e3) at the cuvette temperature and a salinity of 34.1, using the algorithms of [38] . The pH at experimental water temperature was then calculated from the pH measured at 4.2°C using the A T and the [35] equilibrium constants as refitted by [36] . Laternula elliptica sampling L. elliptica were sacrificed from each replicate tank on Days 0, 21 and 120 (27 November, 18 December 2008 and 28 March 2009, respectively) for determination of physiological and/or biochemical parameters. On Day 21, individuals were assessed for gene expression characteristics indicative of stress (i.e., heat shock protein, HSP70 ) and calcification activity ( CHS ). On Days 21 and 120, assessments of overall protein synthesis potential (i.e., total RNA content) were made. Also on Day 120, the individuals used for protein synthesis potential were first used to determine oxygen consumption rates as a proxy for energy use. On Days 0 and 120, physiological condition of the individuals was determined. Different individuals were assessed for physiological and biochemical parameters, respectively, on any one sampling date. Processing of L. elliptica for each of these measured is detailed below. 1. Molecular analyses Mantle and adductor tissue were carefully dissected from each individual, snap frozen in liquid nitrogen and stored at −80°C prior to analysis for (a) HSP70 and CHS gene expression levels and (b) total RNA quantification, respectively, as described below. A. Gene expression analysis The messenger RNA (mRNA) levels in mantle tissue of L. elliptica target genes ( CHS , HSP70 ) were measured by reverse transcription quantitative polymerase chain reaction (RT-qPCR), a highly sensitive method which provides a measure of the relative expression levels of the target mRNA versus a control or reference gene (such as β-actin). Mantle tissue was chosen for analysis as: 1) this is the site of synthesis of calcification proteins, such as our target protein, CHS; and 2) highest expression of HSPs and constitutive expression of HSP70 occurs in this tissue in L. elliptica [39] . Expression analysis was achieved by first isolating total RNA, followed by the generation of cDNA, which was subsequently used in RT-qPCR. As the CHS gene sequence has not previously been determined from L. elliptica , this first necessitated PCR amplification of a partial CHS cDNA fragment using degenerate primers followed by cloning and sequencing prior to the development of a RT-qPCR assay for this gene (ii below). Total RNA was extracted from 100 mg of L. elliptica mantle tissue using TRIzol reagent (Invitrogen Co, Grand Island, NY, USA) and resuspended in DEPC-treated water. The concentration of total RNA was determined by measuring ultraviolet absorbance at 260 nm. RNA purity was checked by determining the A 260/ A 280 ratio, and RNA integrity was checked by agarose gel electrophoresis. RNA samples were treated with Deoxyribonuclease I (Sigma-Aldrich) to remove any contaminating genomic DNA. Single-strand cDNA was reverse transcribed from 100 ng total RNA in a final volume of 10 μl using random primers and MMLV RT as per the SuperScript VILO cDNA synthesis kit (Invitrogen). Reactions were incubated at 25°C for 10 min, then at 42°C for 90 min, and terminated by heating at 85°C for 5 min. cDNA was stored at −20°C until required for cloning or PCR. Degenerate primers for amplifying a circa 800 bp cDNA fragment of the L. elliptica CHS gene were designed on the basis of known molluscan CHS cDNA sequences ( Table 2 ). PCR was performed using 10–100 ng of cDNA as a template in PCR buffer containing 3 mM MgCl 2 , 0.2 mM dNTPs, 0.4 μM of each primer and 1 unit of Platinum Taq Polymerase (Invitrogen) in a total volume of 10 μl. The thermal cycling parameters used were 95°C denaturation for 2 min, followed by 35 amplification cycles of 95°C for 30 sec, 55°C for 30 sec, 72°C for 2 min, and a final extension at 72°C for 10 min. The PCR products were gel-purified and sub-cloned into pCR4 TOPO TA vector (Invitrogen), and sequenced on an ABI Prism 3130 sequencer from both the 5′ and 3′ ends using the ABI PRISM dye terminator cycle sequencing ready reaction kit (Applied Biosystems, Foster City, CA, USA). Complete clone insert sequences were identified using the blast or blastx search programs within the National Center for Biotechnology Information, in order to confirm their identity as CHS . PCR amplifications of target ( CHS , HSP70 ) and reference (β-actin) genes were performed in 25 μl reactions containing: cDNA generated from 200 pg of original RNA template (10 μl of 1∶500 diluted cDNA generated from 100 ng of original RNA template), 0.2 μM of each gene specific primer (GSP, Table 2 ), 10 nM fluorescein, 0.2 mM dNTPs, 3 mM MgCl 2 , 0.33x Sybr Green, 0.15% Triton X-100, 1 U of Platinum Taq polymerase (Invitrogen) and 1X PCR buffer (Invitrogen (20 mM Tris-HCl (pH 8.4), 50 mM KCl)). Note that primers have been previously developed for β-actin and HSP70 qRT-PCR [39] . Amplification was performed and monitored using the Icycler IQ real-time PCR detection system (Biorad, Hercules, CA, USA) with the following thermal cycle protocol: initial denaturation and enzyme activation at 95°C for 2 min, 40 amplification cycles of 94°C for 15 sec, 58°C for 15 sec and 72°C for 15 sec, followed by a melt curve analysis. Normalisation was performed by collecting dynamic well factors using the fluorescein background signal detected during the first few PCR cycles, in order to compensate for any system or pipetting non-uniformity and optimise fluorescent data quality and analysis. Each template was analysed in triplicate. A four point standard curve for each primer pair being assayed, containing 10-fold dilutions of cDNA prepared from stock cDNA covering five orders of magnitude (1, 0.1, 0.001, 0.0001, and also encompassing the target nucleic acid interval), was included in triplicate on each 96-well PCR plate. Data were collected as quantification cycle (C q , formerly cycle threshold (C t )) values using the iCycler iQ Optical System Software Version 3.1. The C q values were normalised to sample quantities using the standard curve method for relative quantification. The primer concentrations (0.2 μM) were empirically determined based on lowest C q values and highest efficiencies. The β-actin gene of L. elliptica was used as a reference gene. Data were normalised against the expression of β-actin (expressed as a ratio) to compensate for any differences in loading or reverse transcriptase efficiency. β-actin, a widely used reference gene in qPCR analysis, has previously been used for qPCR analyses in L. elliptica [40] , and 1-way ANOVA demonstrated this gene did not vary across each time point in response to our pH treatments (data not shown). B. Total RNA content Total RNA on Days 21 and 120 was determined by pulverising freeze dried adductor tissue in a glass mortar and subsampling for RNA quantification. Total RNA was extracted using the TRIzol Reagent (Invitrogen # 15596-018; [41] ), and quantified spectrophotometrically. Approximately 15 mg of tissue was used for the analyses and an additional ethanol wash was added to the manufacturer's protocol to ensure samples of satisfactory purity. Total RNA was quantified by reading the absorption against 0.5% SDS at 260 and 280 nm with a spectrophotometer using quartz cells (1.4 ml volume). In addition, absorption spectra were obtained (200–320 nm) as an indicator of sample purity. To standardise RNA content between individuals of different sizes, we normalised results to the average SL of the L. elliptica used in the experiment, 73 mm, using the following formula: Total RNA ind A adj = [RNA ind A (μg mg −1 )/SL ind A (mm)]×73 Results are expressed as μg RNA mg −1 tissue dry weight. 2. Oxygen consumption Respiration rates of L. elliptica were determined on Day 120. Individuals were placed in closed 600 ml chambers in their equivalent treatment water, with continuous, gentle propeller-driven mixing, after first brushing and rinsing to remove microflora. Chambers were placed in a water bath and temperatures maintained at −1.4°C (range = −1.3 to −1.6°C). After 50 min the chambers were sealed and oxygen consumption measurements were made over 184 min. Fifty min was considered to be an appropriate settling period to eliminate any effects of handling stress, based on trial experiments that we had conducted and other studies (e.g., [42] , [43] ). Measurements were made using hand held probes (YSI55, YSI550A and WTW Multi 340i with CellOx probe), which were first calibrated using fully saturated (100%) water and then sulfite (H 2 S) water (0%). Dissolved oxygen saturation in each chamber was not allowed to fall below 70%. O 2 consumption was recorded every 2 min as % saturation and every 5 min as mg l −1 . These values were later adjusted for the volume of the animal and for consumption rates recorded in chambers without animals. To convert O 2 consumption values to per unit ash free dry weight (AFDW; methods provided below), we derived the following relationship between L. elliptica AFDW and wet tissue weight: AFDW = 0.37014 wet weight 0.55134 ; R 2 = 0.63 As there were no differences in this relationship between animals from different treatments (standard errors of all parameter estimates overlap for all treatments) we considered this to be a good estimator of AFDW for the O 2 consumption animals. 3. Physiological condition On Days 0 and 120, L. elliptica from each experimental replicate were dissected to remove flesh, the flesh wet weight determined, and the flesh and shell oven dried (60°C) and air dried, respectively, to constant weight (hereafter ‘FW’ and ‘SW’, respectively). The dry flesh was then ashed in a muffle furnace at 470°C for 24 h to determine AFDW (AFDW = FW- ashed weight). Two physiological condition indices were calculated: the ratio of FW to SL (CI FW:SL ), and the ratio of SW to SL (CI SW:SL ). We also examined the change in each condition index over the 120 day experiment (Day 120 – Day 0). Statistical analyses A 2-way ANOVA with Day as a random factor and Treatment as a fixed factor was initially conducted to confirm that SL of the animals did not vary between treatments. Where simple monotonic relationships between each response variable and pH were indicated, the significance of the relationship was tested by correlation analysis (either Pearson's or Spearman's) due to the well known insensitivity of ANOVA to monotonic gradients in responses [44] – [46] . For response variables exhibiting non-monotonic relationships, 1-way ANOVAs were used to test for differences between treatments, after determining whether normality and homogeneity of variances assumptions were met. When this was not the case and transformations did not help, a Kruskal-Wallis test was used. Antarctic field conditions Background information on the water chemistry in McMurdo Sound was determined from day time water samples and longer term instrument deployments. On 10 November 2008 water samples were collected for C T and A T analyses from 3 m below the surface at the Granite Harbour collection site and later analysed to determine the Antarctic control treatment pH level. Near bottom (18 m deep) water samples for C T and A T analyses were collected from New Harbour on seven occasions in the following year (6 th and 8–14 November 2009). All samples were immediately preserved with HgCl 2 . A T was analysed as described above, and C T was determined using coulometric measurement of the CO 2 evolved from an acidified sample [34] . In situ pH, pCO 2 , Ω Ar and [CO 3 2− ] were calculated from the measured values using the refitted [35] equilibrium constants [36] . Information on temperature and salinity required for these calculations was obtained at both sites from Seabird Electronics SBE-37 microcat deployments.

Results Experimental conditions The chemical characteristics of our experimental treatments and Antarctic field sites are given in Table 1 . Also provided for comparison are characteristics of water reported in other relevant polar studies from the literature. These values show good agreement with our experimental conditions. Of interest are the high pCO 2 levels of our McMurdo Sound water samples (410 and 440 μatm at Granite Harbour and New Harbour, respectively). Laternula elliptica response There was no mortality over the 120 day experiment, and L. elliptica appeared to behave normally, extending siphons and feeding. There was also no measurable increase in SL of these adult L. elliptica over the experiment. Sizes of individuals did not differ significantly between treatments (p>0.05). 1. Molecular analyses A. HSP70, heat shock protein gene expression L. elliptica mRNA expression levels of HSP70 showed a curvilinear response to seawater pH, with lowest levels in the Antarctic control treatment after 21 days. There was considerably higher variability in HSP70 levels of individuals from the elevated pH treatment than in those of the other pH treatments ( Figure 1 ). There was no significant correlation between pH and HSP70 gene expression relative to reference gene β-actin expression. However, HSP70 expression showed some indication of differences between treatments (Kruskal-Wallis Chi-Square = 5.3450, p = 0.0691), with the Antarctic control differing from both the elevated and lowered pH treatments (unadjusted p-values were 0.0451 and 0.0039, respectively). B. CHS cloning, sequence determination and gene expression There was a significant negative correlation between mRNA expression levels of CHS relative to β-actin reference gene expression in L. elliptica and pH (log 10 transformed) on Day 21 of the experiment (Pearson's R −0.61, p = 0.0060, n = 18), with an increase in CHS with decreasing pH ( Figure 2 ). The L. elliptica CHS partial cDNA sequence of 953 nucleotides (Genbank accession number HQ186262) is provided in Figure S1 overlaid with the deduced protein sequence of 317 amino acids. Highly conserved motif regions 1–4 found in many family 2 glycosyltransferases (GTF2) enzymes, including CHS, are indicated. BLAST analysis showed that the L. elliptica CHS nucleotide sequence gene shares homology with other known CHS genes, indicating that the cloned gene encodes CHS protein ( Figure S2 ). A multiple sequence alignment of the deduced protein sequence of L. elliptica CHS with chitin synthases of other known bivalves is provided in Figure S2 . L. elliptica CHS shows highest homology in its sequence to Atrina rigida (Japanese pearl oyster) and Pinctada fucata (stiff penshell oyster), with 72% identity at the nucleotide level and 82% and 80% identity over the deduced protein sequences, respectively ( Table S1 ). C. Total RNA Total RNA levels were lowest in L. elliptica from the Antarctic control pH treatment after both 21 and 120 days. Levels in the elevated and lowered pH treatments were higher, and similar to each other ( Figure 3 ). This curvilinear response, similar to that noted for HSP70 , was observed on each Day, although neither was statistically significant (Day 21: F = 0.40, p = 0.6815; Day 120: F = 0.37, p = 0.6975). 2. Oxygen consumption O 2 consumption rates were significantly higher in individuals from the elevated and lowered pH treatments compared with those from the Antarctic control at the end of the experiment (F = 5.53, p = 0.0159; Figure 4 ). There was no significant relationship between O 2 consumption and L. elliptica SL (F = 1.17, p = 0.295, R 2 = 0.067). 3. Physiological condition Physiological condition of L. elliptica as measured by CI FW:SL on Day 120 did not differ between treatments, but overall condition had increased relative to Day 0 in all treatments, indicating the experimental conditions were favourable for maintenance and survival of the L. elliptica ( Figure 5 ; 2-way ANOVA, Treatment: P = 0.3245; Day: P = 0.0443; Treatment*Day: P = 0.9434). While no significant difference between treatments on Day 120 was detected, the magnitude of average change in condition between the beginning and end of the experiment did not appear to be random ( Figure 6 , Pearson's R = −0.48, p = 0.0450, n = 18). The average increase in CI FW:SL over the experiment was lowest at elevated pH (7.7%; Figure 6 ). The CI SW:SL index on Day 120 was not significantly different between treatments, nor was there any significant difference between Days 0 and 120 ( Figure 5 ; Treatment: p = 0.9956; Day: p = 0.6332; Treatment*Day: p = 0.7730). Although Figure 6 suggests the magnitude of average change in CI SW:SL between the beginning and end of the experiment may not be random, no significant correlation with pH was observed due to the high within-treatment variation, mainly due to the presence of a single high value in both the Antarctic control and low pH treatments ( Figure 6 , Pearson's R 0.29, p = 0.2655, n = 18). Preliminary examination of the outer surfaces of the L. elliptica shells did not reveal any obvious signs of degradation. As there was no increase in shell length of the slow growing adult L. elliptica over the experiment, any change in CI FW:SL or CI SW:SL can be attributed to a change in flesh weight or shell weight, respectively.

Discussion Functioning of adult Laternula elliptica was affected by pH levels predicted for the Southern Ocean in coming decades (pH 7.78) after as little as 21 days exposure. Although no mortality occurred, L. elliptica in this lowered pH treatment were more stressed ( Figure 1 ) and exhibited significantly higher basal metabolic rates ( Figure 4 ) than L. elliptica from the Antarctic control (pH 7.99). Interestingly, this response was also noted for L. elliptica from the elevated pH treatment (pH 8.32), thus demonstrating a negative response by this bivalve to a change in pH in general. Most importantly, we noted a differential response of L. elliptica to lowered compared to elevated pH for CHS gene activity (as indicated by mRNA transcript abundance, Figure 2 ). CHS expression increased with decreasing pH, indicating an effect on the shell formation process of L. elliptica which appears specific to a decrease in pH. Total RNA content and HSP70 gene expression level did not show a statistically significant response to our experimental treatments, although the pattern of both responses mirrored that of the basal metabolic rates (cf. Figures 1 , 3 and 4 ). The HSP70 response was close to being significant, with clear differences between individuals from the Antarctic control and both the elevated and lowered pH ( Figure 1 ). The different response variables measured were influenced by pH in differing ways, indicating the importance of assessing a variety of different factors to determine the likely effect of pH change on organism functioning. While these effects on L. elliptica did not translate to statistically significant differences in either of the physiological condition indices we measured over the 120 day experiment ( Figure 5 ), we suggest that the sustained cost of increased stress, metabolic and calcification rates may well affect L. elliptica condition (and thus growth and reproduction) in the longer term [47] . Metabolism and growth Reduced metabolism is a recognised strategy in Antarctic invertebrate fauna to minimise energy expended during routine ‘maintenance’ and thus have more energy available to invest in reproduction and growth [48] . In this experiment we have used oxygen consumption rates as a proxy for basal metabolism (e.g., [29] , [49] ). In less than optimal environmental conditions, we may expect to see basal metabolic rates increase as the organism works harder to maintain the status quo, and this may in turn result in changes in overall physiological condition. In our experiment oxygen consumption rates of L. elliptica in the Antarctic control were 2.2 μmol g −1 AFDW h −1 , 2.3 and 2.1 times lower than in the lowered and elevated pH treatments, respectively. These results illustrate that while L. elliptica can function at pH's different from those they currently experience in Antarctica it is energetically more stressful for them to do so, indicating they are optimally adapted to their present environment. Metabolic rates (also measured as O 2 consumption) in the bivalve Yoldia eightsi increased under higher temperatures [50] , and a similar pattern has been noted for a range of other Antarctic marine invertebrates [51] , [52] . Seasonal increases of 3 and 3.7 times between winter and summer have been reported for Antarctic Peninsula L. elliptica [53] , [54] , a likely response to changes in temperature and food availability. As no field measurements of oxygen consumption by McMurdo Sound (or indeed, Ross Sea) L. elliptica have been made, we are unable to comment on the magnitude of the differences between our experimental treatments relative to seasonal or spatial differences in Ross Sea L. elliptica . However, we do note that the O 2 consumption rates in our Antarctic control animals were considerably lower than those recorded for L. elliptica from the Antarctic Peninsula region. [55] measured average rates of 73.3 and 49.6 μmol O 2 h −1 10.6 g −1 AFDW for L. elliptica from Rothera and Signy Island, respectively, at 0.0°C. When converted to enable direct comparison to our measurements, these rates are considerably higher than the average 2.2 μmol h −1 g −1 AFDW noted in the −1.8°C Antarctic control treatment of this experiment (i.e., Rothera 6.9, Signy 4.7 μmol h −1 g −1 AFDW), perhaps reflecting the influence of temperature on respiration. Furthermore, these differences in L. elliptica basal metabolism between Antarctic locations indicate that the functional response of this bivalve to ocean acidification may vary depending on their Antarctic environment. Metabolic differences can also be reflected in short term growth rates as measured by total RNA (protein synthesis potential; [56] ). Total RNA levels provide a measure of proteins synthesised during tissue growth as well as during physiological stress responses, and were relatively high in L. elliptica from the elevated and lowered pH treatments compared with the Antarctic control ( Figure 3 ). Effects on total RNA levels were variable and, consequently, were not statistically significant on either Day 21 or Day 120 ( Figure 3 ). We expect that they may have become significant given a longer experimental duration. Adductor muscle total RNA levels in our experiment were within the range of those found in L. elliptica collected at three other McMurdo Sound locations (i.e., Dunlop Island: 8.20±1.17 μg mg −1 , n = 6; Spike Cape: 4.41±0.22 μg mg −1 , n = 10; Cape Evans: 3.96±0.31 μg mg −1 , n = 6; Joanna Norkko, unpublished data). Stress An increase in protein synthesis (including production of heat shock proteins, HSPs) has also been noted in L. elliptica in response to another stressor, temperature [39] , [40] , [57] . HSPs in the normal cell state assist in the folding of native polypeptides, and their induction under stress conditions prevents production of cytotoxins and stabilises denatured proteins (e.g., [32] , [58] , [59] ). HSPs can be produced in response to a large variety of environmental stressors (e.g., freshwater input in the intertidal [60] ; osmotic stress [61] ; presence of oxygen radicals and toxicants [62] ), and we may expect a similar response to pH change. We have demonstrated a substantial up-regulation (increased production) of HSP70 gene transcript levels in L. elliptica mantle tissue in response to pH levels that are both lowered and elevated relative to existing Antarctic conditions ( Figure 1 ). This up-regulation is not unexpected as either pH directionality is likely to register as a stress to L. elliptica , stimulating induction of HSPs. HSP mRNA levels typically rapidly rise within hours following introduction to a stressor and generally reach their maxima following three hours of post-stress recovery [63] , [64] . Our analysis shows that at 21 days following the initiation of the stress exposure (high or low pH), HSP70 levels in these bivalves remain elevated relative to individuals from the Antarctic control. The magnitude of the HSP70 up-regulation we observed is in line with experimental observations of the heat shock response (HSR [64] ) in L. elliptica in response to the more classically studied stressor, temperature. L. elliptica constitutively express HSP70 [39] , a common phenomenon for Antarctic species, and further sustained up-regulation in response to pH as a stressor may in fact lead to deleterious effects on other cellular and organismal processes (e.g., [65] ). There is an energetic cost associated with induction of the HSR and the HSR system in L. elliptica is likely to have evolved under strong trade-off constraints [59] . Thus, in this experiment, the increased respiration rates we observed at pHs either elevated or decreased relative to the Antarctic control may relate in part to the extra energy required to mount and maintain the HSR over an extended period of time. Shell (mineralogy and dissolution) As L. elliptica shell is comprised purely of aragonite, one of the most soluble forms of CaCO 3 , it may be considered particularly susceptible to ocean acidification (e.g., [5] , [66] ). Aragonite was undersaturated in our lowered pH treatment (pH 7.78, Ω Ar = 0.71) and the % change (loss) in CI SW:SL after 120 days was the highest of all treatments (4.4%; Figure 6B , or 0.037% per day), although this correlation was not statistically significant. [9] reported a 2.767±0.607% loss in shell weight for L. elliptica shells held at 7.4 pH (Ω Ar = 0.47) for 63 days, the equivalent of 0.044% per day. While a greater dissolution rate may be expected in the more undersaturated Ω Ar conditions of the [9] experiment, we would urge caution in direct comparison of daily shell weight loss between experiments given that [9] studied empty shells and that dissolution may not be linear with time. After just 21 days we found that lowering pH resulted in a significant increase in expression (up-regulation) of the CHS gene, which codes for a key enzyme involved in synthesis of bivalve shells [28] , [67] , thus indicating the animals were working harder to calcify. In contrast, elevated pH resulted in decreased CHS gene expression (down-regulation). Chitin is a major component of the bivalve shell [28] , and forms the organic ‘framework’ within which CaCO 3 minerals are subsequently deposited [67] . The enzymes involved in chitin synthesis, and chitin synthase in particular, are important not only in providing mechanical strength, but also in coordinating the shell formation and mineralisation process [67] . Inhibition of chitin synthesis has been clearly demonstrated to negatively affect survival and increase abnormal shell development rates in early Mytilus galloprovincialis larvae [67] , [68] . In addition, larval shells formed under chitin synthesis-inhibited conditions were shown to be more soluble in distilled water [67] . Although CHS gene expression is not a direct measure of calcification, given that it encodes for an enzyme that is important in the shell mineralisation processes, effects on CHS gene expression could potentially affect the basic structure as well as the mineralisation, solubility and thus integrity of the shell. This is the first study examining changes in expression of CHS in response to ocean acidification (or other stressors) in any mollusc. The up- and down-regulation of CHS gene expression observed in our lowered and elevated pH treatments, respectively, is an important finding as it provides evidence for biological control over this process in response to changing environmental conditions, which may afford the organism a mechanism for some degree of acclimation or adaptation to future ocean acidification scenarios. Experimental vs Antarctic conditions The conditions of our laboratory experiment were as close as possible to those of the Antarctic situation. Our pH and Ω Ar measurements from Granite Harbour and New Harbour, and the pH reported for Cape Armitage by [23] show high agreement with the experimental values of our Antarctic control treatment ( Table 1 ). The A T of the seawater used in our experiment was naturally lower than that measured in the McMurdo Sound environment (by approximately 80 μmol kg −1 ; Table 1 ), although it was similar to that reported by [10] for cold, high latitude waters (2271±28.3). At -1.76°C, our experimental water temperatures were only slightly warmer than ambient spring water temperatures recorded for several locations in McMurdo Sound (−1.92°C; [69] , [70] ), and were very similar to those recorded in Terra Nova Bay where L. elliptica are also common (−1.8°C, authors' unpublished data). Salinities were also similar to those measured in McMurdo Sound in this study ( Table 1 ), and by [70] . The experimental conditions were generally favourable for the maintenance and survival of L. elliptica , as shown by the increase in physiological condition CI FW:SL over the 120 day experiment in all treatments ( Figure 5 ). Because there was no measurable increase in SL of the adult L. elliptica over the experiment, this change was purely due to an increase in flesh weight. There was no significant difference in absolute CI FW:SL between treatments on Day 120 ( Figure 5 ), although the magnitude of this gain in condition between Day 0 and Day 120 increased significantly with decreasing pH ( Figure 6 ). However, the significance of this was driven by the lower gain in condition of individuals in the elevated pH treatment, with little real difference between individuals from the lowered and Antarctic control pH treatments ( Figure 6 ). For logistical reasons, L. elliptica were not held in sediments during these experiments, thus these experiments have not allowed for any buffering of the response to changing pH by burial. However, as pH within the sediment column is generally lower than the water column above [71] , there will still be direct contact between the overlying seawater and the infaunal organism during feeding and respiration, and the CaCO 3 content of the Granite Harbour sediments was very low, we do not anticipate that our results have overestimated the magnitude of the L. elliptica response. In ocean acidification experiments, organisms are subjected to often-large pH changes which, of necessity, occur at a much faster rate than that predicted for their natural environments (i.e., years-decades). This has led to concern that these are in fact ‘shock-response’ experiments. It is also worth pointing out that in coastal and estuarine environments [72] and areas of oceanic upwelling [73] , pH may change markedly over time scales more akin to those of experiments (i.e., hours to days). Of note are the high pCO 2 levels of our McMurdo Sound water samples (410 and 440 μatm at Granite Harbour and New Harbour, respectively). The distance between these two coastal locations (ca. 90 km) and the fact that they were sampled in different years, indicates that the pCO 2 levels of these high latitude coastal sites is already high in spring/early summer. Data are urgently needed to determine spatial and temporal variation in pCO 2 and pH in Antarctic coastal regions, to put results of experiments into context of natural environmental conditions. Concluding comments L. elliptica are a long lived species: their life span is estimated at 36 years [74] , and they take about 20 years to reach 100 mm SL [26] , [75] . The duration of this experiment is just a very small portion of their life span, and the effect on functioning of L. elliptica at a lowered pH of 7.78 is of concern given the predicted changes in Southern Ocean pH and aragonite saturation for the coming decades [19] . While many studies have focussed their investigations on juveniles due to the increased susceptibility of early life stages to environmental perturbations (e.g., [76] ), examining effects on adults is also important, particularly for long lived Antarctic species which will experience this change in ocean chemistry within their generation. We have shown significant effects on some crucial functions of L. elliptica at a pH only 0.2 units below current levels. Importantly, the observed changes in L. elliptica CHS gene expression provides evidence for biological control over the shell formation process, which may provide a mechanism for adaptation or acclimation to future changes in seawater carbonate conditions. We anticipate, however, that the energetic costs of maintaining these responses may have more serious implications for the condition of this key Antarctic bivalve in the longer term. In addition, increases in temperature predicted for Antarctic waters over the next 100 years, and associated changes in food supply (e.g., [77] ), are likely to modify any effect of pH reductions alone on the behaviour and functioning of key benthic invertebrates. Future investigations should study synergistic effects, and should incorporate a range of response variables to build a more comprehensive picture of the likely ecological impacts of impending environmental change.

Conceived and designed the experiments: VC VM ST JH KC. Performed the experiments: VC NB PH NH VM AVR SB JN AG CM. Analyzed the data: VC JH KC VM AVR JN RS AG SB. Contributed reagents/materials/analysis tools: VC KC NB PH NH VM CM. Wrote the paper: VC VM KC JH ST JN AVR SB. Current address: Plant and Food Research, Canterbury Agriculture and Science Centre, Christchurch, New Zealand Ocean acidification is a well recognised threat to marine ecosystems. High latitude regions are predicted to be particularly affected due to cold waters and naturally low carbonate saturation levels. This is of concern for organisms utilising calcium carbonate (CaCO 3 ) to generate shells or skeletons. Studies of potential effects of future levels of pCO 2 on high latitude calcifiers are at present limited, and there is little understanding of their potential to acclimate to these changes. We describe a laboratory experiment to compare physiological and metabolic responses of a key benthic bivalve, Laternula elliptica , at pCO 2 levels of their natural environment (430 μatm, pH 7.99; based on field measurements) with those predicted for 2100 (735 μatm, pH 7.78) and glacial levels (187 μatm, pH 8.32). Adult L. elliptica basal metabolism (oxygen consumption rates) and heat shock protein HSP70 gene expression levels increased in response both to lowering and elevation of pH. Expression of chitin synthase ( CHS ), a key enzyme involved in synthesis of bivalve shells, was significantly up-regulated in individuals at pH 7.78, indicating L. elliptica were working harder to calcify in seawater undersaturated in aragonite (Ω Ar = 0.71), the CaCO 3 polymorph of which their shells are comprised. The different response variables were influenced by pH in differing ways, highlighting the importance of assessing a variety of factors to determine the likely impact of pH change. In combination, the results indicate a negative effect of ocean acidification on whole-organism functioning of L. elliptica over relatively short terms (weeks-months) that may be energetically difficult to maintain over longer time periods. Importantly, however, the observed changes in L. elliptica CHS gene expression provides evidence for biological control over the shell formation process, which may enable some degree of adaptation or acclimation to future ocean acidification scenarios.

Supporting Information

We thank Malcolm Reid (University of Otago), and Jeremy Bulleid and Claire Coppard for advice on and construction of the spectrophotometer, respectively, Sarah Allen for her work in establishing the containment facility, and the K082 dive team (all NIWA) for Laternula elliptica collection. Antarctica New Zealand are thanked for their excellent logistical support on the ice, and we are particularly grateful to Paul Woodgate for his efforts in ensuring the safe importation of L. elliptica to NZ. This manuscript was improved by the helpful comments of two anonymous reviewers.

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e16069

oa_package/11/a2/PMC3016332.tar.gz

PMC3016333

21246032

Introduction Along with the rising prevalence of obesity worldwide [1] , [2] , an increasing focus on the role of obesity-related inflammation has evolved, primarily confined to its role in development of metabolic complications to obesity such as type 2 diabetes and cardiovascular disease [3] . Plasma C-reactive protein (CRP) is the obesity-related inflammatory marker that has been most consistently associated with cardiovascular risk [4] , [5] . However, a range of other systemically measurable inflammatory markers have been suggested to associate not only with obesity [6] , but putatively also with morbidity and mortality [4] , [7] , [8] , and these markers are therefore also of clinical interest. There is an obvious lack in the understanding of causes behind and modifiers of this obesity-related inflammatory activity and environmental as well as genetic factors may be considered to play a role in the aetiology of this condition. Hitherto, the A-allele of the FTO rs9939609 is the genetic variant most strongly associated with common obesity [9] – [11] . A recent study including 2,415 participants from a middle-aged German population reported that the FTO rs9939609 A-allele causes variation in CRP levels independent of its effect on fatness [12] . In addition to CRP, it is thus of interest to investigate whether the FTO rs9939609 is associated with a range of other inflammatory markers. The aim of the present study was to investigate whether the FTO rs9939609 A-allele has fatness-independent effects on circulating levels of the inflammatory markers hs-CRP, interleukin (IL)-1β, IL-6, IL-10, IL-18, tumor necrosis factor-alfa (TNF-α), soluble tumour necrosis factor α receptor antagonist (sTNFα-R1), transforming growth factor-beta (TGF-β), macrophage inflammatory proteins (MiP)-1α, and MiP-1β. Several of these markers are secreted from the adipose tissue, so-called adipokines, which also includes the non-inflammatory marker leptin. Leptin was included in the analyses as an adipocyte marker.

Methods The participants included in this study were originally identified among a total of 362,200 Danish men who underwent the mandatory draft board examination in Copenhagen during 1943–1977 and in the remainder of Sjælland during 1964–1977 [13] – [15] . Among these, a study population consisting of a randomly selected 1.0% sample (n = 3,601) and all men with a body mass index (BMI) equal to or above 31.0 kg/m 2 (n = 1,930) was sampled manually from the draft board files during the 1970s. The latter group represented all at least 35% overweight men according to a national standard in use at the time of sampling (all of whom were above the 99 th percentile of the BMI distribution). This sampling design provided coverage of the entire distribution of BMI in the population with a 100-fold relative over-sampling of the obese, in order to allow particular focus on obesity and related aspects. All obese and half of the controls were invited to participate in the examination programme of the Copenhagen City Heart Study in 1981–83 and again in 1991–93. Details of these surveys are published elsewhere [16] , [17] . A last thorough whole-day examination was conducted in 1998–2000 among subjects who participated in the Copenhagen City Heart Study in 1991–93. Exclusion criteria were age above 65 years (to avoid bias by aging), residence too far away from Copenhagen to allow one-day examination, refusal in the past to participate in follow-up examinations, regular medication and known disease. Details of this survey is published elsewhere [11] , [18] . In total, 551 men, hereof 231 of the obese were re-examined, including genotyping and measurement of fasting circulating inflammatory markers at an average age of 49 years (range: 39 through 65 years). Genotyping Genotyping of FTO rs9939609 (Taqman allelic discrimination; KBiosciences, Cambridge, UK) was successful in 97% of the samples with a genotype error rate of 0.27% based on 1,464 duplicate samples. All genotype groups obeyed the Hardy–Weinberg equilibrium and the A-allele frequency was 0.41 in controls and 0.51 in obese individuals. Molecular genetic analysis, including genotyping of the FTO SNP rs9939609, was conducted on 551 men, hereof 231 of the originally obese. At present, the SNPs that are strongest associated with BMI belong to a linkage disequilibrium (LD) block encompassing parts of the first two introns of FTO [19] . Since the SNPs are in tight LD, the study of one of them will convey the effect. Phenotypic and biochemical measurements Objective measures of height and weight were obtained [20] . BMI was calculated as weight per height squared. Furthermore, fat mass was assessed by dual-energy x-ray absorptiometry (DXA-IQ DEXA; Lunar, Madison, WI). Fat-BMI is equivalent with BMI, but calculated as weight of body fat per height squared. Hip and waist circumference were measured to the nearest 0.5 cm with the subjects standing. Hip circumference was measured at the maximal width over the greater trochanters, and waist circumference was measured midway between the iliac crest and the rib cage. Blood samples were obtained in the morning after an overnight 12-hour fast, and hs-CRP was measured in plasma by high sensitivity enzyme-linked immunosorbent assay (ELISA) [21] , with a lower detection limit of 0.005 ìg/dL. The concentration of the cytokines in serum, IL-1β, IL-6, IL-10, IL-18, TNF-α, sTNFα-R1, TGF-β, MiP-1α, and MiP-1β, were measured using the Luminex® xMAP technology as described in Skogstrand et al. [22] . Serum leptin was assessed by RIA (human leptin RIA kit, Alta Diagnostica, Marburg, Germany). The serum levels were measured in the same lab and in the same batches for all samples. The lower and upper detectable concentration limits, as well as the number of measures within these limits, can be seen in Table S2 . Statistical analysis The inflammatory markers were modelled as the response variable in both the linear and the logistic regression analyses. In the linear regressions only detectable measurements were included in the analyses. The distribution of the inflammatory markers were visually tested via Q-Q plots, and as they all departed from a normal distribution (Shapira-Wilk test, all p<0.0001), they were log(e)-transformed [23] . For the logistic regression analyses the measures either above or below the detection limits of the assays were recoded to correspond to these limits, and then all the measurements of the markers, whether within or beyond the detection limits, were dichotomized by the median (the number of measures above or below these limits can be seen in Table S2 ). BMI was also log-transformed (base 1.1) to pull in the long right tail, and hereby obtain a better model fit in regard to the underlying assumptions of linearity in both the linear and logistic regression models [23] . Waist (for given BMI) was modelled per the 10 cm increase, and waist-to-hip ratio (WHR) was log-transformed (base 1.1). Both the likelihood ratio test and the Akaike information criterion (AIC) showed that an additive model of FTO rs9939609 gave the best model fit in both the linear and logistic regression models as compared to dominant and recessive models. Due to the skewed inflammatory markers their characteristics are presented as geometric means (which is the antilog of the arithmetic mean of the logged data) with 95% confidence intervals (CIs). After checking for linearity, all analyses were run for the obese and randomly selected men combined, which was possible because BMI was used only as an independent covariate. We investigated if hs-CRP and the 10 other adipokines were associated with the FTO rs9939609 and with the anthropometric obesity measures in our cohort, both in crude analyses, and when mutually adjusted. Then we investigated if an association between BMI and the inflammatory markers differed across FTO rs9939609 genotype. The results from the linear regressions are presented as effect estimates with 95% CI, which represent a factor of change in the inflammatory marker relative to the measurement unit in the predicting covariate (i.e. an estimate of 1.17 for FTO rs9939609 genotype in relation to hs-CRP is to be interpreted as a 17% higher hs-CRP-level per additional A-allele). The results for the logistic regressions are presented as odds ratios (OR) with 95% CI. Interactions between BMI and FTO rs9939609, both modeled as continuous variables (0, 1 and 2 for the FTO alleles), were investigated. The likelihood ratio test assessed whether the model with the product term provided a better fit than the model without the product term. All models were adjusted for age. All analyses were two-tailed and a significance level was accepted at p<0.05. Analyses were carried out with Stata (version 9.2; Stata Corporation, College Station, Texas). Ethics statement The Danish Data Protection Agency and the Ethical Committees of Copenhagen and Frederiksberg municipalities approved the study, which was in accordance with the Helsinki Declaration II. All participants signed written consent before participating.

Results In Table 1 the phenotype characteristics of the two cohorts of men are presented according to the FTO genotype. Men belonging to the obese cohort were obviously heavier, had larger waist circumference and WHR than the randomly selected men. The obese were also younger, which reflects the increasing prevalence of obesity in more recent time [15] . From Table 1 we calculated the OR of having a BMI = 31.0 kg/m 2 at the draft board examination. Using the TT genotype as the reference group we found that the OR for the TA genotype was 1.29 (95% CI: 0.87–1.93) and for the AA genotype 2.22 (95% CI: 1.38–3.57). The circulating levels of hs-CRP and the adipokines are presented as geometric means in Table 2 for the two cohorts combined (and for the two cohorts separately in Table S1 ). There was no overall pattern in the distribution of these across FTO genotype. For the adipokines IL-1β, IL-10, IL-18, mip1α, mip1β, sTNFα-R1, TGF-β, and TNF- α there were no association with either FTO rs9939609 or BMI (analysis not shown, see Table S1 ). The same results were obtained by the logistic regression analyses of the dichotomized values of the inflammatory markers ( Table S2 ). There was a trend across FTO genotype in relation to hs-CRP, IL-6 and leptin ( Table 2 ); hence, these three adipokines were investigated in further detail. Positive correlations between hs-CRP and fat mass (kg) were observed both among the randomly sampled and the obese men (Spearman's ρ equal to 0.29 (p<0.001) and 0.31 (p<0.001), respectively). Similarly, positive correlation between leptin and fat mass (kg) was observed (Spearman's ρ equal to 0.87 (p<0.001) in the randomly sampled and 0.86 (p<0.001) in the obese). There were no correlations between IL-6 and fat mass (kg) (Spearman's ρ equal to 0.02 (p = 0.81 in the randomly sampled and 0.03 (p = 0.70) in the obese)). The results for the linear regressions of hs-CRP, IL-6 and leptin on FTO genotype and the anthropometric obesity measures, respectively, are presented in Table 3 . FTO genotype and hs-CRP were positively associated, e.g. the hs-CRP level increased by 17% per additional A-allele (p = 0.05). However, this association disappeared when BMI and the other obesity markers were taken into account. Similarly, the leptin level increased by 10% (p = 0.04) per additional A-allele, but BMI, WHR and waist circumference accounted for this association. Further, positive associations were seen between BMI, WHR and waist (for given BMI) and hs-CRP and leptin, respectively, and these remained significant when FTO genotype was taken into account. As an example, hs-CRP level increased by 20% per 10% increase in BMI (p<0.001), and by 19% (p<0.001) when FTO was adjusted for. There were no association between IL-6 and FTO genotype and the anthropometric obesity measures, respectively. Furthermore, no interactions between BMI and FTO genotype in regard to hs-CRP, IL-6 and leptin, respectively, were observed ( Figures 1 , 2 , 3 ). The analyses were repeated with fat-BMI instead of BMI; however, this did not change the associations. Further adjustment for smoking habits, alcohol consumption, and physical activity also did not change the associations. Regarding hs-CRP, additional analyses were performed based on exclusion of individuals with hs-CRP values above 10 mg/l (31 randomly selected and 34 obese men), leaving all associations fairly unchanged. Finally, virtually the same pattern of results was obtained by logistic regression analyses of the dichotomized values of the inflammatory markers ( Table S2 ).

Discussion In the present study of middle-aged Danish men positive associations between hs-CRP, leptin and BMI were observed. The associations did not differ across FTO rs9939609 genotype. For IL-6, there was a non-significant tendency to lower levels by the FTO A-allele, i.e. opposite to the expectation, independent of the fatness variables. There were no associations between IL-1β, IL-10, IL-18, mip1α, mip1β, sTNFα-R1, TGF-β, and TNF-α and FTO genotype or BMI, respectively, in this cohort. Generally, the confidence intervals were so narrow that it seems fair to suggest that all together the FTO rs9939609 A-allele does not have important fatness-independent effects on systemic inflammatory markers or adipokines. A fatness-independent association between the FTO rs9939609 A-allele and hs-CRP was observed in a recent German study [12] , and the effect was persistent even when prevalent myocardial infarction, stroke and diabetes were taken into account. The participants in the German study were sampled among the general population, with a mean body mass index (BMI) of 26.8 kg/m 2 among men and 25.7 kg/m 2 among women. We did not find an association between FTO rs9939609 and systemic inflammatory markers among generally healthy middle-aged men representing a broad range of BMI. The restriction in the inclusion of men to those without known diseases or regular medication may explain why we did not find an association between BMI and a range of adipokines generally known to associate with increased BMI [6] . Whether it also influenced our findings in relation to hs-CRP and FTO rs9939609 is only speculative. Regarding the circulating hormone leptin, a study on diabetics of both sexes found that FTO rs9939609 affected circulating leptin levels, yet the effect was accounted for by BMI [24] . In our data of relatively healthy subjects the FTO rs9939609 A-allele also did not have fatness-independent effects on the circulating levels of leptin. Thus, there is no reason to believe that the FTO rs9939609 A-allele has any functional impact on leptin in addition to the effect on fatness. Finally, we were surprised that the FTO A-allele tended to lower IL-6 levels, since we would have expected obesity to lead to increased IL-6 levels [3] . We have previously shown that FTO rs9939609 had a fatness-independent effect on mortality and morbidity when analysed in a dominant model. We found that carriers of the A-allele had nearly twice the mortality compared to non-carriers irrespective of BMI [25] . In relation to vascular diseases, we observed a tendency towards an increased risk among A-allele carriers illustrated by a hazard ratio of prevalent vascular disease at time of death of 1.43 (95% CI, 0.87–2.37; p = 0.15) vs. non-carriers [25] . A Scottish study of nearly 5000 diabetics of both sexes found that A-allele carriers of the FTO rs9939609 had more than twice the risk of a myocardial infarction or cardiovascular death compared with non-carriers, when adjusted for age, gender, BMI, smoking and history of myocardial infarction [26] . Human adipose tissue is heterogeneous in its metabolic activity, which lead the authors of the German study to suggest that the sites in the adipose tissue that are more sensitive to infiltration of immune cells might be expanded preferentially among A-allele carriers, resulting in a increased risk of cardiovascular disorder [12] . However, our results do not support the hypothesis that the increased morbidity and mortality could be explained by a fatness-independent effect of the FTO rs9939609 A-allele on systemic inflammation. There are several strengths of this study, but also limitations that should to be taken into consideration. The study is based on a unique, well-defined large background study population of white men with the obese and random sample derived from the same genetically quite homogenous population, hence eliminating population stratification. The study design and size of the study population implied that the confidence limits on the non-significant associations were fairly narrow, meaning that presence of any true major associations related to the FTO rs9939609 genotype are very unlikely. A limitation of our study relates to the fact that the biology of the FTO rs9939609 is not fully known. Further, the investigated SNP is located in the first intron of the gene, which according to the current knowledge may not be biologically functioning, so even though only speculative, there is a possibility that this SNP is in linkage disequilibrium with a functional variant within the FTO or nearby genetic regions [19] . Moreover, the inflammatory markers were measured only once, which implies that the levels were subject to random variation due to fluctuations over time. Repetitive measurements would potentially provide a more stable mean measure for the single individual. However, random variation would only introduce bias if this variation were dependent on genotype, which we do not suspect. In summary, we found positive associations between hs-CRP, leptin and BMI; however, the associations did not differ across FTO rs9939609 genotype. There were no associations between IL-1β, IL-6, IL-10, IL-18, mip1α, mip1β, sTNFα-R1, TGF-β, and TNF-α or FTO genotype and BMI, respectively, in this cohort. All together our results do not support the tentative evidence that the FTO rs9939609 A-allele have fatness-independent effects on systemic inflammation.

Conceived and designed the experiments: EZ TJ TIAS. Performed the experiments: EZ TJ TIAS. Analyzed the data: EZ. Contributed reagents/materials/analysis tools: KS DMH AA TH OP. Wrote the paper: EZ TJ TIAS. Critically revised the manuscript for intellectual content: KS DMH AA TH OP. Background A recent study reported that the fatness associated A-allele of FTO rs9939609 increased plasma high sensitivity C-reactive protein (hs-CRP) levels independent of fatness. We aimed to investigate if this gene variant had fatness-independent effects on plasma hs-CRP and 10 additional circulating obesity-related adipokines throughout a broad range of body mass index (BMI) among Danish men. Methodology/Principal Findings In a population of 362,200 young men, examined for military service between 1943 and 1977, two groups were identified: 1) a random 1% sample and 2) all obese men (BMI = 31.0 kg/m 2 , all of whom were above the 99 th percentile of this population). At an average age of 49 years (range: 39 through 65 years), 551 men, hereof 231 of the obese, were re-examined, including genotyping and measurement of the fasting circulating inflammatory markers hs-CRP, IL-1β, IL-6, IL-10, IL-18, mip1α, mip1β, sTNFα-R1, TGF-β, TNF-α and leptin. Men with known disease were excluded from the examination. All the inflammatory markers were log-transformed to approximate a normal distribution. Genotype-phenotype relationships were studied using linear regression analyses with the inflammatory markers as the response variable. Significant positive associations between hs-CRP, leptin and a broad range of BMI were observed, but the associations did not significantly differ across FTO rs9939609 genotype. There were no significant associations between the other inflammatory markers, FTO rs9939609 genotype or BMI, respectively. Conclusion No fatness-independent effects of the FTO rs9939609 A-allele on a series of inflammatory markers were observed in this cohort of healthy middle-aged men representing a broad range of fatness.

Supporting Information

CC BY

no

2022-01-13 08:14:21

PLoS One. 2011 Jan 5; 6(1):e15958

oa_package/f0/25/PMC3016333.tar.gz

PMC3016334

21246033

Introduction Medical schools are responsible for preparing undergraduate students to become professionals. In this context, a powerful connection between teaching and learning is essential for an adequate education. Most medical schools assess students' performance by conferring scores. Recent studies have reported the use of large score databanks to infer future professional performance [1] . Although the association between performance in medical courses at school and professional performance is controversial [2] , a relationship has been found between the scores of medical students in basic medical science courses taken in the first year of medical school and the subsequent performance of these students during and beyond medical school [3] . Thus, student performance in early stages of medical education could be a good predictor of their overall success in other courses. Furthermore, many medical students commonly present learning difficulties and need some pedagogical intervention to improve their performance. Identifying these students at early stages of their medical studies may invite extra attention or necessary efforts to improve their performance. Students' scores must also discriminate between students who need supplemental effort for their education and those who do not in order to achieve their maximum analytical potential [3] . Each medical course develops its own evaluation system for students. Different assessment patterns could impact the need for changes in teaching methods over time. Even if not intended, evaluation performance may vary according to substitution of teachers throughout the years and also the role of each teacher in a given course. The scores conferred in the same course over time may help to discriminate intended from unintended changes in teaching methodology and/or evaluation systems. It may also help to identify external factors that may influence student performance. Properly conducted overall student assessment is a great tool to highlight student skills. An ideal evaluation system should be as precise as possible to measure differences among students at different levels of achievement [4] . The aim of this study is to use a large score databank to identify students with low performance in the early semesters of medical education. These students may benefit from special attention to their needs to improve learning and yield better results on their scores. Additionally, this study performs an analysis of the pattern of mean scores of each course over a period of 10 years, aimed at identifying the trend for each course. This approach may help the department involved to evaluate the changes introduced during the semester. These findings may be important for guiding individualized assistance for most students who need help. They may also suggest innovative educational polices to improve overall performance in the medical course.

Methods In this report we present an observational, retrospective study from 1994 to 2003 of all students regularly registered in the mandatory courses of the curriculum of the School of Medicine, Federal University of Bahia (UFBA), which is a very traditional Brazilian medical school that registers 80 new medical students per semester. Within this medical school, the assessment of student performance is registered using scores from zero to ten, which are stored in the collegiate databank. The data collected consists of the registration numbers and scores in courses taken between the first semester and eighth semester, which is the final period before the four semesters of medical internship. The study identification number allowed for the identification of all the scores obtained from a single individual in each course without revealing the name of the student. Since an individual informed consent was not obtainable, university authorities' permission was sought and granted after an ethics analysis. Medical courses in Brazil last for an average of twelve semesters. The first eight are intended for theory-practice learning. The basic science subjects are concentrated in the first period. The last four semesters are performed as supervised practice, known as surgical and clinical rotations, in different medical specialties. During the first, fifth, sixth and seventh semesters students are enrolled in four courses per semester; in the second, third and eighth semesters there are five courses in each one; three courses comprise the fourth semester. Every course lasts for one single semester. Within the medical school in which the present study was performed, the same method of assessment is used to evaluate all the eight semesters in the medical school. Such method is based on theoretical open written questions or objective multiple-choice questions. In those courses in which some degree of practical activity is a matter of assessment, such as anatomy or histology, standardized practical evaluations are also performed, however with lower impact in the final score. During the period of this study, the same style of assessment was maintained, as well as there were no changes in the medical curriculum. During the study period, a total of 2,398 medical undergraduate students were registered in mandatory courses from the initial eighth semesters. The first objective of the present study was to test the prediction of the students' academic performance in the second and eighth semesters using scores obtained in the first semester. Thus, all the students who had scores from the first, second and eighth semesters registered in the score databank within the years of the study were included. This sub-sample totalized scores from 1,071 students (44.66% of all the medical students registered). The cohorts of students followed up have been selected by the same style of admission tests to start graduation, so changes in the student body were negligible. In addition, the mean summated scores of students in each academic year for a summation of all courses/semesters were relatively stable, ranging from 6.5 to 7.0. We tested correlations between the mean scores of these students in the first semester and their scores in the second or eighth semester to demonstrate if the performance in the beginning was consistent with the performance in the remainder of the course. The correlations were calculated using the Pearson coefficient. The students were grouped according to their scores into two categories based on the possible need for individual assistance. Students whose ratings remained in the lower quartile in two or more courses were classified as the low-performance group and those who had no courses or only one course in the lower quartile were nominated for the high-performance group. Students in each of these groups had their performance evaluated in the second and eighth semesters to assess the correlation between student performances at the beginning and at the end of the theoretical part of medical studies. Therefore, linear regression was used for the estimation of the causal relation (relative risk) between the performances in the first and second or eighth semesters. In an additional approach, we calculated the cut-off values for the scores obtained in the first semester that could predict a low performance in the second or eighth semesters using Receiver-Operator Characteristics (ROC) curves with C-statistics. In an attempt to build the ROC curves, we estimated the students' performances in each semester by averaging the scores from the courses of each period. The cut-off values represent the score mean in the first semester that presents the highest sensitivity, specificity and likelihood ratio for predicting a low performance in the second or eighth semesters (scores within the lower quartile). The second objective of this study was to follow each course's long-term pattern, the mean of scores conferred in a semester was compared to the overall course mean (±1 standard deviation, SD) over 10 years. Hence, data from all the 20 registered classes were included (n = 2,398), as this additional approach does not require follow up of a specific student. A summary of the sampling approaches used for the different analyses are shown in the Figure 1 . All data were analyzed using the GraphPad Prism 5.00 (GraphPad software, San Diego, CA). Differences were considered significant at p<0.05.

Results Considering all the registered students, there was a strong and statistically significant correlation between the mean scores obtained by students in their first semester and those in the second semester ( Figure 2A ; r = 0.5851, 95% CI:0.5507–0.6175; p<0.0001). Interestingly, there was a positive correlation between the scores in the first semester and those in the eighth semester ( Figure 2B ; r = 0.4144, 95% CI:0.3616–0.4647; p<0,0001). We decided then to evaluate on an individual basis the relationship between the performance in the first semester to that in the second and the eighth semesters in terms of the relative performance of other students in the same period. For such an assessment we considered two groups of students: those who stayed in the three highest quartiles in all or all-but-one course, and those who remained in the lowest quartile in two or more courses. The performance of medical students in their first semester of the Medical School correlated well to their performance in the second or eighth semesters, as shown in Table 1 . Individuals whose performance appeared in the lower quartile in at least two courses of the first semester had a high risk of being in the lower quartile of at least one course in the second semester, with a relative risk of 3.907 (95% CI: 3.378–4.519). Furthermore, students who were in the lower quartile in their first semester also had a higher risk of being in the lower quartile in the eighth semester with a relative risk of 2.873 (95% CI: 2.495–3.308). An additional analysis using ROC curves revealed that an average grade in the first semester of 7.188 can serve as a good cut-off value to predict a low performance in the second semester (Area Under Curve: 0.8098; sensitivity: 71%; specificity: 75%; likelihood ratio: 2.87; p<0.0001) and also in the eighth semester (Area Under Curve: 0.7380; sensitivity: 70%; specificity: 65%; likelihood ratio: 2.0; p<0.0001) ( Figure 3A and 3B ). When scores conferred by single courses were compared over a period of 10 years (most with a total of 13 groups of students), three time-trend patterns emerged. A group of courses had a consistent upward trend in mean scores conferred, as shown in Figure 4A . Moreover, in another group of courses, no clear pattern was noted (erratic trend; Figure 4B ). It is noteworthy that no downward trend was observed. Yet in another group of courses, very low variations in terms of the mean of attributed scores were exhibited ( Figure 4C ). Finally we assessed the courses in which students failed in the second and eighth semesters when they had failed in the first semester. The students identified as having a low performance in the first semester failed in different courses in the second and eighth semesters, with no clear pattern. These courses from the second and eighth semesters presented no direct relationships and the assessments were also not similar, with some courses presenting the low variation pattern while other presented erratic or upward trends. This observation suggests that the students with low performance do not have a specific deficiency with one aspect of a certain subject.

Discussion Upon entering medical school, students face a new academic environment that requires a new way of life, as well as time and dedication. However, many cannot adapt to this change and end up performing poorly. The present report shows that low performance in the first semester serves as an indicator of lower achievement in later semesters of the medical course. Therefore, identification of students with learning difficulties is essential to provide them with special pedagogical attention. Frequency of attendance was related to academic performance in an introductory structure-function course for first-year students [5] . Frequent attendees earned higher scores on the comprehensive examination and higher final scores than those of their sporadic attendee counterparts [5] . The databank used by us did not allow for this analysis, but the frequency of attendance is certainly a point that needs to be evaluated for students with lower scores in several courses. Predicting the academic performance of medical students is an object of intense interest. It has been shown that 23% of variance in medical school performance can be explained by previous academic performance and 6% of variance in postgraduate performance can be explained by previous academic performance [2] . Many educators consider the US Medical College Admissions Test (MCAT) to be the “gold standard” for predicting success in medical school [6] , [7] . However, in a different study, MCAT scores did not demonstrate a significant correlation with the final score, pretest scores, or first-year through second-year score averages [5] . Furthermore, correlations of the admissions measures in MCAT with clinical performance during graduation were quite weak and couldn't be a reliable predictor of clinical skills [8] . The Graduate Australian Medical School Admissions Test (GAMSAT) has also been considered a poor predictor of academic performance [9] . As mentioned above, performance on admission tests does not seem to be a good predictor of medical students' performance during their studies. In this context, there is also evidence that the language and the pre medical school educational system can influence the student's performance in both admission tests and during the medical school [10] . The present report does not address the issue of selecting students for medical schools. We also did not address the issue of predicting professional capacity based on academic performance, which is also a relevant problem [11] , [12] . Our approach was focused on possible advantages of an early identification of individuals who may benefit from special educational measures. Gauging the performance of medical students at the beginning of their medical education can provide better parameters to predict their yield during the remainder of their studies. Teaching methods and pedagogical arrangement might not be uniformly efficient and may unequally benefit some types of students, preventing them from reaching their maximal potential. Low student performance could be correlated with poor study skills. Learning style covers both motivations for learning and the processes by which the student approaches the task of learning [2] . An individualized approach can help otherwise capable students to overcome their learning difficulties. Another point to consider is the reported significant positive association between the use of strategic learning and final scores [13] , [14] , [15] , as well the association of a consistent learning with performance in examinations [15] , [16] . Additionally, students who emphasize the deductive method tend to perform better than those with any other style [13] , [17] . It may therefore be useful for medical educational programs to inform students how to successfully use their study skills [18] , [19] . Students in both groups, high and low performance, should be evaluated to verify if there are differences in their methods of studying and learning. Stimulating and encouraging students in the low-performance group to seek a more efficient method of studying can be of great help, as seen with peer-assisted learning, which motivates the learner to spend more time preparing, possibly resulting in deeper learning [20] . As demonstrated here, the overall performance of medical students in the first semester can predict their performances in subsequent semesters with a considerable relative risk (3.907 for the second and 2.873 for the eighth semester). To validate this observation, ROC curves were built using scores from the first semester and the performance outcomes in the following semesters. The analysis using ROC curves led us to identify a high cut-off score (7.188) in the first semester that was able to differentiate students with high performance from the ones with low performance during further semesters of medical school. This fact demonstrates that the medical students included in our sample had an overall good performance. One of the goals of medical school is to encourage students to develop a maximum theoretical learning ability and practical performance. A cut-off value can help the departments of the medical schools to identify those students who need special pedagogical intervention. Nevertheless, a cut-off value does not necessarily establish the capability or quality of their professional future. Another aspect to highlight is the possibility of monitoring course scores, as mean and deviation, over a period of time. If the students' assessment in a course is monitored over time, the teachers and coordinators can discriminate between desirable and unwanted changes. We have shown that some courses have a constant increasing mean score over time, suggesting either a gradual softening in assessment, or a benefit of a maintained increase in the quality of teaching. In other courses, the students' average is always in the narrow range (±1SD of global media). This trend may be seen as a constant and equitable pattern of assessment, but it may also reflect the lack of a policy for improvement from the teachers. A marked oscillation of data around a global average suggests that there are periods with milder and others with more rigorous evaluations, as well as the absence of a steady teaching pattern. In any case, it is clear that courses must follow a stringent standard, maintain a good quality of teaching and assessment of the students, and observe time trends as a means to help implement adequate changes. This study presents some limitations. The course content and assessment difficulty may have changed over the 10-year period. It limits the grouping of scores from students of different semesters. Performance of the students in course examinations often does not reflect their ability and professional skill. Thus, performance in the first semester is just a predictor of student performance in the rest of the theoretical part of the course and does not reflect necessarily their future medical performance. Our study has at least two implications. Firstly, students' low performance in courses offered in the first semester of course is an acceptable parameter to predict their performance in subsequent courses. We propose that students with low performance at the beginning of their course merit special attention to minimize the risks of a continuous limitation. Secondly, an evaluation of the time trend of scores conferred by courses may help departments to monitor changes in personnel and methodology that may affect a student's performance.

Conceived and designed the experiments: VRRdM BBA AA MB-N. Performed the experiments: VRRdM BBA. Analyzed the data: VRRdM BBA AA MB-N. Wrote the paper: VRRdM BBA MB-N. Current address: Laboratory of Parasitic Diseases, National Institutes of Allergy and Infectious Diseases (NIAID), National Institutes of Health, Bethesda, Maryland, United States of America Background Basic courses in most medical schools assess students' performance by conferring scores. The objective of this work is to use a large score databank for the early identification of students with low performance and to identify course trends based on the mean of students' grades. Methodology/Principal Findings We studied scores from 2,398 medical students registered in courses over a period of 10 years. Students in the first semester were grouped into those whose ratings remained in the lower quartile in two or more courses (low-performance) and students who had up to one course in the lower quartile (high-performance). ROC curves were built, aimed at the identification of a cut-off average score in the first semesters that would be able to predict low performances in future semesters. Moreover, to follow the long-term pattern of each course, the mean of all scores conferred in a semester was compared to the overall course mean obtained by averaging 10 years of data. Individuals in the low-performance group had a higher risk of being in the lower quartile of at least one course in the second semester (relative risk 3.907; 95% CI: 3.378–4.519) and in the eighth semester (relative risk 2.873; 95% CI: 2.495–3.308). The prediction analysis revealed that an average score of 7.188 in the first semester could identify students that presented scores below the lower quartiles in both the second and eighth semesters (p<0.0001 for both AUC). When scores conferred by single courses were compared over time, three time-trend patterns emerged: low variation, upward trend and erratic pattern. Conclusion/Significance An early identification of students with low performance may be useful in promoting pedagogical strategies for these individuals. Evaluation of the time trend of scores conferred by courses may help departments monitoring changes in personnel and methodology that may affect a student's performance.

The authors thank Mr. Aleksey Novikov for critical review of the manuscript and Dr. Jorge Guedes for helpful discussions. We thank Dr. Antonio Augusto Silva for help with the statistical analysis.

CC BY

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15695

oa_package/8e/5c/PMC3016334.tar.gz

PMC3016335

21246034

Introduction Pulmonary arterial hypertension is a heterogeneous group of disorders characterized by increased pulmonary artery pressure and resistance as a result of pulmonary vascular remodeling, active vasoconstriction, and in-situ thrombosis. The increased pressure load results in right ventricular hypertrophy representing an initial stage which can progress to failure and death [1] . The current treatments with prostaglandin (PGI 2 ) analogues (e.g. epoprostenol), endothelin receptor antagonists (e.g. bosentan), and phosphodiesterase 5 inhibitors (e.g. sildenafil) have markedly improved the prognosis of the disease [2] . Direct inotropic effects of epoprostenol on the right ventricle was described [3] , [4] and that bosentan reduces cardiac fibrosis and hypertrophy [5] . However, these treatments are mainly vasodilatory and antiproliferative, and hence targeting the vasculature rather than the right ventricle in pulmonary hypertension. Parameters of right ventricular function such as cardiac index and mean right atrial pressure are the most important determinants of survival in pulmonary arterial hypertension [1] . Multiple signal transduction pathways are known to be involved in the remodeling of the heart [6] . Although it is a matter of debate which mechanisms are important for transition to failure and hypertrophy, failure of the left ventricle are characterized to varying degrees by changes in extracellular matrix composition, energy metabolism, contraction, adrenergic signaling, and calcium handling [7] . In the right ventricle from chronic hypoxic rats gene expression studies have suggested a switch of metabolic genes suggesting that the hypertrophic right ventricle changes from fatty acid to glucose oxidation [8] , and a recent microarray study of the right ventricle from rats with monocrotaline-induced pulmonary hypertension suggested that pro-apoptotic pathways and intracellular calcium handling enzymes play a role for development of failure [9] – [11] while growth genes such as mitogen activated protein kinase (MAPK) are pivotal in compensated hypertrophy [9] . However, in contrast to the thick-walled left ventricle, the right ventricle has a concave thin wall opposite to the convex interventricular septum, and the anatomic response to pressure overload of the right ventricle is different from the left ventricle [1] , hence suggesting that other signaling pathways may play a role for development of right ventricular hypertrophy in response to pressure load. Global gene analysis has been employed to map the expression profile of cardiac hypertrophy in man [12] and in the lungs and peripheral blood cells from patients with severe pulmonary arterial hypertension [13] , [14] as well as in lungs of mice with hypoxic pulmonary hypertension [15] . These types of global gene analyses are believed to be of significant value both for understanding and predicting disease processes also in pulmonary hypertension [16] . The present study investigated the changes in global gene expression by gene chip analysis during the development of right ventricular hypertrophy induced by chronic hypoxic pulmonary hypertension in rats. Most of the regulated genes in the hypoxic model were expected to be associated to the adaptive response to sustain right ventricular output, but some may be exclusively associated to hypoxia. Therefore, gene expression changes were also analyzed in rats undergoing pulmonary trunk banding (PTB), another animal model for pressure loading of the right ventricle. The alterations in expression of a subset of genes were confirmed by quantitative realtime polymerase chain reaction (qPCR), immunoblotting, and immunohistochemistry.

Methods Ethics Statement All animal procedures followed the revised NIH publication no. 86–23, entitled: “Principle of laboratory animal care”, and were performed according to the Danish legislation with permission (no. 2006/561–1160) from the Animal Surveillance Committee, The Danish Ministery of Justice. Animal models of right ventricular hypertrophy 48 male Wistar rats (10 weeks, weight approximately 290 grams) were divided into two groups of 24 animals. Each group was divided into subgroups of six animals. Four subgroups were maintained at hypobaric, hypoxic pressure. Four subgroups were maintained at normobaric, normoxic pressure and served as controls ( Table 1 ). The hypoxic group was placed in a hypobaric chamber, where the ambient pressure was continuously held at 500 mbar which was equivalent to rats breathing 10% oxygen at standard pressure. The temperature in the chamber was maintained at 21–22°C and the chamber was ventilated with air at approximately 45 l min −1 through an inlet valve with the aid of a vacuum pump. The four subgroups were maintained in these hypoxic, hypobaric conditions for 1, 2, 3 or 4 weeks and studied immediately after removal from the chamber. The hypoxic chamber was opened once a week for approximately 30–40 minutes for cleaning and supplying purposes. Age-matched controls were maintained in similar but normoxic, normobaric chambers. All animals were provided with chow and water ad libitum. For the PTB experiment, 46 male Wistar rats were divided into two groups. One group underwent PTB operation, and the other group underwent sham operation. Sham operated animals underwent the same procedure as PTB rats except for banding of the pulmonary trunk. Animals were subdivided into 4 groups and examined 2, 3, 5 and 6 weeks after the operation. 4 Sham and 4 PTB animals were randomly selected for gene chip analysis ( Table 1 ). The animal model for pressure loading of the right ventricle using PTB has been described in detail in a previous publication [17] . In the experiment with chronic hypoxic rats, body weight (BW) and systemic systolic blood pressure (SSBP) of the animals were measured on the day of sacrifice. SSBP was measured using the tail cuff method with a plethysmograph (Digital Pressure Meter LE 5000 with tail cuff). Before measurements, rats were preheated for 20 minutes at 35°C in their cages. Rats were then placed in a heated Plexiglas tube (35°C). In each rat, three subsequent measurements within a difference of 5 mmHg were made, and the mean values were used. For measurement of right ventricular systolic pressure (RVSBP), rats were fully anaesthetized with an intraperitoneal injection of Midazolam 0.825 ml/kg, Fentanyl 0.825 ml/kg and sterilized water 0.825 ml/kg. For maintenance, intraperitoneal injection of 100 μL Fentanyl was given every 30 minutes. The external jugular vein was isolated by blunt dissection and ligated (using Seralon® 3/0 – Serag Wiessner). A small hole was cut in the vein and through this a catheter (30 cm. Tygon micropore, OD: 0.76 mm, ID: 0.25 mm, Norton performance plastics, OH, USA) slightly bended in one end, was inserted and led into the right ventricle through the right atrium (marks on the catheter at four to five cm indicates this location). The pressure profile was simultaneously registered via a pressure transducer (MLT0699 Disposable BP Transducer, ADInstruments, CO, USA), an amplifier (ML118G Quad Bridge Amp, ADInstruments) and through a signal box (PowerLab 4/20, ADInstruments) registered on a computer (software: Chart v5.0, ADInstruments). Once the catheter was in place, the pressure was left to stabilize over a period of approximately 5 to 10 minutes. The rats were sacrificed by decapitation. The hematocrit was measured post mortem by centrifuging the blood (Micro-Hematocrit Centrifuge, model MB, 11,700 RPM, 13,700 × g, International Equipment Company, MA, USA) for 10 minutes. In the PTB experiment, invasive pressure measurements were performed as previously described [17] . Assessment of right ventricular hypertrophy Immediately after the rat was decapitated, the heart was removed and placed in diethylpyrocarbonate (DEPC) treated water. For gene expression analysis a piece offour times four mm of the free wall of the right ventricle was cut out, and placed in RNA later (Ambion Inc., TX, U.S.A.). The atria of the heart were removed and the right ventricle separated from the left ventricle and septum. Right ventricle (including the piece cut out for gene expression analysis) and left ventricle was weighed separately. Right ventricular weight to body weight ratio (RV/BW) and the entire heart weight to body weight (HW/BW) were calculated. Morphometric measurements Tissue from the right ventricle was stained with reticulin that is useful to outline the architecture. The myocardial cell was considered to represent a near cylinder with a nucleus placed in the centre of the cell. Measurements were restricted to the nuclear areas. The smallest diameter was chosen as the one representing the actual diameter. The measurements were done on as many cells as possible in a field of vision. Only cells represented by a nucleus were measured. The fields of vision were randomly picked. Gene chip analysis The gene expression profiles was obtained by the use of Affymetrix GeneChips (GeneChip® Rat Genome 230 2.0 Array, Affymetrix Inc., CA, U.S.A.). Protocols for the analysis of Affymetrix GeneChips and the evaluation of the sensitivity and quantitative aspects of the method have been previously described [18] . The raw data files are available in European Molecular Biology Laboratory, European Bioinformatic Institute, Microarray Informatics, ( http://www.ebi.ac.uk/miamexpress ). 293 genes were chosen for analysis in the hypoxic and PTB experiment (see details in the statistical analysis section). qPCR analysis In the present project we used an Applied Biosystems 7000 Real-Time PCR System (Applied Biosystems, CA, U.S.A.) and probes composed of LNATM molecules (Exiqon, Denmark). LNA (Locked Nucleic Acid) is a bicyclic nucleic acid where a ribonucleoside is linked between the 2′-oxygen and the 4′-carbon atoms with a methylene unit (CH 2 ). As with the DNA probes each of the 90 LNA probes in the ProbeLibraryTM kit is labeled with fluorescein ( reporter ) and a non-fluorescent quencher . Primers (forward and reverse) and probes used were: adrenergic receptor, alpha 1B (α 1B -AR): 5′-cgtatccttgggtgccagt-3′ , 5′-cacggccggtaggtgtaa-3′ , probe #22; aquaporin 7: 5′-tccccggttcttc-actttc-3′ , 5′-acccaccaccagttgttcc-3′ , probe #15; triadin 1: 5′-gtggactacaaaaacttttcagca-3′ , 5′-cagcatcgttcactagttttagagg-3′ , probe #20; CD151 antigen: 5′-acggaacctgttacgcttgt-3′ , 5′-cagca-atgatctccagaagga-3′ , probe #5; tissue transglutaminase (tTG): 5′-agctggagagcaacaagagc-3′ , 5′-gcctggtcatccaggactc-3′ , probe #9; transforming growth factor, beta 1 (TGF-β1): 5′-cctgcc-cctacatttgga-3′ , 5′-tggttgtagagggcaaggac-3′, probe #73; Ubiquitin C: 5′-caggacaaggagggcatc-3′ , 5′-gccatcttccagctgctt-3′ , probe #90; myosin regulatory light chain: 5′-cttcgcttgcttcgatgag-3′ , 5′-gtgagcagctccctcaggt-3′ , probe #31; clathrin, light polypeptide (Lcb): 5′-tggagagaggagc-agaagaaa-3′ , 5′-cactcctgttcggtcacctt-3′ , probe #89; Similar to C11orf17 protein: 5′-ctcagactc-ggggcacag-3′ , 5′-atgcttcctggaccaacaga-3′ , probe #40. Immunoblotting Frozen samples of the right ventricle from chronic hypoxic and normoxic rats were homogenized, extracted in 300 μl lysis buffer (20 mM Tris-HCl, 5 mM EGTA, 150 mM NaCl, 20 mM glycerolphosphate, 10 mM NaF, 1% Triton X-100, 0.1% tween-20, 0.02 mM ortho-vanadate, 40 nM PMSF, PIM) and centrifuged at 3000 rpm at 4°C for 15 minutes. The supernatant was removed and protein concentration was measured by double determination of the samples against a standard curve with known concentration of albumin using Bio-Rad Protein Assay (Bio-Rad Laboratories, CA, USA). Protein lysate was mixed with sample buffer (350 mM Tris-HCl, 10% dithiothreitolsodium-laurylsulfate, 30% glycerol, 0.123% bromphenol blue). Samples, containing equal amounts of protein, and Protein Stain marker (Bio-Rad Laboratories) were loaded to the gel (Criterion XT Bis-Tris gel 4–12%, Bio-Rad Laboratories) and separation was carried out in the Criterion Electrophoresis System (Bio-Rad Laboratories) using XT-MOPS as a running buffer. The proteins were transferred from the gel to a polyvinylidine difluoride membrane for 1 hour at 100 V. The membrane was blocked in 5% milk for 2 hours at 20°C and afterwards incubated with primary antibody diluted in 5% milk over night at 4°C. The membrane was washed four times in TBS-T (stock: 10 mM tris/base, 2M NaCl, 1 mM EDTA 0.1% tween-20–200 ml stock, 1800 ml MQ-water, pH adjusted to 7.5) and then incubated in secondary antibody for 2 hours at 20°C. Then again washed four times in TBS-T and finally the blots were detected by enhanced chemiluminiscence system (ECL Plus Western Blotting Detection System, GE Healthcare UK Ltd., UK). Differences in protein abundance were determined by densitometry (Image Quant TL; Amersham Biosciences, UK). The known amount of loaded protein was used as loading control. Antibodies Primary antibodies used were: anti-acetyl-Coenzyme A acyltransferase 2 (ACAA2) (cat.nr. 11111-1-AP, Ptg.Lab, IL, USA), anti-hydroxyacyl-Coenzyme A dehydrogenase/3-ketoacyl-Coenzyme A thiolase/enoyl-Coenzyme A hydratase (trifunctional protein), alpha subunit (HADHA) (cat.nr. 10758-1-AP, Ptg.Lab, IL, USA), anti-aquaporin 7 (cat.nr. AB3076, Chemicon International, MA, USA), anti-monoamine oxidase A (MAOA) (cat.nr. 10359-1-AP, Ptg.Lab, IL, USA), anti-tissue transglutaminase (tTG) (TGase II Ab-2 (TG-100), Thermo Scientific, CA, USA), anti-endothelin receptor B (ET B ) (cat.nr. ab1921, Abcam, UK). Secondary antibodies used were: horse radish peroxidase (HRP) conjugated goat anti-rabbit IgG (Santa-Cruz biotechnology, CA, USA), HRP conjugated goat anti-mouse IgG (Zymed, CA, USA), HRP conjugated rabbit anti-chicken (ab6753, Abcam). Immunohistochemistry Pieces from the right ventricle from normoxic and hypoxic rats were stored in formalin 4% until they were embedded into paraffin. Slices of approximately 3 μm were made from each piece. Removing paraffin from the slices was done by washing in different solutions each for 5 minutes following the schedule: 2 x xylene (Xylene – mixture of isomers (AppliChem, Germany)), 2×99% ethanol, 96% ethanol, 50% ethanol and water. The slices were incubated in 3% H 2 O 2 for 10 minutes and afterwards washed 2×5 minutes in Coon's buffer (Na 2 H 2 PO 4 (2H 2 O), NaH 2 PO 4 (H 2 O), NaCl – ad. 10,000 ml H 2 O). Slices were then pretreated with either citrat buffer (10 mM (Tri-sodiumcitrate dihydrat 5 mM, dinatriumhydrogencitrat 5 mM, pH adjusted to 6.0)) or TEG buffer (Tris 10 mM, EGTA 0.5 mM, pH adjusted to 9.0) for 2×5 minutes in microwave at 650 W and afterwards washed in Coon's buffer. 10% bovine serum in solution of 1% bovine serum albumin dissolved in Coon's Buffer (BSA1%) was added to avoid unspecific binding of the antibodies. The serum was removed and the primary antibody, the same as used for immunoblotting, diluted in BSA1% was added and incubating over night at 4°C in a moisture chamber. The slices were washed in Coon's buffer and then incubating with secondary antibody (Bionylated Link - Universal LSABTM2 Kit/HRP, Rabbit/Mouse, Dako, CA, USA) for 20 minutes at 20°C in a moisture chamber. Again washing with Coon's buffer and incubation with streptavidin (Streptavidin, Dako) for 20 minutes at 20°C in a moisture chamber. After washing with Coon's buffer DAB (DAB Chromogen (tablets dissolved in Coon's buffer), Dako) was added and incubating for 5 minutes. Slices were incubated with Mayer's hematoxylin for 1 minute, washed in H 2 O and incubating for 5 minutes in each solution: 50% ethanol, 96% ethanol and 99% ethanol. Negative controls were made exactly indentical but without incubation with primary antibody. Statistical analysis The data obtained in the hypoxic experiment was firstly filtered to include probe sets with at least one present call in all 28 chips and secondly logarithmic transformed. A t-test for difference between the hypoxic and the normoxic groups, that adds the effects from the four time points (allowing for different general levels at the four time points) was calculated. The corresponding P-value from the appropiate t-distribution was calculated for each gene. The calculated P-values were used to rank the genes within each test. To estimate how many of the called genes that were falsely positively called, we estimated a false discovery rate (FDR), avoiding the assumptions of the data being independent and normally distributed. The FDR was calculated by comparing the original data to a reference distribution based on permutated data. Permutation consists in permutating group labels while keeping the size of the two groups. In this experiment we generated alternative arrangements by exchanging two individuals between the control and the intervention group for each of the four time points. For the whole experiment, there were a total of 104.976 different permutations. 500 randomly picked permutations were chosen, and for each of the 500 permutated experiments a t sum perm. test was calculated, and the corresponding P-values were found. FDR was found by calculating the ratio between the median number of genes based on the permutated datasets with P<0.0001 and the number of genes based on the original dataset with P<0.0001. In the PTB experiment, the genes that were found to be significantly differentially expressed in the hypoxic experiment, were selected in the gene expression dataset belonging to the PTB experiment, and differences between the two groups were analysed using Student's t test. Results from immunoblotting, qPCR, morphometric and hemodynamic measurements from both the PTB and hypoxic experiment were analyzed separately by using two-way analysis of variance (ANOVA) comparing hypoxia or PTB to time. In case of significance a Bonferroni post hoc test was made. Statistical calculations were made by using GraphPad Prism version 4.03 (GraphPad Software Inc, San Diego, CA, USA). All over, P-values of less than 0.05 were considered significant.

Results Functional Data Hematocrit, SSBP, BW and HW/BW are listed in Table 2 . The hematocrit was 39% in the normoxic group. Hypoxia increased the hematocrit to 50–54%. SSBP was unaltered by hypoxia. The normoxic animals gained weight during the four weeks, while the hypoxic rats had a significant lower body weight after four weeks of experiment. HW/BW in the hypoxic group showed a significant increase at week two, three and four. RVSBP was constant in the normoxic group (18–25 mmHg) but raised by hypoxia from 32 mmHg at week one to 40–43 mmHg at week two, three and four ( Figure 1A ). RV/BW was almost unaltered in the normoxic group, while in the hypoxic group RV/BW was increasing during the four weeks ( Figure 1B ). There was a strong positive correlation between RVSBP and RV/BW in the hypoxic experiment (n = 48, R 2 = 0.66, P<0.0001). The diameter of the cardiomyocytes increased significantly (31 to 64%) in the hypoxic group but within each group there was no significant difference between the four different time points ( Figure 2A ). Staining with reticulin did not reveal any other differences between hypoxic and normoxic rats ( Figure 2B ). In the PTB experiment, RVSBP was approximately constant in the sham group (26 to 38 mmHg). RVSBP was significantly raised in the PTB group at all time points (76 to 118 mmHg) ( Figure 3A ). Temporal change in right ventricular hypertrophy was assessed by RV/BW ratio. The degree of right ventricular hypertrophy in the PTB group was increased significantly at all time points compared to the sham group ( Figure 3B ). There was a strong correlation between RVSBP and RV/BW (n = 46, R 2 = 0.76, P<0.0001). Gene expression data Table 3 contains results for 49 genes from gene chip analysis for genes up- or down-regulated with a fold change greater than two. The validation of the gene expression profiles was done by qPCR. We chose four upregulated (CD 151 antigen, tTG, TGF-β1 and Similar to C11orf17 protein), three downregulated (α 1B -AR, aquaporin 7, and triadin 1) and two genes without change (myosin regulatory light chain and clathrin – light polypeptide). Ubiquitin C was used for normalization. The qPCR results were in accordance with the gene chip data ( Figure 4A,B ), exemplified by aquaporin 7 and tTG. The correlation between the gene chip and qPCR results was tested by comparing 3 up-regulated, 4 down-regulated and 2 genes with no change in expression according to the gene chip tsum analysis. There is a strong positive correlation between gene chip detection (normalized to 0) values and the qPCR detection values (normalized to 0) and in the hypoxic experiment (n = 72, R 2 = 0.70) ( Figure 5 ). The signal log ratio estimates the magnitude and direction of change of a transcript when two arrays are compared (experiment versus control). Signal log ratios from the hypoxic and the PTB experiments were plotted for 288 genes (5 genes were excluded of the original 293 genes tested because they showed absent calls in the PTB experiment). 266 were regulated in the same direction or were unchanged in the two experiments (Wilcoxon signed rank test: P<2.2×10 −16 ). Of these genes, 172 were non-expressed sequence tags (non-EST) involved in apoptosis, inflammation, heart function, and growth and the remaining 94 were EST. Of the 22 genes regulated differently, 11 were up-regulated and 5 unaltered in the PTB experiment but down-regulated in the hypoxic experiment, 3 were down-regulated in PTB and up-regulated in hypoxia and 3 were EST ( Table 4 ). There is a strong positive correlation between the log ratios of gene expression in the hypoxic and the PTB experiment (R 2 = 0.69, P<0.05, n = 288) ( Figure 6 ). Limiting the analysis to genes where the fold change is larger than 2 yielded an even stronger correlation (R 2 = 0.84, P<0.05, n = 125) of gene expression in the hypoxic and PTB experiment. Immunoblotting Figure 7 shows the representative immunoblots. Six proteins were chosen for immunoblotting based on the gene chip that showed three genes were downregulated (ACAA2, HADHA and aquaporin 7) and three were upregulated (tTG, MAOA and ET B ). By use of antibodies against these proteins we showed that some of the changes found on gene levels are also reflected at protein level. ACAA2, HADHA and aquaporin 7 were not significantly downregulated, but the tendency is clear ( Figure 8A–C ). MAOA and tTG were significantly upregulated in hypoxic rats after 2, 3 or 4 weeks when compared to normoxic controls ( Figure 8D–E ). Finally, ET B was only significantly upregulated after 3 weeks of hypoxia ( Figure 8F ). Immunohistochemistry By incubating slides from the right ventricle from hypoxic and normoxic rats with antibodies used for immunoblotting we were able to evaluate the localization of the proteins. Controls were made without primary antibody incubation ( Figure 9A ). For the proteins participating in beta-oxidation, ACAA2 and HADHA, the localization is intracellular in the cardiomyocyte ( Figure 9B–C ). Aquaporin 7 was also located in the cardiomyocyte at the cellular membrane ( Figure 9D ). MAOA was also upregulated on the immunostainings when compared to normoxic controls and the location is intracellular ( Figure 9E ). Finally, tTG was also located to the cardiomyocyte mainly in the cytosol ( Figure 9F ).

Discussion The main findings of the present study are addressing gene expression common for the pressure loading of the right ventricle in both chronic hypoxic rats and rats with banding of the pulmonary trunk. The present study revealed alterations in expression of 172 genes involved in apoptosis, inflammation, heart function, and growth. A small subset of differentiated genes in the hypoxia and PTB groups suggests pressure load as the main contributer to development of right ventricular hypertrophy. GeneChip analysis of the right ventricle was confirmed by qPCR for a subgroup of genes and was further substantiated by measuring protein expression showing a marked upregulation of tTG due to right ventricular hypertrophy. Role of hypoxia versus pressure load in right ventricular hypertrophy in pulmonary hypertension The hypobaric hypoxia model has been used in several studies for induction of pulmonary hypertension in rats and has shown increase of RVSBP and right ventricular weight [19] . Another approach to induce high RVSBP and to increase right ventricular weight is by PTB [17] , [20] . By comparing these two animal models, it is possible to distinguish the isolated effect of respectively pressure load and hypoxia. There was a significant correlation of genes changed in the right ventricle of chronic hypoxic rats with the changes observed in the PTB experiment and this correlation was even stronger when only genes regulated more than two-fold were considered. These findings suggest that pressure load is the main factor altering gene expression in the right ventricle in rats with pulmonary hypertension induced by hypoxia. Comparing chronic hypoxia and PTB identified a small subgroup of genes changing in opposite directions suggesting that hypoxia alone has an impact on the gene expression in the right ventricle. Thus, c-kit receptor tyrosine kinase and DEAD (asparatate-glutamate-alanine-asparatate) box polypeptide 25 were upregulated in hypoxia and downregulated in the PTB experiments, while insulin-like growth factor binding protein 3 and regulator of G-protein signaling protein 2 were upregulated. Both c-kit receptor tyrosine kinase and insulin-like growth factor binding protein 3 have previously been described to be involved in hypoxia-induced angiogenesis [21] , [22] , and insulin growth factor 1 stimulation in mouse fibroblasts resulted in upregulation of DEAD [23] . Therefore, hypoxia seems to directly affect gene expression in the right ventricle. However, in all cases the fold changes in gene expression comparing the PTB and hypoxic experiments with regard to differentially expressed genes are small ( Table 4 ), and it cannot be excluded that differences in pressure load comparing chronic hypoxic rats and PTB rats may also contribute to the differential gene regulation. Previous studies have also provided evidence suggesting that mechanical load of the right ventricle from rats with pulmonary hypertension influences gene expression [11] . Thus, atrial natriuretic peptide expression, probably induced by stretch of the myocardium, was upregulated in the right ventricle from rats with pulmonary hypertension induced by either moncrotaline or hypoxia [8] , [9] , and in agreement with these findings, both natriuretic peptide precursor type A and B were markedly increased in the present study. Genes involved in cell proliferation, the cyclin family of genes and BCl2, were upregulated in the right ventricle of rats with pulmonary hypertension induced by monocrotaline [9] , [10] , and the same was the case for cyclin D1 and D2 as well as BCl2 in the present study. In addition, several signaling processes involving fetal gene re-expression, activation of protein translocation, increase in mass, and enlargement of cell size/volume have been identified as markers of hypertrophy as a response to hemodynamic overload [24] . In the present study the diameter of the cardiomyocytes was increased, and alpha-actin expression was upregulated together with four and a half LIM domains 1 (FHL), and enigma (LIM domain protein). FHL is contained in a complex within the cardiomyocyte sacromere and mice lacking FHL displayed a blunted hypertrophic response suggesting FHL1 to mediates hypertrophic biomechanical stress responses in the myocardium [25] , while the Enigma protein family are Z-line proteins at the border between two sarcomers [26] . Thus, upregulation of a series of genes in the present study also suggest that mechanical load regulate gene expression and results in right ventricular hypertrophy. Role of specific signal pathways changed in right ventricular hypertrophy During development of right ventricular hypertrophy the myocardium changes metabolism to avoid ischemia [27] . Normally the major substrate for heart metabolism is free fatty acids that account for 60–80%. The remaining part comes from metabolism of carbohydrates, but during development of left ventricular hypertrophy and heart failure the ratio alters towards increased carbohydrates as cardiac fuel substrate and augmented mitochondrial respiratory capacity which is considered to play a central role in hypoxia-mediated cardioprotection [27] . A study of gene expression from chronic hypoxic rats showed increased expression of genes associated to glucose metabolism and they also found changes in the left ventricle, which indicates that not only myocardial hypertrophy causes changes, but also chronic hypoxia contributes to altered gene expression [8] . Indeed, in the present study genes encoding for enzymes participating in beta-oxidation of fatty acids (ACAA2 and HADHA) were downregulated in right ventricles from hypoxic rats. The tendency was reflected at protein level, although not significantly and supports that pressure load by itself is able to cause a shift in genes related to myocardial metabolism from free fatty acids to carbohydrates. Aquaporin 7 is a water and glycerol channel that has been found especially in adipocytes and skeletal muscle cells in the human body. The overall function of aquaporins is to maintain cellular water homeostasis [28] , [29] . Studies of aquaporin 7 showed that it is expressed in cardiac tissue from mice, rats and humans [30] . Our results confirmed these findings both by gene chip, qPCR and immunoblotting. Moreover, we found that hypoxia decreases gene expression for aquaporin 7, although this was not confirmed at protein level. Skowronski et al. found that aquaporin 7 is only localized in small vessels in cardiac tissue, and these observations agree with our findings [31] . A downregulation of aquaporin 7 in hypoxic rats may reflect reduced glycerol transport as a consequence of a shift of the metabolism from fatty acids to carbohydrates. Hypertrophy of the ventricle also leads to remodeling of the ventricular wall and altered expression of structural proteins in the myocardium and in the surrounding tissue. Studies of tTG in left the ventricle show association between the expression of tTG and development of ventricular hypertrophy [32] , [33] . The mechanism is primarily through its action as TGase leading to structural changes of actin and myosin, but also more or less through the GTPase activity. tTG is coupled to the α 1B -AR as Gα-protein [34] . Overexpression of α 1B -AR is known to induce cardiac hypertrophy [35] and studies of the expression of the α 1B -AR have shown that it is downregulated on mRNA level in vascular smooth muscle cells from chronic hypoxic animals [36] , and that knock-out of the receptor did not alter development of right ventricular hypertrophy and the increase in RVSBP [37] . The conclusion of these studies is that α 1B -AR is associated to vascular smooth muscle cell proliferation. Our findings show that the α 1B -AR is downregulated in the right ventricle at mRNA level, while the potential coupling protein tTG is markedly upregulated and associated to right ventricular hypertrophy in rats with pulmonary hypertension. The exact role of α 1B -AR is still unknown but it seems to play an adaptional role to avoid development of cardiac hypertrophy according to pulmonary hypertension. Transforming growth factor beta 1 (TGF-β1) is thought to be associated with proliferation of cells during development of hypertrophy and cell division. Studies of rats with pulmonary hypertension and right ventricular hypertrophy induced by monocrotaline showed by qPCR analysis increased levels of TGF-β1 in the right ventricle but not in the left ventricle indicating association to right ventricular hypertrophy [38] . Also immunoblottings of pulmonary arteries from chronic hypoxic rats showed association between TGF-β1 and increased proliferation of vascular smooth muscle cells [39] . These findings indicate that TGF-β1 is associated both to right ventricular hypertrophy and vascular smooth muscle cell proliferation. Our studies support that TGF-β1 appears to play a role in development of right ventricular hypertrophy. MAOA is an enzyme located to the mitochondria of the cardiomyocytes and metabolizes epinephrine, norepinephrine, and serotonin (5-HT). Studies have shown that 5-HT is associated to ventricular hypertrophy by binding to its receptor 5-HT 2B , and that it induces oxidative stress and apoptosis [40] . It has been found that blocking of the 5-HT 2B receptor only partly inhibited the effect of 5-HT, and that inhibition of MAOA prevented the hypertrophic effect of 5-HT [41] . Overexpression of the 5-HT 2B receptor leads to left ventricular hypertrophy. The localization of MAOA has been found to be intracellular [42] , [43] . Our findings indicate an association between right ventricular hypertrophy and the expression of MAOA. Moreover, we evaluated the localization of MAOA and found that it is located to the cardiomyocytes and probably to the mitochondria, which are highly expressed in cardiomyocytes and is the place where catecholamines and 5-HT are metabolized. Reactive oxygen species (ROS), a product from oxidation of 5-HT catalyzed by MAOA, is related to right ventricular hypertrophy and ROS has been found to be located to the mitochondria [44] . This indicates that metabolization of 5-HT and thereby MAOA is located here. The effects of endothelin are mediated by two distinct receptors termed ET A and ET B , where 90% of endothelin receptors belong to the ET A subtype in cardiomyocytes, and their stimulation has a positive inotropic effect [45] . Cardiac ET B receptors may contribute to clearance of circulating endothelin and together with the ET A to cardiac fibrosis and cardiomyocyte hypertrophy [45] , [46] . In the present study only the ET B receptor expression was elevated in the right ventricle as well as expression of several collagens e.g. collagen type 1 alpha 1 and collagen type V alpha 1 ( Table 3 ). The dual ET A /ET B receptor antagonist, bosentan reduces right ventricular hypertrophy in pulmonary hypertension in chronic hypoxic rats [47] , but at present it is unclear whether the block of endothelin clearance and pulmonary vascular dilation by ET B receptors outweigh the beneficial effects of blocking both the ET A and ET B receptors in pulmonary hypertension due to hypoxia. In conclusion, we have found that several genes are altered during development of right ventricular hypertrophy induced by pulmonary hypertension in chronic hypoxic rats. In case of the metabolic genes the effect of high pressure on the right ventricle appears compensated at the protein level, while both expression of genes and proteins of importance for myocardial function and remodelling are altered by the increased pressure load of the right ventricle. These findings imply that treatment of pulmonary hypertension, in addition to reduction of pulmonary vascular resistance, should also aim at reducing right ventricular pressure or by direct effects on the heart limit the organ damaging effects of high pulmonary pressure.

Conceived and designed the experiments: LHM JDB US MK TFØ. Performed the experiments: JDB LHM CDP UKS. Analyzed the data: LHM JDB CDP UKS NEM JLJ US. Contributed reagents/materials/analysis tools: US JLJ MK TFØ. Wrote the paper: LHM JDB CDP UKS NEM JLJ MK TFØ US. Background The present study investigated whether changes in gene expression in the right ventricle following pulmonary hypertension can be attributed to hypoxia or pressure loading. Methodology/Principal Findings To distinguish hypoxia from pressure-induced alterations, a group of rats underwent banding of the pulmonary trunk (PTB), sham operation, or the rats were exposed to normoxia or chronic, hypobaric hypoxia. Pressure measurements were performed and the right ventricle was analyzed by Affymetrix GeneChip, and selected genes were confirmed by quantitative PCR and immunoblotting. Right ventricular systolic blood pressure and right ventricle to body weight ratio were elevated in the PTB and the hypoxic rats. Expression of the same 172 genes was altered in the chronic hypoxic and PTB rats. Thus, gene expression of enzymes participating in fatty acid oxidation and the glycerol channel were downregulated. mRNA expression of aquaporin 7 was downregulated, but this was not the case for the protein expression. In contrast, monoamine oxidase A and tissue transglutaminase were upregulated both at gene and protein levels. 11 genes (e.g. insulin-like growth factor binding protein) were upregulated in the PTB experiment and downregulated in the hypoxic experiment, and 3 genes (e.g. c-kit tyrosine kinase) were downregulated in the PTB and upregulated in the hypoxic experiment. Conclusion/Significance Pressure load of the right ventricle induces a marked shift in the gene expression, which in case of the metabolic genes appears compensated at the protein level, while both expression of genes and proteins of importance for myocardial function and remodelling are altered by the increased pressure load of the right ventricle. These findings imply that treatment of pulmonary hypertension should also aim at reducing right ventricular pressure.

CC BY

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15859

oa_package/6e/cf/PMC3016335.tar.gz

PMC3016336

21246035

Introduction Use of pesticides in crop production has been an important practice in modern agriculture, especially in the Central Valley of California, the most dynamic agricultural region in the world. Pesticide use can lead to severe environmental problems due to their toxicity to humans and many ecosystem organisms. Synthetic pyrethroids have become increasingly popular following outright bans or limitations on the use of cholinesterase-inhibiting insecticides, such as organophosphates (OPs). Previous studies have indicated that the decrease in OP use in California was related to the substitution with pyrethroids [1] . Pyrethroid insecticides are associated with selective potency in insects and relatively low potency in mammals. However, results of exposure monitoring and pesticide illness surveillance suggested that field residues of pyrethroids can cause irritant respiratory symptoms, nausea and headache [2] . Furthermore, pyrethroids are very acutely toxic to fish and invertebrates, with the 10-day LC50 values ranging from 2–140 ng/L in water ( Americamysis bahia and Ceriodaphnia dubia ) and 4–110 ng/g in sediment ( Hyalella azteca ) [3] . Surface water monitoring indicated widespread presence of pyrethroids and associated toxicity in agricultural and urban waterways in California [4] , [5] , [6] . Identifying the distribution of pyrethroids in surface waters and their effects on aquatic organisms is very important in pesticide regulation and water management of pyrethroids. Monitoring data are usually insufficient to characterize the spatial distribution and the main sources of pesticide residues. Therefore, mathematical models are used to simulate the effects of pesticide use, management practices, and environmental factors on pesticide fate and distribution. In addition, the regulatory burden has evolved to currently consider negative impacts of pesticides on aquatic organisms. Detailed information on pesticide residues, such as the magnitude, timing and frequency of peak concentrations, are required to examine the overall ecosystem exposure by the use of pesticides. Therefore, continuous modeling at the field scale is essential for decision making processes to adequately meet regulatory requirements and improve management practices. Recent developments in GIS technology enable the application of field-scale models on a large landscape by incorporating spatially distributed simulations of water and chemical movement in river networks. The integrated systems with field-scale models routing algorithms have been successfully applied to simulate pesticide fate and behaviors in streams. Most of those models were originally designed for the simulation of pesticides in the dissolved phase, indicating appropriate model applications on pesticides with lower adsorption coefficients. With octanol-water partition coefficients (K OW ) values of 10 5 –10 7 , pyrethroids tend to adsorb to soil and sediment rather than remain in the dissolved phase [7] . Therefore, accurate prediction of pyrethroid fate and transport must incorporate the simulations of soil erosion, in-stream sediment transport, and pyrethroid partitioning. Due to inadequate representation of the above hydrologic and transport processes, most of the existing field-scale models and in-stream routing models are not appropriate for predicting environmental behaviors of pyrethroids. For example, RZWQM (Root zone water quality model) were developed for water flow and solute transport, and thus do not simulate soil erosion and adsorbed pesticide removal. Consequently, edge-of-field pesticide fluxes are underestimated, especially for those with strong sediment/soil sorption [8] . In the GLEAMS (Groundwater loading effects of agricultural management system) model, solid-bound pesticide concentration in eroded soil is determined based on a prescribed soil mass per unit runoff volume regardless of the actual soil erosion rate [9] . In addition, many popular routing models are not able to sufficiently capture the dynamics of pesticide partitioning and transport. They either assume a steady state hydraulics (e.g., River and stream water quality model, QUAL2K [10] ), or utilize prescribed suspended solid concentrations (e.g., River water quality model, RIVWQ [11] ). PRZM (Pesticide root zone model) estimates soil erosion based on a modified Universal Soil Loss Equation. In our previous study [12] , PRZM model was coupled with a linear routing model for assessing pesticide dynamics and distribution in crop fields and stream networks. The coupled system provided a suitable modeling platform for determining environmental concentration and toxicity of pyrethroids. However, some model improvements were required. For example, a minor deficiency has been identified in the PRZM algorithm for adsorbed pesticide removal [13] . This paper presents an improved modeling system based on our previous study [12] for simulating the environmental fate and dispersion of pyrethroid insecticides. The specific purposes for the proposed model were to: (a) account for pyrethroid entry into surface water via soil erosion, (b) predict dynamics and distribution of pyrethroids in channel flow and bed sediment, and (c) characterize the toxicity by pyrethroids to sediment-dwelling organisms. This is one of the first studies on the dynamic modeling of pyrethroids at watershed scale, responding to the emerging research need for pyrethroid reevaluation and watershed management planning. Simulation capability of the developed model was demonstrated by applying it to the Orestimba Creek Watershed ( Figure 1 ), an agriculturally dominated watershed in the California's Central Valley, with four pyrethroids of bifenthrin, λ -cyhalothrin, esfenvalerate, and permethrin as test agents.

Materials and Methods PRZM application at watershed scale A geo-referenced modeling system has been developed in our previous study [12] for tracking pesticide transport from its field application to the receiving waters. Pesticide discharges from the soil-canopy system were simulated by PRZM model. PRZM [18] is a one-dimensional dynamic model, primarily designed to predict the influence of climate, land/soil properties, and agricultural management on the physical and biochemical dynamics of pesticides in the environment. PRZM was selected based on its ability to simulate relevant governing processes of pesticide transport and its preferential use by the USEPA for pesticide-associated risk assessment [19] . Pre-calibrated PRZM parameters were recommended in the USEPA Standard Tier 2 scenarios for the major crops throughout the United States [14] . GIS technology was used to extend the PRZM capability for geo-referenced parameterization and application at a watershed scale. Based on a linear routing model, edge-of-field fluxes of water and pesticides predicted by PRZM were routed through stream channels to a downstream location, e.g., a monitoring site. For water transport, stream flows at the routing destination were calculated as the summation of convolutions between PRZM-predicted runoff and corresponding watershed unit hydrograph in each simulation zone. The hydrologic response was presented by a flow-path redistribution function ( U ) based on the first passage time distribution [20] , [21] , where i is a running index for simulation zone, T (s) is the lag time in the flow-path, Δ i (dimensionless) represents the shear and storage effects on the flow, and K i (dimensionless) is the loss factor accounting for evaporation and transmission losses. The same flow-path redistribution functions were also applied in the transport simulation of dissolved pesticides. A pesticide dispersion coefficient was determined as the sum of molecular diffusivity and flow diffusivity for the corresponding flow paths, and pesticide decay rate was calculated from its aquatic half-life. Modeling nodes in the channel network were selected to correspond with tributary/drainage junctions and monitoring locations. The developed model was applied to the Orestimba Creek watershed during 1990 through 2006, with diazinon and chlorpyrifos as test agents. The model yields reasonable agreements with measured data for the stream flow and dissolved pesticide loads [12] . Pesticide transport with eroded soil PRZM is known to inadequately predict pesticide transport associated with soil erosion [13] . In PRZM, soil column was divided into compartments according to user-defined numerical simulation interval of soil depth. Adsorbed pesticide is only removed from the top-most compartment. Therefore, the removal of pesticide in adsorbed phase is primarily a function of soil compartment depth. Small depths of soil compartments, which are likely to be applied by model users to improve the numerical calculations, will result in significantly less pesticide mass removed by erosion, especially for high-sorbing compounds such as pyrethroids. In this study, we improved the PRZM simulation algorithm by introducing a soil-interaction depth (D E ). It is assumed that all soil layers from the ground to the depth of D E were subjected to the soil erosion process. This concept is similar to the extraction model used in transport model for estimating dissolved chemicals in surface runoff. For example, PRZM estimates the amount of dissolved pesticide runoff based on the average concentration of dissolved pesticide concentration weighted by an exponential curve for all compartments from the surface to a depth of 2 cm. The depth D E , which could be initialized and calibrated by users, is independent from the compartment size for numerical calculation, and remains a fixed value during each PRZM simulation run. Weighted average concentration of pesticide adsorbed on soil particles subject to erosion (C S,E , g/g) for all compartments within the depth of D E was first determined as: where j is a running index for compartments, N E is total compartments within D E , C s (g/g) is the concentration of soil-bound pesticides, and w is a return-to-unit weighting function, i.e., sum(w) = 1. The amount of adsorbed pesticide transported out of the field (J ER , g/day) is calculated by: This equation was the same as Eq. (6.13) in the PRZM manual [18] , with X e (ton/day) as the erosion sediment loss, r om as the enrichment ratio for organic matter, and p as a units conversion factor (g/ton). To implement the above equation, source codes of PRZM were modified and the new procedure for determining pesticide removal with eroded soil was: [1] initialize the soil-interaction depth (D E ), and define a weighting curve as a function of depth (w); [2] determine the affected compartments within D E , and calculate weighting factors for each compartment; [3] adjust “B” term in the PRZM numerical solution as: where B (day −1 ) is the diagonal element in the tri-diagonal matrix solution (Thomas algorithm) utilized by the PRZM code for the governing equations of pesticide transport, ELTERM (day −1 ) is the erosion loss term for pesticide balance, and DELT (day) is the simulation time step. Sediment and pesticide transport in stream network Chemical partitioning and degradation in channel transport have been formulated in the linear routing model as described in our previous study [12] . In this study, improvements were made mainly for sediment routing and partitioning/transport of pesticide associated with suspended solids and bed sediment. A concept of sediment transport capacity was applied in this study to predict sediment deposition in the channels. By following the algorithm of Soil and water assessment tool (SWAT) [22] , the maximum sediment concentration (C ss,max , kg/m 3 ) that can be transported from a reach segment is calculated as: where V (m/s) is the peak channel velocity, and SPCON and SPEXP are coefficients to be determined. Sedimentation flux was determined by comparing the initial concentration of suspended solids in a reach at a time step (C ss,0 , kg/m 3 ) to C ss,max . For instance, if C ss,0 >C ss,max , the exceeding amount of suspended solids and associated pesticides in the adsorbed phase would be transported into bed sediment. The resulting sedimentation flux was used to adjust the initial concentration of suspended solids. A similar methodology was applied in the calculation of resuspension fluxes of bed sediment and pesticides by introducing factors for channel erodibility and channel cover [22] . The predicted concentrations of suspended solids were applied to determine pesticide partitioning between dissolved and adsorbed phases in the water column: where F wd (dimensionless) is the fraction of total pesticide of the water column in dissolved phase, K d (L/kg) is the pesticide partition coefficient, and C ss (kg/m 3 ) is the predicted concentration of suspended solids. Pesticide simulation in bed sediment was only conducted for the active sediment layer with user-defined depth. Based on the solid-liquid partitioning, the fraction (F dd , dimensionless) of total sediment pesticide in the dissolved phase was calculated as: with Φ (dimensionless) denoting sediment porosity and ρ s (g/m 3 ) particle density. Pesticide decay and burial in bed sediment were combined and simulated according to first-order kinetics. Pesticide transport flux by sedimentation was calculated as the product of previously determined sedimentation flux of suspended solids and the pesticide concentration in suspended solids. Similarly, pesticide resuspension flux was based on the sediment resuspension flux and the concentration of sediment-bound pesticide. Therefore, sedimentation and resuspension processes for both suspended solids and solid-bound pesticides were simulated dynamically rather than being prescribed. Pesticide diffusion flux (J diff , kg/m 2 /day) between the water and bed sediment was formulated using a multimedia environmental fate modeling approach [23] : where subscripts w and d are for water compartment and sediment compartment, respectively, D wd (kg/Pa/day) is the Mackay-type mass transfer coefficient, the Z's (mol/Pa/m 3 ) are the fugacity capacity of pesticide, δ's (m) are boundary layer depths at the water-sediment interface, and K's (m 2 /day) are pesticide diffusivities. As suggested by the CalTox model [24] , the boundary layer thickness in water side (δ wd_w ) was set as 0.02 m, while that in sediment side was estimated as 318K d 0.683 . Site Description The modeling system newly developed in this study was applied to the field conditions of the Orestimba Creek watershed of California ( Figure 1 ). Located in western Stanislaus County, the creek originates in the mountainous areas of the Coast Range, and discharges into the San Joaquin River. Characterized by heavier textured soils and greater slopes relative to eastside watersheds of the San Joaquin River, the Orestimba Creek watershed represents a worse-than-average condition for pesticide contamination in surface water. Climate and landscape characteristics for the studied watershed were summarized in the previous studies [12] , [25] . The lower reach of the Orestimba Creek flows through agricultural lands in California's Central Valley, the most dynamic agricultural region in the world. Pyrethroids are applied to control a myriad of pests, and in this study the most important crops receiving these insecticides are orchards and row crops. Sediment from this creek was also found to be toxic to sediment-dwelling organisms, most likely because of high levels of pyrethroids [26] . Based on sediment sampling in the Orestimba Creek during 2004 irrigation season, high sediment concentrations of bifenthrin and λ-cyhalothrin were reported with acute toxicity to sensitive aquatic species [27] . In the 2010 Clean Water Act 303(d) report of California, the Orestimba Creek was listed for sediment toxicity, however the source pollutants were not yet fully identified in the report [28] . Pesticide Data Acquisition The case study was based on two monitoring studies of pyrethroid concentrations and aquatic toxicity in streambed sediments of the Orestimba Creek watershed. Our previous monitoring study included 20 sampling sites throughout the San Joaquin River Valley, with 3 sites situated within the Orestimba Creek watershed [29] ( Figure 1 and Table 1 ). Field measurements were conducted for 9 pyrethroids during the irrigation season of 2007. In an associated study, Ensminger et al. [30] collected monthly water and sediment samples at the site OCRR from December 2007 through June 2008, to determine concentrations of organophosphate and pyrethroid insecticides. In addition to chemical analyses, both studies conducted sediment toxicity tests with Hyalella azteca , following the standard USEPA protocols [31] . More details on experimental design, analytical and sediment toxicity methods, and monitoring results for the two studies can be found in Domagalski et al. [29] and Ensminger et al. [30] . Among all analyzed pyrethroids, only those detected at least twice in the two sampling studies, including bifenthrin, λ -cyhalothrin, esfenvalerate, and permethrin, were selected for model application in this study. Table 2 lists the physicochemical properties, reaction half-lives, and toxicity benchmark (as 10-d median lethal concentration, LC50, for Hyalella azteca in sediment) of the simulated pyrethroids. Pesticide application data were retrieved from the Pesticide Use Reporting (PUR) database maintained by California Department of Pesticide Regulation [32] . The PUR database records daily pesticide use by active ingredient and crops for each Meridian-Township-Range-Section (MTRS, or section) following the United States Land Survey System. Simulation Design The improved PRZM and routing simulations were performed to simulate water, sediment, and pesticide transport processes in the Orestimba Creek watershed at a daily time step for the period 1990–2008. The watershed was delineated into sections for the convenience of incorporating pesticide use data from the PUR database. Multiple fields were simulated in each section, based on the contemporary land use mapping in the study area [33] . A soil interaction depth of 1 cm and uniform weighting factors for each soil compartment were used in the case study, as suggested by SWAT documentation [22] . Individual conservation practices were not included in the model configuration. Instead, the model was calibrated based on the field measurements of water flow, sediment loading, and pesticide concentrations. Therefore, the model parameterization and simulation results reflected the overall reduction of pesticide use due to various best management practices (BMPs) implemented in the study area. Channel parameters, including Manning's roughness coefficient, flow diffusivity, coefficients for sediment transport capacity, and depth for active sediment layers, were taken from previous studies in the Orestimba Creek watershed [17] , [25] . Automatic calibration was conducted for the USLE crop factor (USLE_C) to match the measured suspended solid concentrations at the watershed outlet. The calibrated model was assumed to establish a reliable hydrologic framework for the study area, and applied to the dynamic simulation of pyrethroids. Model Evaluation The modeling system has been validated in our previous study for its simulation capacity for stream flow and organophosphate pesticides in the dissolved phase [12] . In this study, therefore, model evaluation was emphasized on transport simulation of the suspended solids and absorbed pesticides. The location of the site OCRR is also gauged by a USGS station (#11274538) for stream flow, suspended solids, and organic carbon in suspended solids [34] . Flow-weighted concentrations of suspended solids and associated organic carbon on a monthly basis were calculated from the measurements at sampling days. In the chemical analysis of pyrethroids, reported results were associated with different method detection limits (MDLs) ( Table 2 ), and chemicals with concentrations lower than MDLs were reported as zeros. In addition, most of the reported concentrations of detectable pyrethroids in the study area were below the nominal reporting limit [29] . Therefore, it is not appropriate to directly compare predicted and observed concentrations for individual pyrethroids. In this study, predicted and observed pyrethroid concentrations were first converted into toxic units (TU), which is based on the assumption of toxicity additivity and is widely used as an estimate for aquatic toxicity. For each sample, the TU value was calculated as a summation of concentrations normalized by the corresponding sediment LC50 on an organic carbon (OC) basis. When no pyrethroids were detectable, the TU value was set as 0.01 for plotting convenience. To evaluate the model efficiency in predicting pyrethroid transport, the predicted TU values were compared with the measured values for each monitoring day at the three sites. It is important to note that the predicted TU values were calculated based on daily average predictions of pyrethroid concentrations, while measured values were from instantaneous samples. The model evaluation also compared the predicted TUs and cumulative mortality of Hyalella azteca to collected bed sediment samples. The cumulative mortality reflected the actual sediment toxicity by all chemicals, including those not analyzed or not detected, in the bed sediment. Thus, comparisons between predicted TUs and mortality were anticipated to provide useful information on the toxicity identification evaluations.

Results and Discussion Evaluation of the improved algorithm The improved algorithm for adsorbed pesticide removal in PRZM was evaluated in a melon crop field with historical bifenthrin applications in the Orestimba Creek watershed. This field is close to the monitoring site OCER (Orestimba Creek at Eastin Road, Figure 1 and Table 1 ), with an average annual bifenthrin use of 0.08 kg per treated hectare. The concentration of eroded bifenthrin was reported as the total amount of eroded bifenthrin divided by the total amount of eroded soil during the simulation period. Results of the improved PRZM in this study were generally invariant with the depth of soil compartment used in the numerical analysis ( Figure 2 ). This confirmed that the improved algorithm removed adsorbed pesticide from all compartments within the soil-interaction depth (D E , see Materials and Methods ), thus the resulting removal was not dependent with the depth of each compartment. As discussed in Materials and Methods , the original PRZM considers only the top-most soil compartment for adsorbed pesticide removal; therefore, the results were very sensitive to the depth. The original PRZM generated similar results as the improved one when the depth of soil compartment was close to D E (1 cm in this case study). Figure 2 demostrated PRZM simulations for soil erosion and associated pesticide removal with depth of soil compartment up to 1 cm. In the real PRZM modeling, however, small depth was required for accurate numerical simulation of water and chemical movement in the soil. For example, all crop scenarios for PRZM require depth of 0.1 cm or less for the top soil horizon developed by U.S. Environmental Protection Agency (USEPA) [14] . With small depth of soil compartment, the original PRZM significantly underestimated the adsorbed pesticide removal. The improved PRZM should be applied for consistence estimations of adsorbed pesticide release from the applied field. Sediment Loadings Due to their high adsorption coefficients (K OW ), pyrethroids are typically adsorbed to soil particles and transported with suspended solids in surface runoff and stream flows. Therefore, a reasonable estimation of sediment concentration in a stream is the first necessary step in simulating pyrethroid partitioning between dissolved and particulate phases. Figure 3 shows the flow-weighted suspended solid concentration on a monthly basis observed and predicted at site OCRR (Orestimba Creek at River Road, close to the watershed outlet, Figure 1 and Table 1 ). The temporal trend of predictions followed the measured data (R 2 = 0.536), indicating a satisfactory simulation of suspended solid transport processes based on the model evaluation guidelines by Moriasi et al. [15] . High concentrations of suspended solids were observed during the irrigation season, especially in July. According to the USGS sampling results, in-stream concentrations of particulate organic carbon (OC) and suspended solids were strongly correlated (r = 0.90, p<0.001). Therefore, agreement between the predicted and observed concentrations of suspended solids also indicated a reasonable simulation for the particulate OC concentrations in the study area. Pyrethroid Toxicity Significant correlation (r = 0.72, p = 0.004) was observed between the predicted and measured toxicity units (TU) of pyrethroids in 15 samples at the three monitoring sites during 2007 and 2008 ( Figure 4a ). This indicated that the model generally captured the spatial variability and seasonality of the pyrethroid distribution in bed sediment. Based on the model prediction and field measurements, the OCRR site was generally associated with low pyrethroid TU relative to the other sites. Samples with undetected pyrethroids (plotted at 0.01 for measured TU in the figure) were all collected at the OCRR site during dormant seasons or early irrigation seasons. Located at the outlet of the Orestimba Creek, OCRR has larger drainage areas and a longer transport path for pesticides compared to other two sites. Pesticide residues have been largely decayed and diluted before reaching the water-sediment system at this site. For those undetected samples in the OCRR site, corresponding model results yielded TU values of 0.16–0.21, suggesting that the actual toxicity level was undetectable by the analytical methods applied in the sampling projects. This might also be the reason why the model overestimated measured data with low TUs. In the range of higher toxic levels (TU> = 0.3), the model had better agreement with the measurements, and the predicted and measured TUs approached the 1∶1 line on the plot. Figure 4b compares the predicted TUs based on the simulated pyrethroids to the measured cumulative mortality of Hyalella azteca . For mortality values <40%, the predicted TUs are significantly correlated to the measured cumulative mortality (r = 0.75, p = 0.005). By fitting a toxicity-mortality curve in logistic form [16] , the resultant R 2 was 0.578, suggesting the model satisfactorily captured the dose-expose relationship in the evaluated sampling site. This also supported the hypothesis that sediment toxicity in the study area was mainly associated with pyrethroid concentrations. However, the correlation was not as strong among the samples with higher mortality values of >40%. For those samples, predicted TUs were about 0.2 based on the modeled pyrethroid concentrations in this study. This deficiency may have arisen due to the substantial contributions of other pesticides or other toxic compounds which were not modeled in this study to the measured sediment toxicity. This possibility was confirmed by the fact that, for those samples with high mortality, the measured TUs in sediment were also low, ranging from non-detected to 0.1. Another issue was that some samples with relatively high measured TUs were associated with low mortality. For instance, the sample at site OCMR in September 2007 had a measured TU of 0.4 and mortality of 6%. In this case, the respective model prediction of TU = 0.23 gave a more reasonable match to the observed mortality. Characterization of Pyrethroid Exposures In the Orestimba Creek watershed, there was a general increasing trend of total pyrethroid use during 1990–2004 ( Figure 5 ). This increase was mainly attributed to esfenvalerate for years before 2000, and to bifenthrin and λ-cyhalothrin after 2000. After 2004, the amount of pyrethroids used has decreased, except for 2007 when reported permethrin use was very high. Figure 6 shows the predicted TUs at the OCRR site on a monthly basis, presenting sub-chronic risks of benthic organisms to pyrethroid exposures. Similar temporal trends of the predicted pyrethroid TUs were shared at the three sites in this study. Before 2000, the predicted sediment toxicity in the study was low, with maximum monthly TUs less than 0.5. Esfenvalerate was the major contributor to TUs during this period. The use of bifenthrin was started from July 1992 and explained a significant portion of the predicted TU. However, there was a general decreasing trend for both bifenthrin use (from 23.4 to 3.2 kg/year) and predicted TUs during 1992–1999. After 2000, bifenthrin use was increased again and predicted TU in sediment was substantially elevated and peaked in 2002–2004. Another potential reason for the elevated TUs was the change in application timing of bifenthrin. While it was mainly applied in July and August before 2000, significant amounts of bifenthrin were also applied during the late irrigation months of September and October and subject to the significant runoff events induced by winter precipitation. Consequently, bifenthrin became the major contributor to sediment toxicity for the last 10 years, accounting for 50–85% TU in sediment. Esfenvalerate and λ-cyhalothrin were also important contributors, especially during the irrigation season when they explained up to 50% of TU in sediment. During the recent years of 2004–2008, λ-cyhalothrin accounted for 38% of total pyrethroid used in the study area, followed by permethrin (32%), esfenvalerate (17%), and bifenthrin (10%). It is noteworthy that the use of λ-cyhalothrin was first reported in this area in 1998 and its use has been significantly increased since 2004, with an annual rate of about 180 kg in recent years. However, λ-cyhalothrin has a relatively short half-life in sediment (12 days), limiting its persistence and toxicity in aquatic ecosystems. During the study period, about 50% of annual pyrethroids are applied in July and August, and 70% during June to September. However, high concentrations and TU values of pyrethroids in sediment were predicted during rainfall seasons. Predicted TUs from November to January were significantly higher than the annual average. Therefore, no linear relationship was confirmed for pyrethroid uses and predicted TUs on either monthly or annual bases. In previous studies, however, such correlations were reported for organophosphate pesticides [12] , [17] . With relatively high adsorption coefficients, the off-site movement of pyrethroids is mainly associated with soil erosion from agricultural fields and suspended solid transport in channels. Bifenthrin is highly persistent in soil and sediment with half-lives in aerobic soils of 85 days and in aquatic sediments of 251 days used in the modeling ( Table 2 ). The applied pyrethroids might persist in soil and sediment long enough, and be available for the subsequent winter storms. Therefore, it is possible that predicted TUs in sediment reflected pyrethroid uses in the previous growing season. Significant time-lagged correlations were detected between the annual maximum TU and total prior bifenthrin use during the late irrigation season of September and October (r = 0.74, p = 0.022) for years when bifenthrin usage in September and October were observed ( Table 3 ). Monthly precipitation corresponding to the maximum TU was identified as the second most important factor. Further analysis indicated that the precipitation and September+October bifenthrin use explained 90.2% of the variance of the annual maximum TUs, among which 54.9% was contributed by the bifenthrin use and 35.3% by the precipitation. This finding suggested two potential measures to efficiently reduce sediment toxicity by pyrethroids in the study area: [1] limiting bifenthrin use immediately before rainfall season; and [2] implementing conservation practices to retain soil on cropland, which would mitigate suspended soild transport to surface water bodies during early rainfall season.

Results and Discussion Evaluation of the improved algorithm The improved algorithm for adsorbed pesticide removal in PRZM was evaluated in a melon crop field with historical bifenthrin applications in the Orestimba Creek watershed. This field is close to the monitoring site OCER (Orestimba Creek at Eastin Road, Figure 1 and Table 1 ), with an average annual bifenthrin use of 0.08 kg per treated hectare. The concentration of eroded bifenthrin was reported as the total amount of eroded bifenthrin divided by the total amount of eroded soil during the simulation period. Results of the improved PRZM in this study were generally invariant with the depth of soil compartment used in the numerical analysis ( Figure 2 ). This confirmed that the improved algorithm removed adsorbed pesticide from all compartments within the soil-interaction depth (D E , see Materials and Methods ), thus the resulting removal was not dependent with the depth of each compartment. As discussed in Materials and Methods , the original PRZM considers only the top-most soil compartment for adsorbed pesticide removal; therefore, the results were very sensitive to the depth. The original PRZM generated similar results as the improved one when the depth of soil compartment was close to D E (1 cm in this case study). Figure 2 demostrated PRZM simulations for soil erosion and associated pesticide removal with depth of soil compartment up to 1 cm. In the real PRZM modeling, however, small depth was required for accurate numerical simulation of water and chemical movement in the soil. For example, all crop scenarios for PRZM require depth of 0.1 cm or less for the top soil horizon developed by U.S. Environmental Protection Agency (USEPA) [14] . With small depth of soil compartment, the original PRZM significantly underestimated the adsorbed pesticide removal. The improved PRZM should be applied for consistence estimations of adsorbed pesticide release from the applied field. Sediment Loadings Due to their high adsorption coefficients (K OW ), pyrethroids are typically adsorbed to soil particles and transported with suspended solids in surface runoff and stream flows. Therefore, a reasonable estimation of sediment concentration in a stream is the first necessary step in simulating pyrethroid partitioning between dissolved and particulate phases. Figure 3 shows the flow-weighted suspended solid concentration on a monthly basis observed and predicted at site OCRR (Orestimba Creek at River Road, close to the watershed outlet, Figure 1 and Table 1 ). The temporal trend of predictions followed the measured data (R 2 = 0.536), indicating a satisfactory simulation of suspended solid transport processes based on the model evaluation guidelines by Moriasi et al. [15] . High concentrations of suspended solids were observed during the irrigation season, especially in July. According to the USGS sampling results, in-stream concentrations of particulate organic carbon (OC) and suspended solids were strongly correlated (r = 0.90, p<0.001). Therefore, agreement between the predicted and observed concentrations of suspended solids also indicated a reasonable simulation for the particulate OC concentrations in the study area. Pyrethroid Toxicity Significant correlation (r = 0.72, p = 0.004) was observed between the predicted and measured toxicity units (TU) of pyrethroids in 15 samples at the three monitoring sites during 2007 and 2008 ( Figure 4a ). This indicated that the model generally captured the spatial variability and seasonality of the pyrethroid distribution in bed sediment. Based on the model prediction and field measurements, the OCRR site was generally associated with low pyrethroid TU relative to the other sites. Samples with undetected pyrethroids (plotted at 0.01 for measured TU in the figure) were all collected at the OCRR site during dormant seasons or early irrigation seasons. Located at the outlet of the Orestimba Creek, OCRR has larger drainage areas and a longer transport path for pesticides compared to other two sites. Pesticide residues have been largely decayed and diluted before reaching the water-sediment system at this site. For those undetected samples in the OCRR site, corresponding model results yielded TU values of 0.16–0.21, suggesting that the actual toxicity level was undetectable by the analytical methods applied in the sampling projects. This might also be the reason why the model overestimated measured data with low TUs. In the range of higher toxic levels (TU> = 0.3), the model had better agreement with the measurements, and the predicted and measured TUs approached the 1∶1 line on the plot. Figure 4b compares the predicted TUs based on the simulated pyrethroids to the measured cumulative mortality of Hyalella azteca . For mortality values <40%, the predicted TUs are significantly correlated to the measured cumulative mortality (r = 0.75, p = 0.005). By fitting a toxicity-mortality curve in logistic form [16] , the resultant R 2 was 0.578, suggesting the model satisfactorily captured the dose-expose relationship in the evaluated sampling site. This also supported the hypothesis that sediment toxicity in the study area was mainly associated with pyrethroid concentrations. However, the correlation was not as strong among the samples with higher mortality values of >40%. For those samples, predicted TUs were about 0.2 based on the modeled pyrethroid concentrations in this study. This deficiency may have arisen due to the substantial contributions of other pesticides or other toxic compounds which were not modeled in this study to the measured sediment toxicity. This possibility was confirmed by the fact that, for those samples with high mortality, the measured TUs in sediment were also low, ranging from non-detected to 0.1. Another issue was that some samples with relatively high measured TUs were associated with low mortality. For instance, the sample at site OCMR in September 2007 had a measured TU of 0.4 and mortality of 6%. In this case, the respective model prediction of TU = 0.23 gave a more reasonable match to the observed mortality. Characterization of Pyrethroid Exposures In the Orestimba Creek watershed, there was a general increasing trend of total pyrethroid use during 1990–2004 ( Figure 5 ). This increase was mainly attributed to esfenvalerate for years before 2000, and to bifenthrin and λ-cyhalothrin after 2000. After 2004, the amount of pyrethroids used has decreased, except for 2007 when reported permethrin use was very high. Figure 6 shows the predicted TUs at the OCRR site on a monthly basis, presenting sub-chronic risks of benthic organisms to pyrethroid exposures. Similar temporal trends of the predicted pyrethroid TUs were shared at the three sites in this study. Before 2000, the predicted sediment toxicity in the study was low, with maximum monthly TUs less than 0.5. Esfenvalerate was the major contributor to TUs during this period. The use of bifenthrin was started from July 1992 and explained a significant portion of the predicted TU. However, there was a general decreasing trend for both bifenthrin use (from 23.4 to 3.2 kg/year) and predicted TUs during 1992–1999. After 2000, bifenthrin use was increased again and predicted TU in sediment was substantially elevated and peaked in 2002–2004. Another potential reason for the elevated TUs was the change in application timing of bifenthrin. While it was mainly applied in July and August before 2000, significant amounts of bifenthrin were also applied during the late irrigation months of September and October and subject to the significant runoff events induced by winter precipitation. Consequently, bifenthrin became the major contributor to sediment toxicity for the last 10 years, accounting for 50–85% TU in sediment. Esfenvalerate and λ-cyhalothrin were also important contributors, especially during the irrigation season when they explained up to 50% of TU in sediment. During the recent years of 2004–2008, λ-cyhalothrin accounted for 38% of total pyrethroid used in the study area, followed by permethrin (32%), esfenvalerate (17%), and bifenthrin (10%). It is noteworthy that the use of λ-cyhalothrin was first reported in this area in 1998 and its use has been significantly increased since 2004, with an annual rate of about 180 kg in recent years. However, λ-cyhalothrin has a relatively short half-life in sediment (12 days), limiting its persistence and toxicity in aquatic ecosystems. During the study period, about 50% of annual pyrethroids are applied in July and August, and 70% during June to September. However, high concentrations and TU values of pyrethroids in sediment were predicted during rainfall seasons. Predicted TUs from November to January were significantly higher than the annual average. Therefore, no linear relationship was confirmed for pyrethroid uses and predicted TUs on either monthly or annual bases. In previous studies, however, such correlations were reported for organophosphate pesticides [12] , [17] . With relatively high adsorption coefficients, the off-site movement of pyrethroids is mainly associated with soil erosion from agricultural fields and suspended solid transport in channels. Bifenthrin is highly persistent in soil and sediment with half-lives in aerobic soils of 85 days and in aquatic sediments of 251 days used in the modeling ( Table 2 ). The applied pyrethroids might persist in soil and sediment long enough, and be available for the subsequent winter storms. Therefore, it is possible that predicted TUs in sediment reflected pyrethroid uses in the previous growing season. Significant time-lagged correlations were detected between the annual maximum TU and total prior bifenthrin use during the late irrigation season of September and October (r = 0.74, p = 0.022) for years when bifenthrin usage in September and October were observed ( Table 3 ). Monthly precipitation corresponding to the maximum TU was identified as the second most important factor. Further analysis indicated that the precipitation and September+October bifenthrin use explained 90.2% of the variance of the annual maximum TUs, among which 54.9% was contributed by the bifenthrin use and 35.3% by the precipitation. This finding suggested two potential measures to efficiently reduce sediment toxicity by pyrethroids in the study area: [1] limiting bifenthrin use immediately before rainfall season; and [2] implementing conservation practices to retain soil on cropland, which would mitigate suspended soild transport to surface water bodies during early rainfall season.

Conceived and designed the experiments: YL MZ. Performed the experiments: YL. Analyzed the data: YL. Contributed reagents/materials/analysis tools: YL MZ. Wrote the paper: YL MZ. Synthetic pyrethroid insecticides have generated public concerns due to their increasing use and potential effects on aquatic ecosystems. A modeling system was developed in this study for simulating the transport processes and associated sediment toxicity of pyrethroids at coupled field/watershed scales. The model was tested in the Orestimba Creek watershed, an agriculturally intensive area in California' Central Valley. Model predictions were satisfactory when compared with measured suspended solid concentration (R 2 = 0.536), pyrethroid toxic unit (0.576), and cumulative mortality of Hyalella azteca (0.570). The results indicated that sediment toxicity in the study area was strongly related to the concentration of pyrethroids in bed sediment. Bifenthrin was identified as the dominant contributor to the sediment toxicity in recent years, accounting for 50–85% of predicted toxicity units. In addition, more than 90% of the variation on the annual maximum toxic unit of pyrethroids was attributed to precipitation and prior application of bifenthrin in the late irrigation season. As one of the first studies simulating the dynamics and spatial variability of pyrethroids in fields and instreams, the modeling results provided useful information on new policies to be considered with respect to pyrethroid regulation. This study suggested two potential measures to efficiently reduce sediment toxicity by pyrethroids in the study area: [1] limiting bifenthrin use immediately before rainfall season; and [2] implementing conservation practices to retain soil on cropland.

The authors acknowledge the two anonymous reviewers who have helped improve the present article with their most appropriate suggestions.

CC BY

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15794

oa_package/9b/aa/PMC3016336.tar.gz

PMC3016337

21246036

Introduction Ixodes scapularis and Ixodes ricinus ticks transmit pathogens such as Borrelia, Babesia, Anaplasma and selected flaviviruses [1] . In order to acquire a successful blood meal, these ticks engorge for several days on a mammalian host and counter the haemostatic system and immune responses of the host by spitting tick saliva into the skin [2] . Tick saliva contains proteins that inhibit T-cells [3] , B-cells [4] , the complement system [5] , [6] , [7] , [8] , dendritic cells [9] and the coagulation system [10] , [11] , [12] , [13] . Even though ticks modulate and dampen host immune responses to ensure successful feeding, upon repeated tick infestations some animals develop an immune response resulting in tick rejection. This phenomenon, referred to as ‘tick immunity’, was first described by William Trager in 1939, when he observed that Dermacentor variabilis ticks were not able to efficiently engorge on guinea pigs that had previously been exposed to several tick infestations [14] . Parameters of tick-immunity include decreased numbers of ticks feeding on the host, delayed time of engorgement, a reduction in tick weight, the inability to molt and decreased fecundity. Mast cells, basophils, eosinophils [15] , and antibodies [16] against exposed and concealed [17] tick proteins play a role in tick-immunity. In contrast to animals such as guinea pigs and rabbits, mice, do not develop the hall marks of tick-immunity upon repeated infestations with Ixodes scapularis ticks [18] . The mechanism underlying this difference remains to be understood. However, immune responses directed against tick proteins was shown to reduce Borrelia transmission when infected ticks fed on mice that were repeatedly infested with ticks [18] . Borrelia transmission in mice passively administered serum from tick-immune rabbits was also reduced when challenged with Borrelia burgdorferi -infected I. scapularis nymphs [19] . These observations uncouple tick feeding from pathogen transmission and suggest that while the tick-immune serum is unable to thwart tick feeding in mice, tick-immune serum contains antibodies directed against tick salivary proteins critical for Borrelia transmission to mice. Repeated exposure to tick bites is also associated with fewer episodes of Lyme disease in residents living in areas where B. burgdorferi infection is endemic [20] . Therefore, identification of tick salivary antigens that react with tick-immune serum would provide the preamble for a molecular understanding of tick feeding as well as pathogen transmission and also provide novel vaccine targets both to block tick feeding and pathogen transmission [21] . Immunoscreening of cDNA expression libraries using a phage display approach has identified several tick salivary proteins that react with tick-immune serum [22] , [23] . A limitation with phage-displayed proteins is that they lack eukaryotic post-translational modifications that might contribute to critical epitopes, and preclude the identification of such antigens by phage display screening. Therefore, additional screening efforts that exploit novel high-throughput approaches would be essential to generate a comprehensive array of salivary antigens that react with tick-immune sera. Such a detailed catalog would help develop and distill a critical subset of tick salivary antigens that might serve as vaccines to block tick feeding and impair pathogen transmission. Towards this goal, we adapted the Yeast Surface Display (YSD) approach [24] , that allows eukaryotic proteins to be displayed in a near-native form [25] . While YSD has been traditionally applied to study protein-protein interactions, we have in this report utilized the YSD approach to identify a subset of salivary proteins from I. scapularis nymphal stage that react with nymph-immune rabbit sera.

Materials and Methods Ticks and animals I. scapularis adults, nymphs and larvae were obtained from a tick colony at the Connecticut Agricultural Experiment Station in New Haven CT, USA. Ticks were maintained at 23°C and 85% relative humidity under a 14 hour light, 10 hour dark photoperiod. For the immunization studies, 6 week old inbred New Zealand white rabbits (Charles River Laboratories) were used. The work reported in this study is fully compliant with and approved by institutional policies pertinent to biosafety and animal care protocols. The protocol for the use of mice and rabbits was reviewed and approved by the Yale Animal Care and Use Committee (protocol number 2008-07941, approval date is 03/31/10 to 3/31/11). Construction of an I. scapularis salivary gland cDNA library RNA was purified from the salivary glands from 1000 I. scapularis nymphs fed to repletion (repletion achieved between 72 and 96 h), and cDNAs directionally cloned into the EcoR I and Not I sites of the yeast expression vector pYD1 (Invitrogen, CA) to generate a salivary gland expression library wherein that tick salivary proteins were expressed as Aga2 fusion proteins on the yeast surface (Invitrogen, CA). PstI digestion of plasmids purified from 24 clones of the pYD1-salivary gland library, showed an average insert size of 2.1 kb and 100% of the clones contained inserts. The unamplified library titre was 0.5×10 6 cfu/ml. Growth of transformant yeast cells and induction of recombinant protein production was done essentially as detailed by Chao et al . [50] . Briefly, fresh Saccharomyces cerevisiae EBY100 cells (Invitrogen, CA) with 5 μg of DNA were electroporated and subsequently grown in SDCAA medium (2% dextrose, 0.67% yeast nitrogen base, 0.5% bacto amino acids, 30 mM NaHPO 4 , 62 mM NaH 2 PO 4 ). Induction of surface protein expression was done as described below. Purification and conjugation of IgG from tick-immune rabbit sera Two I. scapularis nymph-immune rabbits were generated as described earlier [19] and sera tested by western blot analysis to confirm reactivity with nymph SGE. IgG was purified from the nymph-immune rabbit sera using the Melon Gel IgG purification kit (Thermo Fisher Scientific inc, Rockford, IL). IgG concentration was measured using the BCA protein assay kit (Thermo Fisher Scientific inc., Rockford, IL). Rabbit IgG was labeled with Alexa-488 using the Alexa Fluor® 488 Protein Labeling Kit (Invitrogen, CA) according to the manufacturer's protocol. Selection of yeast cells expressing immunogenic I. scapularis nymph salivary proteins To induce surface protein expression, approximately, 1×10 10 transformed yeast cells were grown for 24 hours at 28°C and induced with galactose and selected by 4 rounds of MACS sorting, as described earlier [50] . To induce surface protein expression, approximately, 1×10 10 transformed yeast cells were grown for 24 hours at 28°C in 100 ml of SGCAA medium (2% galactose, 0.67% yeast nitrogen base, 0.5% bacto amino acids, 30 mM NaHPO 4 , 62 mM NaH 2 PO 4 ). After induction with galactose, surface expression was demonstrated by indirect immunostaining with an antibody against the Xpress-epitope located on the N-terminal part of the expressed salivary protein on the yeast cell surface [50] . Ten thousand cells were examined on a FACSCalibur flow cytometer (Beckton Dickinson, Franklin Lakes, NJ) and data analyzed using the FlowJo software (Tree Star, Ashland, OR). The first round of selection was done using AutoMACS TM (Miltenyi Biotec, Auburn, CA). 2×10 10 transformed yeast cells were washed 3 times with cold MACS (0.5% BSA, 2 mM EDTA) buffer and pelleted at 600 x g for 10 minutes. Next, cells were resuspended in cold MACS buffer and incubated with 30 μg/ml of purified nymph-immune rabbit IgG and incubated with gentle rotation for 30 minutes at 4°C. Subsequently, cells were washed 2 times and resuspended in 15 ml MACS buffer. 1 ml of goat-anti rabbit microbeads (Miltenyi Biotec, Auburn, CA) was added and incubated for 15 minutes at 4°C. Cells were washed 3 times, resuspended in 150 ml of MACS buffer and subjected to magnetic separation. The sorted cells were grown in SDCAA medium with Pen/Strep for 24 hours at 30°C. Subsequently, the cells were further selected by 3 rounds of MidiMACS sorting under the same conditions as described above. For screening of individual clones, 1×10 7 induced yeast cells were incubated on a shaking incubator with 33 μg/μl Alexa-488 conjugated nymph-immune rabbit IgG in MACS buffer for 45 minutes at room temperature. The cells were washed and resuspended in MACS buffer and analyzed on a FACSCalibur flow cytometer as described above. Plasmid DNA was isolated from individual positive clones using the Zymoprep TM II Yeast Plasmid Miniprep kit (Zymo research, Orange, CA), transformed into E. coli DH5α (Invitrogen, CA) and plated on LB plates containing 100 μg/ml ampicillin. Plasmid DNA was then isolated from bacterial colonies using the Plasmid Miniprep kit (Qiagen, CA), digested with Xho I and BamH I (New England Biolabs, MA) to assess insert sizes and unique clones sent for sequencing (Keck Facility, Yale University). Production of recombinant salivary proteins p8, p19 and p23 cDNAs were cloned in frame into the pMT/Bip/V5-HisA plasmid containing a His tag, V5 epitope, and a blasticidin resistance gene (Invitrogen, CA), and validated by sequencing. Drosophila melanogaster S2 cells were transfected with the plasmids containing p8, p19 or p23 and the blasticidin selection vector pCOBlast using the Calcium Phosphate Transfection Kit (Invitrogen, CA) to generate stable transfectants and protein expression induced in 500 ml cultures with copper sulfate as described by the manufacturer (Invitrogen, CA). The supernatant was filtered using a 0.22-μm filter (Millipore, MA). rP8, rP19 and rP23 were purified from the supernatant by means of the Ni-NTA Superflow column chromatography (Qiagen, CA) and eluted with 250 mM imidazole. The eluted fractions were filtered through a 0.22-μm filter and concentrated with a 5-kDa concentrator (Sigma-Aldrich, MO) by centrifugation at 4°C, washed and dialyzed against PBS. The purity of rP8, rP19 and rP23 was checked by Coomassie blue staining after electrophoreses on SDS 12% polyacrylamide gel and the concentration was determined by BCA protein assay kit (Thermo Fisher Scientific inc., IL). Serological analysis by immunoblotting Equal amounts of purified recombinant salivary proteins (100 ng), were electrophoresed on a SDS 12% polyacrylamide gel and transferred to nitrocellulose membranes. The membranes were blocked with PBS containing 5% milk powder and the immunoblots were probed with a 1∶250 dilution of serum. To demonstrate that sera from immunized rabbits recognize tick salivary proteins, nymphal SGE (2 μg) prepared as described earlier [19] was electrophoresed and blotted as positive control. Immunoreactive bands were visualized using horseradish peroxidase conjugated goat anti-rabbit secondary antibodies (Sigma-Aldrich, MO) and the enhanced chemiluminescence Western Blotting Detection System (GE Healthcare, NJ). Detection of glycosylation modifications on recombinant proteins by Periodic acid-Schiffs (PAS) staining Periodic acid-Schiff (PAS) staining of glycoproteins after electrophoreses of the 3 proteins (25 μg) on SDS 12% polyacrylamide gel [32] was performed with Salp15 as a positive control [51] and BSA as a negative control according to the manufacturer's specifications (Sigma-Aldrich, MO). Tick RNA isolation and quantitative RT-PCR Ticks were fed to repletion on experimental and control animals. Larval ticks were pooled (5 ticks), nymphs and adults were dissected and salivary glands and midguts were pooled (3 ticks), homogenized and RNA was extracted using the RNeasy minikit (Qiagen, CA). The same procedure was done with unfed ticks. cDNA was synthesized using the iScript RT-PCR kit (Biorad, CA) and analyzed by quantitative PCR for the expression of tick actin, p8, p19, p23 and p32 and genes using the primers listed in Table 3 . Quantitative real-time PCR was performed for all tick genes using the iQ Syber Green Supermix (Biorad, CA) on a MJ cycler (MJ Research, CA). Assay for detection of complement-mediated killing of Borrelia spirochetes The serum-sensitive Borrelia garinii strain A87S was used (10 7 spirochetes ml −1 ) to determine complement-mediated killing as described earlier [8] . Spirochetes (2.5×10 5 ) were pre-incubated with bovine serum albumin (BSA), Salp15, rP19, rP23 or rP8 (0.24 μg/μl) respectively for 30 min at 33°C. They were then incubated with 12.5% normal human pooled serum (NHS) or heat-inactivated NHS and examined after 1.5 h and 4.5 h. Serum samples were checked for the absence of Borrelia -specific antibodies by western blot analysis. Heat inactivation of NHS was performed by incubation at 56°C for 30 minutes. Borreliacidal effect was recorded by screening for immobilization and bleb formation of the spirochetes. Immotile spirochetes were considered dead, as described previously [31] . The percentages of non-viable spirochetes from 200 spirochetes per well were assessed. Thrombin generation Thrombin generation was initiated by recalcification of human pooled normal plasma in the presence of 1 pM recombinant human tissue factor (Innovin, Siemens Healthcare Diagnostics, Germany), 4 μM phospholipids, 417 μM thrombin substrate (z-Gly-Gly-Arg-AMC) and Salp15, rP23, rP8 or rP19 respectively. A calibrated automated thrombogram was used to assay the generation of thrombin in clotting plasma using a Fluoroskan Ascent microtiter plate reading fluorometer (Thermo Fisher Scientific, MA) and Thrombinoscope software (Thrombinoscope BV, Netherlands) according to the manufacturer's instructions. Thrombin formation was followed for 40 min and measurements were taken at 20 second intervals. The endogenous thrombin potential (ETP), lag time and time-to-peak were calculated using the Thrombinoscope software. Experiments were performed in duplicate and repeated three times. Immunization of rabbits with recombinant nymphal I. scapularis proteins Rabbits were immunized subcutaneously with 3 doses containing 50 μg of each purified recombinant protein separate, or with a cocktail composed of 30 μg of each recombinant protein, emulsified with Complete Freund's Adjuvant (first dose) and two subsequent booster injections emulsified in Incomplete Freund's Adjuvant at 3-week intervals. Control rabbits were inoculated with adjuvant and OVA (50 μg and 90 μg for the single and cocktail immunizations, respectively). Two weeks after the last immunization, rabbits were infested with 50 I. scapularis nymphs on the ear of each rabbit and ticks kept in place using socks over each ear. Nymphs that had fed to repletion and detached were weighed. In an independent control experiment, nymphal ticks were weighed and allowed to molt individually at 23°C and 85% relative humidity under a 14 hour light, 10 hour dark photoperiod. Statistical analysis The significance of the difference between the mean values of the groups was analyzed using the Student t test with Prism 5.0 software (GraphPad Software, USA). p ≤0.05 was considered statistically significant.

Results Identification of antigenic I. scapularis salivary proteins from the nymphal stage A YSD expression library of I. scapularis salivary gland cDNAs was probed with purified IgG from pooled sera from nymph-immune rabbits. After 4 rounds of magnetic-activated cell sorting (MACS) screen, a ∼110-fold enrichment of yeast cells expressing salivary proteins recognized by rabbit nymph-immune serum ( Fig. 1 A,B ) was obtained. The library sorted with IgG purified from non-immune serum did not show binding and did not provide enrichment of yeast cells ( Fig. 1 A ). Two hundred fifty colonies from the 4 th sort were individually tested for their ability to bind to rabbit nymph-immune antiserum by fluorescence-activated cell sorting (FACS) analysis. Recombinant plasmids were isolated from 98 positive colonies and insert sizes compared by restriction digestion analysis. Clones with similar insert sizes were grouped and representative clones sequenced. Five unique clones encoding tick salivary proteins were identified and provided a unique identifier based on their putative molecular mass as shown in Table 1 . YSD was also useful in comparing the specificity and avidity of the antigen-antibody interaction between different clones using varying amounts of IgG from immune rabbits ( Fig. 1 C ). In silico analysis of P23 and P32 protein sequences revealed homology with putative secreted salivary gland proteins of I. scapularis , P19 was found to share homology with proteins from Rhipicephalus annulatus and Heamaphysalis qinghaiensis [26] , [27] . P8 showed similarities with the I. scapularis anticoagulant protein, Salp14, and the putative anticoagulant Salp9pac. P40 was identical to I. scapularis Transducing (beta)-like 2 protein ( Table 1 ). With the exception of P40, all other antigens identified contained a canonical secretory signal sequence. With an initial focus on secreted salivary proteins, we chose to exclude P40 in this study. P8, P19 and P23 were produced as recombinant proteins in a Drosophila expression system and purified using Ni-NTA Superflow chromatography ( Fig. 2 A ). Escherichia coli DH5α cells transformed with the pMT/Bip/V5-HisA plasmid containing p32 were not viable, which precluded plasmid generation for transfection of S2 cells and rP32 protein production. The three recombinant salivary proteins were recognized avidly by antibodies in nymph-immune rabbit serum ( Fig. 2 B , panel 1 ), and did not react with normal rabbit serum ( Fig. 2 B , panel 2 ). Analysis of the amino acid sequences of the proteins ( www.cbs.dtu.dk/services/NetNGlyc/ ) revealed that P8 and P19 have 1 predicted N -glycosylation site and P23 has 3 predicted N -glycosylation sites. Indeed, consistent with the in silico glycosylation predictions, glycoprotein staining with periodic acid-Schiff (PAS) indicated that rP8 and rP19 were glycosylated, albeit to a lesser extent compared to rP23 ( Fig. 2 C ). Expression of the p19 , p23 , p32 and p8 genes in different I. scapularis stages p19 and p23 were expressed in larval, nymphal and adult ticks, while p32 and p8 were primarily expressed in nymphs ( Fig. 3 ). As expected all 4 genes were expressed in tick salivary glands and significantly induced upon feeding. In addition, p19 and p23 showed additional expression in the tick gut in selected developmental stages and p8 and p32 were preferentially expressed in the nymphal salivary glands. rP8 protects Borrelia from complement-mediated killing Since no functional domains were found by in silico analysis, we examined each of the three recombinant proteins in assays to test for predominant biochemical activities represented in tick saliva, i.e. anticomplement [6] , [8] , [28] and anticoagulant activity [10] , [11] , [12] , [13] , [29] , [30] . Unlike, B. burgdorferi sensu stricto, B. garinii is a human complement sensitive strain [31] . Therefore, we utilized a B. garinii killing assay [8] to investigate whether rP8, rP19 and rP23 were able to protect spirochetes from or inhibit the human complement system. rP8 significantly reduced complement-mediated killing of B. garinii A87S ( Fig. 4 ) after 1.5 hours ( Fig. 4 A ; 90±0.9% for BSA versus 31±1.2% for rP8, p<0.0001) and 4.5 hours ( Fig. 4 B ; 97±0.6% for BSA versus 25±0.9% for rP8, p<0.0001), in a dose dependent manner. B. garinii spirochetes incubated with heat-inactivated NHS remained viable at all time points examined ( Fig. 4 ). None of the other recombinant proteins provided protection against complement-mediated killing ( Fig. 4 ). The I. scapularis salivary protein Salp15 was used as a positive control [8] . rP23 inhibits coagulation of human plasma Calibrated automated thrombography (CAT) was used to assess the effect of the recombinant I. scapularis salivary proteins on tissue factor initiated thrombin generation. In normal human pooled plasma, recombinant rP23 delayed thrombin generation ( Fig. 5 A ), with significant prolongation of lag time and time to peak ( Fig. 5 B,C ) in a dose dependent manner, suggesting that rP23 may influence the initiation phase of coagulation. In addition, determination of the total amount of thrombin formed (Endogenous Thrombin Potential, ETP) showed that rP23 significantly reduced thrombin generation by 18% compared to ETP in the absence of rP23 ( Fig. 5 D ). rP8 and rP19 at similar concentrations did not affect thrombin generation ( Fig. 5 ). Immunization with recombinant P8, P19 and P23 impairs nymphal tick feeding on rabbits Three rabbits were immunized with a cocktail of rP8, rP19 and rP23. Nymphal salivary gland extract (SGE) probed with immune sera from each of the rabbits recognized the native proteins ( Fig. 6 A , panel 1 ). A 45 kDa band appeared in all blots most likely due to binding of anti-rabbit IgG to host globulins present in the fed SGE [32] . Immune sera from 3 control rabbits immunized with OVA did not react with proteins in the nymphal SGE ( Fig. 6 A , panel 1 ). Upon challenge of the immunized rabbits with I. scapularis nymphs, comparable numbers of nymphs fed to repletion on the control and experimental animals (127 nymphs on the immunized group versus 137 on the control group). Nymphs feeding on rabbits immunized with the cocktail of rP8, rP19 and rP23 were significantly lighter than nymphs feeding on the OVA control rabbits ( Fig. 6 B ; 3.7±0.1 mg and 4.0±0.1 mg, respectively, p = 0.03). An independent control experiment showed that, after feeding on normal rabbits, nymphs with a weight of 3.3 mg and below consistently molted into males, while the heavy group of nymphs (3.4 mg and above) molted into female adult ticks ( Table 2 ). Therefore, engorged ticks were also divided into subgroups (light and heavy) ( Fig. 6 C–D ), Analysis of the subgroups showed a significant reduction in engorgement weights in the heavy group of nymphs fed on the rP8, rP19 and rP23 cocktail immunized rabbits compared to the heavy group of nymphs fed on control rabbits (4.7±0.1 mg and 4.9±0.1 mg, p = 0.01) ( Fig. 6 C ). There was no significant difference between the light group of nymphs feeding on immunized rabbits versus controls ( Fig. 6 D ). Finally, when three rabbits per group were immunized with single proteins, nymphal SGE probed with immune sera from each of the rabbits recognized the native proteins ( Fig. 6 A , panel 2 ), but there was no difference in nymph weights, compared to the control OVA-immunized animals in both heavy ( Fig. 6 E ) and light groups ( Fig. 6 F ).

Discussion Ixodes ticks transmit numerous pathogens, including bacteria, protozoa and selected flaviviruses [33] . Anti-tick vaccines could potentially prevent transmission of Borrelia , as well as other pathogens, from the tick to the host. Acquired resistance to tick feeding can also impair pathogen transmission [34] , [35] . Several studies have shown a critical role for the humoral response in tick immunity [16] , [36] and provided the impetus to identify antigens that elicit tick-immunity by exploiting approaches including immunoscreening of cDNA expression libraries generated in routine prokaryotic expression systems [22] , [23] . However, the prokaryotic systems are not capable of displaying the antigens with eukaryotic post-translational modifications. When these modifications are critical determinants of antigenic epitopes, prokaryotic expression systems preclude identification of such antigens. It is therefore critical to exploit additional strategies to circumvent these limitations. It is also important to note that the proteome of the developmental stages of ticks are likely different [37] and antigens critical for one stage might not necessarily be critical for all stages. Therefore, it is important to develop viable and high-throughput strategies to screen for salivary antigens that react with stage-specific immune sera. This would help build a comprehensive catalog of salivary antigens that react with tick-immune serum against different developmental stages of Ixodes scapularis , and facilitate the development of functional vaccines targeting critical developmental stages. Since, I. scapularis nymphs are critical for disease transmission to humans [38] , in this study we advance our understanding of nymphal salivary proteins that react with nymph-immune rabbit sera. It is expected that these antigens might play a role in tick feeding and immunity against these antigens might impair tick feeding. Such a phenotype would therefore also affect the success of nymphal molting to adults and would be conductive to control tick populations. Decreased feeding might also lead to decreased pathogen transmission. Additionally, these antigens might provide functions critical for pathogen transmission [6] , [8] , [39] . Therefore, we used the YSD approach, to identify salivary antigens that react with nymph-immune serum and examine their role in the context of nymphal feeding success. We demonstrate that YSD technology is a robust tool for the identification of antigenic salivary gland proteins. The major advantage of YSD over other expression systems such as phage display, is that yeast cells have eukaryotic machinery allowing the display of eukaryotic proteins with post-translational modifications like glycosylation, phosphorylation and correct disulphide bond formation [40] and hence, utilized for several applications including protein affinity maturation, epitope mapping, and cell adhesion molecule engineering [25] , [41] . To date, there are only a few reports describing the utility of YSD for cDNA library screening and have addressed human cDNA libraries [42] , [43] . This study extends the utility of YSD to perform high throughput immunoscreening of a YSD library of nymphal I. scapularis salivary gland cDNAs to enrich for yeast cells expressing antigenic tick proteins ( Fig. 1 ). The YSD library of I. scapularis salivary gland cDNA was prepared from RNA isolated from replete nymphs. We are cognizant that the salivary gland proteome is dynamic, and the protein profiles change during feeding. The current screening effort is therefore not likely to identify antigens expressed specifically at early time points of feeding. Five novel antigenic tick salivary proteins were identified using the YSD approach ( Table 1 ). While P23 and P32 only showed homology with putative secreted salivary gland proteins, with no known functions. P8 showed homology with the putative anticoagulant Salp9pac and anticoagulant Salp14 family of proteins. Salp14 was identified earlier by Das et al [22] by immunoscreening a phage expression library of I.scapularis nymphal cDNAs using tick-immune sera [10] . P19 was homologous to a larval immunogenic protein, Ba05 from Rhipicephalus annulatus [27] and the protein Hq05 from Haemaphysalis qinghaiensis [26] (93 and 82% identity respectively). Expression analysis of Hq05 showed expression specifically in salivary glands of nymphs and adults, but not in eggs and larvae of H. qinghaiensis [26] . P40 showed similarities with a putative G-protein from I. scapularis (XP_002416461.1) and with Transducin beta-like 2 proteins (NP_001008084) from several species including mammals. P40 contains 7 copies of the WD40 domain, which is found in several eukaryotic proteins that are involved in diverse functions including pre-mRNA processing, signal transduction, cytoskeleton assembly and cell cycle control [44] , [45] , [46] . P40 lacked a canonical secretory signal sequence and possibly not secreted in tick saliva. However, based on the reactivity of P40 with tick-immune sera ( Table 1 ), we speculate that P40 might be secreted in tick saliva by other/novel secretory mechanisms and warrants detailed verification and therefore not included in this study. Four of the five identified proteins, i.e. P8, P19, P23 and P32, had a predicted secretory signal sequence as assessed by the SignalP 3.0 signal prediction server ( www.cbs.dtu.dk/services/SignalP/ ) and were therefore selected for further analysis. As expected, the identified genes p8, p19, p23 and p32 encoding for the antigenic tick proteins were all induced upon nymphal feeding ( Fig. 3 ). While p19 and p23 were expressed both during larval and adult stages, p8 and p32 were preferentially expressed in the nymphal stage, indicating that some salivary proteins might play a stage-specific role. Next, recombinant P8, P19 and P23 were made in S2 cells using a Drosophila expression system ( Fig. 2 ). Attempts to generate stable S2 cell lines expressing rP32 were unsuccessful. All three recombinant proteins showed glycosylations as seen by a positive reaction using the Periodic Acid Schiff method ( Fig. 2 C ), suggesting that these proteins might also be glycosylated in I. scapularis in vivo and might account for the slightly increased molecular masses of the native P8 and P23 ( Fig. 6 A ). In silico analysis of the proteins for potential glycosylation sites using the NetNGlyc server ( www.cbs.dtu.dk/services/NetNGlyc/ ) also suggested one and 3 potential N-glycosylation sites on P8 and P23, respectively. Despite the homology between P8 and the anticoagulant Salp14 [10] , [47] , rP8 did not have any effect on thrombin formation in human plasma. We have previously shown that Salp9pac, also homologous to Salp14, did not have anticoagulant properties [10] . The anticoagulant Salp14 has a positively charged stretch of 20 amino acids at the C-terminal tail compared to rP8 and Salp9pac, which is most likely important for interaction with the coagulant proteins [10] . rP23 demonstrated anticoagulant activity and significantly delayed thrombin generation in human plasma ( Fig. 5 ). Interestingly, P23 does not contain any known anti-coagulant domains and might represent a novel type of anticoagulant. The mechanism of anticoagulation by rP23 remains to be determined. To investigate the three proteins for anti-complement activity we used a B. garinii killing assay, since B. garinii is a serum sensitive strain compared to B. burgdorferi sensu stricto [8] , [31] . We have previously demonstrated that the serum-sensitive B. garinii A87S strain is killed by the human complement system after formation of the C5b-9 membrane attack complex and that Ixodes Salp15 provided protection against this complement-mediated killing [8] . When Borrelia was incubated with normal human serum, rP8 was also able to protect Borrelia against complement-mediated killing ( Fig. 4 ). Whether, rP8 provides this protection by virtue of an anticomplement activity or by physically binding to spirochetes and shielding the spirochetes from complement attack remains to be examined. We next examined if immunity against these recombinant salivary proteins would recapitulate tick immunity in rabbits. When rabbits were immunized with a cocktail of rP8, rP19 and rP23 and challenged with pathogen-free nymphs, feeding efficiency was modestly, but significantly, decreased compared to nymphs that fed on OVA-immunized rabbits, as determined by engorgement weights ( Fig. 6 B ). Previous experiments in our laboratory, consistent with observations by Hu et al [48] , have shown that the heavy group of nymphs (3.4 mg and higher) consistently molted into adult females, while the light group of nymphs (3.3 mg and lighter) molted into adult males ( Table 2 ). The impaired feeding phenotype was significant in the to-be-female group ( Fig. 6 C ), since when fed nymphs were separated into heavy (to-be-female) and light (to-be-male) groups of nymphs, no difference in post-engorgement weights was found in the to-be male group of nymphs fed on the rP8, rP19, rP23 immunized rabbits compared to controls ( Fig. 6 D ). In addition, two weight populations were seen in the to-be male group of nymphs fed on the immunized animals ( Fig. 6 D ), possibly due to to-be female ticks that showed impaired feeding due to immunization with the recombinant tick proteins controls. It is possible that the to-be male group of ticks differentially express salivary proteins while feeding on the host compared to the to-be female group of nymphs. P8, P19 and/or P23 might play a redundant role in feeding in the to-be male nymphs. Future studies will determine whether the cocktail vaccine (rP8, rP19 and rP23) might also impair pathogen transmission from the tick to the host. In conclusion, these data show that cocktail immunization targeting multiple (functional) salivary proteins resulted in impaired tick feeding, while immunizations with individual proteins did not have an effect. It is conceivable that targeting 3 antigens simultaneously provided neutralization of antigens important for suppressing several arms of the host defense (in this case, host coagulation and complement) and resulted in impaired tick feeding. Further, the past decade has demonstrated that the tick transcriptome elaborates an array of functional paralogs [49] . Several anticomplement [6] , [8] , [28] and anticoagulant proteins [10] , [11] , [12] , [13] , [29] , [30] have been described and characterized in Ixodes ticks. This functional redundancy is central to the tick's ability to feed successfully. To efficiently block tick feeding, it might be important to immunize animals with cocktails of several anticomplement or anticoagulant proteins to circumvent fall-back strategies of the tick. Since tick-immunity effectively disables the ability of the tick to feed, presumably, tick-immune serum targets the critical subsets of functional paralogs. Thus, intensive screening using the YSD approach would provide a comprehensive list of antigens that react with tick-immune serum. This would enable us to group subsets of functional paralogs critical for feeding and facilitate the development of an effective cocktail vaccine to block tick feeding.

Conceived and designed the experiments: TJS SN SD CvV JCMM EF. Performed the experiments: TJS SN SD KD KB. Analyzed the data: TJS SN SD JWRH CvV KB JCMM APvD EF. Contributed reagents/materials/analysis tools: ETB JCMM TvdP. Wrote the paper: TJS SN EF. Repeated exposure of rabbits and other animals to ticks results in acquired resistance or immunity to subsequent tick bites and is partially elicited by antibodies directed against tick antigens. In this study we demonstrate the utility of a yeast surface display approach to identify tick salivary antigens that react with tick-immune serum. We constructed an Ixodes scapularis nymphal salivary gland yeast surface display library and screened the library with nymph-immune rabbit sera and identified five salivary antigens. Four of these proteins, designated P8, P19, P23 and P32, had a predicted signal sequence. We generated recombinant (r) P8, P19 and P23 in a Drosophila expression system for functional and immunization studies. rP8 showed anti-complement activity and rP23 demonstrated anti-coagulant activity. Ixodes scapularis feeding was significantly impaired when nymphs were fed on rabbits immunized with a cocktail of rP8, rP19 and rP23, a hall mark of tick-immunity. These studies also suggest that these antigens may serve as potential vaccine candidates to thwart tick feeding.

CC BY

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15926

oa_package/de/6c/PMC3016337.tar.gz

PMC3016338

21246037

Introduction Traditionally, attitudes have been measured by having consumers respond to an attitude object (or entity) on self-report rating scales. In these scales, consumers rate a particular object (e.g., a product or a brand) on dimensions such as “good/bad”, “like/dislike”, or “pleasant/unpleasant”. Yet, consumers often find it difficult to report on these scales. They may not have attitudes readily available for reporting on them (in an explicit way), or may even find it difficult to retrieve them [2] , [3] . Indirect measures, in particular the popular Implicit Association Test (IAT) by Greenwald, McGhee, and Schwartz [1] , constitute a viable alternative avoiding some of the problems associated with direct measures (e.g., lack of attitude availability/accessibility, social desirability bias). In this article, we introduce a procedural extension of the IAT, a multi-dimensional Implicit Association Test ( md -IAT). In contrast to the regular IAT, which is utilized as a procedure that allows assessment on a single dimension only, the md -IAT comprises six dimensions, thus allowing for a more detailed, multi-dimensional assessment of attitudes. More fine-grained attitudes/associations have been assessed in several studies but were confined to a single administration and thereby also to a single dimension in the IAT: for example, to measure gender stereotypes (i.e., men–women/warm–cold [4] ), self-concepts (i.e., self–other/anxious–calm, [5] ), or even abnormal pedophilic tendencies (i.e., children–adults/sex–no-sex, [6] ). The additional information offered by this multi-dimensional measure can be of particular value in marketing and consumer research, allowing for example—in the same way as with direct measures—to easily create more complex and differentiated profiles of products and brands (cf. [7] ). Tapping consumer insights in such a way more appropriately captures the richness of consumers' perceptions, feelings, and attitudes toward a brand. For example, the IAT can indeed provide important information about consumers' general attitude toward a specific brand or product (consumers' likes and dislikes), but it does not elucidate the different components contributing to this global attitude. Any kind of intervention, however, depends on clear diagnostics: the specific aspects consumers like or dislike or the specific properties they associate with the product [8] . The contribution of the present research is both of theoretical and practical relevance: our results show that the md -IAT procedure is a methodologically sound extension of the IAT that—unlike the latter—also allows for multi-dimensional assessment of brand attitudes. This in turn opens up numerous possibilities for researchers to test constructs such as brand or product personality [9] , [10] , or more generally, consumers' brand associations or attitudes on any kind of multi-dimensional scale [11] . In addition, we show that the md -IAT, just like the IAT, is not affected by the specific stimuli selected to represent a brand. The three brand identifiers used in the present studies (logos, signatures, and product pictures) all yielded similar results, therefore rendering the md -IAT rather suited as a conceptual (as opposed to perceptual) measure of brand attitudes. The structure of the paper is as follows: We start by briefly reviewing different forms of attitude measurement—distinguishing between indirect and direct measures. We then turn to the IAT itself before introducing its multi-dimensional extension (the md -IAT) and its application in two within-subjects repeated measurement studies. Indirect versus Direct Measures Indirect measures differ from direct measures in that they do not rely on verbal self-reports as a way of inferring attitudes [12] . Instead, they rely on rather indirect means of assessing an attitude, for example differences in reaction times, facial expression, or specific brain activation. Indirect measures can be further distinguished into physiological or latency based measures. Physiological measures include techniques such as electro-dermal activity (EDA; [13] ), pupillometry [14] , eyetracking [15] , electromyography (EMG; [16] ); or various brain imaging techniques, such as functional magnetic resonance imagining (fMRI; [17] ), which allow direct observation of brain activity during mental tasks. While promising in their own right, these physiological measures do not yet offer standardized forms of attitude assessment (for advances in this domain, see, [16] , [18] ). In addition, they require (very) expensive equipment and a considerable expertise in the domain of cognitive neuroscience, which make most of these research techniques inaccessible and/or ill-suited for any kind of more applied research. This is much less the case for indirect measures based on response latencies (or reaction times). Measures such as affective priming [19] , the Extrinsic Affective Simon Task [20] , the Go/No-Go Association Task [21] , and particularly the Implicit Association Test (IAT; [1] ), are fairly standardized forms of attitude assessment requiring little more than a computer and a testing environment void of external distractions. Attitude Measurement and the Implicit Association Test (IAT) The IAT is a method of estimating evaluative associations that underlie implicit attitudes, which draws on differences in reaction times in a rapid computerized categorization task. Introduced more than a decade ago by Greenwald, McGhee, and Schwartz [1] , it is one of the most widely used indirect attitude measures. The IAT is considered superior to the other latency-based techniques mentioned above, showing moderate-to-high correlations with self-report attitude measures in the consumer domain [2] , [22] – [26] and satisfactory split-half reliabilities [22] , [27] , [28] . The IAT has also shown to be quite robust with regard to stimulus artifacts. That is, stimulus specifics, for example in the visual domain, seem to be of little importance as long as category membership remains unambiguous [12] , [29] . Brunel et al. [2] tested the applicability of the IAT in consumer research and concluded that the IAT is a valid measurement instrument for capturing consumer attitudes. In two studies, they showed that the IAT was sensitive to individual differences in attitude accessibility and that the IAT can capture automatic associations that are distinct from explicit measures. Conscious and Less Conscious Manifestations of Attitudes Up until the late 1990s research in the domain of attitudes largely involved assessing attitudes by means of direct measures. Direct measures require participants to consciously or deliberately think about a certain attitude object and subsequently report their attitudes in the form of verbal self-reports, for example, on semantic differential scales or Likert-scales [30] – [32] . By means of such explicit introspective processing, participants arrive at an attitude toward an object, either by retrieving it from memory or by constructing it on the spot. In contrast, indirect measures try to measure participants' implicit attitudes, which Greenwald and Banaji [33] describe as “introspectively unidentified (or inaccurately identified) traces of past experience that mediate favorable or unfavorable feeling, thought, or action toward social objects” (p. 8). Greenwald and Banaji introduced this implicit–explicit dichotomy to attitude research. Since then the implicit-explicit terminology has become popular for referring both to the form of measurement (indirect vs. direct) and the form of representation in memory (unconscious vs. conscious). Greenwald and Banaji note that attitudes—in addition to their conscious manifestations—might also operate in an indirect, unconscious, or implicit mode. Such implicit attitudes are activated automatically, not necessarily requiring conscious thought or attention [19] . Whether or not implicitly measured attitudes are also (truly) unconscious is widely debated since participants might be unaware that their attitudes are being assessed, but this does not necessarily imply unawareness of possessing those attitudes [12] , [34] . Despite these reservations, indirect measures seem particularly useful when consumers do not have readily available attitudes that they could consciously report on—attitudes consumers may not be aware of, able to express, or willing to share with the researcher [2] , [3] . Design of the IAT The IAT has shown to be a flexible and fairly easy-to-use tool in assessing strengths of associations between different concepts, contributing notably to its attractiveness and widespread use in research [27] . Typically, the IAT engages subjects into a sorting task requiring them to quickly sort stimuli (e.g., pictures or words) into one of four categories. The categories themselves are referred to as target categories and attribute categories; for example, in an IAT assessing cultural stereotypes and prejudice, one could employ the categories “American” and “European”, “pleasant” and “unpleasant”, respectively. The category names are displayed in the top corners of the computer screen, whereas the stimuli (e.g., pictures of famous Americans/Europeans and words with a clear pleasant/unpleasant connotation) appear in the center. The IAT comprises five consecutive tasks: the target discrimination task (task 1), the attribute discrimination task (task 2), the initial combined task (task 3), the reversed target discrimination task (task 4), and the reversed combined task (task 5). Throughout tasks 1–5 subjects respond by pressing either one of two keys; that is, the “left key” for stimuli belonging to a category on the left side of the screen, and the “right key” for stimuli belonging to a category on the right side of the screen. The first two tasks are intended to familiarize the subjects with both the stimuli and the overall assignment. Subjects are either required to sort target category stimuli to the target categories (task 1) or attribute category stimuli to the attribute categories (task 2). Unlike tasks 1, 2, and 4, which assign each key to only one category, the combined tasks assign each key to two categories. Referring to our example, “American” and “pleasant” might be assigned to the “left key” for the first combined task, requiring “European” and “unpleasant” to be assigned to the “right key” (or vice versa). The second combined task is identical to the initial combined task, except for the target categories (i.e., “American” and “European”) being reversed. Due to the change in target categories, subjects need to unlearn the previous key assignments and rehearse the new key assignments in an intermediate task (task 4). The dependent measure (i.e., the “IAT-effect”) is calculated as a difference score by subtracting the average response time of the initial combined task from the average response time of the reversed combined task. A positive IAT-effect is interpreted as a stronger association for the category pairing in the initial combined task—for attitude-IATs it may as well be interpreted as a preference for one concept over the other [1] . IAT scripts are usually based on a seven-block (seven-task) structure. This is because earlier research essentially employed a seven-task model in which each of the combined tasks was preceded by a combined task practice block that was shorter but otherwise identical. Originally, these preceding practice blocks were not used for computing the IAT-effect. Although Greenwald, Nosek, and Banaji [35] proposed a new scoring algorithm, the D measure , which draws also on data from the combined practice blocks for computing the IAT-effect, most scripts for analyzing IAT-effects still use traditional routines for dividing the analysis into seven blocks. The multi-dimensional Implicit Association Test ( md -IAT) The main idea behind the present research was to extend the IAT procedure to allow for a valid multi-dimensional assessment of attitudes that is also economically feasible (i.e., the diagnostic value in proportion to the time and effort invested). Instead of employing just one IAT, using for example good–bad as the single attribute dimension (as typical for attitude-IATs), the multi-dimensional Implicit Association Test ( md -IAT) consists of several IATs, each aimed at measuring different aspects of a more abstract, general attitude. Most definitions of attitudes consider affective-evaluative components to be most essential in attitudes. Attitude measures typically ask participants to evaluate an attitude object along attribute dimensions such as good–bad or favorable–unfavorable [31] , [36] . By having participants evaluate two target concepts (in our case automobile brands) on several distinct attribute dimensions rather than just a single overall attribute dimension, it is possible to obtain a more detailed and differentiated account of consumers' associations with a brand, similar to that of brand (personality) profiles generated by semantic differential scales known from the tradition of explicit measures. Naturally, in introducing a new measure or—as in this case—an extension to an existing measure, it is important to address its methodological appropriateness. Reliability of the md -IAT was assessed by calculating the IAT-effects separately for odd and even trials and correlating these two scores (for each IAT in the md -IAT procedure) using a Spearman-Brown correction (see [22] ). Of particular interest was whether participants could handle six IATs in a row, that is, whether the md -IAT, despite requiring multiple administrations, would preserve the same level of reliability. Validity of the md -IAT was assessed in two ways: First, by comparing the results from the IATs to direct (or explicit) ratings of the same six attribute dimensions; and second, by adding the factor brand cue , which involved brand stimuli varying by their level of abstraction. Based on previous findings that identified the IAT to be more driven by the target category labels than by the actual stimuli in the sorting task [12] , [27] , [29] , differences due to this factor were not expected. Obtaining similar results, regardless of the brand cue used, may thus be interpreted as evidence for its external validity—making the md -IAT better suited for conceptual brand assessment and less prone to idiosyncrasies in the perceptual domain. Thus, compared to a regular IAT, the main benefit of the md -IAT lies in its more detailed and differentiated assessment of consumers' brand attitudes. With such a method in hand, practitioners can easily create brand profiles based on indirect measures that provide more information than simply how good or bad a brand is. This, in turn, will also provide more opportunities for specific intervention in practice. In this article, we draw on the results of two within-subjects repeated-measurement studies to provide evidence both for the methodological appropriateness and practical utility of the extended, multi-dimensional IAT procedure.

Materials and Methods Study 1 Participants Thirty volunteers (15 women) participated in the study. The sample consisted of adults from the Vienna Metropolitan Area, both students and young professionals between the ages of 20–40 (median age = 27.0 years). Two female subjects were excluded prior to the analysis after reporting difficulties with the task upon debriefing. An additional two subjects (one male, one female) were excluded after the analysis of the reaction time data because of an average total error rate of more than 10% across all IATs. Among the remaining participants, 76.9% (20) were car owners. The average overall interest in cars showed to be low among the participants ( M = 2.0, SD = 1.93). Overall interest in cars was assessed by six yes–no questions (“I buy and read car magazines”; “I watch broadcasts about cars on TV”; “I am interested in cars”; “I actively follow the latest developments in the car sector”; “I talk about different car models with friends and/or family members”; “I pay attention to car advertisements”) which were then summed up to form an index (range 0–6). All subjects had normal or corrected-to-normal vision (visual acuity was checked with standard Snellen charts). Materials The present research was interested in indirectly assessing participants‘ brand attitudes toward two automobile brands using a multi-dimensional extension of the IAT, the multi-dimensional -IAT ( md -IAT), as the dependent measure. Instead of employing just one IAT and therefore only one attribute dimension (e.g., pleasant–unpleasant ), the present research was based on a more complex design that involved administering six consecutive IATs, each intended to measure associations on a different attribute dimension. The six bipolar attribute dimensions were selected on the basis of highly relevant properties derived from consumer research [7] , [10] , [37] – [39] : (1) safe–unsafe (2) young–old (3) reliable–unreliable (4) aggressive–peaceful (5) environmentally friendly–non-environmentally friendly (6) innovative–conventional. Each pole (or attribute category) was represented by three word stimuli (see table 1 for a complete list of the word stimuli used in all of the IATs). Additionally, stimuli also varied according to another factor—called brand cue (through stimuli varying in their level of abstraction). This added complexity in the manipulation served the purpose of further testing the validity of the md -IAT. Based on previous findings that identified the IAT to be mostly driven by the category labels and less so by the actual stimuli in the sorting task [12] , [27] , [29] , we expected minor or no differences at all between the different levels of the factor brand cue. The following brand cues served as stimuli for the target categories AUDI and FORD: images of the AUDI/FORD logo , images of the AUDI/FORD signature , and images of the products themselves (i.e., current car models of AUDI/FORD). See figure 1 for target category stimuli used to represent the brands AUDI and FORD. Two stimuli were used to represent each brand (i.e., each target category): a realistic image and an artificial image. Realistic images included real photographs of the logo, the signature, or a specific product model. Artificial images were digitized versions of either the logo or the signature as used in advertising and public relations or simply renderings from computer-aided design drawings of the same product models. All stimuli for the IATs, both words and pictures, were selected in accordance with suggestions by Nosek and colleagues [27] , [40] : First, only stimuli that were clearly and unambiguously associated with a category (or concept) were selected from free association protocols in a pretest. This is a necessary prerequisite to prevent cross category associations from exerting an influence on task performance [27] , [41] . For example, it would be impracticable in a race-related IAT to have a stimulus depicting a person with ambiguous face race markers; clearly, this could cause subjects to sort such a stimulus arbitrarily to either category, or to refuse giving a response entirely. Second, a minimum of two stimulus items per target category and three items per attribute category was used throughout the experiment. Previous research showed that the magnitude of IAT effects, reliability, and correlations with direct measures remained stable for IATs assigning two or more stimulus exemplars per category (cf. [40] , for results on IATs using 1, 2, 4, 6 or 8 items per category). Apparatus The various IATs were administered using PsyScope X (build 46) experimental software [42] —both to present the stimuli and to collect the data. The experiment was run on two identically configured Apple Mac mini computers (1.25 GHz PowerPC G4 chip set, 512 MB RAM) with preinstalled OS X v10.4 (Tiger). Participants sat approximately at a distance of 50–55 cm away from the screen—a 19" BenQ FP93V LCD monitor at a resolution of 1280×1024 pixels with a refresh-rate of 75 Hz. Additionally, a USB button box by ioLab served as the default input device, limiting the inaccuracy in measuring reaction times to < = 1.0 ms. General Procedure and Design The entire experiment required subjects to complete eighteen IATs and a subsequent questionnaire. Data were gathered in three separate test sessions (T1, T2, and T3). The minimum time interval between two sessions was one day. Subjects completed one md -IAT (six IATs) per session, one for each of the six bipolar attribute dimensions, taking them approximately 20–35 minutes. All attribute dimensions were in fixed order throughout the entire experiment: (1) safe–unsafe (2) young–old (3) reliable–unreliable (4) aggressive–peaceful (5) environmentally friendly–non-environmentally friendly (6) innovative–conventional. The three dimensions of the factor brand cue (logo, signature, and product) were counterbalanced across subjects. This was necessary as learning effects could be an issue after several administrations of an IAT. Previous research found the magnitude of IAT effects declining for subjects with prior experience. Yet, this was primarily the case for subjects who had previously completed no more than two IATs (see [43] ). Little or no further decrease was observed for subjects that had completed more than two IATs [35] . Hence, counterbalancing for brand cue also helped minimizing order effects for the factor attribute dimension. After participants had completed the six IATs at T3, they were prompted to fill out a questionnaire, which also included 7-point semantic differential scales as a direct (or explicit) measure of brand attitudes [30] . The semantic differential scales required subjects to rate each brand separately on the same six attribute dimensions also used for the IATs. Half of the subjects first rated AUDI followed by FORD (vice versa for the other half) to control for order effects. Written consent was acquired from each participant prior to the experimental sessions. As this was a non-clinical study without any harming procedure and as all data were collected anonymously, ethical approval was not sought for the execution of this study. Procedure and Design of the md-IATs Following the IAT procedure outlined earlier, the IATs used for the present research were based on the same structure. Each IAT consisted of seven blocks (B1 through B7). Although B3 and B4, and similarly, B6 and B7, were in fact separate blocks, they essentially can be considered one task. There are two reasons for this: first, B3 and B4, and B6 and B7, were identical except for the number of trials used in each block. The number of trials in B3 and B6, and B4 and B7, was 23 and 40, respectively. Second, Greenwald et al. [35] suggested using their new scoring algorithm, the D measure , which involves joint analysis of the data in B3 & B4, and also B6 & B7. Other scoring algorithms do not make use of the data in B3 and B6—for the most part, because these blocks were initially devised as practice blocks for the ensuing combined tasks. Each trial in every block involved subjects sorting just one stimulus, either a word or a picture, to its designated category. The stimuli were presented in the middle of the screen. Each stimulus remained until the subject hit the correct button on the button box. If a subject pressed the wrong button, a red capital X served as error feedback, upon which a subject had to press the other button as fast as possible. The inter-trial interval (ITI), that is, the interval between a correct response to a stimulus and the next stimulus onset was set to 200 ms. Stimuli within the seven blocks were fully randomized, the only restriction being that for the combined tasks a target category stimulus was never followed by another target category stimulus, instead it was always followed by an attribute category stimulus (or vice versa). Finally yet importantly, extraneous effects of task order of the two combined tasks (B3 & B4, B6 & B7) were counterbalanced by two means. First, the display of the target categories (whether AUDI or FORD was first assigned to the left key) was counterbalanced: half of the subjects started with AUDI assigned to the left key and FORD assigned to the right key (vice versa for the other half). For both groups, key assignments for the target categories changed after the initial combined task, with AUDI being assigned to the right key and FORD being assigned to the left key (again vice versa for the other half). Second, the reversed target discrimination task (B5) involved some extra trials in order to provide subjects with the opportunity and the time to unlearn the previous key assignments, and consequently, to learn the new assignments. Nosek, Greenwald, and Banaji [40] provided ample empirical evidence that adding extra trials to the reversed target discrimination task virtually eliminates this unwanted effect of task order. Messner and Vosgerau [44] have recently introduced a new procedure of neutralizing this task order effect by adding iterations of the initial combined task and the reversed combined task to the procedure. This adaptation effectively counteracted the impact of cognitive inertia (i.e., the difficulty in switching between the two tasks) even on the individual level (as opposed to the aggregate level). Study 2 Study 2 was intended to replicate the findings of Study 1 with a different set of brands in the md -IAT. BMW was chosen to replace FORD as the contrasting brand in the comparisons with AUDI. AUDI and BMW are commonly perceived to be highly similar in terms of several key aspects associated with the brand: for example, in ratings of safety, build quality, reliability, and technical innovativeness [45] . Finding reliable differences between these two highly similar brands (i.e., IAT effects of comparable magnitude across the three levels of the factor brand cue) would provide not just evidence of the md -IAT’s reliability but also of its sensitivity. It is evident that finding differences between two highly similar attitude objects asks for a more sensitive measure. Together, Study 1 and Study 2 allow for an assessment of the md -IAT procedure and its methodological appropriateness based on its sensitivity, reliability and validity. Participants Thirty students from the University of Vienna (15 women) participated in the study. Among them a total of 27 received extra undergraduate course credit in return; the remaining three subjects were not associated with the Faculty of Psychology and therefore did not receive anything in exchange. One male subject was excluded due to an unspecified mental condition that impaired his speech and motor behavior. The median age of the remaining twenty-nine subjects (ranging from age 18 to 34) was 22.0 years. An additional three subjects (one woman, two men) were excluded after the analysis of the reaction time data because of an average total error rate of more than 10% across all IATs. Among the remaining participants 26.9% (7) were car-owners. Overall interest in cars was assessed by the same six yes–no questions as in Study 1 and showed to be low among the participants ( M = 1.65, SD = 1.70). All subjects had normal or corrected-to-normal vision (visual acuity was checked with standard Snellen charts). Materials The materials used for Study 2 were identical to the materials used in Study 1, except for the stimuli related to the new target category brand FORD, which were replaced with stimuli related to BMW (see figure 1 ). As in the previous study, brand associations were measured on the same six bipolar attribute dimensions. Each pole (or attribute category) was represented by the same three word stimuli. General Procedure and Design The procedure and design of Study 2 was identical to that of Study 1. Procedure and design of the md-IATs The procedure and design of the md-IATs was identical to that of Study 1.

Results Study 1 Data preparation As noted earlier, IAT effects are based on differences in reaction times between two experimental tasks: the initial combined task(s) (B3 & B4) and the reversed combined task(s) (B6 & B7). This difference, however, may be computed in different ways. Earlier studies were based on an algorithm that involved dropping the first two trials of each block, discarding subjects‘ trials with responses either below 300 ms or above 3,000 ms—and ultimately, log-transforming the resulting values before computing the IAT-effect by subtracting the averaged log-transformed values of B4 from B7. Recently, Greenwald et al. [35] introduced a new scoring algorithm, the D measure , which has since then been adopted by most researchers [22] , [46] , [47] – [50] . Lane, Banaji, Nosek, and Greenwald [51] recommended the new algorithm, as it proved to be superior to the conventional algorithm in minimizing: (1) the correlation between IAT effects and individual subjects‘ average response latencies, (2) the effect of the order of the IAT blocks, and (3) the effect of previously completing one or more IATs on IAT scores, while (4) retaining strong internal consistency and (5) maximizing the correlation between implicit and explicit measures. The present research opted for a variant of the new scoring algorithm that differed exclusively in terms of its outlier treatment. Instead of using an absolute outlier criterion—dropping trials above 10,000 ms as suggested by Greenwald et al. [35] —boundaries for outliers were set dynamically. For each individual on each of the 18 IATs, trials outside the boundary defined by the mean response latency + 2.5 SD s (standard deviations) were excluded from further analysis following the advice of Carbon and Leder [52] . Table 2 gives step-by-step instructions for the adapted D measure algorithm. All of the results reported further below are based on the adapted D measure . Note: analyses relying on the regular D measure (without the dynamic outlier criteria) yielded similar results. Main Results The experiment was based on a 6×3 ( attribute dimension x brand cue ) within-subjects design. Table 3 lists all of the 18 IATs in each factor combination, providing both weighted means in milliseconds and means according to the adapted D measure along with their standard deviations ( SD ) and effect sizes ( d ). The adapted D measure served as input for the statistical analyses. The average effect size across all 18 IATs amounted to d = .34. As mentioned above, two participants had to be excluded because of an above average overall error rate exceeding 10% of total trials. A repeated-measures ANOVA with the two within-subjects variables attribute dimension (i.e., (1) safe–unsafe, (2) young–old (3) reliable–unreliable (4) aggressive–peaceful (5) environmentally friendly–non-environmentally friendly (6) innovative–conventional) and brand cue (i.e., logo, signature, product) revealed a main effect of attribute dimension, F GG (2.60, 64.94) = 7.98, p <.001, η p 2 = .24 (corrected for Greenhouse-Geisser). Mauchly's test of sphericity showed that the assumption of sphericity had been violated, X 2 (14), p <.001; degrees of freedom were corrected according to Greenhouse-Geisser estimates of sphericity (ε = .52). All F -values missing the subscript “ GG ” were not corrected. This main effect, however, was not relevant for the objectives of the present research. Differences for the factor attribute dimension were expected, simply because each dimension was intended to measure unique aspects of the overall attitude. As expected, there was no main effect observed for the other factor, brand cue, F (1.51, 37.77) = 1.05, p = .34, ns. Likewise, we did not find an interaction between attribute dimension and brand cue , F (10, 250) = 1.68, p = .09, ns. Figure 2 shows that the variable brand cue only accounted for relatively minor variations within each of the six attribute dimensions. Reliability and Validity of the md-IAT To calculate split-half reliabilities for each of the 18 IATs in the md -IAT, we followed the procedure by De Houwer and De Bruycker [22] . For each IAT, we first listed all the trials by order of appearance, separately for each stimulus type (AUDI, FORD, positive, negative), test block (AUDI-positive, FORD-positive) and participant. Following this, separate IAT-effects (operationalized by the adapted D measure ) were calculated for odd and even subsets of those trial-response lists. The average split-half reliability in Study 1 was r = .79, SD = .13. Table 4 provides the split-half reliabilities for all of the 18 IATs. To obtain estimates of the md -IAT's convergent validity we compared the results from the IATs to direct (or explicit) ratings of the same six attribute dimensions. The relationship between indirect and direct measures was assessed by several linear regressions—one for each of the six attribute dimensions. IAT-effects were averaged across the three levels of the factor brand cue (following the non-significant main effect in the ANOVA) and subsequently compared to direct measures. Based on previous meta-analyses [53] , [54] , relationships were expected to be positive, varying in magnitude due to factors such as social desirability or ability to introspect. Therefore, all of the p -values reported in Table 5 are based on one-tailed tests of significance. Besides the main interest in the present study to develop and evaluate the md -IAT as an attitudinal, multi-dimensional measure of brand associations, we gained interesting information about the two brands. Ratings derived from the semantic differentials were converted into a difference score in order to make them comparable to the IAT-effect scores. Averaged across the three levels of the factor brand cue , the results showed a small effect for the dimensions young–old ( M = −.10, SD = .42, d = −.23), reliable–unreliable ( M = .09, SD = .39, d = .27), and innovative–conventional ( M = −.07, SD = .39, d = −.20)—with FORD being stronger associated with “young” and “innovative” and AUDI being stronger associated with “reliable”. In addition, the results showed a medium effect for the dimensions aggressive–peaceful ( M = .31, SD = .52, d = .61) and environmental–non-environmental ( M = −.17, SD = .33, d = −.50), with AUDI being stronger associated with “aggressive” and FORD being stronger associated with “environmental”. According to Cohen [55] absolute effect sizes are classified as small, medium, and large, for the following values, d = 20, d = 50, d = 80, respectively. Study 2 Data preparation Study 2 utilized the same algorithm as Study 1. Main Results The experiment was based on a 6×3 ( attribute dimension x brand cue ) within-subjects design. Table 6 lists all of the 18 IATs in each factor combination, providing both weighted means in milliseconds and means according to the adapted D measure , along with their standard deviations and effect sizes. The average effect size across all 18 IATs amounted to d = .51. A repeated-measures ANOVA revealed a main effect of attribute dimension, F GG (2.56, 64.08) = 10.31, p <. 001, η p 2 = . 29). Mauchly's test showed that the assumption of sphericity had been violated, X 2 (14) = 42.82, p <.001; degrees of freedom were corrected according to Greenhouse-Geisser estimates of sphericity (ε = .51). This main effect, however, was not relevant for the objectives of the present research. As in Study 1, differences for the factor attribute dimension were irrelevant (and partly expected), simply because each dimension was intended to measure unique aspects of the overall attitude. As expected, there was no main effect observed for the other factor, brand cue, F(2, 50) <1, p = . 88 , ns. Likewise, we did not find an interaction between attribute dimension and brand cue, F(10, 250) <1, p = . 60 , ns. Similar to Study 1, figure 3 shows that the factor brand cue accounted only for relatively minor variations within each of the six attribute dimensions. Differences were a bit larger for the attribute dimensions 4 (aggressive–peaceful) and 5 (environmental–non-environmental). Reliability and Validity of the md-IAT As in Study 1, reliabilities were calculated based on an odd–even split of the trial-responses, following the procedure by De Houwer and De Bruycker [22] . The average split-half reliability in Study 2 was r = .79, SD = .09. Again, refer to table 4 for the split-half reliabilities for all of the 18 IATs. Estimates of the md -IAT's convergent validity were obtained by comparing the results from the IATs to direct (or explicit) ratings of the same six attribute dimensions following the same procedure as in Study 1. Table 7 below shows the results of the regression analyses. As in Study 1 the results of the md -IAT also revealed interesting information about consumers' brand associations for the two brands. Averaged across the three levels of the factor brand cue , the results showed a small effect for the dimensions safe–unsafe ( M = .21, SD = .46, d = .47), reliable–unreliable ( M = .08, SD = .35, d = .23) and innovative–conventional ( M = .14, SD = .33, d = .44)—with AUDI being stronger associated with the attributes “safe”, “reliable”, and “innovative”. In addition, the results showed a medium effect for the dimensions young–old ( M = .30, SD = .45, d = .68), aggressive–peaceful ( M = −.28, SD = .51, d = −.56) and environmental–non-environmental ( M = .23, SD = .35, d = .66)—with AUDI being stronger associated with “young” and “environmental” and BMW being stronger associated with “aggressive”.

Discussion With the Implicit Association Test Greenwald et al. [1] have radically innovated research on attitudes in general. Over the last decade the IAT has become the most popular indirect measure of attitudes, welcomed by researchers and marketing practitioners alike as a tool to measure attitudes in a rather indirect and implicit way, unlike common explicit measures such as verbal self-reports. The IAT is deemed to be a promising alternative, particularly for measuring attitudes consumers may not be aware of, able to express, or willing to share with the researcher [2] , [3] . The multi-dimensional Implicit Association Test ( md -IAT) constitutes an extension of the IAT procedure that goes beyond measuring attitudes on a single dimension only (e.g., good–bad); that is, with the md -IAT it is possible to measure different nuances of a global attitude (e.g., on scales such as safe–unsafe; young–old; innovative–conventional; etc.). As a consequence, the md -IAT procedure (i.e., multiple measurement on more than just one attribute dimension) yields a more detailed representation of consumers' evaluations of a brand or product. Being of high practical relevance, this gain in dimensionality provides more insight and therefore more opportunities for specific intervention. The results of Study 1 (“AUDI/FORD”) and Study 2 (“AUDI/BMW”) provide strong evidence of the md -IAT's methodological appropriateness. Split-half reliabilities averaged r = .79 (n = 2×18 IATs) for both studies. For comparison, in a meta-analysis Hofmann et al. [26] reported the same mean reliability of r = .79 ( n = 50) for the IAT. Regarding the md -IATs convergent validity, regression analyses of the six md -IAT dimensions and the direct measures revealed that, except for one dimension (reliable–unreliable), R -values (for simple regressions R -values are identical with the correlation coefficients) were of close to average or above average magnitude: Hofmann et al. [26] reported an average indirect–direct correlation of .34 for consumer research related studies (based on n = 11 independent studies). Considering this meta-analytic finding, the results of the present studies fit well into the overall picture. As a test of external validity we varied stimuli in the perceptual domain (i.e., through the three levels of the factor brand cue : logo, signature, product). In both studies, the factor brand cue was not significant and therefore accounted only for minor variations of the adapted D measure means within each of the six attribute dimensions. These results show that the md -IAT can be rather seen as a conceptual measure of brand associations—widely unaffected by perceived stimulus variations (characteristics) in the perceptual domain of a brand [56] . While this is—in most cases—viewed as an advantage, the md -IAT is therefore less suited for testing the impact of specific (product) designs (e.g., visual identifiers) on brand associations. As a last indicator of methodological appropriateness, sensitivity of the md -IAT can be regarded as reasonable. Despite the fact that the two brands used in Study 2 (“AUDI/BMW”) are commonly perceived to be highly similar, which could make finding differences difficult, we did not find any decrease in sensitivity compared to the two brands used in Study 1 (“AUDI/FORD”). On the contrary, effect sizes averaged d = .34 in Study 1 and d = .51 in Study 2 across all 6×3 IATs part of the md -IAT, indicating a small average and a medium average effect, respectively. Finally, brand attitudes as revealed by the md -IAT indicate that FORD is judged “slightly younger”, “slightly more innovative”, “more environmental”, “less aggressive”, but also “less reliable” than AUDI. Study 2 revealed that AUDI is judged as “slightly safer”, “slightly more reliable”, “slightly more innovative”, “younger”, “more environmental”, and “less aggressive” than BMW. Conclusions Based on the results of the present research, the multi-dimensional Implicit Association Test ( md -IAT) has shown to be a reliable, valid, and sensitive indirect measure of brand attitudes. Regular one-dimensional IATs are useful if one is only interested in an overall brand attitude (e.g., are people's attitudes more favorable toward AUDI or to BMW?). The main advantage of the md -IAT lies in its more detailed, multi-dimensional assessment. Marketing practitioners in particular might value the additional information offered by the md -IAT, for example allowing them to easily create complex and differentiated brand profiles, and thus distinguishing between different components of an overall brand attitude (i.e., tapping into the multifaceted nature of consumers' brand associations). Similarly, academics might find the md -IAT useful for testing constructs such as brand or product personality [9] , [10] also with indirect measures. Just as the IAT, its multi-dimensional extension ( md -IAT) is better suited for measuring attitudes consumers are not consciously aware of, able to express, or willing to share with the researcher [2] .

Conclusions Based on the results of the present research, the multi-dimensional Implicit Association Test ( md -IAT) has shown to be a reliable, valid, and sensitive indirect measure of brand attitudes. Regular one-dimensional IATs are useful if one is only interested in an overall brand attitude (e.g., are people's attitudes more favorable toward AUDI or to BMW?). The main advantage of the md -IAT lies in its more detailed, multi-dimensional assessment. Marketing practitioners in particular might value the additional information offered by the md -IAT, for example allowing them to easily create complex and differentiated brand profiles, and thus distinguishing between different components of an overall brand attitude (i.e., tapping into the multifaceted nature of consumers' brand associations). Similarly, academics might find the md -IAT useful for testing constructs such as brand or product personality [9] , [10] also with indirect measures. Just as the IAT, its multi-dimensional extension ( md -IAT) is better suited for measuring attitudes consumers are not consciously aware of, able to express, or willing to share with the researcher [2] .

Conceived and designed the experiments: VG C-CC. Performed the experiments: VG. Analyzed the data: VG. Contributed reagents/materials/analysis tools: VG C-CC. Wrote the paper: VG C-CC MCS. Background The authors present a procedural extension of the popular Implicit Association Test (IAT; [1] ) that allows for indirect measurement of attitudes on multiple dimensions (e.g., safe–unsafe; young–old; innovative–conventional, etc.) rather than on a single evaluative dimension only (e.g., good–bad). Methodology/Principal Findings In two within-subjects studies, attitudes toward three automobile brands were measured on six attribute dimensions. Emphasis was placed on evaluating the methodological appropriateness of the new procedure, providing strong evidence for its reliability, validity, and sensitivity. Conclusions/Significance This new procedure yields detailed information on the multifaceted nature of brand associations that can add up to a more abstract overall attitude. Just as the IAT, its multi-dimensional extension/application (dubbed md -IAT) is suited for reliably measuring attitudes consumers may not be consciously aware of, able to express, or willing to share with the researcher [2] , [3] .

Parts of this article are based on the unpublished diploma thesis of VG at the University of Vienna, supervised by CCC . The authors would like to thank Oscar Person, Katarina Hellén and one anonymous reviewer for their helpful comments on the manuscript, and Judith Forstner for collecting parts of the data. Furthermore, the authors would like to thank the assigned editor Susanne Hempel.

CC BY

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15849

oa_package/be/0e/PMC3016338.tar.gz

PMC3016339

21167059

Background Treatments of inner ear diseases with traditional strategies have had limited success and the goal of sensory organ and nerve repair or regeneration has yet to be achieved. One of the greatest challenges has been the limitations to passage through the blood-inner ear barriers and uncontrollable delivery of treatment agents after either intravenous or oral administration (Figure 1 ) [ 1 ]. Multifunctional nanoparticles (MFNPs) are a promising new means for the delivery of gene or drug to the inner ear for the treatment of inner ear diseases including sensorineural hearing loss (SNHL) and Meniere's disease. Eight types of MFNPs are currently under development within the European Union consortium, Nanoear [ 2 ] with the intention that they can be applied with an intratympanic, minimally-invasive approach in the clinic. MFNPs can be functionalized with the application of surface ligands to improve cellular or nuclear internalization to improve the targeted delivery of therapeutic agent. In some of our preliminary studies lipid nanocapsules were shown using confocal microscopy to become distributed in different cochlear cell populations and liposome nanoparticles functionalized with TrkB ligand were internalized into cochlear cells after round window membrane permeation [ 3 , 4 ]. In the process of optimizing the functionalization of each type of nanoparticle, it is essential to be able to trace their passage through the cochlea. However, it is problematic to track the MFNPs in the inner ear by histological means. First, the appearance of autofluorescence from lipofuscin granules, which are found in the cochleae of humans, chinchillas, guinea pigs and rats, impairs the detection of the fluorescence signals from the MFNP targets [ 5 - 10 ]. Second, the inner ear is composed of fluids, soft tissue, and bone [ 1 ] (Figure 1 ). This results in signal loss in histological study, which is incapable of retaining the inner ear fluids. Finally, it is inconvenient to observe the dynamics of MFNPs in the inner ear in histological studies. MRI is an excellent tool to trace MFNPs that are labeled with a contrast agent through the inner ear by detecting their signal changes in both the inner ear fluids and soft tissue in vivo . The critical regions of the cochlea, including the lateral wall, spiral ganglion cells, and cochlear nerve, are also demonstrated with the resolution of a 4.7 T MRI [ 11 , 12 ]. Gadolinium- tetra-azacyclo-dodecane-tetra-acetic acid (Gd-DOTA) was shown to be efficiently taken up into rat inner ears and detected at 60 min post-transtympanic injection with greater signal intensity seen in the scala vestibuli than in the scala tympani and no uptake into the endolymph after four hours [ 11 ]. In the present study, amphiphilic MRI-traceable liposome nanoparticles were first developed from 1,2-ditetradecanoyl-sn-glycero-3-phosphoethanolamine-N-diethylenetriaminepentaacetic acid (gadolinium salt) [DMPE-DTPA (Gd)] (LPS-DTPA-Gd). The contribution of effective Gd (water accessible Gd) to MR signal was evaluated by disassembling Gd chelate lipid nanoparticles with 5% sodium dodecyl sulfate (SDS). Finally, effective MRI traceable liposome nanoparticles were developed by encapsulating Gd-DOTA in the nanoparticles. These nanoparticles, which are abbreviated LPS+Gd-DOTA, were finally visualized in vivo in the rat inner ears with MRI.

Methods Reagents and animals Sphingosine (Sph), egg phosphatidylcholine (EggPC), and 1, 2-distearoyl- sn -glycero-3-phosphoethanolamine-N-[methoxy(polyethyleneglycol)-2000] (ammonium salt) [DSPE-PEG2000], 1,2-ditetradecanoyl- sn -glycero-3-phosphoethanolamine-N-diethylenetriaminepentaacetic acid (gadolinium salt) [DMPE-DTPA (Gd)], were from Avanti polar lipids (Alabaster, AL). Gd-DOTA (DOTAREM) was from Guerbet, Cedex, France. Hepes, and EDTA, were from Sigma. The purity of lipids was checked by thin-layer chromatography on silicic acid coated plate (Merck, Darmstadt, Germany) developed with a chloroform/methanol/water mixture (65:25:4, v/v/v). Examination of the plates after iodine staining, and when appropriate, upon UV illumination revealed no impurities. Lipid concentrations were determined gravimetrically with a high precision electrobalance (Cahn, Cerritos, CA). The other chemicals were of analytical grade and from standard sources. Twenty-six male Wister rats, weighing between 270 g and 440 g, were provided by the Experimental MRI Laboratory, Department of Neurology, Helsinki University Central Hospital, FIN-00029 HUS, Helsinki, Finland. All animal experiments were approved by the Ethical Committee of the University of Tampere (permission: LSLH-2006-4143/Ym23). Animal care and experimental procedures were conducted in accordance with European legislation. Animals were divided into 3 groups: transtympanic injection of LPS+Gd-DOTA (TTI), intracochlear administration of LPS+Gd-DOTA with open perilymph circulation (IC-OPC), and intracochlear administration of LPS+Gd-DOTA with close perilymph circulation (IC-PPC) (Table 2 ). In TTI group, 2 rats received injection of unpurified LPS+Gd-DOTA and 12 rats were injected purified LPS+Gd-DOTA. All experimental procedures were performed under general anaesthesia, induced by intraperitoneal injections of medetomidine hydrochloride (0.5 mg/kg, Domitor, Orion, Finland) and ketamine (75 mg/kg, Ketalar, Pfizer, UK) and maintained at half dosage of the agents with the animal's eyes protected by Viscotears ® (Novartis healthcare A/S, Danmark). Preparation of liposomes containing Gd-DTPA-DMPE (LPS-DTPA-Gd) Appropriate lipids along with DMPE-DTPA (Gd) (Gd chelate lipid) were mixed and solvent was evaporated under nitrogen to obtain a thin lipid film which was subsequently dried overnight under nitrogen. Lipid film was hydrated with buffer and resulting MLV's were then sequentially extruded 400 (five times), 200 (five times), and 100 nm (seven times) polycarbonate Millipore filters to yield unilamellar liposomes containing gadolinium in both leaflets (Figure 10A ). The size of LPS-DTPA-Gd was 130 ± 20 nm. Preparation of liposomes encapsulating Gd-DOTA (LPS+Gd-DOTA) Lipids were dissolved in chloroform and subsequently mixed in this solvent so as to obtain the desired compositions. The solvent was removed using a gentle stream of nitrogen whereafter the dry lipid residues were maintained under reduced pressure for at least 24 h to remove trace amounts of chloroform. The lipid film was hydrated with Gd-DOTA solution (500 mM, Guerbet, Cedex, France) at 60°C for 60 min. The resulting lipid suspensions were extruded (Avestin, Ottawa, Canada) through 400 (five times), 200 (five times), and 100 nm (seven times) polycarbonate Millipore filters to yield large unilamellar vesicles (LUV's) containing Gd-DOTA both inside as well as outside of the liposomes (Figure 10B ). After extrusion, external gadolinium was removed by Sephadex G-50 (fine) quick spin columns (Roche Diagnostics GmbH, Mannheim, Germany). Before and after removal of the external Gd-DOTA average particle size (Z av , using DLS, Zetasizer Nano ZS, Malvern Instruments Ltd., UK) of 300 ± 20 and 130 ± 20 nm was obtained (Figure 11 ). All LPS+Gd-DOTA nanoparticles contain 500 mM Gd-DOTA. Liposomes containing Gd-DOTA were stable even after one week of preparation as confirmed by DLS, revealing no changes in Z av . Phantom study A 4.7 T MR scanner with bore diameter of 155 mm (PharmaScan, Bruker BioSpin, Germany) was used in both phantom and in vivo MR measurements. The maximum gradient strength was 300 mT/m with an 80-μs rise time. A dedicated rodent head coil (linear bird cage coil) with diameter of 38 mm was used for the phantom and animal studies. To evaluate T1 and T2 relaxation times of LPS-DTPA-Gd nanoparticles and analyze the contribution of water-accessible Gd-DTPA to MR signal, the nanoparticle solutions were prepared with physiological saline at a concentration of (1 mM containing 0.5 mM Gd). 5% SDS was applied to disassemble the nanoparticles and expose all the lipid-bond Gd-DTPA to the water. This was done to detect MRI signal from all of the Gd-DTPA associated with the nanoparticles, including that amount within the nanoparticles that could be shielded by the lipid from proton interaction. Gd-DOTA diluted with physiological saline at a concentration of 0.5 mM was used as a positive control. Plastic phantom tubes (400 μl, AgnTho's AB, Sweden) were filled with the solutions of Gd-DOTA, LPS-DTPA-Gd nanoparticles, or LPS-DTPA-Gd nanoparticles plus 5% SDS. In order to measure relaxivities, r1 and r2, of LPS+Gd-DOTA, solutions of various concentrations were placed into plastic phantom tubes arranged concentrically within a 50 ml syringe. Dilutions were made with artificial perilymph containing 145.5 mM NaCl, 2.7 mM KCl, 2.0 mM MgSO4, 1.2 mM CaCl2, 5.0 mM HEPES, pH adjusted to 7.4 [ 16 ]. Negative controls were prepared using either plain artificial perilymph or HEPES buffer. Each sample was prepared in duplicate. T2 relaxation time was determined using a multi-slice multi-echo sequence (MSME), based on CPMG (Carr-Purcell Meiboom-Gill) spin echo (SE) [repetition time (TR) 1500 ms, echo time (TE) 7-229 ms, 32 echo times, matrix size 128 × 128, single slice, slice thickness 2.0 mm, field of view (FOV) 5.0 cm, resolution 0.098 × 0.130 mm2, number of excitation (NEX) 3]. T1 relaxation times were determined by rapid acquisition with relaxation enhancement (RARE) sequence with variable TR (TR 100, 432, 859, 1458, 2472, 7500 ms, TEeff 8.7 ms, RARE factor 2, NEX 1, matrix size 128 × 128, FOV 5 cm, single slice with slice thickness 2.0 mm). Transtympanic injection of LPS+Gd-DOTA LPS+Gd-DOTA nanoparticles were administered through transtympanic injection under an operating microscope as previously reported [ 11 ]. An incision of approximately 1 mm length was made in the anterior superior quadrant of the tympanic membrane by a 25 gauge needle to release any potential air bubble. Either purified or unpurified 0.04 ml LPS+Gd-DOTA (1 mM LPS+Gd-DOTA containing 500 mM Gd-DOTA) were injected into the middle ear cavity through the posterior superior quadrant of the tympanic membrane. As a positive control, Gd-DOTA (100 mM) was transtympanically injected as above. Intracochlear administration of LPS+Gd-DOTA to open perilymphatic compartments The bulla was exposed through a post-auricular approach. Working under the operating microscope, a hole was drilled through the bulla with a 2 mm diameter burr. The round window membrane was identified superior to the stapedial artery as reported previously [ 1 ]. The scala tympani of the basal turn was opened with a 0.5 mm cutting burr inferior to the stapedial artery and the round window. A second opening was made in the scala vestibuli of the basal turn. After wicking away the perilymph that egressed through the opening, two pieces of gelatin sponge soaked in LPS+Gd-DOTA (the size of each piece was proximately 3 mm 3 after soaking in the LPS+Gd-DOTA solution) were placed against the openings in the middle ear cavity. The bulla opening was covered by muscle and the wound was sutured closed. Intracochlear injection of LPS+Gd-DOTA into closed perilympatic compartments After drilling into the scala tympani of the basal turn as above, polyimide tubing (MicroLumen Inc., Tampa, USA) was inserted intraluminally into a polyethylene tubing (I.D. 0.28 mm, O.D. 0.61 mm, Becton Dickinson, Franklin Lakes, USA) and fixed with nail polish. The tubing was filled with LPS+Gd-DOTA solution. After opening the scala tympani of the basal turn with a 0.5 mm cutting burr, the high performance polyimide tubing tip (MicroLumen, Tampa, USA), which was connected to polyethylene tubing (I.D. 0.28 mm, O.D. 0.61 mm, Becton Dickinson and Company, USA) and primed with LPS+Gd-DOTA solution was inserted into the scala tympani and was sealed circumferentially with Histoacryl (enbucrilate) glue (Aesculap AG, Tuttlingen, Germany). LPS+Gd-DOTA nanoparticles were initially applied by bolus injection of 5 μl, but they failed to demonstrated any MRI detectable signal for uncertain reasons. Although it is a physiologically excessive volume, 50 μl was injected and produced satisfying MR signal. The remaining observations were performed with 50 μl LPS+Gd-DOTA. The bulla opening was covered by muscle and the wound was sutured closed. In vivo MRI observation of LPS+Gd-DOTA in the inner ear The body temperatures of the rats were maintained by circulating warm water and their respirations were recorded with Physio Tool-1.0.b.2 program (Bruker, Germany). Rats were placed in the magnet with the ears positioned at the isocenter. 2D and 3D MRI measurement was performed. T1-weighted 2D images were acquired with RARE sequence (TR/TEeff 500/10 ms, RARE factor 4, matrix size 256 × 192, slice thickness 0.5 mm, FOV 2.5 × 2.5 cm2, resolution 0.098 × 0.13 mm2, NEX 33). T1-weighted 3D images were acquired with RARE sequence (TR/TEeff 500/12 ms, RARE factor 16, matrix size 64 × 64 × 64, FOV 0.89 × 0.89 × 0.89 cm3, resolution 0.139 × 0.139 × 0.139 mm3, NEX 2). For the groups who had either intracochlear administration of LPS+Gd-DOTA with open perilymph circulation or transtympanic injection of LPS+Gd-DOTA, MRI scanning commenced immediately after the procedure. After setting up the geometry with T2-weighted 2D imaging, T1-weighted 2D, and T1-weighted 3D images were acquired at different time points. In the group of intracochlear injection of LPS+Gd-DOTA with closed perilymph circulation, LPS+Gd-DOTA nanoparticles were injected after setting up the MRI geometry. Then T1-weighted 2D, and T1-weighted 3D images were acquired as above. MRI measurements were followed up to 48 h in the group of transtympanic injection of LPS+Gd-DOTA and for one hour in the group of intracochlear administration of LPS+Gd-DOTA with open perilymph circulation (Table 2 ). 3D volume rendering of the inner ear MRI OsiriX v3.3.2 (OsiriX Foundation, Geneva, Switzerland) software was applied for 3D rendering of the 3D raw stacked MR images of the inner ear. 3D volume renderings were created for visualisation of the inner ear structures with the ability to rotate the images into various orientations and to create photographs. Quantification of signal intensity and statistics ParaVision PV 4.0 (Bruker, Germany) software was used for post-processing of MR images and calculations of T1 and T2 relaxation times in phantom and signal intensity measurements in the inner ear compartments. Signal intensities in the region of interest were measured in the middle ear cavity, the vestibule, the scala vestibuli and scala tympani in the higher basal turn, and the brainstem using ParaVision 4.0. Differences in the signal intensities between 3 h and 6 h time points among these compartments were compared with student t-test. Signal intensity ratio of the vestibule, the scala vestibuli, and scala tympani over the middle ear was defined as the indicator of transport efficacy and compared between LPS+Gd-DOTA and Gd-DOTA groups with student t-test. A difference was considered to be statistically significant at P < 0.05. Photoshop CS3 software was used for labeling and demonstration of the inner ear anatomy in MR image.

Results Contribution of effective Gd to MR signal intensity of LPS-DTPA-Gd In phantom testing, LPS-DTPA-Gd showed characteristics of a T1-contrast agent, but the maximum available concentration of the nanoparticles was unable to induce enough signal intensity for in vivo application. Nanoparticles dissociated by 5% SDS demonstrated T1 and T2 relaxation times similar to that of Gd-DOTA with the same molar concentration of Gd (Table 1 ). As the ratio of Gd chelate to lipid within the material was already maximized, an alternative strategy was pursued to enhance the r1 of the nanoparticles. LPS+Gd-DOTA nanoparticles were developed by encapsulating Gd-DOTA into the aqueous core of liposomes to obtain nanoparticles better visible by MR imaging. T1 relaxation time of purified LPS+Gd-DOTA at 160 folds dilution was close to that of LPS-DTPA-Gd at 1 mM concentration while T1 relaxation time of unpurified LPS+Gd-DOTA at 320 folds dilution was still significantly shorter than both of them (Table 1 ). Unpurified LPS+Gd-DOTA in artifical perilymph showed r1 of 2.2 S -1 mM -1 and r2 of 4.4 S -1 mM -1 while the purified LPS+Gd-DOTA generated r1 of 0.033 S -1 mM -1 and r2 of 0.037 S -1 mM -1 in both artifical perilymph and HEPES. Transportation of LPS+Gd-DOTA through the middle-inner ear barriers When applied to the middle ear cavity through transtympanic injection, 1 mM LPS+Gd-DOTA nanoparticles containing 500 mM Gd-DOTA were detected in the middle ear cavity with strong T1-weighted signal which decayed between 3 h to 6 h post administration (Figure 2 and Figure 3 ). In the inner ear, LPS+Gd-DOTA nanoparticles were efficiently detected in the perilymph at 3 h post-administration, which reached 21.1% (average of the vestibule, scala vestibuli, and scala tympani) of that in the middle ear cavity (Figure 2 and Figure 3 ). After 6 h, the singal intensity in the inner ear slightly increased though statistic significance was not achieved indicating that more LPS+Gd-DOTA nanoparticles were gained in the inner ear. 3D rendering image showed that abundant nanoparticles retained in the vestibule and perilymphatic compartments of the basal lower turn (Figure 4 ). Within the cochlea, additional nanoparticles had diffused into the higher turns of the cochlea, which made the perilymph in the second turn and apex more visible. The signal intensity in the brainstem was also slightly enhanced at 6 h in comparison to 3 h though without statistic significance (Figure 2 and Figure 3 ). An important phenomenon was noticed, in that the uptake of purified LPS+Gd-DOTA in the inner ear was influenced by the posture of the animal after transtympanic injection. In order to be able to scan immediately following injection, two rats were placed into the MRI machine in the prone position and LPS+Gd-DOTA nanoparticles were not detected in their inner ears at 3 hours post-administration. However, when rats were laid in the lateral position with the exposed ear upwards for about 2.5 h before scanning, obvious uptake of LPS+Gd-DOTA occurred in the inner ear at 3 h post-administration (Figure 2 ). LPS+Gd-DOTA nanoparticles were not detectable in the inner ear after neither 24 h nor 48 h in either group. Unpurified LPS+Gd-DOTA nanoparticles were efficiently detected in the perilymph of the inner ear at 40 min post-transtympanic injection (Figure 5 ). When Gd-DOTA (100 mM) alone was injected into the middle ear cavity, universal distribution in the inner ear was observed within 1 h (Figure 6 ). When same regions of interest were quantified, transport efficacy of Gd-DOTA across the middle-inner ear barriers into the inner ear compartments was significantly higher than that of LPS+Gd-DOTA at 3 h post-administration (p < 0.01, student t-test) (Figure 7 ). Distribution of LPS+Gd-DOTA in the inner ear after intracochlear injection Intracochlear administration of LPS+Gd-DOTA with a volume 5 μl (1 mM LPS+Gd-DOTA containing 500 mM Gd-DOTA) in a closed perilymph circulation system through a catheter failed to produce visible signal in the inner ear. 10 μl volume of the same LPS+Gd-DOTA induced bright signal on T1-weighted images mainly in the region adjacent to the injection, that is, in the basal turn of the scala tympani, scala vestibuli, and modiolus immediately after intracochlear injection, (Figure 8 ). However, no further diffusion to the distal locations of the inner ear was visualized with prolonged observation time. When 50 μl was injected, using higher pressure to drive the nanoparticles, a broader distribution of LPS+Gd-DOTA in the inner ear was visualized, including the cochlea and the adjacent scala vestibuli. An abundance of LPS+Gd-DOTA nanoparticles were demonstrated in the modiolus at the level of the basal turn. In the scala vestibuli, the signal intensity was lower than in the scala tympani (Figure 8 ). At 56 min, LPS+Gd-DOTA nanoparticles diffused to distal sites, such as the apex of the cochlea and the semicircular canals in the vestibular organ (Figure 8 ). Intracochlear administration of LPS+Gd-DOTA nanoparticles through gelatin sponge applied to an open perilymph system, with windows created in both the scala vestibuli and scala tympani, induced highly intense bright signal in the inner ear on T1-weighted images acquired at 30 min (Figure 9 ). At this time point, the signal intensity in the perilymph of the vestibule was higher than in the perilymph of the cochlea, which was in turn higher than the modiolus. Homogeneous distribution of LPS+Gd-DOTA nanoparticles was observed throughout the perilymphatic compartments of the cochlea. A dynamic reduction of signal in the hook region of the scala tympani and the modiolus of the cochlea was demonstrated, which became faint at 120 min post-administration (Figure 9 ). In the vestibular organ, gradual enhancement occurred in the semicircular canals over 120 min while no visible change was found in the vestibule, which remained bright throughout this period (Figure 9 ).

Discussion In the present work, we demonstrated that multifunctional liposome nanoparticles, which were tagged with gadolinium, were visualized by MRI. T1 signal enhancement correlates to the amount of water accessible to the gadolinium, shown by the phenomenon that SDS treatment induced T1 relaxation time of LPS-DTPA-Gd equivalent to that of Gd-DOTA at the same concentration. It was reported that Gd-DOTA and Gd-DTPA showed similar relaxivities [ 13 ]. This suggested that each LPS-DTPA-Gd molecule became water accessible upon dissolution by SDS. Because the ratio of Gd to lipid is nonadjustable in LPS-DTPA-Gd , there was difficulty in generating sufficient MR signal. Alternatively, encapsulation of Gd-DOTA into the liposomes was an improved technique to increase the amount of gadolinium available for MR imaging. With this method it was possible to encapsulate 500 mM Gd into the liposomes compared to LPS-DTPA-Gd where the maximum concentration of incorporated Gd was 0.5 mM. Moreover, highly monodisperse NPs were obtained when Gd-DOTA was encapsulated into the liposomes. This is critical for quality control in pharmaceutics, and correlates with treatment efficacy. It is important to note that the gadolinium within LPS+Gd-DOTA nanoparticles is carried in the aqueous core and the shell can be manufactured to be identical to that of the treatment nanoparticles. Therefore, this MRI study is directly translatable for predicting the destination of these liposome nanoparticles in clinical use to carry genes or drugs. Gd-DOTA-tagged liposome nanoparticles were demonstrated within the inner ears of rats in vivo using MRI. After intracochlear injection, nanoparticles diffused throughout the inner ear efficiently and immediately reached the modiolus, which includes the spiral ganglion cells. Therefore, the multifunctional liposome nanoparticle is a promising vector to deliver treatment agents to inner ear targets from a cochlear implant, where the spiral ganglion cells are a major concern. After intracochlear injection, distribution of nanoparticles in distal destinations, such as the semicircular canals, within one hour indicated that liposome nanoparticle can also be applied in drug delivery to the vestibular organ. The efficacy of delivery was enhanced by injecting a larger volume, likely due to the increased driving force, or by opening a second window into the scala vestibuli to avoid the pulsatile egress of cerebrospinal fluid (CSF) from the scala tympani. This reduced distribution with lower injection volumes suggests that diffusion of the hydrophobic liposome nanoparticles in the hydrophilic perilymph may be problematic. In the clinic practice of cochlear implantation, the constant release of nanoparticles from an advanced drug delivery electrode could resolve this diffusion issue. Application of therapy into the middle ear, as a minimally invasive technique for inner ear drug delivery, is a favorable approach for the treatment of inner ear diseases that are not suitable for cochlear implantation. The outcome of intratympanic treatment would depend in a large part upon the efficacy of transportation from the middle ear to the inner ear. The existence of at least two pathways, including the round window membrane and the oval window, from the rat middle ear to the inner ear were recently supported by an in vivo MRI study [ 11 ]. There are a number of additional studies concerning round window membrane passage in the literature. The round window membrane appears well adapted for active transport. There are microvilli on the surface of outer epithelium and abundant organelles, such as mitochondria, rough endoplasmic reticulum, and a well-developed Golgi complex, that provide metabolic support and mechanisms for transport activity [ 14 ]. Our own unpublished data showed the appearance of clathrin in the round window membrane of rats, which suggests a specific pinocytosis. There is discontinuity in the basement membrane of the inner epithelium and loose junctions between inner epithelial cells that provide potential openings to support diffusion.. Both the surface characteristics and the size of the nanoparticles are important factors that affect the passage from the middle ear to the inner ear. One μm latex microspheres were reported to pass through the round window membranes of chinchillas and Rhesus monkeys [ 15 ]. In that study, the quantity of latex microspheres that passed through the round window membrane was unknown, but this is an important factor to define in studies of therapy directed toward inner ear diseases. In the present observations, liposome Gd nanoparticles of 130 nm passed the middle ear-perilymph barriers in quantities sufficient for MR imaging in 3 h and reached a signal intensity of above 1/5 of that in the middle ear cavity. Our preliminary histological study demonstrated augmented distribution of LPS+Gd-DOTA into the utricle of the vestibule than that in the spiral ligament capillary and spiral ganglion cells of the cochlea at 8 h following transtympanic injection of TRITC labeled LPS+Gd-DOTA in specimens taken from the rat after MRI study (data not shown). This implies the involvement of oval window in transporting LPS+Gd-DOTA. The oval window pathway was proved by abolish of immediate Gd-DOTA uptake in the vestibule and scala vestibuli induced by oval window sealing (data not shown). Potential uptake of TRITC labeled LPS+Gd-DOTA in other organs such as the brain, liver, and kidney should be investigated in a future study. These same advanced MFNPs showed targetability of their distribution in rat cochleae when functionalized with TrkB ligand [ 4 ] after intratympanic administration.

Conclusions In the present study, novel MRI visible multifunctional liposome nanoparticles were designed by encapsulating Gd-DOTA inside the hydrophilic core of the nanoparticles. Acceptable r1 values were realized in the final products. The Gd-DOTA-tagged liposome nanoparticles were visualized in the rat inner ear in vivo after both intracochlear and transtympanic injections. The intracochlear approach induced stronger signals on T1-weighted images. MRI also proved to be an excellent tool to monitor the purity of the Gd-DOTA-tagged liposome nanoparticles. There are potentially broad applications for these novel types of nanoparticles in future biomedical investigations.

Background Treatment of inner ear diseases remains a problem because of limited passage through the blood-inner ear barriers and lack of control with the delivery of treatment agents by intravenous or oral administration. As a minimally-invasive approach, intratympanic delivery of multifunctional nanoparticles (MFNPs) carrying genes or drugs to the inner ear is a future therapy for treating inner ear diseases, including sensorineural hearing loss (SNHL) and Meniere's disease. In an attempt to track the dynamics and distribution of nanoparticles in vivo , here we describe manufacturing MRI traceable liposome nanoparticles by encapsulating gadolinium-tetra-azacyclo-dodecane-tetra-acetic acid (Gd-DOTA) (abbreviated as LPS+Gd-DOTA) and their distribution in the inner ear after either intratympanic or intracochlear administration. Results Measurements of relaxivities (r1 and r2) showed that LPS+Gd-DOTA had efficient visible signal characteristics for MRI. In vivo studies demonstrated that LPS+Gd-DOTA with 130 nm size were efficiently taken up by the inner ear at 3 h after transtympanic injection and disappeared after 24 h. With intracochlear injection, LPS+Gd-DOTA were visualized to distribute throughout the inner ear, including the cochlea and vestibule with fast dynamics depending on the status of the perilymph circulation. Conclusion Novel LPS+Gd-DOTA were visible by MRI in the inner ear in vivo demonstrating transport from the middle ear to the inner ear and with dynamics that correlated to the status of the perilymph circulation.

Competing interests The authors declare that they have no competing interests. Authors' contributions JZ participated in the design of the study and performed the MRI measurement. RS and SR participated to the design of the liposomes and prepared the liposomes. DP participated in MRI measurements. UAR was responsible for the sequences of MRI. PK supervised the design and preparation of liposomes. IP supervised the study. All authors have read and approved the final manuscript.

Acknowledgements The authors gratefully acknowledge financial support from the European Community 6th Framework Program on Research, Technological Development and Demonstration (Nanotechnology-based Targeted Drug Delivery. Contract number: NMP4-CT-2006-026556, Project acronym: NANOEAR).

CC BY

no

2022-01-12 15:21:45

J Nanobiotechnology. 2010 Dec 18; 8:32

oa_package/20/a9/PMC3016339.tar.gz

PMC3016340

21144059

Background Carcinoid tumor was first detected as multiple tumors in the ileum by Lubarsch during autopsy more than 100 years ago [ 1 ]. The concept of this disease is an endocrine cell tumor with low-grade atypia, which is generally a low-grade malignant cancer with a good prognosis. About 70% of carcinoid tumors occur in the gastrointestinal tract, including the rectum, appendix, stomach, and ileum. The rest of them arise mainly from neuroendocrine cells in the pancreas, gonads and pulmonary bronchi, and they rarely occur in the kidney. Metastatic renal carcinoid is even rarer, with only one reported case [ 2 ]. Here, we present our experience of a patient with metastatic renal carcinoid with a review of the literature.

Discussion Carcinoid tumors are generally considered to arise from neuroendocrine cells, and secrete different substances depending on their primary site. Such tumors characteristically secrete serotonin, although some of them are inactive. Accordingly, urinary secretion of 5-hydroxyindole acetic acid (5-HIAA), a metabolite of serotonin, increases. Other hormones and biochemical substances secreted from carcinoid tumors include corticotropin, histamine, dopamine, substance P, neurotensin, prostaglandins and kallikrein. Secretion of these vasoactive substances, including serotonin, may cause symptoms of palpitations, hot flushes, wheezing, diarrhea, and skin flushing, as well as carcinoid syndrome such as right-sided valvular disease. Since serotonin and other similar substances are metabolized mainly in the liver, carcinoid syndrome is observed only when such substances flow directly into the systemic circulation from metastatic lesions in the liver, lung, gonads etc., without involving the liver. From the perspective of a histogenetic pathway, carcinoid is a benign tumor originating from epithelial stem cells, and highly malignant carcinoids are referred to as endocrine cell carcinoma. For differentiation of malignancy, immunostaining for p53 is useful. Nishikura et al . conducted a study in a total of 98 patients with gastric endocrine cell carcinoma or carcinoid tumor, and reported that patients with carcinoid tumor did not have any p53 protein overexpression or genetic mutation [ 3 ]. Moyana et al . investigated the association between differentiation of gastrointestinal carcinoid and MIB-1, p53, and bcl-2 in 58 patients, and reported that MIB-1 and p53 could be prognostic factors [ 4 ]. Nearly 100 cases of primary renal carcinoid have been reported since Resnick et al . reported the first case in 1966 [ 5 ]. There has been no established view on factors involved in the occurrence of renal carcinoid, but the following mechanisms have been hypothesized: 1) Neuroendocrine cells that do not normally exist in the kidney undergo neoplastic changes, 2) stem cells induce neuroendocrine differentiation, 3) the transitional epithelium of the renal pelvis develops intestinal metaplasia, and renal carcinoid originates from neuroendocrine cells present at this site, and 4) renal carcinoid arises from the migration of intestinal or bronchial epithelium into the kidney [ 6 ]. Romero et al . reviewed 56 case reports of renal carcinoid [ 7 ]. The median age of the patients was 49 years, with no sex difference. Horseshoe kidney was present in 17.8% of patients, and renal carcinoid was incidentally diagnosed in 28.6% of patients. The most common symptom was abdominal pain, and carcinoid syndrome was seen in 12.7% of patients. Many (73.6%) of the tumors were 4 cm or more in diameter, and as many as 45.6% of the patients already had metastasis at initial diagnosis. Although lymph node metastases were detected in as many as 47% of patients, renal carcinoid was found to have a relatively good outcome, with a mean disease-free interval of 43 months during follow-up. However, some of the patients with metastasis to solid organs such as the liver, contralateral kidney or bone had a poor outcome and died within a few months. Krishnan et al . reported that renal carcinoid often complicates horseshoe kidney, polycystic kidney or renal teratomata [ 8 ]. In particular, horseshoe kidney is known to be associated with a 62% higher risk of carcinoid; such carcinoids tend to arise from a solid teratoma, and are considered to have a better outcome than those complicating other underlying renal diseases. Renal carcinoid is mostly detected incidentally by imaging, and is rarely accompanied by symptoms such as flank pain, hematuria or a palpable mass. Imaging findings of renal carcinoid are nonspecific, while angiography often reveals a hypovascular mass, and ultrasonography, CT, and MRI visualize a well-defined mass. This disease should be differentiated most importantly from renal cell carcinoma, as well as from benign renal tumors including oncocytoma, angiomyolipoma or malacoplakia. As in cases of other carcinoid, surgical resection is the mainstay of treatment for renal carcinoid. Kobayashi et al . reported the first case of renal carcinoid treated by laparoscopic partial nephrectomy [ 9 ]. To our knowledge, there has been only one reported case of metastatic renal carcinoid (metastasis from the lung to the right kidney) by Tal et a l [ 2 ]. This patient was a 64-year-old woman who underwent resection of a pulmonary carcinoid and then was incidentally found to have a renal carcinoid 2 years later during detailed investigation of a left neck mass. Abdominal CT revealed a hepatic mass and a renal mass measuring 5 × 8 cm. The renal mass had invaded the renal pelvis and vein. The patient did not have any other noteworthy abnormalities or signs of carcinoid syndrome. She underwent partial hepatectomy and right nephrectomy. Microscopically, immunostaining was positive for chromogranin and neuron-specific enolase. Indium-111 octreotide scan was performed and showed positivity along with the left neck mass. The somatostatin analog octreotide, which is expressed in 80% or more of carcinoid tumors, was found to be useful for diagnostic staging. But it's impossible to use octeride at our institution. In this case, we should choose and recommend laparoscopic partial nephrectomy even if the patient had strong wish to laparosopic radical nephrectomy. Because, Simmons et al . reported that in appropriate patients with Stage T1b-T3 tumor >4 cm, laproscopic partial nephrectomy provides equivalent oncologic efficacy and superior renal functional outcomes compared with laparoscopic radical nephrectomy [ 10 ].

Conclusions The carcinoid tumor of the kidney in our patient, who had a history of liver metastasis from rectal carcinoid, was considered metastatic based on the pathological findings. In view of the tumor being p53-negative and no apparent recurrence to date, he seemed to have a good prognosis.

Background Carcinoid is an endocrine cell tumor with low-grade atypia, which is generally a low-grade malignant cancer with a good prognosis. Metastatic renal carcinoid is even rarer than primary carcinoids. Case presentation We present our experience of a patient with metastatic renal carcinoid from the gastrointestinal tract. Conclusions The carcinoid tumor of the kidney in our patient, who had a history of liver metastasis from rectal carcinoid, was considered metastatic based on the pathological findings.

Case presentation A 56-year-old man underwent resection of a rectal carcinoid twelve years ago, and subsequently underwent resection of liver metastases of rectal carcinoid twice, twelve and ten years ago. Two years ago, he was found to have a 3-cm mass in the left kidney on ultrasonography during a general checkup, and attended our department. Results of routine blood chemical analysis and urinalysis were unremarkable. Ultrasonography revealed a 3-cm relatively well-defined mass in the lower pole of the left kidney, and the interior of the mass was low echoic, unclear of margin, and hypovascular. Contrast-enhanced abdominal CT demonstrated a poorly enhanced, well-defined mass at the above-mentioned site in both the arterial and venous phases (Figure 1A ). MRI showed a well-defined mass containing irregular high-signal-intensity areas on T1-weighted images and relatively low-signal-intensity areas on T2-weighted images. Bone scintigraphy did not reveal any obvious bone metastases. Since these findings did not rule out malignancy, left renal cell carcinoma T1aN0M0 was diagnosed. According to the patient's strong wishes, left nephrectomy was performed laparoscopically through the retroperitoneal approach. Examination of the resected specimen revealed a 2 × 2 × 3 cm well-defined, solid tumor with a white capsule in the inferior pole of the kidney (Figure 1B ). Histopathologically, hematoxylin and eosin staining revealed cells uniformly arranged in a trabecular or funicular pattern. The cytoplasm was ill-defined and showed eosinophilic staining. The nucleus was ovoid to short fusiform, with no obvious mitoses. The tissue was partly necrotic. Immunostaining was positive for synaptophysin and chromogranin A, and negative for p53 (Figure 2A, B, C ). These findings were consistent with those of hematoxylin and eosin staining of specimens of the rectum and liver reported by the patient's former physician. Electron microscopy was not performed. Based on the pathological findings and the history of liver metastasis from rectal carcinoid, we diagnosed metastatic renal carcinoid. Postoperatively, blood tests, urinalysis, imaging, gastrointestinal endoscopy and other examinations showed no findings suggestive of recurrence of carcinoid while two years. Competing interests The authors declare that they have no competing interests. Authors' contributions YK drafted the report, and approved the final version of the manuscript. YY, GN, TY, KZ, RK, MT, SA, TT and NH cared for the patient and approved the final version of the manuscript. KN drafted the report, cared for the patient and approved the final version of the manuscript. Authors information Department of Urology, Aichi Medical University School of Medicine, Nagakute, Aichi 480-1195, Japan. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2490/10/22/prepub

Acknowledgements Written consent was obtained from the patient and his family prior to publication of this case report.

CC BY

no

2022-01-12 15:21:45

BMC Urol. 2010 Dec 14; 10:22

oa_package/23/eb/PMC3016340.tar.gz

PMC3016341

21246038

Introduction The mouse has become the most popular model organism for investigating the molecular mechanisms regulating energy metabolism and body weight (BW). However, a quantitative understanding of energy expenditure in mice remains lacking as highlighted by recent articles addressing problems with the interpretation of indirect calorimetry measurements [1] – [4] . Indeed, it is often unclear whether an observed BW change in mice is a result of altered energy intake (EI), energy expenditure (EE), or both. While we know that diet and EE impact metabolic fuel selection and body fat change over time, their quantitative relationship is uncertain. From a physiological perspective, a proper understanding of the metabolic phenotypes of various mouse models requires quantitative integration of these variables and how they change over time. To begin addressing these issues, we present a mathematical model of EE and metabolic fuel selection in male C57BL/6 mice. Our EE model incorporated the influence of body fat mass (FM), fat-free mass (FFM), the energy cost of tissue deposition, physical activity, and diet-induced thermogenesis (DIT). We combined the EE model with a mathematical model of energy partitioning to predict changes of BW, FM, and respiratory quotient (RQ) in response to measured changes of food intake. The model was validated by accurately predicting the BW and FM data from an independent set of experiments in C57BL/6 mice without adjusting any model parameters. The mathematical model demonstrates the complex relationships between metabolic fuel selection, diet composition, energy imbalance, and body composition change and provides a quantitative framework for investigation of murine energy metabolism.

Methods Modeling Energy Expenditure and Body Composition Change We begin with the law of energy conservation, also known as the energy balance equation: where = 9.4 kcal/g and = 1.8 kcal/g are the energy densities for changes in FM and FFM, respectively [5] . EI is the total metabolizable energy intake rate corrected for spillage. We assumed that the calculated metabolizable energy intake based on food intake measurements adequately accounted for any differences of digestibility between the diets. We did not directly measure the energy content of the feces to confirm this assumption. We previously showed that there is a well-defined, time-invariant function, α , that describes the relationship between changes of FFM and FM in adult male C57BL/6 mice: where the parameters c = 0.1, d = 1.89×10 −4 , and k = 0.45 g −1 specify the shape of the empirically measured function α [6] . This function allows us to write equation 1 as a pair of differential equations specifying the rates of change of FM and FFM [6] , [7] : Given measurements of EI, solving equation 3 requires a model of EE which was adapted from published human models [8] – [10] : where K is a thermogenesis parameter which was assumed to be constant for a fixed environmental temperature and λ represents physical activity whose energy cost was assumed to be proportional to BW. The values for K and λ were determined in the model calibration procedure described below. The parameter β accounts for the thermic effect of feeding as well as adaptive changes of EE as a result of diet changes. We will refer to the product β ΔEI as diet-induced thermogenesis (DIT) where ΔEI is the change of energy intake compared to baseline chow and the value β = 0.4 was determined in a recent analysis of rodent feeding and body composition data [11] . The parameters η FM = 0.18 kcal/g and η FFM = 0.23 kcal/g account for the biochemical efficiencies associated with fat and protein synthesis [11] assuming that the change of fat-free mass is primarily accounted for by body protein and its associated intracellular water [5] . The parameters γ FFM and γ FM determine how metabolic rate varies with FFM and FM, respectively. To estimate the values of γ FFM and γ FM , we note that basal metabolic rate (BMR) across species is well described by the Kleiber 3/4 power law of BW: BMR ∼ BW 3⁄4 [12] . Within a species, FFM and FM are proportional to BW. Therefore, and Using equation 6 to rescale the human values of γ FFM = 22 kcal/kg/d and γ FM = 3.6 kcal/kg/d [13] results in mouse values of γ FFM = 0.15 kcal/g/d and γ FM = 0.03 kcal/g/d. Data for Model Calibration The calibration data were obtained from a previously described study, the results of which are depicted in Figure 1 [14] . Briefly, we studied 47 three-month-old male C57BL/6 mice that were individually housed at a temperature of 22°C and randomly assigned to five groups: 1) C group (N = 12) on a chow diet (24% protein, 12% fat, and 64% carb.); 2) HF group (N = 12) on a high fat diet (14% protein, 59% fat, 27% carb.); 3) EN group (N = 11) on a high fat diet plus liquid Ensure® (14% protein, 22% fat, 64% carb.); 4) HF-C group (N = 6) switched from high fat to chow after 7 weeks; 5) EN-C group (N = 6) switched from high fat plus Ensure® to chow after 7 weeks. All animals received free access to water and food throughout the study. The high fat diet was provided using Rodent CAFÉ feeders (OYC International, Inc., MA), and liquid Ensure was provided in a 30-ml bottle with a rodent sip tube (Unifab Co., MI) and liquid intake was measured every day. Solid food intake was corrected for any visible spillage and was measured every day for the high fat diet and every other day for the chow diet using a balance with a precision of 0.01 g (Ohaus model SP402). Body composition was measured using 1H NMR spectroscopy (EchoMRI 3-in-1, Echo Medical Systems LTD, Houston, TX) and was recorded longitudinally throughout the study along with food intake and BW. The BW and FM at the beginning of the study were used as the initial values for the model inputs. We certify that all applicable institutional and governmental regulations concerning the ethical use of animals were followed during this research. All procedures were approved by the National Institute of Diabetes and Digestive and Kidney Diseases Animal Care and Use Committee under protocol K009-LBM-07. Model Calibration Procedure We assumed that the physical activity parameter λ depended on the diet as described by the following differential equation: where the initial value λ(0) = λ C0 was the physical activity for the baseline chow diet group, C. For the HF, EN, HF-C and EN-C mice that were switched to a high energy diet at time t switch1 , the physical activity parameter was assumed to change immediately to the value λ H (as implemented using the Dirac delta function, δ ) and to remain at this level during the high energy diet (as implemented using the Heaviside function, Θ). For the HF-C and EN-C mice switched back to the chow diet at time t switch2 we assumed that they immediately changed their physical activity to the value λ C1 which then exponentially relaxed to a value of λ C2 with a time constant of τ = 14 days. The model equations were numerically solved with the ODE45 solver using MATLAB software version R2008a ( http://www.mathworks.com ). While most model parameters were defined from analysis of previous data as stated above and tabulated in Table 1 , the parameters K , λ C0 , λ C1 , λ C2 , and λ H had to be estimated using data from our calibration experiment. The parameter values were determined using the Markov Chain Monte Carlo method where random guesses for the parameter values are proposed along with a simple rule for updating the parameter values depending on how closely the model results match the data [15] . Specifically, the model was run for 100,000 rounds for each group of mice and the first 30000 were discarded as a burn-in period with one fifth of the subsequent rounds were retained. Parameter sets were drawn from a proposal density that was normally distributed and centered on the previous value. The variance of the proposal density was tuned for an average acceptance rate of around 0.25 during the burn-in period. The convergence of the chain was assessed both by visual inspection of the trace plots for all the parameters and through the Geweke test [16] . At each sampling, the probability of accepting the new parameter set given current parameter set was min(1, r), where r is the Metropolis ratio. The energy intake in each group of animals was normally distributed with a standard error of 0.39, 0.39, 0.41, 0.55, and 0.55 Kcal/d for the C, HF, EN, HF-C, and EN-C groups, respectively. The 95% confidence intervals of the predicted energy output were obtained by calculating the 2.5th and 97.5th percentiles of the posterior distribution of energy output. Model validation The model was validated using data from an independent study where 30 three-month-old male C57BL/6 mice were individually housed at a temperature of 22°C and randomly assigned to five ad libitum fed groups depicted in Figure 2: 1 ) the control group on a chow diet (N = 6); 2) the 7HF-C group on a high fat diet for 7 wk followed by a switch to chow (N = 6); 3) the HF-C-HF-C on a high fat diet for 7 wk followed by a switch to chow for 3 wk, back on the high fat diet for 10 wk and followed by a switch back to chow (N = 6); 4) the 20HF-C on a high fat diet for 20 wk followed by a switch to chow (N = 6); 5) the 4HF-C group on high fat diet for 4 wk followed by a switch to chow (N = 6). The chow and high fat diets were identical to the diets used in the calibration study. The body composition ( Tables S1 and S2 ) and energy intake data ( Table S3 ) are provided as Supporting Information. The measured food intake rates were used as model inputs to predict BW and FM in the validation experiment and no model parameter values were adjusted. The energy intake in each group of animals was normally distributed with a standard error of 0.42, 0.48, 0.50, 0.53, and 0.55 kcal/d for the five groups, respectively. The 95% confidence intervals of the predicted energy output were obtained by calculating the 2.5th and 97.5th percentiles of the posterior distribution of EE. Metabolic Fuel Selection and the Respiratory Quotient Since we are also interested in metabolic fuel selection, we considered the fates of dietary macronutrients including their oxidation rates, storage in the body, as well as major inter-conversion fluxes de novo lipogenesis (DNL) and gluconeogenesis (GNG). The following macronutrient balance equations represented these changes: where P is body protein, G is glycogen, and I F , I P and I C are the metabolizable intake rates of dietary fat, protein and carbohydrate, respectively. The oxidation rates of fat, protein, and carbohydrate (FatOx, ProtOx, and CarbOx, respectively) sum to EE. To simplify the macronutrient balance equations, we note that glycogen stores are small, especially when compared with daily carbohydrate intake rates [6] . Thus, over the time-scale of interest the system is in a state of average carbohydrate balance: Therefore, where the net oxidation rates were defined as: Assuming that FFM is proportional to body protein: then equations 3, 8 and 9 resulted in the following expressions for the net macronutrient oxidation rates based on the model variables: The respiratory quotient, RQ, is the carbon dioxide production rate divided by the oxygen consumption rate and was approximated by: The calculated RQ may have slight inaccuracies during rapid transitions immediately after diet switches but will be reasonably accurate thereafter since the net carbohydrate oxidation rate is approximately equal to the carbohydrate intake rate on long time scales (several days in mice). Using the measured food intake rates along with the model calculated values of EE and FM, we applied equations 13 and 14 to calculate the dynamic changes of metabolic fuel selection over time. Note that these calculations did not influence the main model predictions for BW and FM. We compared the calculated RQ values with the food quotient, FQ, which is a measure of the expected carbon dioxide production rate divided by the oxygen consumption rate had the food itself been combusted directly: In a state of energy and macronutrient balance, RQ = FQ and any deviations from this equality reflect a situation where the metabolic fuel mixture differs from the diet and body composition changes will therefore result.

Results Model Calibration The comparison between the model simulations and the measured values for BW and FM from the calibration study is presented in Figures 1A and 1B , respectively. As previously described [14] , the C group that only ate chow slightly increased BW and FM (solid circles in Figures 1A and 1B , respectively) throughout the study. The animals on the high fat diet (HF group shown in solid blue triangles) gained a substantial amount of weight and fat and the animals supplemented with the liquid Ensure (EN group shown in solid magenta squares) gained even more. Both the HF-C and EN-C groups (open red triangles and green squares, respectively) rapidly lost weight and fat after the switch from the high energy diets to the chow diet and reached a stable BW and FM four weeks after the diet switch at a level significantly higher than the C group. The model simulations, depicted by the solid curves in Figure 1 , agreed with the measurements for both the BW and FM measurements thereby demonstrating that the model was able to accurately describe the observations in the calibration study. The best-fit model parameters for physical activity were λ H = 0.13 kcal/g/d for the HF and HF-C groups and λ H = 0.16 kcal/g/d for the EN and EN-C groups (note that the latter value for λ H was irrelevant for the validation study since Ensure was not used). The baseline physical activity was λ C0 = 0.22 kcal/g/d for all groups, and following the switch to chow in the HF-C and EN-C groups λ C1 = 0.27 kcal/g/d and λ C2 = 0.19 kcal/g/d. Thus, introduction of the high fat diet resulted in a predicted drop of physical activity by ∼40% below the baseline chow diet. When the diet was switched back to chow, there was an immediate increase of physical activity, reaching ∼20% higher than baseline, which subsequently relaxed back to a value close to baseline. Since all mice were individually housed and kept at the same 22°C environmental temperature, the thermogenesis parameter K = 2.1 kcal/d was constant across all groups. The EE and RQ predictions closely matched the previously published estimates of these variables (not shown) [6] . Model Validation Using the same parameter values determined in the calibration procedure, Figure 3 shows that the model (curves) accurately predicted the measurements (•) of BW and FM of the five validation groups shown in the left and right columns, respectively. The 95% confidence intervals predicted by the model (dashed curves) and the error bars of the measurements overlapped throughout the entire study. Moreover, most of the model predictions fell within one standard error of the observed data for both BW and FM. The dynamics of EI and EE for the various groups are presented in the left column of Figure 4 . The measured EI data (•) were first fit by the solid black curves which were used as the model inputs. The simulated EE dynamics are depicted using solid blue curves with the 95% confidence intervals shown using dashed blue curves. The control group EE was only slightly lower than EI, which is consistent with the slow increase of BW and FM in this group ( Fig. 4A ). High fat feeding decreased EE at the beginning of the experiment compared with the control group which, along with increased EI, resulted in significant weight gain that was accompanied by a gradual increase of EE. In the 7HF-C and HF-C-HF-C groups ( Fig. 4B and 4C ), switching from the high fat diet to chow at week 7 caused a small transient increase of EE due to increased physical activity which, along with the dramatic fall of EI, gave rise to the rapid BW losses shown in Fig. 3B and 3C . In contrast to the small increase of EE observed at week 7 upon switching to chow in the 7HF-C and HF-C-HF-C groups, the model predicted a substantially greater increase of EE after the diet switches at week 20 in the HF-C-HF-C and 20HF-C groups ( Fig. 4C and 4D ). This was predicted to result from the same transient increase of the physical activity parameter, λ, leading to a greater increase of EE since the BW was higher at week 20 versus week 7 and the energy cost of physical activity is given by the product λ×BW. Furthermore, the drop of EI was not as great upon switching to chow at week 20 versus week 7 and therefore the decrease of DIT did not offset the increased physical activity to the same extent as it did at 7 weeks. In the HF-C-HF-C group, although an increase in EI was observed at week 10 when the animals were switched from chow diet back to high fat diet, the EE remained relatively unchanged which resulted in rapid weight regain. The relative stability of EE at week 10 was predicted to result from a balance between the decreased physical activity on the high fat diet which was completely offset by the increased DIT. The predicted dynamics of RQ (red) and FQ (black) are presented in the right column of Figure 4 . The control group had RQ approximately equal to FQ = 0.92 corresponding to the chow diet. The high fat diet had a lower FQ = 0.80 and RQ immediately dropped at the onset of high fat feeding to a value slightly higher than FQ. Subsequently, RQ gradually fell as FM increased. When the animals were switched from the high fat diet to chow at week 7, a transient decrease of RQ was predicted right after the diet switch in the 7HF-C and HF-C-HF-C groups ( Fig. 4B and 4C ) and a similar transient decrease of RQ was seen in the 4HF-C group at week 4 ( Fig. 4D ). However, such a decrease of RQ was not predicted when switching from high fat to chow at week 20 in the 20HF-C and HF-C-HF-C groups because their higher FM had already dropped the RQ to a lower value at the time of the diet switch and the decrease of EI was not as great at week 20. When the HF-C-HF-C group was switched from the chow diet to the high fat diet at week 10, an immediate decrease of RQ was predicted similar to the initial onset of high fat feeding at week 0. These results highlight the complex interplay between diet composition, the degree of energy imbalance, and the body composition in determining metabolic fuel selection.

Discussion To our knowledge, there is only one previous report of a mathematical model of mouse metabolism and BW regulation and that model focused on the role of leptin to influence both EI and EE [17] . While that previous model led to interesting theoretical insights regarding feedback control of BW, the mathematical model was not validated and did not address the issue of metabolic fuel selection [18] . Our previous mathematical analysis used measured BW and food intake data as model inputs to estimate EE and fuel selection dynamics [6] . Here, we extended our previous analysis by explicitly modeling the determinants of EE. The only model input was food intake and this information allowed our model to accurately predict the observed changes of BW and FM in five independent groups of animals. Therefore, our model is the first validated mathematical model of mouse energy metabolism. The model also predicted dynamic changes of RQ and EE over the entire course of the study based on the principle of energy conservation. However, we were unable to directly confirm these predictions since we did not have corresponding indirect calorimetry data. Such measurements would be prohibitive over the entire time course of the study, and measurements at isolated time points would likely lead to behavior change in the animals as they are removed from their home cages. Indeed, a recent elegant study demonstrated that the indirect calorimetry procedure caused weight loss mice that had previously been gaining weight on either a high fat or chow diet [19] . Weight gain only occurred when mice previously fed chow were provided with a novel high fat diet in the indirect calorimetry chamber. While the authors presented a simple statistical method to adjust the indirect calorimetry measurements based on the observed weight changes [19] , the fact remains that the behavior of the animals during the procedure was clearly not representative of the extended study duration. Our previously described methodology for estimating RQ and EE avoids this difficulty [6] , and the comparison of our model predictions for RQ and EE closely matched our previous estimates (not shown). Regardless of the likely behavior changes introduced by the procedure, indirect calorimetry may be useful for investigating the contribution of FFM and FM to EE and we compared our mathematical model of EE with the results from a recently published mouse study [4] . Our equation 4 predicts that the coefficients describing how total EE varies with FFM and FM have values of ( γ FFM + λ) = 0.23 cal/g/min and ( γ FM +λ) = 0.14 cal/g/min averaged across all of the diets used in the present study. These values agree reasonably well with the regression coefficients of 0.269 cal/g/min and 0.144 cal/g/min found by Kaiyala et al. using mice fed various diets where total EE was regressed against FFM and FM, respectively [4] . Thus, these data lend further independent support to our model which also includes the influence of DIT, tissue deposition costs, and physical activity in addition to the dependence on FM and FFM. We assumed that physical activity depended on the diet composition which is in accordance with previous studies showing that high fat diets result in substantial decreases of physical activity [14] , [20] . Interestingly, to match the BW data from our calibration study our model predicted that high fat feeding caused an immediate and sustained decrease of physical activity whereas switching back to a chow diet causes a transient overshoot of physical activity, which together with a dramatic reduction of EI, resulted in rapid weight loss. This is consistent with previous studies showing that caloric restriction leads to increased activity in mice [21] , but in our study the mice voluntarily restricted their own intake of chow after the high fat diet was removed. The degree of increase of total EE upon switching to the chow diet was found to depend on BW at the time of the diet switch. Furthermore, the decrease of EI caused a decrease of DIT that offset the effect of increased physical activity. Our model has several limitations. First, the time scale of the model is days, weeks and months and therefore does not address within-day dynamics such as the transition from fed to fasted states. Second, physical activity behavioral changes need to be either input directly or calibrated from previous data for mice in similar environments. Therefore, applying the model calibrated for individually housed mice without running wheels will not necessarily represent the behavior of mice housed under different conditions (e.g., with running wheels). Similarly, all of our studies were conducted using individually housed mice kept at the same 22°C environmental temperature, but it is well-known that temperature can significantly impact EE in mice [22] , [23] . This effect could be incorporated in our model by replacing the constant thermogenesis parameter K with a decreasing function of temperature until thermo neutrality is reached. It may also be necessary to make other model parameters temperature dependent. For example, the DIT parameter β and the dependence of metabolic rate on body fat, γ FM , since it is possible that UCP-1 activation in brown adipose tissue is temperature dependent and mediates DIT [24] , although this effect is controversial [25] . We anticipate that modification of the model parameters will be required to appropriately represent other strains of mice, genetic knockouts, or transgenic mouse models. Indeed, the process of determining the parameter modifications required to accurately simulate different mouse models will provide important quantitative information regarding their integrative metabolic phenotypes and the differences between mouse models. Our mathematical model provides a quantitative framework for integrating murine data on food intake, body weight, and body fat to help understand the complex dynamic relationships between diet, expenditure, body composition and metabolic fuel selection. In the future, we hope to validate our RQ and EE model predictions by comparing with indirect calorimetry data where mice spend many weeks inside suitably modified metabolic cages. We also plan to incorporate a more mechanistic representation of metabolic flux regulation and the influence on EE and metabolic fuel selection as was recently described in more detailed mathematical models of human metabolism published by our research group [26] , [27] . Finally, we plan to include the potential influence of circulating factors such as insulin and leptin and eventually “close the loop” by modeling the regulation of food intake in mice.

Conceived and designed the experiments: JG KDH. Performed the experiments: JG. Analyzed the data: JG KDH. Contributed reagents/materials/analysis tools: JG KDH. Wrote the paper: JG KDH. The mouse is an important model organism for investigating the molecular mechanisms of body weight regulation, but a quantitative understanding of mouse energy metabolism remains lacking. Therefore, we created a mathematical model of mouse energy metabolism to predict dynamic changes of body weight, body fat, energy expenditure, and metabolic fuel selection. Based on the principle of energy balance, we constructed ordinary differential equations representing the dynamics of body fat mass (FM) and fat-free mass (FFM) as a function of dietary intake and energy expenditure (EE). The EE model included the cost of tissue deposition, physical activity, diet-induced thermogenesis, and the influence of FM and FFM on metabolic rate. The model was calibrated using previously published data and validated by comparing its predictions to measurements in five groups of male C57/BL6 mice (N = 30) provided ad libitum access to either chow or high fat diets for varying time periods. The mathematical model accurately predicted the observed body weight and FM changes. Physical activity was predicted to decrease immediately upon switching from the chow to the high fat diet and the model coefficients relating EE to FM and FFM agreed with previous independent estimates. Metabolic fuel selection was predicted to depend on a complex interplay between diet composition, the degree of energy imbalance, and body composition. This is the first validated mathematical model of mouse energy metabolism and it provides a quantitative framework for investigating energy balance relationships in mouse models of obesity and diabetes.

Supporting Information

We thank Michael Dore, Carson Chow, and Oksana Gavrilova (all at NIDDK, NIH) for their assistance with the experiments.

CC0

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15961

oa_package/8b/a6/PMC3016341.tar.gz

PMC3016342

21246039

Introduction Cardiac glycosides (CGs) are a group of plant-derived compounds that have been used for many years in traditional medicine and that are currently used in treatment of cardiac failure and atrial fibrillation. In parallel to this use, CGs have also received attention as potential drugs in the treatment of various malignant diseases. Epidemiological observations suggest that patients on digitalis medication diagnosed with breast cancer in general present with lower proliferating tumours of smaller size, and subsequently better prognosis than control groups [1] , [2] , [3] , [4] , and that a high concentration of digitoxin could reduce the risk of developing leukaemia, lymphoma or urogenital cancer [5] . In vitro experiments have shown that CGs can induce cell death in several cell lines derived from solid cancers [6] , [7] as well as in leukaemic cell lines [8] , [9] , [10] , [11] . In the myocardium, CGs bind reversibly to the α-subunit of Na + /K + ATPase, leading to a rise in intracellular sodium levels, which then results in an increase of calcium ions in the myocytes. The mechanism of the cytotoxic effects of CGs on tumour cells has been a subject of many studies but it largely remains unanswered. Binding to Na + /K + ATPase is not only a way of regulating ion pumps in the cell membrane but it can also activate several signalling pathways in the cell. For example, calcium-dependent activation of caspases and other hydrolytic enzymes [7] , [12] , [13] , generation of reactive oxygen species (ROS) [14] , topoisomerase inhibition [15] , interference with signal transduction pathways (e.g. Src-mediated phosphorylation of epidermal growth factor receptor (EGFR) and induction of the cell cycle inhibitor p21 Cip1 [16] have all been associated with the anti-tumour effects of CGs. Digoxin-Like Immunoreactive Factors (DLIFs, also termed Digitalis-Like Compounds (DLCs)) are endogenous steroids identified in human tissues and they are identical or similar to plant and amphibian steroids. The DLIFs are believed to be synthesized in the adrenal gland, and to affect ion transport via Na + /K + ATPase. Binding of these compounds to Na + /K + ATPase may activate changes in intracellular Ca 2+ homeostasis and in specific gene expression [17] , and may be associated with the development of malignancies [18] . Interestingly, DLIFs selectively induce apoptosis in a human acute T-cell lymphoblastic leukaemia cell line but not in the myelogenous leukaemia cell line K-562 or healthy human peripheral blood mononuclear cells (PBMCs) [19] . Despite years of interest in these effects and numerous studies in vitro and in animals, it has not yet been possible to utilize the anti-cancerous potential of CGs clinically. Recently, a very discouraging report on this issue was published, suggesting general inhibition of protein synthesis as the main mechanism of the anti-cancerous effects of CGs, and species differences of a magnitude sufficient to explain the results of most preclinical studies [20] . During a routine screening programme carried out in vitro we observed that some samples of acute leukaemia were extremely sensitive to the cytotoxic effects of digitoxin, thus prompting further investigation. Hence this study was undertaken to categorize the activity of some CGs in primary cultures from patients with various leukaemic diagnoses, and to determine if general protein inhibition is the dominant mechanism of action, and if a therapeutic index in vitro exists.

Materials and Methods Patient Samples and Cell Lines Cryopreserved cells from bone marrow or peripheral blood from adult patients with B-precursor or T-acute lymphoblastic leukaemia (ALL), acute myeloid leukaemia (AML) and chronic lymphocytic leukaemia (CLL) were used in the study. Peripheral blood mononuclear cells (PBMCs) from healthy donors were used as controls. Informed consent was obtained from all patients (verbal until 2006 and written thereafter) to save diagnostic samples in a biobank to be used for scientific research. The informed consent was verbal until 2006 (in accordance with the approval from the Ethics Committee) and the Ethics Committee did not demand that the consent should be documented at this time (until 2006). Since 2006 written consent has been obtained. Sampling for drug sensitivity testing was approved by the local Ethics Committee in Uppsala (Regionala etikprövningsnämnden i Uppsala, Sweden, approval number Dnr 21/93). The samples from the biobank used in the study were coded but labelled with diagnosis. The T-lymphoblast-like cell line CEM/VBL 100 (CCRF-CEM) [21] was kindly donated by professor W.T. Beck, St. Jude's Children's Research Hospital, USA and the B-precursor Philadelphia-positive cell line SUP-B15 [22] was obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Germany. The cells were grown in complete medium (CCRF-CEM in RPMI 1640 and SUP-B15 in McCoy's 5A) and split twice weekly. The colorectal adenocarcinoma cell lines Hct116, HT29 and CC20 were used as comparator cell lines as regards effects on protein synthesis. Chemicals and Reagents The CGs digitoxin, digoxin and ouabain were purchased from Sigma (Sigma-Aldrich, Stockholm, Sweden). The compounds were dissolved in DMSO and further dilution was in PBS. The drugs were tested at five concentrations (ten-fold dilution steps) ranging from 100 μM to 0.01 μM or 10 μM to 0.001 μM. The maximum concentration of DMSO did not exceed 1% in the cell cultures. The serially diluted stock solutions were transferred to 384-well microtitre plates (NUNC, Roskilde, Denmark) and control wells were filled with PBS only (5 μl/well). Measurement of Cytotoxic Activity The cytotoxic activity of the CGs was measured by using a fluorometric microculture cytotoxicity assay (FMCA) as previously described [23] , [24] . The method is based on measurement of the fluorescence derived by hydrolysis of fluorescein diacetate (FDA) to fluorescein by cells with intact plasma membranes. Cell suspensions were seeded into drug-prepared 384-well microtitre plates. Wells with medium only served as blanks. The plates were incubated at 37°C for 72 hours; thereafter FMCAs were performed using an automated Optimized Robot for Chemical Analysis (Orca; Beckman Coulter Fullerton, CA) programmed through SAMI software (Beckman Coulter). The plates were washed in physiological buffer and FDA added. After 50 min of incubation at 37°C, fluorescence was measured at 485/520 nm using a Fluostar Optima microplate reader (BMG Technologies, Germany). The fluorescence measured is proportional to the number of living cells in each well. Because a 30–40 nM plasma concentration of digitoxin can be maintained in patients for many days, additional experiments with extended incubation time (6 days) was performed. Leukaemic patient cells and the CCRF-CEM cell line was treated with therapeutically achievable concentrations (10, 30 and 50 nM) of digitoxin in 96-well microtiter plates. To simulate a situation with daily dosing, 50% of the medium (including drug to treated cells) was exchanged every day. Quality control was evaluated for each test by demanding a signal/blank ratio of >10 and a coefficient of variation in controls and blanks of <30%. The proportion of leukaemic cells in primary cultures should exceed 70% on days zero or three when examined morphologically. Measurements of protein and nucleic acid synthesis inhibition Effects on DNA and protein synthesis were monitored in Cytostar-T® plates (available in the “ In Situ mRNA Cytostar-T® assay” kit, Amersham International, Buckinghamshire, UK) using 14 C-labelled thymidine and leucine. A Cytostar-T® plate is a 96-well microtitre plate with scintillants molded into the transparent polystyrene bottom. When labelled substrate is absorbed into the intracellular compartment of the cells at the bottom of the wells, the radioisotope is brought into proximity with the scintillant, thereby generating a detectable signal. Free radiolabelled substrate in the supernatant is unable to stimulate the scintillant [25] , [26] . CCRF-CEM and SUP-B15 cells were suspended in fresh medium containing 14 C-thymidine (111 nCi/ml; for DNA experiments) or 14 C-leucine (222 nCi/ml; for protein experiments), yielding final radioactivity in the wells of 20 and 40 nCi, respectively. Cell suspension (50×10 3 cells in 180 μl) was added to each well; blank wells had isotope-containing medium only. Drugs (digoxin and digitoxin at final concentrations of 10 μM to 1 nM) and PBS in test and control wells were added in duplicate (20 μl per well) 2 hours after cell seeding, when the measured radioactivity in cell-containing wells was at least double compared with blank wells. Radioactivity was measured with a computer-controlled Wallac 1450 MicroBeta® trilux liquid scintillation counter (Wallac OY., Turku, Finland) immediately after addition of the cell suspension and at different time points up to 72 hours. Between measurements, the plates were stored in an incubator at 37°C. During measurement, the plates were covered with a plate sealer to inhibit microbiological contamination.

Results The cytotoxic activities of digitoxin and ouabain were studied in primary leukaemic cells: T-ALL (n = digitoxin 4; ouabain 7), B-precursor ALL (n = 10; 6), AML (n = 11;11), CLL (n = 9; 6) and PBMCs (n = 4; 8) using the FMCA. Similar tests on the activity of digoxin, digitoxin and ouabain were performed in the leukaemia cell lines CCRF-CEM and SUP-B15. All tests were carried out in triplicate. The primary T- and B-precursor ALL cells were significantly more sensitive to digitoxin than CLL cells (Mann–Whitney test, p = 0.02 and 0.006 respectively) and PBMCs (p = 0.02 and 0.005 respectively) ( Figure 1A ). The median IC 50 value regarding digitoxin was 0.07 μM for T-ALL and 0.06 μM for B-precursor ALL cells, compared with 0.44 μM for PBMCs ( Table 1 ). For the AML cells a scattered distribution regarding IC 50 values was observed. As regards ouabain, the IC 50 value was significantly lower (Mann–Whitney test, p = 0.02) for the T-ALL cells than for CLL cells but otherwise no significant differences were observed in the different leukaemic cells or the PBMCs ( Figure 1B and Table 1 ). At 0.1 μM the B-precursor and T-ALL cells were significantly more sensitive to digitoxin than the CLL cells and PBMCs (Mann–Whitney test, B-precursor ALL vs. CLL p = 0.0003, B-precursor ALL vs. PBMCs p = 0.002, T-ALL vs. CLL, p = 0.004 and T-ALL vs. PBMCs, p = 0.03). In addition, the B-precursor ALL cells were more sensitive than the AML cells (p = 0.02) ( Figure 2A ). With extended exposure (6 days) both T- and B-precursor ALL cells appeared sensitive at clinically achievable concentrations ( Figure 2B ). The cell line SUP-B15 was highly sensitive to all tested CGs, with IC 50 values as follows: digitoxin 0.002 μM, ouabain 0.004 μM and digoxin 0.03 μM. The T-lymphoblast-like cell line CCRF-CEM showed effects similar to those in the primary ALL cells, with IC 50 values of 0.04 μM for ouabain, 0.12 μM for digitoxin and 0.22 μM for digoxin. In both cell lines there was a tendency towards a lower sensitivity to digoxin than to the other two CGs. Both digoxin and digitoxin inhibited DNA as well as protein synthesis in CCRF-CEM and SUP-B15 cells. The effects of the specific inhibitors aphidicolin and cycloheximide, used as positive controls, were strong and immediate. As presented in Figure 3 (digitoxin only; digoxin showed similar results), the effects were, however, only observed at relatively high concentrations, and at 100 nM, surprisingly, no significant effects were detected during the 24-h observation period. At the highest concentration tested (1.0 μM), well exceeding the IC 50 value for cytotoxicity, the effects on DNA and protein synthesis were similar in time and magnitude, i.e. the cells tended to “shut down” in an expected manner due to the toxic insult, presumably related to severe ionic imbalance. To put these results into perspective, a comparator cell line, the adenocarcinoma cell line Hct116was also analysed. As the Hct116 cell line is more tolerant to the cytotoxic effects of glycosides, slightly higher CG concentrations were used. Figure 4 shows the effects of various digitoxin concentrations on DNA (A) and protein synthesis (B). In contrast to the results in the leukaemia cell lines, protein synthesis in Hct116 cells was decreased more efficiently (i.e. at concentrations corresponding to cytotoxic activity) and at an earlier time point than DNA synthesis. With 1 μM digitoxin, protein synthesis in Hct116 cells was effectively inhibited at early time points (significant from 6 hours), while DNA synthesis did not appear to slow down until 24 h. This concentration is comparable to the cytotoxic IC 50 value for Hct116 cells measured in the FMCA (0.71 μM, Figure 4C ). The effects in two other colorectal adenocarcinoma cell lines (HT29 and CC20) were similar (not shown). The leukaemic SUP-B15 cell line was approximately 500 times more sensitive to the cytotoxic effects of digitoxin, with an IC 50 value of 1.5 nM ( Figure 4C ). Despite this, exposure of SUP-B15 cells to 10 nM digitoxin had no effect on protein synthesis up to 24 h ( Figures 3D and 4D ), but it effectively reduced viability at 72 h ( Figure 4D ). Results in the leukaemic CCRF-CEM cell line were similar (not shown).

Discussion In this study we have identified primary B-precursor and T-ALL cells as being particularly susceptible to the cytotoxic effects of CGs, being significantly more sensitive than CLL cells and PBMCs. Digitoxin was the most potent of the CGs tested, and the measured IC 50 value was comparable with therapeutic serum concentrations, especially if exposure time was extended to 6 days. The therapeutic concentration range for CGs in clinical use is narrow – for digitoxin 15–40 nmol/l (10–30 ng/ml) [27] . This, of course, raises the question of whether or not there could be a place for CGs in the treatment of acute leukaemia, despite negative results in solid tumours, for example. In ALL maintenance treatment with moderate doses of 6-mercaptopurin and methotrexate for a period of approximately two years has a documented effect on the risk of relapse [28] . The intensity of the treatment is limited by haematological and/or liver toxicity. New drugs with different toxicity spectra could clearly be of substantial benefit in this setting. Therefore, digitoxin as an addition to conventional maintenance chemotherapy in ALL could be proposed as a future clinical study, but only after careful preclinical investigation of possible drug-drug interactions. For example, previous studies have indicated that digitoxin at concentrations commonly found in the plasma of cardiac patients, significantly reduced etoposide and idarubicin-induced topoisomerase II cleavable complexes in K562 leukemia cells [29] . The cytotoxic effect of digitoxin on both primary ALL cells as well as in the extremely sensitive cell line SUP-B15S indicates that a mechanism of inducing cell death other than inhibition of sodium- and potassium-activated adenosine triphosphatase (Na + /K + ATPase) may be possible. Numerous effects of cardiac glycosides in cancer cell lines ultimately leading to apoptosis have been demonstrated previously. Most of these effects are probably mediated through the main target enzyme [30] . However, interpretation of experiments involving established cell lines calls for caution because of possible genetic drift and altered properties. This is illustrated by the extreme sensitivity towards CGs in the cell line SUP-B15, while results in the CEM cell line were similar to those in the primary leukaemic cells. Despite years of interest in these effects and numerous studies in vitro and in animals, it has not yet been possible to utilize the anti-cancer potential of CGs clinically. Several preclinical studies have demonstrated these compounds to be involved in selective control of human proliferation, which support the use of CGs to treat malignancies [30] , [31] . Different glycosides with different profile regarding cellular effects and cardiotoxicity have been developed, and several clinical trials have been performed. For example, UNBS-1450, a semisynthetic cardenolide, with good preclinical activity against NSCLC has entered a phase I clinical trial in Belgium [32] . A phase I trial of AnvirzelTM, an aqueous extract from Nerium Oleander , in patients with refractory solid tumors has been reported [33] . Studies of the addition of digoxin to combination chemotherapy and immunotherapy in patients with advanced malignant melanoma have also been initiated [34] , the study has not been finally reported. There are several possible explanations for this slow progression into clinical practise; one is of course the possibility of in vitro biases, and/or species differences. For example, it has been demonstrated that oleandrin activates MAPK and JNK and also induces expression of FasL, leading to apoptosis in human, but not in murine cells (28). This difference has also been detected at the cellular membrane level, as oleandrin altered its fluidity, inhibiting Na + /K + ATPase activity, and increasing intracellular free Ca 2+ levels, followed by calcineurin activity only in human, but not in murine cells. The results suggested that murine plasma membranes might be different from human membranes, which interact with oleandrin, disturbing the Na + /K + ATPase pump and resulting in calcification, followed by induction of Ca 2+ -dependent cellular responses such as apoptosis [35] . In line with these findings, Perne et al. recently published experimental results suggesting that general inhibition of protein synthesis is the main mechanism of the anti-cancerous effects of CGs in human cells, and that physiological species differences may explain the previously observed sensitivity of human cancer cells in mouse xenograft experiments [20] . It was proposed that inhibition of protein synthesis is directly related to effects on the Na + /K + ATPase pump. Indeed, this protein synthesis-inhibiting mechanism is supported by observations in the colorectal adenocarcinoma cell line Hct116 ( Figure 4 ). With 1 μM digitoxin, protein synthesis is effectively inhibited at early time points (significant from 6 hours), leading to decreased cell numbers at 72 hours. It might be concluded that Hct116 cells die because of an inability to synthesize vital proteins. The results from the leukaemia cell lines appear to be in sharp contrast to this – protein synthesis levels at time points up to 24 h are unaffected by concentrations having severe effects on cellular viability. Thus, it may be concluded that these cells stop synthesizing their proteins because they are dying, rather than the opposite. Conclusions Cardiac glycosides have been used for treatment of congestive heart failure and atrial fibrillation for over a century. During the last twenty years this class of compounds (e.g. digitoxin and oubain) has received much attention as regards their potential as anti-cancer agents, based on epidemiological as well as experimental findings. Despite great efforts the mechanism behind the anti-cancerous effects has been difficult to determine, as have the potential benefits in the clinic. The recently published study by Perne et al. [20] shed some light on this issue, but also strongly discouraged further clinical development of CGs, and derivatives thereof, for use in this therapeutic area. In this study we describe how some subsets of leukaemic cells from patients express high sensitivity towards CGs, particularly digitoxin, and at concentrations that may be achieved in the clinic. Furthermore, it was shown that the suggested mechanism of protein inhibition is valid in tumour cell lines with moderate or high resistance to the cytotoxic effects of CGs, but probably not in highly sensitive cancer cell lines. It is thus suggested that further investigation may be focused on diagnoses like T- and B-precursor ALL.

Conclusions Cardiac glycosides have been used for treatment of congestive heart failure and atrial fibrillation for over a century. During the last twenty years this class of compounds (e.g. digitoxin and oubain) has received much attention as regards their potential as anti-cancer agents, based on epidemiological as well as experimental findings. Despite great efforts the mechanism behind the anti-cancerous effects has been difficult to determine, as have the potential benefits in the clinic. The recently published study by Perne et al. [20] shed some light on this issue, but also strongly discouraged further clinical development of CGs, and derivatives thereof, for use in this therapeutic area. In this study we describe how some subsets of leukaemic cells from patients express high sensitivity towards CGs, particularly digitoxin, and at concentrations that may be achieved in the clinic. Furthermore, it was shown that the suggested mechanism of protein inhibition is valid in tumour cell lines with moderate or high resistance to the cytotoxic effects of CGs, but probably not in highly sensitive cancer cell lines. It is thus suggested that further investigation may be focused on diagnoses like T- and B-precursor ALL.

Conceived and designed the experiments: HH JF AE MF LB RL JG. Performed the experiments: HH JF AE JG. Analyzed the data: HH JF JG. Contributed reagents/materials/analysis tools: HH JG RL. Wrote the paper: HH JF AE MF LB RL JG. Obtained permission for use of cell line: RL. Background Despite years of interest in the anti-cancerous effects of cardiac glycosides (CGs), and numerous studies in vitro and in animals, it has not yet been possible to utilize this potential clinically. Reports have demonstrated promising in vitro effects on different targets as well as a possible therapeutic index/selectivity in vitro and in experimental animals. Recently, however, general inhibition of protein synthesis was suggested as the main mechanism of the anti-cancerous effects of CGs. In addition, evidence of species differences of a magnitude sufficient to explain the results of many studies called for reconsideration of earlier results. Principal Findings In this report we identified primary B-precursor and T-ALL cells as being particularly susceptible to the cytotoxic effects of CGs. Digitoxin appeared most potent and IC 50 values for several patient samples were at concentrations that may be achieved in the clinic. Significant protein synthesis inhibition at concentrations corresponding to IC 50 was demonstrated in colorectal tumour cell lines moderately resistant to the cytotoxic effects of digoxin and digitoxin, but not in highly sensitive leukaemia cell lines. Conclusion It is suggested that further investigation regarding CGs may be focused on diagnoses like T- and B-precursor ALL.

CC BY

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15718

oa_package/96/d1/PMC3016342.tar.gz

PMC3016343

21156075

Introduction Varus foot is often secondary to cerebral palsy and split tibialis anterior (SPLATT) or posterior tibialis tendon transfers (SPOTT) are commonly performed to correct the deformity. In both procedures the distal part of the tendon is splitting longitudinally, half of the tendon is detached from its medial insertion and is reattached to the lateral side of the foot [ 1 ]. The goal of the semi-transfers is to affect the muscle-tendon complex in a way that it neither inverts nor everts the foot maintaining thus its stability and flexibility. The SPLATT as described by Hoffer et al [ 2 ] corrects supination and varus deformity of the midfoot secondary to spasticity of the anterior tibial muscle. Equinovarus hindfoot deformity is most common in children with spastic hemiplegia and is caused by spasticity of the posterior tibial muscle that very often is associated with weakness of the peroneal muscles and tightness of the heel cord [Figure 1 ]. The reported clinical outcomes of SPLATT and SPOTT have been generally good but have been considerably varied [ 2 - 9 ]. SPLATT was first described by Kaufer et al [ 10 ] and was popularized by Green et al [ 4 ] and Kling et al [ 6 ] as a technique that balances the hind part of the foot and maintains the plantar flexion power, but it should be applied only to patients from 4 to 6 years of age due to the potential risk of converting the foot to a valgus deformity in children younger than 4 years. Prerequisites of split tendon transfers is the ability or the potential ability for walking. Contraindications include a fixed bony deformity, and severe contraction mainly concerning the anterior tibialis, as the transferred semi-tendon can not reach the cuboid bone.

Materials and methods Written parental permission was obtained to allow the use of information held in the hospital records to be used in this review as Institutional Review Board (IRB) does not exist in our country. The cohort of the study consisted of 48 consecutive ambulant or potentially ambulant patients (52 feet) with spastic paralysis and dynamic equinovarus foot deformity that underwent split anterior (SPLATT) or split posterior (SPOTT) tendon transfer. The hemiplegic patients were 32, the diplegic 12 and the quadriplegic 4. Our inclusion criteria were: 1. ambulatory or potentially ambulatory patients with cerebral palsy, 2. age no less than 6 years at the time of the operation, 3. varus deformity of the hind foot during gait (stance and swing phase), 4. flexible varus foot deformity, and 5. follow-up at least 4 years. Eigtheen feet presented equinus hind foot deformity due to triceps shortening. According to the deformity, the feet were divided in two groups (Group I with predominant forefoot and midfoot inversion and Group II with predominant hindfoot varus). The deformities were flexible in all cases in both groups. The first group consisted of 11 patients (11 feet, 9 female-2 male) all unilateral, 10 of them presenting hemiplegia and one quadriplegia. They also presented prominent forefoot and midfoot inversion due to overactivity of the anterior tibial tendon (AT), associated with a mild cavus component [Figure 2 , 3 ]. Patients in this group underwent the SPLATT (Hoffer's procedure). The second group consisted of 34 patients (38 feet, 24 female-10 male). The hemiplegic patients were 20, the diplegic 11 and the quadriplegic 3. They presented prominent varus hindfoot which persisted during the entire gait cycle due to the overactive PT. Patients in this group underwent the split PT tendon transfer (Green's procedure). Eighteen feet presented also equinus hind foot deformity, requiring concomitant Achilles cord lengthening. Three patients (3 feet), two hemiplegic and one diplegic that were not included in the groups, underwent both procedures because both muscles contributed to a combined deformity and simultaneous SPLATT and SPOTT were performed. Clinical evaluation was based on the inspection of the patients while standing and walking, the range of motion of the foot and ankle, callus formation and the foot appearance using the clinical criteria of Hoffer et al [ 1 ] in Group I and of Kling et al [ 6 ] in Group II. According to Hoffer [ 1 ], the result was considered very good when there was no deformity postoperatively, total foot contact on the ground and proper shoe wearing. Satisfactory was considered when there was mild varus, valgus or equinus deformity, small foot contact and overnight braces were used. Poor was considered when there was overcorrection, undercorrection or equinus > 5° and braces were available. According to Kling [ 6 ] excellent results were graded when the child managed to walk with a plantigrade foot, without fixed or postural deformity, in a regular shoe having no callosities. Patients and parents were pleased with the result and no brace was required post-operatively. Results were graded good in children who walk with less than 5° varus, valgus, or equinus posture of the hind foot, wearing regular shoes, having no callosities and were satisfied with the outcome. Feet with recurrent equinovarus deformity, or overcorrected into a valgus or calcaneovalgus deformity were considered as poor results. The position of the hind foot was evaluated according to the criteria of Chang et al [ 11 ] for the surgical outcome. Severe varus was defined when the hind foot was in > 10° varus and additional operations were required, mild varus when the hind foot was in 5° to 10° of varus and no additional operation was required, neutral when the hind foot was in neutral position or in less than 5° of varus or valgus, mild valgus when the hind foot is in 5° to 10° of valgus with no additional operations and severe valgus when the hind foot was in more than 10° of valgus and additional operations were required. Surgical procedures In SPLATT, the first incision exposed the insertion of the tendon which was split longitudinally as far as through the musculotendinous junction. The medial half of the tendon was left attached to the first metatarsal and first cuneiform, but the lateral half was detached from its insertion. The split lateral half of the tendon was passed subcutaneously into the incision made over the cuboid and then was inserted into the holes made in the bone and either sutured to itself under moderate tension or if the length of the stump was not sufficient, anchoring was carried out with any other technique (pull-out wire, anchoring to the periosteum, etc.). For SPOTT, four separate incisions were used according to Green et al [ 4 ]. The first incision two centimetres long was positioned over the insertion of the posterior tibialis tendon on the navicular. The distal end of the tendon was identified and its sheath was opened. The tendon was split longitudinally and the plantar half was dissected from its insertion. The free end was grasped and the tendon was split longitudinally as far proximally as possible. The second incision begun at the level of the medial malleolus and continued for approximately six centimetres. The free half of the tendon was transferred into the proximal incision and the longitudinal split in the tendon was continued to the musculotendinous junction. A third incision is made directly posterior to the lateral malleolus beginning at the proximal tip of the malleolus and continuing proximally. The peroneus brevis was identified and its sheath was split longitudinally. The distal stump of the split posterior tibial tendon in the second incision was threaded into a tendon-passer that passed the split portion directly posterior to the tibia and fibula and anterior to all the neurovascular and tendinous structures so as to enter laterally to the opened sheath of peroneus brevis. The fourth incision was made along the peroneus brevis and begun distal to the lateral malleolus and continued distally just proximal to the insertion of the peroneus brevis on the base of the fifth metatarsal. The distal part of the sheath was opened and the split posterior tibial tendon was sutured to the peroneal brevis tendon onto the cuboid by fish-mouth technique. The tension should be adjusted so that the hind part of the foot will rest in neutral position, by holding the foot in neutral and pulling hard on the posterior tibial tendon and slightly reducing the pull. The heel-cord lengthening was usually performed prior to the procedure and a long cast was applied with the knee extended and the foot in neutral position. The patient could bear weight on the cast as tolerated and four weeks later the cast was changed and a short walking cast was applied. If the patient was able to dorsiflex the foot and ankle to neutral, no postoperative brace was used.

Results Evaluation of the results was carried out using the clinical criteria of Hoffer [ 1 ] in group one and Kling and Kaufer [ 6 ] in group two (Tables 1 , 2 ). In the former, very good results were obtained in 8 feet and satisfactory in 3. In the later one, 22 feet were excellent, 12 good and 4 poor. The 3 feet that underwent simultaneously both of the procedures presented 1 excellent and 2 satisfactory results. In the first group due to mild cavus foot component supplementary operations were performed at the same time with the index procedure [Figure 3 , 4 ]. Plantar soft tissue releases (open release of the plantar aponeurosis+release of the plantar muscles from their insertion into the calcaneus) were performed in 11 feet, transcutaneous flexor tenotomies in 8, and Jones procedure in 5 feet (Table 2 ). The mean range of motion at the last follow-up was 10-20° of dorsiflexion, 30-40° of plantarflexion, 25-30° of foot inversion and 15-20° of foot eversion. No overcorrection or undercorrection was reported. In the second group, 23 feet presenting concomitant cavus foot component that underwent supplementary operations performed at the same time with the index operation. Plantar soft tissue releases were performed in 15 feet, Jones procedure in 5, long extensor tendons transfer to the metatarsals in 2, as well as transcutaneous flexor tenotomies in 23 feet (Table 3 ). It has also been required concomitant Achilles cord lengthening in 18 feet due to the equinus position of the hind foot. None of the feet presented mild or severe valgus postoperatively, while 4 feet presented severe varus deformity and underwent calcaneocuboid fusion sixteen and eighteen months after the index operation. The mean value of mild varus was (-14,5 ± 12,2°) and concerning the feet with the hind foot in neutral position the mean value was 5.0 ± 7.4°. The results in patients with hemiplegic pattern were better and significantly different than the diplegic and quadriplegic ones (p = 0.005), by using the chi-square analyses as statistical significant involvement at p < 0.01 in the second group (Table 1 ). All patients with an excellent result were brace free at the last follow-up with significant improvement in gait, able to walk with plantigrade feet, use of regular shoes and parent's satisfaction with the outcome. The patients with good results continued to use a night brace (AFO). All of them had good correction of the hind foot equinus and the ankle was able to dorsiflex to at least 90°. The patients presenting a poor result required continued bracing because of the severe varus residual deformity, appearing excessive weight bear on the lateral border of the foot and having painful callosities. These patients required further foot realignment (calcaneocuboid fusion).

Discussion It is generally accepted that overactivity of the AT is responsible for varus-inversion forefoot deformity, whereas the overactivity of PT causes equinovarus hind-foot deformity. Between these two conditions the second is much more common. However, the cause of the deformity can not be clinically identified in some cases, especially when Achilles shortening co-exists. The use of dynamic electromyography and gait analysis can be helpful, but it can not be available in every Institution. Twenty-seven out of thirty-eight feet in the second group presented concomitant cavus foot component and underwent supplementary operations performed concomitant with the index operation. Plantar soft tissue release was the most common out of these procedures, as the release constitutes a keystone procedure for lengthening the shortened base of the foot, and its contribution to the successful outcome for the correction of the cavus component cannot be overemphasized [Figure 5 , 6 ]. Our results in hemiplegic patients were better and significantly different than the diplegic and quadriplegic ones, indicating that the underlying neurologic impairement affect the results of the surgery [Figure 7 , 8 ]. One of our prerequisite for split tibialis tendon transfers was the ability of the patients for walking or the potential of standing and ambulation [Figure 9 , 10 , 11 and 12 ]. The simple lengthening of the posterior tibialis tendon weakens the muscle, and if the tendon and the heel cord are lengthened, then plantar-flexion strength is significantly reduced. All reports regarding split tendon transfers showed favourable results [ 4 - 8 , 11 , 12 ] with our study supporting the same. Two factors that should be considered to perform the procedures are: the flexibility of the deformity and the dorsiflexion of the ankle to at least 5°- 10° beyond neutral. The fixed bony deformity prevents complete correction of the equinovarus position of the foot and if it co-exists, a bone procedure should be considered before the tendon transfer, to prevent persistent varus. In the present series, only flexible deformities passively corrected were included. The extensor tendons transfer to the metatarsals aimed to improve the metatarsophalangeal dysfunction, to enhance ankle dorsiflexion and in association with the transcutaneous flexor tenotomies in several toes to correct the clawing. Although overcorrection is more difficult to treat [ 13 ] we did not observe any, while the poor results in SPLATT included severe varus deformity in 4 feet that required calcaneocuboid fusion. One of our inclusion criteria was the age of the patients at the time of the surgery to be more than 6 years, as it is an important factor for the final outcome. Ruda and Frost [ 14 ] reported that after intramuscular posterior tendon lengthening in 29 patients, the reccurence in varus observed in two, being less than 6 years of age. Lee and Bleck [ 15 ] reported a reccurence of 29% in patients less than 8 years at the time of the operation, as the spastic muscle tends to retain its contractile properties even if it is weakened or transferred at an age younger than 8 years. In conclusion, the rapid bone growth in children that underwent split tibialis tendon transfers in less than 6 years of age, may lead to reccurence. We selected a follow-up period of more than 4 years as the failure rate increases with the postoperative period and the final results can be estimated only after the skeletal growth [ 11 ]. Residual varus deformity in our series were attributed to technical intraoperative errors in balancing the tension between the medial and lateral tendon halves. Four feet underwent bone procedure for the correction of the hindfoot deformity at a later stage. In 3 feet, some technical difficulty was encountered in suturing the split posterior tibial tendon to the peroneus brevis, as the split half being short. Although we performed concomitant intramuscular lengthening of this part of the tendon so as to be sufficient for transfer, we do not recommend it as the tendon loose more of its power. Additionally, in 3 feet both muscles contributed to a combined deformity, which was defined only intraoperatively, and therefore simultaneous SPLATT and SPOTT were satisfactorily performed [Figure 13 , 14 ]. The purpose of doing split tendon transfers as opposed to whole tendon transfers in children with cerebral palsy has been considered to be more effective, as it distributes equally the muscle power, eliminating the possibility of residual deformity or overcorrection. Anterior transposition or rerouting of the posterior tibial tendon has been previously described [ 16 ] but calcaneus deformity may be a result in spastic muscles, by converting the PT into an ankle dorsiflexor [ 17 ]. The anterior transfer of the posterior tibial muscle through the interosseous membrane is an attractive procedure as it removes the actual deforming force and balances the weak or absent anterior tibial and peroneal muscles, but is effective in patients with nonspastic paralytic equinovarus deformities of the foot [ 18 ] A continuously spastic posterior tibial muscle will maintain its spasticity when transferred and if a heel cord lengthening is performed concomitant with the anterior transfer of the posterior tibialis, a calcaneus deformity may likely result. In the majority of our cases, the deforming force was successfully determined preoperatively and the final results justified the applied operative procedure. In complex deformities, other supplementary procedures may be required to achieve the best possible outcome.

Background Overactivity of anterior and/or posterior tibial tendon may be a causative factor of spastic varus foot deformity. The prevalence of their dysfunction has been reported with not well defined results. Although gait analysis and dynamic electromyography provide useful information for the assessment of the patients, they are not available in every hospital. The purpose of the current study is to identify the causative muscle producing the deformity and apply the most suitable technique for its correction. Methods We retrospectively evaluated 48 consecutive ambulant patients (52 feet) with spastic paralysis due to cerebral palsy. The average age at the time of the operation was 12,4 yrs (9-18) and the mean follow-up 7,8 yrs (4-14). Eigtheen feet presented equinus hind foot deformity due to gastrocnemius and soleus shortening. According to the deformity, the feet were divided in two groups (Group I with forefoot and midfoot inversion and Group II with hindfoot varus). The deformities were flexible in all cases in both groups. Split anterior tibial tendon transfer (SPLATT) was performed in Group I (11 feet), while split posterior tibial tendon transfer (SPOTT) was performed in Group II (38 feet). In 3 feet both procedures were performed. Achilles tendon sliding lengthening (Hoke procedure) was done in 18 feet either preoperatively or concomitantly with the index procedure. Results The results in Group I, were rated according to Hoffer's clinical criteria as excellent in 8 feet and satisfactory in 3, while in Group II according to Kling's clinical criteria were rated as excellent in 20 feet, good in 14 and poor in 4. The feet with poor results presented residual varus deformity due to intraoperative technical errors. Conclusion Overactivity of the anterior tibial tendon produces inversion most prominent in the forefoot and midfoot and similarly overactivity of the posterior tibial tendon produces hindfoot varus. The deformity can be clinically unidentifiable in some cases when Achilles shortening co-exists producing foot equinus. By identifying the muscle causing the deformity and performing the appropriate technique, very satisfying results were achieved in the majority of our cases. In three feet both muscles contributed to a combined deformity and simultaneous SPLATT and SPOTT were considered necessary. For complex foot deformities where the component of cavus co-exists, supplementary procedures are required along with the index operation to obtain the best result.

Competing interests The authors declare that they have no competing interests. Consent Written informed consent was obtained from the patient for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Authors' contributions MV: Analysis and interpretation of data, preparation of the manuscript, acquisition of pictures and additional materials, corresponding author. DD: Approved the final manuscript. All authors read and approved the final manuscript.

CC BY

no

2022-01-12 15:21:45

J Foot Ankle Res. 2010 Dec 14; 3:28

oa_package/13/aa/PMC3016343.tar.gz

PMC3016344

21138547

Introduction Ovarian cancer is the second most common malignancy of the female genital tract in the UK. Cancer statistics from 2007 reveal that 4,317 UK women died from ovarian cancer, accounting for around 6% of all female deaths from cancer [ 1 ]. Despite advances in chemotherapy, ovarian cancer mortality rates in the UK since the early 1970 s, have remained stable at ~10-12 per 100,000 women. This is in part due to the asymptomatic nature of the disease with most women presenting at a late stage [ 1 ]. Current treatment with platinum based chemotherapy results in clinical remission in 75% of patients but the median progression free survival is only 16 to 21 months [ 2 ]. Thus, there is a clear need for the development of novel therapies to improve conventional treatments and identify new prognostic markers for survival. Ion channels are pore-forming proteins that help establish and control voltage gradients across the plasma membranes of all living cells by allowing the flow of ions down their electrochemical gradient [ 3 ]. Voltage gated potassium (K + ) channels have recently generated great interest due to their involvement in cell proliferation in various cancers [ 4 ]. Moreover, K + channel blockers have been shown to inhibit proliferation of the ovarian cancer cell line A2780 [ 5 ], identifying voltage-gated K + channels as potential therapeutic candidates for the treatment of cancer. Four main K + channel subtypes (Kv1.3, K2p9.1, Eag and HERG) are found to be overexpressed in a number of tumour types [ 4 ]. K + channels have been suggested to be involved in cancer through the action on membrane potential and regulation of cell volume [ 6 ]. Hyperpolarisation of the cancer cells mediated by K + channels not only leads to increased Ca 2+ influx [ 7 ] a well known factor for regulation of cell proliferation but also maintains the driving force for Na + dependent nutrient transport and influencing intracellular pH [ 6 ]. K + channels have also been shown to affect cell proliferation due to their regulation of intracellular concentration of solute involved in DNA synthesis or activating a cell cycle regulating protein through the effect on cell volume, in fact rat glioma cells show optimal proliferation in a small range of cell volume [ 8 ]. Eag (Ether-a-go-go, Kv10.1) was first isolated from the fruitfly Drosophilia melanogaster as the leg shaking phenotype induced under ether anesthesia [ 9 ]. Eag has a restricted distribution limited to the central nervous system [ 10 ] and expressed transiently in myoblasts [ 11 ]. Chinese Hamster Ovary (CHO) cells transfected with the Eag gene show increased proliferation, growth factor independence and loss of contact inhibition compared to normal CHO cells [ 12 ]. Implantation of Eag-transfected cells in severe combined immune deficient mice resulted in tumour formation. Eag expression has also been detected by RT-PCR in cell lines from different organs including as He-La (carcinoma of cervix), SH_SY5Y (neuroblastoma) and various mammary cell lines (COLO-824, EFM-19, BT_474). Inhibition of Eag expression in EFM-19, HeLA, MCF-7, and SH-SY5Y cell lines with antisense oligonucleotides reduced their growth, demonstrating a role for Eag in cell proliferation [ 12 ]. Eag channel expression has also been demonstrated in various clinical tumours [ 10 ] and cervical cancer [ 13 ]. HERG (Human Ether-a-go-go related gene), also belonging to the Eag family, plays a fundamental role in cardiac excitability by regulating action potential repolarisation. It has been implicated in the molecular basis of familial Long QT 2 syndrome [ 14 ]. A potential role of these channels in cancer was first demonstrated in neuroblastoma cell lines [ 15 ] and they have since been described in cells lines derived from a plethora of malignant tissues and organs [ 16 ]. HERG expression has also been demonstrated in clinical cancers that include endometrial carcinoma [ 17 ], colorectal cancer [ 18 ] and acute myeloid leukaemia [ 19 ]. HERG channels are important determinants of the acquisition of an invasive phenotype in colorectal cancers [ 18 ]. These studies indicate that both Eag and HERG channels demonstrate the potential to be used as tumour markers for ovarian cancer. Further, specific blockers of these channels in conjunction with carboplatin, may be used as novel treatments in order to reduce the dosage required for treatment and associated side effects. They may also be considered as a treatment for patients with resistant or recurrent disease. This is the first study to examine the expression of Eag and HERG channels in ovarian cancer biopsies, with subsequent relation to prognostic factors. We also sought to examine a role for Eag and HERG channels in the proliferation of ovarian cancer cells.

Materials and methods Cell culture The SK-OV-3 cell line (Passage 13) was kindly donated by Dimitra Dafou, Translational Research Laboratories, University College Hospital, London in July 2008, from an authenticated stock provided by the American Type Culture Collection (ATCC). The cell line was cultured in Dulbecco's Modified Eagle's Media (DMEM) supplemented with 10% fetal bovine serum (FBS) and antibiotics (100 u/ml penicillin and 100 μg/ml streptomycin; Invitrogen, Paisley, UK) Antibodies Rabbit anti-human Eag antibody was purchased from Alomone Labs (Israel). Rabbit anti-human HERG antibody was from Abcam laboratories (UK) while secondary anti-rabbit fluorescein isothiocyanate (FITC) conjugated antibodies were obtained from Sigma (Poole UK). The antibodies were specific for the Eag and HERG protein, as demonstrated by specific staining of Eag and HERG protein in ovarian cancer tissue and SKOV-3 cells on Western blotting (Figure 1 ). Cell proliferation assay The proliferation of SK-OV-3 cells was examined in the presence and absence of voltage gated K + channel blockers by (3-(4, 5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt) MTS assay using the CellTiter 96 aqueous non-radioactive cell proliferation assay kit (Promega, UK). The assay was performed according to the manufacturer's instructions to assess the effect of the voltage gated K + channel blockers, 4-aminopyridine (4-AP) and tetraethylammonium (TEA) on proliferation of the SK-OV-3 cell line. 5000 cells were added to each well of a 96 well plate (Perkin Life Sciences) and incubated for 24 hours in 5% CO 2 /air at 37°C. Thereafter, the media was aspirated and replaced with 100 μl fresh media containing either 4-AP or TEA at various concentrations or media only to serve as a control. The cells were incubated for 96 hours and proliferation assessed daily by the addition of MTS (3-(4, 5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt) reagent. Plates were then read at 490 nM using a Victor 1427 multilabel counter (Wallac). All observations were performed in triplicate and experiments repeated thrice. Immunofluorescence SK-OV-3 cells were grown on cover slips in 24 well plates and incubated for 24 hours until 80% confluent. Cells were then fixed with 4% paraformaldehyde for 20 min and treated with 0.5% Igepal before blocking with 3% BSA/5% glycine in PBS and 10% goat serum in PBS. Eag and HERG (both 1: 50 and 1:100 in10% goat serum/PBS) antibodies were added to the cover slips then incubated overnight at 4°C. Wells containing 10% goat serum without the primary antibody were used as a negative control. After incubation, FITC conjugated secondary antibody at 1 in 50 dilution in 10% goat serum/PBS appropriate to Eag and HERG was added and cells incubated at room temperature for 90 min. Images of the cells were captured and analysed using Cell F software (Olympus UK). Immunohistochemistry The ovarian cancer tissue microarray (TMA) (SDLREC Ref 0205/495) was prepared from 336 patients who underwent surgery for ovarian cancer from 1 st January 1982 to 31 st December 1998. Clinicopathological characteristics recorded included age at diagnosis, International Federation of Gynecologists and Obstetricians (FIGO) stage, grade, histological subtype, details of adjuvant treatment, extent of cytoreduction and Disease Specific Survival (DSS) which was then used for analysis. DSS was calculated from the date of the operation until 31 Dec 2005, when any remaining survivors were censored. We calculated that a sample size of 336 patients allowed an 80% chance of detecting a hazard ratio of ≥ 1.40 and ≤0.71 (N Query Advisor software). The TMA were prepared by Ian Spendlove, University of Nottingham as described previously [ 20 ], and were kindly donated to us. For each tumour, 5 μM section, slides were stained with H&E to locate representative areas of viable tissue. Needle core biopsies (0.6 mm) from the corresponding areas on the paraffin embedded tumour blocks were then placed in prespecified coordinates in recipient paraffin array blocks using a manual tissue arrayer (Beecher instruments). Five copies of the array were assembled using different points within the representative tumour area. 5 μM thick sections from each TMA block were placed on coated glass slides to allow for the procedure to be performed. Thus all the biopsies in the TMA were representative of the sample and were validated. After dewaxing in xylene, tissue sections were rehydrated in a graded alcohol series and endogenous peroxidase activity blocked by immersing the slides in 2% hydrogen peroxide in PBS for 20 min. Antigen retrieval was performed by microwaving the slides in 0.1 M sodium citrate buffer at 800 W for 15 min followed by 400 W for 10 min. After blocking with horse serum, the slides were incubated with either HERG antibody (1:200) for 1 hour at room temperature or Eag antibody (1:600) overnight at 4°C. Staining was developed using the Vectastain Elite kit and di-aminobezidine (DAB) chromagen (Dako, Hamburg, Germany) as per manufacturer's instructions. Slides were counterstained with Harris's hamatoxylin and dehydrated in alcohol before mounting in Depex for viewing. The staining intensity was graded as: 0-no staining, 1-low, 2-intermediate and 3-high by three independent observers (VA, HS and GVS) who had no knowledge of the pathological variables or the disease outcome for the dataset. Statistical analysis Analysis was performed using SPSS (version 17.0). Continuous data were analysed using median, interquartile range (IQR) and 95% confidence intervals (CI). Fisher's exact tests were used for comparative analysis of categorical data. Kaplan Meier survival estimates using log-rank testing were determined for all baseline factors specifically stage, grade, residual disease, adjuvant chemotherapy, histological type, and Eag and HERG immunostaining. In all cases, a P value of <0.05 was considered statistically significant. The cell proliferation assay data was analysed using one way analysis of variance with Dunnett's multiple comparison test (Graphpad Prism 5).

Results 1. Eag and HERG potassium channels are expressed in the SK-OV-3 cell line There is demarcated staining of the plasma membrane of cells incubated with Eag antibody while nuclear staining was seen when the cells were incubated with HERG antibody (n = 3; Figure 2A and 2B , higher magnification is shown in inset). Diffuse staining of the cytoplasm and nuclei was seen with both Eag (Figure 2A ) and HERG antibody (Figure 2B ). No immunoreactivity was observed in negative control experiments where primary antibody was replaced with goat serum (Figure 2C ). 2. K + channel blockers inhibit cell proliferation of SK-OV-3 cells 4-aminopyridine (4-AP) inhibited cell proliferation significantly (P < 0.001) at concentrations of 1, 2 and 5 mM compared to control cells at 96 hours with maximum inhibition noted at 5 mM (n = 3). Tetraethylammonium (TEA) inhibition of proliferating cells was also noted at all concentrations at 96 hours (n = 3; P < 0.001) (Figure 2C and 2D ) compared to untreated cells. 3. Eag is a significant prognostic marker for ovarian cancer Clinicopathological characteristics along with Eag and HERG staining of 336 patients with ovarian cancer are summarised in Table 1 . The median age at diagnosis and overall survival was 62 years (IQR 24-90) and 19.5 months (IQR 0-271) respectively. Nearly half of the patients (49.4%) had FIGO stage 3 disease and two thirds (65.8%) had grade 3 tumour. The predominant histological type observed is serous (49.4%) followed by endometriod (11.7%) and mucinous (9.7%). The intensity of Eag and HERG staining was stratified into two groups of high and low/intermediate staining (Figure 3A and 3B ). The survival curves for these groups are shown in Figure 3C and 3D . Biopsies with an insufficient number of cancer cells to produce a reliable result were not included in the final analysis. Patients with high Eag staining had significantly poor survival of 13.8 months (CI 7.2-20.3) compared with 24.1 months (18.9-29.2); (P = 0.016) for the low/intermediate group (Table 2 ). HERG staining did not show significant correlation with survival (P = 0.586). Other prognostic markers for ovarian cancer such as stage, grade, residual disease, histological type and adjuvant therapy also showed significant differences in median survival (Table 2 ). Following identification of a significant association of Eag staining with survival, further evaluation was carried out between the two groups of patients with Eag staining and other prognostic makers. There was no difference in the proportion of patients with stage of disease (P = 0.587), histological type (P = 0.622) or receiving adjuvant chemotherapy (P = 0.594), but there was significant correlation between high Eag staining, grade of tumour (P = 0.014) and residual disease left after the operation (P = 0.011) (Table 3 ). On Cox multivariate analysis (Table 4 ) patients with high Eag staining did not show a significant effect (P = 0.215) on overall survival compared to other prognostic markers like FIGO stage and residual disease. Staining of normal ovary (n = 6) demonstrated low level expression of both Eag and HERG on the surface epithelium compared to ovarian cancer (shown by arrow in Figure 3C ).

Discussion Voltage gated K + channels have been implicated in cell proliferation and the control of cell cycle progression. They also show cell and tissue-specific expression, and impaired expression of K + channels has been detected in a number of cancer and tumour cells [ 21 ]. Both K + channel blockers, 4-AP and TEA, significantly inhibited the proliferation of Eag and HERG-positive SK-OV-3 cells, confirming data by Zhanping et al ., on the A2780 ovarian cancer line [ 5 ]. Both TEA and 4-AP are non selective K + channel blockers capable of inhibiting Eag and HERG as well as other voltage gated K + channels. The therapeutic potential of blocking Eag and HERG channels non-specifically may be limited by cardiotoxicity toxicity resulting from impaired HERG channel activity in the heart. However, recently a monoclonal Eag antibody has been developed which specifically inhibits proliferation based on its Eag blockade with no interaction with HERG channels and has been postulated as a potential therapeutic for the treatment of cancer [ 22 ]. We show here that high Eag expression is associated with poor prognosis in patients with ovarian cancer. Hemmerlein et al. , have reported the presence of Eag channels in ovarian cancer tissue but no expression with normal ovarian follicular epithelium [ 10 ]. High Eag expression is also associated with high grade ovarian cancers and the presence of residual disease at surgery, which in turn is associated with poor outcome, but no correlation between tumour stage, histological type or adjuvant chemotherapy was demonstrated. In contrast to our findings, no association has been shown between Eag expression and different grades of tumour in soft tissue sarcoma patients [ 23 ] which may be due to the non epithelial origin of these tumours. Eag channels are expressed on the surface of the cell membrane are attractive targets for novel treatments in management of ovarian cancer. HERG expression did not show any difference in survival indicating that HERG channels as targets may not have a therapeutic role in ovarian cancer. Though both Eag and HERG channels belong to the same Eag family and despite blockade by the voltage gated blockers and similar positivity on staining, these two channels give rise to very different electrical currents and have different functional role in human tissues. This difference in functionality may be one of the reasons that Eag is linked to poor survival while HERG is not. This is likely to be mediated through cellular mechanisms that link to changes in membrane potential which we intend to explore further. Eag showed significant survival difference on univariate analysis, but there was no significant difference on Cox regression analysis which may be due to insufficient number of patients and having a larger cohort may also change the difference between the two markers.

Conclusion In summary, we have demonstrated for the first time that Eag and HERG K + channels are overexpressed in ovarian cancer and that high Eag staining is associated with significantly poorer survival, identifying Eag as a putative prognostic marker. Moreover inhibiting proliferation of ovarian cancer cells using K + channel blockers indicates that they have a role in ovarian cancer cell proliferation. Further research into specific Eag blockers or using RNA interference to silence the Eag gene is required in order to elucidate the role of Eag channels in ovarian cancer progression and their link to clinical outcome.

Background Ovarian cancer is the second most common cancer of the female genital tract in the United Kingdom (UK), accounting for 6% of female deaths due to cancer. This cancer is associated with poor survival and there is a need for new treatments in addition to existing chemotherapy to improve survival. Potassium (K + ) channels have been shown to be overexpressed in various cancers where they appear to play a role in cell proliferation and progression. Objectives To determine the expression of the potassium channels Eag and HERG in ovarian cancer tissue and to assess their role in cell proliferation. Methods The expression of Eag and HERG potassium channels was examined in an ovarian cancer tissue microarray. Their role in cell proliferation was investigated by blocking voltage-gated potassium channels in an ovarian cancer cell line (SK-OV-3). Results We show for the first time that high expression of Eag channels in ovarian cancer patients is significantly associated with poor survival (P = 0.016) unlike HERG channel expression where there was no correlation with survival. There was also a significant association of Eag staining with high tumour grade (P = 0.014) and presence of residual disease (P = 0.011). Proliferation of SK-OV-3 cells was significantly (P < 0.001) inhibited after treatment with voltage gated K + channel blockers. Conclusion This novel finding demonstrates a role for Eag as a prognostic marker for survival in patients with ovarian cancer.

Financial support Derby Hospitals Charitable funds, Gynaecology Oncology Research Fund No 21470. Competing interests The authors declare that they have no competing interests. Authors' contributions VA performed the experiments and wrote the manuscript, RK,AV guided in performing experiments and helped in correction of manuscript, RS, AB helped in collection of tissue samples, GVS and HS helped in reviewing of slides and correction of manuscript. All authors read and approved the final manuscript.

CC BY

no

2022-01-12 15:21:45

Diagn Pathol. 2010 Dec 7; 5:78

oa_package/6d/8f/PMC3016344.tar.gz

PMC3016345

21144044

Background Parkinson's disease (PD) and related disorders known as synucleinopathies are characterized by abnormal accumulation of α-synuclein (α-Syn). Such filamentous aggregates in neuronal perikarya and processes are referred to as Lewy bodies and Lewy neurites, respectively [ 1 , 2 ]. Most synucleinopathies are sporadic, but some are linked to missense mutation or multiplications of the gene encoding α-Syn [ 3 - 8 ]. In order to understand how and why α-Syn accumulation affects neuronal survival, several experimental models have been generated. In some cell-based studies overexpression of α-Syn did not affect cell survival but sensitized the cultures (mutant α-Syn in particular) to insults [ 9 , 10 ]. In others, the results were different. Infection of primary mesencephalic neuronal cultures with lentiviral constructs containing A53T mutant α-Syn led to reduced cell viability [ 11 ]. Transient co-expression of mutant α-Syn and synphilin-1 in H4 cells was reported to cause formation of cytoplasmic α-Syn inclusions and cytotoxicity. These changes were alleviated by treatment with Agk2, a histone deacetylase (HDAC) inhibitor [ 11 ]. Transient expression of wild-type α-Syn in the nuclei of SH-SY5Y cells, but not cytoplasm, was reported to cause α-Syn aggregation and cytoxicity that can be prevented by a 48-hour (h) treatment with 10 mM sodium butyrate (SB) [ 12 ]. It remains unknown whether benefits comparable to those observed in transient transfectants are attainable in stable transfectants overexpressing α-Syn and whether comparable outcomes can be achieved with exposure to other HDAC inhibitors, since there are multiple classes of HDAC. Some of these issues were examined in the present studies using cell cultures, referred to as 3D5, of human neuroblastoma BE2-M17D cells overexpressing wild-type human α-Syn [ 13 ]. The overexpression of human α-Syn in 3D5 cultures is regulated by the tetracycline-off (TetOff) inducible mechanism. These cells develop neuritic processes and express neuronal markers in response to all-trans-retinoic acid (RA) treatment [ 14 ]. 3D5 cells that have been treated with RA for 10 days (ds) followed by 28 ds of induced α-Syn expression are capable of forming α-Syn inclusions that reproduce many morphological and biochemical characteristics of Lewy bodies [ 14 ]. Similar α-Syn inclusions were not readily detected in 3D5 cultures after 14 ds of induced α-Syn expression, although the cultures were biochemically demonstrated to contain α-Syn oligomers. It has been well documented that acetylation and deacetylation of histone proteins play important roles in the epigenetic regulation of transcription and other cellular functions [ 15 , 16 ]. Deacetylation of histone proteins leads to the silencing of gene expression. At least four major classes of HDACs have been identified in mammalian cells [ 17 - 21 ] and several HDAC inhibitors have been developed. Some HDAC inhibitors are more specific to certain subclasses of HDAC than others [ 17 , 18 ]. For example, fatty acid derivatives such as sodium butyrate (SB) and valproic acid (VPA) are capable of inhibiting class I (HDAC8 in particular) and class II HDAC [ 22 ]. In comparison, Agk2 is more specific to sirtuin 2, a member of class III HDAC [ 11 ]. In this study we used RA-differentiated 3D5 cultures, with or without induced α-Syn expression, and evaluated their responses to SB, VPA or Agk2 exposure. We began with cultures that have 10 ds of induced α-Syn expression. The goal was to establish background information on the drug dosage and duration of treatment and use the information in a subsequent cell-based study with an extended duration (e.g 28 ds) of α-Syn overexpression. We unexpectedly found that SB or other HDAC inhibitors treatment of 3D5 cells with induced α-Syn expression caused an increase of α-Syn aggregate accumulation in a time- and dose-dependent manner. Such changes were not detected in the SB-treated, non-induced cultures or in 3D5 cells treated with sodium acetate (SA). More important, 3D5 cultures with α-Syn induction were less viable than those without the induction when exposed to HDAC inhibitors. Similar results were obtained using primary neuronal cultures generated from α-Syn transgenic (TG) mice and their non-transgenic (NT) littermates. The TG mice express wild-type human α-Syn at levels much lower than that displayed by the TetOff induced 3D5 cultures. These findings prompted us to examine whether α-Syn aggregation is closely associated with apoptosis or upstream pathways.

Materials and methods Conditional transfectant 3D5 Alpha-Syn transfectant 3D5 was derived from a human neuroblastoma cell line (BE2-M17D) and characterized previously [ 13 , 14 ]. The transfectant expresses wild-type human α-Syn and displays neuronal phenotypes upon TetOff induction and incubation with RA [ 14 ]. Cultures of 3D5 were maintained in DMEM/10% fetal bovine serum with 2 μg/mL Tet (referred to as non-induced or α-Syn-) or without (referred to as induced or α-Syn+) at 37°C and 5% CO 2 . Those intended for biochemical analysis were seeded at 0.5 × 10 6 cells/plate (100 × 20 mm, BD Biosciences, San Jose, CA) or 1.5 × 10 5 cells/well in 6-well plates. For immunofluorescence or spectrophotometric assay, 3D5 cells were seeded at 2 × 10 4 cells/well on coverslips in 24-well plates and 1 × 10 4 in 48-well plates (Bellco Glass Inc, Vineland, NJ), respectively. Media were replaced the next day with Neurobasal medium (Invitrogen, Carlsbad, CA), 2% B-27 supplement (antioxidant free, Invitrogen), 2 mM L-glutamine (Sigma) and 10 μM RA (Sigma-Aldrich, St Louis, MO). After 10 ds of differentiation and induced α-Syn overexpession, cells were treated with HDAC inhibitors [SB (sigma), VPA (Sigma) and Agk2 (EMD Biosciences, La Jolla, CA)], a derivative of short-chain fatty acid [sodium acetate (SA)], an endoplasmic reticulum (ER) stress inducer [tunicamycin (TM, Sigma)], an inhibitor of phosphatases that act on the eukaryotic translation factor 2 subunit alpha (EIF2α) [salubrinal (Sal, Tocris Bioscience, Ellisville, Missouri)], a protein synthesis inhibitor [cycoheximide (CHX, Sigma), or a pan caspase inhibitor (CI) [Z-VAD (OMe)-FMK (EMD Biosciences)]. The cultures were treated with different drug concentrations for durations described in the Result section. Some cultures were treated with two of the aforementioned drugs or with SB and tetracycline (2 μg/ml, for blocking α-Syn induction). Primary neuronal cultures were prepared from hippocampal and cortical brain tissues of postnatal (day 1) mice (see below) with or without transgenic expression of wild-type human α-Syn, according to a protocol reported previously [ 61 ]. They were plated at 6 ~ 7 × 10 5 cells per well on poly-D-lysine (Sigma) coated 6-well plate and maintained for 7 ds before treatment in the absence or presence of different drugs for different durations. Animals All animal procedures were approved by the Mayo Clinic Institutional Animal Care and Use Committee (IACUC) and were in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 80-23) revised 1996. Alpha-Syn PAC transgenic mice Mice were provided by Dr. Farrer's Laboratory and will be described in detail elsewhere. Briefly, transgenic mice were generated using a P1 artificial Chromosome (PAC) RP-1 27 M07, which was previously identified to contain the entire wild-type human α-Syn gene and regulatory sequences PAC DNA [ 62 ]. Purified DNA was injected into FVB/N (Taconic, Germantown, NY) fertilized oocytes and transplanted into pseudo-pregnant ICR (Harlan, Indianapolis, IN) female mice. Transgenic founders were genotyped by PCR and analyzed for gene expression by quantitative RT-PCR and western blotting. The founder line with the highest expression was expanded and maintained on FVB/N background (Taconic). Human protein expression in 6 month-old mice is approximately 2 fold of endogenous murine α-Syn. The expression level is highest in the hippocampus followed by cortex, olfactory tubercle and striatum. Antibodies To detect α-Syn, we used monoclonal antibody Synuclein-1 (BD Biosciences) and polyclonal antibody Ab98 [ 63 ]. For other proteins we used rabbit polyclonal antibodies to total and cleaved Caspase-3, Caspase 9 and PARP (Cell Signaling, Danvers, MA), phosphorylated EIF2α (pEIF2α, Invitrogen), GADD153/CHOP10 (Sigma), XBP-1 and ATF6α (Santa Cruz Biotechnology, Santa Cruz, CA). We also used monoclonal antibodies to GRP78 (BD Biosciences), total EIF2α (T-EIF2α, Cell Signaling) and GAPDH (1D4, Covance Research Products, Berkeley, CA). The binding of primary antibody to protein of interest was detected using peroxidase-conjugated goat anti-rabbit or anti-mouse IgG antibodies (Chemicon, Temecula, CA) and Alexa594 or Alexa488 goat anti-mouse antibodies (Invitrogen). Western blot analysis Cell cultures were harvested and centrifuged at 200 × g for 15 min. The pellet was resuspended in a lysis buffer [20 mM MES, pH 6.8; 80 mM NaCl, 1 mM MgCl 2 , 2 mM EGTA, 10 mM NaH 2 PO 4 , 20 mM NaF, phenylmethylsulfonyl fluoride (PMSF, 1 μg/ml and leupeptin, 10 μg/mL], homogenized and centrifuged at 180 × g for 15 min at 4°C. The cell lystates were mixed with Tricine-SDS sample buffer (Invitrogen) and 2% β-mercaptoethanol, boiled for 5 min and resolved by SDS-PAGE using 10-20% Tris/Tricine gel (Bio-rad, Hercules, CA). Precision plus protein standards (Bio-Rad) were included as references. Proteins separated by SDS-PAGE were transferred onto nitrocellulose and processed for immunolabeling for proteins of interest. Membranes probed with GAPDH antibody were used as loading controls. Immunoreactivity was visualized with enhanced chemiluminescence (ECL plus, Amersham Pharmacia Biotech, Buckinghamshire, UK) or SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific, Rockford, IL) and analyzed as before [ 64 ]. Fractionation Studies of Culture Specimens Cell lysates were fractionated to derive buffer extractable (SN1), 5 M NaCl/1% sarkosyl-soluble (SN2) and 1% sarkosyl-insoluble fractions (SKI) as reported [ 14 ]. The SKI fraction was finally resuspended in 50 mM Tris (pH 8.0). All fractions and lysates were analyzed by Western blotting. A portion of the SKI sample from TetOff induced cultures with or without SB treatment and corresponding sample from cultures without α-Syn induction were analyzed further (see below). Electron and immunoelectron microscopies Sakosyl-insoluble samples were processed for electron and immunoelectron microscopies using protocols described previously [ 64 ]. Cultures with or without SB treatment were processed as before [ 14 ] for electron microscopy. Immunocytochemistry Cells grown on coverslips were rinsed with PBS, fixed in 4% paraformaldehyde and permeabilized with 0.1 M Tris-buffered saline (TBS, pH 7.6) containing 0.5% triton X-100 for 5 min. They were subsequently blocked with 3% goat serum in TBS, incubated with antibody Synuclein-1 in TBS containing 1% goat serum overnight at 4°C then incubated with Alexa594 goat anti-mouse secondary antibody for 1 h. For Thioflavin S staining the coverslips were immersed in 0.0005% thioflavin S/PBS/formalin and washed 3 times with 70% ethanol and TBS. After immuno- and thiofavin S staining, the coverslips were stained with DAPI (Invitrogen) for 10 min to locate the nuclei and evaluated by confocal fluorescence microscopy (Zeiss LSM 510; Carl Zeiss MicroImaging, Thornwood, NY). Cell viability The viability of 3D5 cells was assessed by incubation of cultures with 2 μM calcein AM (Invitrogen), according to the manufacturer's recommendation. Fluorescence emitted from live cells was measured at 495 nm (excitation)/530 nm (emission) using a Cary Eclipse Fluorometer and Spectra Max M2 and Soft Max Protein 4.6 software (Molecular Devices, Sunnyvale, CA). All measurements were performed in triplicate. Viability of primary neuronal cultures was assessed by counting cells in live cultures. We randomly captured the image of cultures at 10-15 fields (200×). Each image was evaluated independently by two-individuals for intact cells to derive the number of cells in all fields as well as the average cell counts per field. After image capturing, the cultures were processed for Western blot analysis. Statistical analysis Data from at least 3 sets of independent experiments were analyzed by one-way Anova with Dunnett's post hoc test or Student's t test to determine statistical significance.

Results SB exposure causes accumulation of α-Syn oligomers and activation of caspase 3 in a dose- and time-dependent manner We exposed 3D5 cultures to between 2 and 20 mM SB for 24 hrs and determined the effects of these exposures on α-Syn oligomer accumulation. This drug is commonly used at a millimolar range and has been shown to inhibit histone deacetylation in cell cultures at 2 mM or higher concentrations [ 23 ]. Prior to the drug treatment, 3D5 cultures were differentiated with RA and induced to express α-Syn for 10 ds. Compared to non-treated cultures, the drug treatment caused accumulation of different sized (≥34 kDa) α-Syn oligomers in a dose-dependent manner as demonstrated by Western blotting using α-Syn antibodies Synuclein 1 (Figure 1A ) and Ab 98 (data not shown), which recognize different epitopes. This increase in α-Syn oligomer levels was accompanied by increased amounts of active caspase 3 (Figure 1A ). We also analyzed cell lysates from cultures treated with 10 mM SB for different durations. We found that the levels of α-Syn oligomers and active caspase 3 increased in a time-dependent manner (Figure 1B ). Similar results were demonstrated in 3D5 cultures exposed to VPA or Agk2 (Additional file 1 ). In contrast, exposure of 3D5 cells to sodium acetate (SA, also a derivative of short-chain fatty acid) did not increase α-Syn oligomer accumulation (Additional file 2 ). Cultures with SB exposure accumulate α-Syn oligomers with different solubility To characterize α-Syn oligomers further, we fractionated total lysates (TL) from non-treated (control) and SB-treated 3D5 cultures to generate buffer-soluble (SN1), salt plus sarkosyl-extractable (SN2) and sarkosyl-insoluble (SKI) samples and then analyzed using Western blotting (Figure 2 ). All fractions from SB-treated cells contained α-Syn oligomers (Figure 2 ), although the SKI fraction displayed more large-sized oligomers, some of which were retained at the top of separation gel. After correction for differences in sample loading, the amount of α-Syn oligomers recovered in SKI fraction was about the same as that recovered in the SN2 or SN1 fraction. In comparison, samples from non-treated cells contained much less sarkosyl-insoluble α-Syn oligomers. Morphological assessment of cultures and SKI samples Immunocytochemistry To determine whether SB exposure during α-Syn induction (α-Syn+) promotes accumulation of α-Syn aggregates and subsequent fibrillogenesis, 3D5 cells were subject to immunocytochemistry with α-Syn antibodies and Thioflavin S staining. Many SB-treated cells displayed both α-Syn and Thioflavin positive inclusions (Figure 3 , SB-treated, marked by arrowheads), however, such a labeling pattern was neither demonstrated in α-Syn+ cultures without SB exposure nor in α-Syn-cultures with SB treatment (Figure 3 , Control). Electron microscopy The SB-treated α-Syn+ cells contained two groups of filamentous structures, which appeared in clusters or bundles (Figures. 4A-D ). These filaments had a diameter of either 8-10 nm or about 5 nm (Figures. 4B & 4D ), and are far less abundant in α-Syn+ cultures without the SB treatment (Figures. 4E & 4F ). Immunoelectron microscopy To determine whether SB exposure resulted in assembly of filamentous α-Syn, we probed SKI samples prepared from cultures with either α-Syn antibodies or without, followed by secondary antibodies with immunogold labeling and viewed at the ultrastructural level. Filamentous elements of diameter 8-10 nm were detected in samples from the SB-treated samples only, and they often appear as small bundles (Figure 5A-C ) positive for α-Syn (Figure 5D , arrows marked immunogold labeled filaments). In contrast, 5 nm filaments were not detected in these samples by electron or immunoelectron microscopy. SB exposure does not affect the level of α-Syn in 3D5 cells with induced α-Syn expression It has been reported that a major cellular effect of SB is modulation of gene expression [ 15 , 16 ]. Therefore, the accumulation of α-Syn oligomers in SB-treated 3D5 cultures may be due to an increase of α-Syn expression. This possibility was tested by comparing the amount of total, monomeric and oligomeric α-Syn in the SB-treated and non-treated cultures, which had been subjected to 10 ds of differentiation plus 9 ds of induced α-Syn expression or 10 ds of differentiation plus 5ds of induction (started on the 5 th day of differentiation). By dot blotting (Figure 6 ), we detected very few differences in α-Syn expression between the 5 d-induced cultures with and without SB treatment, but found higher levels of α-Syn in the 9 d-induced cultures. Based on 3 independent experiments, in which the ratio of α-Syn to GAPDH was set to 1 for 5d-induced samples, the ratio is 0.97 ± 0.09 (average ± stand error of mean) and 1.58 ± 0.06, for the 5d-induced with SB treatment and the 9-d induced cultures, respectively. Statistical analysis demonstrated a significant difference (P < 0.001) between the 5d- and 9d-induced samples but no difference (P > 0.8) between the 5d-induced and its SB-treated counterpart. More α-Syn oligomers were observed by Western blot in the 5 d-induced, SB-treated cultures than the 9 d-induced, non-treated. The results indicate that a 24-h exposure of 3D5 cells to SB does not affect α-Syn levels, suggesting that other factors, besides α-Syn expression, may be involved in the accumulation of α-Syn oligomers in cultures with SB treatment. To demonstrate further that SB treatment did not affect oligomeric α-Syn accumulation via an increase of α-Syn synthesis, we analyzed 3D5 cells that have been exposed to SB, SB plus Tet (2 μg/ml, for blocking of α-Syn induction), SB plus cycloheximide (CHX, 20 μM, for inhibition of protein synthesis), CHX or Tet for 36 hs by Western blotting. The analysis showed the co-treated cultures contained only slightly fewer α-Syn oligomers than those treated with SB, but much more than those treated with CHX or Tet (Additional file 3 ). The results support the possibility that oligomer accumulation in SB-treated cells is unlikely caused by an increase of α-Syn expression. Caspase inhibitor treatment has no affect on α-Syn oligomer accumulation To determine the effect of SB treatment on cell viability, we treated subsets of differentiated and induced 3D5 cultures with SB or SB plus a pan caspase inhibitor (CI). Cultures without any drug treatment or treated only with CI were included as controls. The SB exposure caused a significant decrease in cell viability (Figure 7A ). Such effect was significantly attenuated in cultures co-treated with CI, indicating a role for apoptosis in SB cytotoxicity. Treatment of cultures with CI alone did not affect their viability. We next determined whether the SB-treated cultures differed from non-treated in terms of apoptosis and whether apoptosis per se is responsible for accumulation of α-Syn oligomers (Figure 7B ). The SB-treated cultures were found to contain far more active forms of capases 3, caspase 9 and cleaved PARP than the non-treated, indicating the treated underwent apoptosis. In cells co-treated with SB and CI, the amount of both active capases 3 and 9 was reduced, whereas the level of α-Syn oligomers was not, indicating α-Syn aggregation is likely an upstream event of caspase activation. Exposure of α-Syn overexpressing 3D5 cultures to SB or other HDAC inhibitors induces ER stress response leading to accumulation of α-Syn oligomers and apoptosis Exposure to HDAC inhibitors such as SB has been reported to cause not only apoptosis, but also lipid peroxidation [ 24 ], oxidative stress [ 24 - 27 ], mitochondrial dysfunction [ 28 , 29 ] and ER stress [ 30 - 32 ]. Since accumulation of α-Syn oligomers in 3D5 cells was not blocked by treatment with pan caspase inhibitor, we decided to test whether it can be reduced by treatment with inhibitors capable of blocking several upstream events leading to caspase activation. Cell cultures were treated with SB or co-treated with different dosages/durations of vitamin C, vitamin E, cyclosporine A and salubrinal (Sal), which can counter oxidative stress [ 33 ], lipid peroxidation [ 34 ], cytochrome C release [ 35 ] and EIF2α dephosphorylation [ 36 ], respectively. Lysates from these cultures were compared for α-Syn oligomer accumulation. Among the agents tested, Sal was the only one capable of preventing the SB-treated cells from both apoptosis and accumulation of α-Syn aggregates (Figure 8 ); hence, ER stress response became the focus of our subsequent studies. As reported before, Sal treatment alone had no adverse effects on 3D5 cells. To determine whether ER stress plays a role in α-Syn oligomer accumulation, we used a known ER stress inducer, tunicamycin (TM). Similar to SB-treated cells, the TM-treated cultures displayed higher levels of α-Syn oligomer than their non-treated counterparts (Figure 8B ). Moreover, the increase was impeded by co-treatment with Sal, thus establishing a close link between ER stress and α-Syn oligomer accumulation. To verify the effects of drug treatment on ER stress, lysates from α-Syn expressing 3D5 cells (marked as α-Syn+ in Figure 8 ), treated with or without SB, were probed with antibodies to several markers relevant to ER stress response (i.e. GRP78, p-EIF2α, XBP-1, ATF6α and CHOP) and with antibodies to GAPDH to serve as loading control (Figure 8B ). The SB exposure caused an increase in the ratio of p-EIF2α to total EIF2α and that of CHOP and other ER stress markers to GAPDH (Additional file 4 , compared SB to Con). α-Syn+ cultures co-treated with Sal and SB displayed lower levels of CHOP (a molecule involved in ER stress-induced apoptosis), XBP-1 (Additional file 4 ), but higher levels of p-EIF2α and GRP78 when compared to the SB-treated. The results are consistent with those reported previously, indicating that Sal is an inhibitor of EIF2α dephosphorylation and that such inhibition leads to an increase in molecular chaperones such as GRP78 and a decrease in CHOP [ 37 - 39 ]. Studies of 3D5 cultures exposed to other HDAC inhibitors (i.e. Agk2, VPA) yielded comparable results (Additional files 5 & 6 ). Overexpression of α-Syn elicits ER stress response Previous studies reported that induced expression of A53T mutant α-Syn led to PC12 cell death and that the toxicity is mediated via ER stress and mitochondrial cell death pathways [ 40 ]. However, it was unknown whether induced expression of wild-type α-Syn in 3D5 cells also elicits an ER stress response. We investigated this issue by probing 3D5 cell lysates with or without induced α-Syn expression for ER stress markers. The induced cultures (α-Syn+) displayed a higher ratio of either GRP78 to GAPDH or p-EIF2α to total EIF2α than their non-induced (α-Syn-) counterparts, indicating the overexpression of wild-type α-Syn per se can cause an ER stress response (Figure 8B & Additional file 4 ). It is worth noting that without drug treatment neither α-Syn- nor α-Syn+ cultures had detectable amounts of CHOP, XBP-1 or ATF6α. Induced expression of α-Syn sensitizes 3D5 cultures to SB or TM toxicity via ER stress response Cultures from α-Syn+ and α-Syn- were either treated with SB, TM or not for 36 hrs followed by cell viability assay (Figure 8A ). At this time point cell death occurred in both cultures with drug treatment. The drug exposure significantly caused more cell loss in the α-Syn+ cultures than their α-Syn- counterparts. This difference in drug response was significantly reduced in cultures exposed to SB plus Sal or to TM plus Sal. Similar to what we observed in α-Syn+ cultures, exposure of α-Syn- cultures to SB, VPA, Agk2 or TM caused an increase in ER stress response (Figure 8B , Additional files 4 & 5 ). However, the drug treatment was less effective in altering the level of most ER stress markers and activation of caspase 3 in α-Syn- than α-Syn+ cultures. Together, the results indicate that α-Syn overexpression elicits ER stress response leading to the sensitization of 3D5 cultures to insults. α-Syn- cultures co-treated with Sal and SB or other aforementioned agents also led to increase in the level of EIF2α phosphorylation and that of GRP78. However, the amount of ATF6 was not affected. The result is consistent with a recent report in which Sal treatment did not affect ATF6 ER stress pathway [ 41 ]. ER stress response, α-Syn aggregation and apoptosis in primary neuronal cultures exposed to SB and other HDAC inhibitors To determine whether the effects of HDAC inhibitors treatment on α-Syn aggregation and ER stress response are unique to 3D5 cultures, we next compared primary cultures from α-Syn PAC mice (also referred to as TG mice) to those of non-transgenic (NT) littermates for their responses to the drug exposure, focusing first on cell viability. The TG mice expressed low levels of wild-type human α-Syn and did not have α-Syn aggregates or neurodegeneration at any age (up to 24 months). Cultures from both TG and NT mice had similar cell numbers before their exposure to the drug. After 2 ds of SB exposure (Figure 9A ), significant cell loss was demonstrated in TG cultures whereas such cell loss in NT cultures was detected after 4 ds of drug treatment (Figure 9B ). In comparison, a significant cell loss was observed after 24 hrs of the drug exposure in SB-treated 3D5 cultures. Consistent with what we observed in 3D5 cultures, co-treatment with Sal significantly protected primary neuronal cultures from the cytotoxicity of SB (Figure 9 ) or Agk2 (Additional file 7 ). Overexpression of α-Syn in TG specimen was confirmed by Western blot analysis of cell lysates from primary cultures. The level of α-Syn monomers in TG cultures is about 1.6 times of that detected in NT controls (Figure 10 , compared lanes 1 and 5). In the absence of SB exposure, α-Syn oligomers were not readily detected in TG or NT cultures. Upon SB treatment, both cultures accumulated α-Syn of a size consistent with that of dimer (i.e. ~34 kDa), but the amount appeared to be greater for the TG cultures. The increase in α-Syn dimer did not result in the generation of inclusions detectable by immunocytochemical staining with α-Syn antibodies or Thioflavin S labeling. Co-treatment of primary cultures with SB and Sal resulted in reduced level of α-Syn dimers. The results are consistent with those demonstrated in 3D5 cultures. Treatment of primary neuronal cultures with SB also resulted in an ER stress response and activation of caspase 3 (Figure 10 ). Importantly, such response was more intense in the SB-treated TG cultures than similarly treated NT controls and less intense with Sal co-treatment (Additional file 8 ).

Discussion Many studies have been devoted to identifying the mechanisms involved in α-Syn assembly and determine whether the accumulation of α-Syn assemblies is cytotoxic or protective. Although the results are far from conclusive, it is reasonable to suggest that the physicochemical property of α-Syn oligomers determines whether they are cytotoxic and what relation the resultant cytotoxicity has to aggregate formation. In this regard, the 3D5 transfectant is of great value for identification of pathways that modulate the risk of α-Syn-mediated cytotoxicity and agents capable of inhibiting accumulation of α-Syn and its assemblies, due to its ability to accumulate α-Syn assemblies without compromising cell survival [ 14 ]. A number of agents have been tested in previous cell-based studies and shown to reduce α-Syn aggregation and its toxicity [ 11 , 12 , 42 , 43 ]. In the present study, we investigated the effect of SB, Val or Agk2 treatment on α-Syn assembly/accumulation in 3D5 cultures. We demonstrated that exposure to each of these drugs reduced the viability of differentiated, α-Syn induced 3D5 cells and enhanced α-Syn oligomer accumulation, and that overexpression of wild-type α-Syn alone is not cytotoxic. This is in contrast to what was observed in previous studies [ 12 ], in which overexpression of α-Syn in the nuclei, but not cytoplasm, of neuroblastoma cells, SH-SY5Y, resulted in cell death. Treatment of SH-SY5Y cultures with 10 mM SB for 48 hs was shown to protect cells from toxicity caused by nuclear accumulation of α-Syn, but did not adversely affect those with cytoplasmic α-Syn accumulation. Treatment of H4 cells overexpressing A53T α-Syn with Agk2 was shown to reduce cell death caused by cytoplasmic α-Syn aggregates. In contrast, treatment of α-Syn overexpressing 3D5 cells with Agk2 caused cell death. The discrepancy may reflect differences in cell type, as supported by the results from our analysis of non-induced 3D5 cultures and non-transfected SH-SY5Y cells (Additional file 9 ). Cell death was noted in 3D5 cultures within 24 hs of SB (10 mM) treatment. In comparison, cell death was not observed up to 48 hrs after treatment of SH-SY5Y cells, indicating they are less sensitive than 3D5 cells to SB cytotoxicity. Therefore, to evaluate whether drug treatment has beneficial effects or not in cell-based studies, it is important to examine more than one cell type. For this reason, we included primary neuronal cultures in the present studies. This also allowed us to substantiate the pathobiological relevance of findings from cell-line based studies. We showed that exposing 3D5 and primary cultures to HDAC inhibitors caused an accumulation of α-Syn oligomers, an increase in ER stress response and a decrease in cell survival. In this regard, several recent studies have demonstrated the induction of ER stress response in cell cultures following treatment with different HDAC inhibitors [ 30 - 32 ]. In our studies, such stress response was greater for cultures with α-Syn overexpression than those without. We considered that the difference is at least in part, if not entirely, due to differences in the basal level of ER stress markers. This is in light of our results that (i) α-Syn overexpression alone, regardless of whether it led to the assembly of filamentous α-Syn [ 14 ] or not, elicited ER stress response, (ii) inhibition of EIF2α dephosphorylation via Sal treatment protected cultures from undergoing apoptosis caused by HDAC inhibitors exposure, (iii) treatment of cultures with ER stress inducer TM also resulted in α-Syn oligomer accumulation as in the HDAC inhibitors treated, and (iv) such accumulation was reduced by co-treatment of cultures with Sal. It has been reported that overexpression of A53T mutant α-Syn or wild-type α-Syn phosphorylated at serine 129 caused ER stress and cell death in PC12 or SH-SY5Y cells [ 40 , 44 ], and that expression of human α-Syn in yeast cells led to blocking of ER-Golgi traffic [ 45 ]. Recently, induced α-Syn expression was shown to increase the vulnerability of PC12 cells to stressors such as thapsigargin, TM and 1-methyl-4-phenyl-pyrisinium (MPP+) [ 46 ]. We also have data demonstrating differences between the induced and non-induced 3D5 cultures in sensitivity to neurotoxin treatment (i.e. MPP+, rotenone, 6-hydroxydopamine(6-OHDA), paraquat) (Additional file 10 ). Other investigators have also provided evidence supporting the important role of ER stress response in PD pathogenesis. For example: (i) activation of ER stress response was detected in dopaminergic neurons in the substantia nigra of postmortem PD cases [ 47 ], and (ii) PD animal or cell models generated via exposure to neurotoxins displayed ER stress response [ 48 , 49 ]. We do not know how HDAC inhibitor treatment leads to ER stress in neuronal cells. However, recent studies of non-neuronal cells demonstrated that (i) exposure of breast cancer cells to panobinostat (a pan HDAC inhibitor) led to deacetylation of GRP78, dissociation of GRP78 from PERK and increase the levels of pEIF2α, ATF4 and CHOP [ 32 ], (ii) exposure of colorectal carcinoma cells to tricostatin A (a pan-HDAC inhibitor) affected the binding of HDAC to GRP78 promotor [ 30 ] and (iii) the C-terminal region of RTN-C protein, an ER membrane protein, can interact with HDAC8 in vitro [ 50 ]. Our studies showed that 3D5 and primary cultures have different sensitivities to HDAC inhibitor treatment. When the same dosage of SB or VPA was used to treat these cultures, it required a longer duration of incubation for primary cultures to display cytotoxicity. In contrast, 3D5 cells were less sensitive than primary cultures to Agk2 exposure. The cause(s) of such differences in drug sensitivity remains unknown. We noted that a previous study has demonstrated a protective effect of Agk2 treatment on rat primary neuronal cultures overexpressing A53T α-Syn via lentiviral infection of constructs containing mutant α-Syn [ 11 ]. The discrepancy between the previous study and our primary neuronal culture study may be due to differences in experimental paradigm. Low levels of overexpressing wild-type human α-Syn in mice had no impact on the viability of mouse primary neuronal cultures. In comparison, overexpression of A53T α-Syn by viral infection reduced cell survival. We showed that 3D5 and primary cultures treated with HDAC inhibitors have different quantity/quality of α-Syn oligomers. Most α-Syn oligomers appearing in primary neuronal cultures are dimers and relatively soluble, whereas those in 3D5 cells are larger oligomers with different solubility (ranging from buffer-extractable to sarkosyl-insoluble). The distinction may reflect difference in α-Syn expression levels. This is supported by our findings that the level of α-Syn in TetOff induced 3D5 cells is far greater than that expressed in primary neuronal cultures from TG mice (based on equal sample loading (Additional file 11 ). Moreover, it has been reported that a critical concentration of α-Syn is required to initiate α-Syn assembly in test tubes, and that formation of dimer is an early step of α-Syn assembly [ 51 ]. The difference between cell types in α-Syn oligomers production may also reflect the rate of α-Syn assembly. In this regard, several molecules have been shown to promote α-Syn assembly in vitro and in vivo [ 52 - 56 ]. It is possible that the amount/type of endogenous inducer for α-Syn assembly may vary in different cell types. Our finding that an overexpression of wild-type human α-Syn in both 3D5 and primary cultures can elicit ER stress response is noteworthy. According to some studies, the accumulation of α-Syn in the human brain elevates significantly during aging [ 57 , 58 ]. There is also ample evidence that the unfold protein response is compromised with aging [ 59 ]. Therefore, there is a potential risk of using HDAC inhibitors as therapeutics for elderly subjects with neurodegenerative disorders [ 60 ].

Background Accumulation of filamentous α-synuclein as Lewy bodies is a hallmark of Parkinson's disease. To identify the mechanisms involved in α-synuclein assembly and determine whether the assemblies are cytotoxic, we developed a cell model (3D5) that inducibly expresses wild-type human α-synuclein and forms inclusions that reproduce many morphological and biochemical characteristics of Lewy bodies. In the present study, we evaluated the effects of several histone deacetylase inhibitors on α-synuclein aggregation in 3D5 cells and primary neuronal cultures. These drugs have been demonstrated to protect cells transiently overexpressing α-synuclein from its toxicity. Results Contrary to transient transfectants, the drug treatment did not benefit 3D5 cells and primary cultures. The treated were less viable and contained more α-synuclein oligomers, active caspases 3 and 9, as well as ER stress markers than non-treated counterparts. The drug-treated, induced-3D5 cells, or primary cultures from transgenic mice overexpressing (<2 fold) α-synuclein, displayed more α-synuclein oligomers and ER stress markers than non-induced or non-transgenic counterparts. Similar effects were demonstrated in cultures treated with tunicamycin, an ER stressor. These effects were blocked by co-treatment with salubrinal, an ER stress inhibitor. In comparison, co-treatment with a pan caspase inhibitor protected cells from demise but did not reduce α-synuclein oligomer accumulation. Conclusions Our results indicate that an increase of wild-type α-synuclein can elicit ER stress response and sensitize cells to further insults. Most importantly, an increase of ER stress response can promote the aggregation of wild type α-synuclein.

List of Abbreviations α-Syn: alpha-synuclein; CHX: cycloheximide; CI: caspase inhibitor; ER: endoplasmic reticulum; HDAC: histone deacetylase; NT: non-transgenic; RA: retinoic acid; Sal: salubrinal; SKI: sarkosyl-insoluble; SA: sodium acetate; SB: sodium butyrate; SN1: buffer-soluble, SN2: salt plus sarkosyl-extractable; α-Syn+: cultures with induced α-Syn expression; α-Syn-: cultures without induced α-Syn expression; TG: transgenic; TetOff: tetracycline-off, TL: total lysates; TM: tunicamycin; VPA: valproic acid; Competing interests The authors declare that they have no competing interests. Authors' contributions PJ designed the study, executed most of the experiments and analyzed the data. GM executed some of Western blots and involved in data analysis. ABE maintained animals and involved in the preparation of primary neuronal cultures. W-LL performed ultrastructural analysis. HM provided transgenic animals. SHY oversaw the study, analyzed the data and wrote the paper. All authors have read and approved the final manuscript. Supplementary Material

Acknowledgements We thank Mei Yue (Mayo Clinic) for assistance in animal experiments. This study was supported by the National Institute of Health (P50-NS40256), the Mayo Foundation (Yen) and the Smith Fellowship Awards from Mayo Clinic (Jiang).

CC BY

no

2022-01-12 15:21:45

Mol Neurodegener. 2010 Dec 13; 5:56

oa_package/3f/2b/PMC3016345.tar.gz

PMC3016346

21246040

Introduction The primary infection of plants caused by the entry of viruses after mechanical damage to the cell wall and plasma membrane is mostly confined to a single cell. Infection spreads to adjacent cells with the help of viral encoded specialised proteins, called movement proteins (MPs) [1] . The cell to cell movement of viruses is a complex and dynamic process which involves functional contribution from many proteins of viral and host origin [2] . A number of RNA viruses have been identified which require an MP and two or more additional proteins for effective cell-to-cell spread of their genetic material. For example, in the case of potyviruses, the coat protein (CP), an RNA helicase (CI), and a helper component-proteinase (HC-pro) are essential for the movement function [3] , [4] , [5] , [6] . In certain other viruses like hordeivirus and potexviruses, three evolutionarily conserved triple gene block (TGB) encoded TGBp1, TGBp2, and TGBp3, proteins are involved in the process [7] , [8] , [9] . Another important and conserved feature of many MPs is their nucleic acid binding property, required for transport of viral genome from cell to cell [10] . However, the question of specificity of genome recognition for in vivo transport with in the host remains mostly. Not much is known about the molecular aspects of cell–to–cell movement in sobemoviruses and the ancillary proteins have not been identified. Earlier studies have shown that the protein encoded by ORF1 of Rice yellow mottle virus (RYMV) [11] , Cocksfoot mottle virus (CfMV) [12] and Southern cowpea mosaic virus (SCPMV) [13] is essential for cell to cell movement. Further, ORF1 encoded product of RYMV is implicated as a RNA silencing suppressor [14] , [15] . The genome of Sesbania mosaic virus (SeMV), a member of Sobemovirus genus is 4149 nucleotides in length and encodes four overlapping ORFs. The ORF 1 codes for the MP, ORF 2a for the poly protein 2a, and ORF 2b for the RNA dependent RNA polymerase (RdRp) that is translated by a frame shift mechanism to yield poly protein 2ab. The 3′- proximal ORF codes for the CP which is expressed via a sub genomic RNA [16] . The domain arrangement in poly protein 2a and 2ab was recently shown to be membrane anchor- protease (Pro)- viral protein genome linked (VPg)- 10 KDa protein (P10) – 8 KDa protein (P8) and membrane anchor- Pro- VPg- RdRp respectively [17] . The structure and the assembly of the virus have been described earlier [18] , [19] , [20] , [21] . The Pro, RdRp, VPg and the P8 domains have been functionally characterised and the latter two are shown to be natively unfolded [22] , [23] , [24] , [25] . In an earlier study SeMV MP was shown to interact with SeMV CP via the N terminal domain [26] . In the present study, the viral encoded ancillary proteins that may be essential for the viral movement in addition to ORF 1 encoded MP were identified. For this purpose, gene segments corresponding to all the domains encoded by SeMV were cloned in Matchmaker yeast 2 hybrid (Y2H) system and their interactions with SeMV MP were studied. Two proteins, namely VPg and P10 were shown to interact specifically with SeMV MP, in addition to CP. The domains in MP involved in these interactions were also identified. These observations, for the first time, implicate that the interaction of MP with VPg which is linked to the 5′ end of the viral genome might provide the specificity for the transport of viral RNA from cell to cell and that the interaction of MP with P10, which is an ATPase [23] might fulfil the energy requirement for the movement function.

Materials and Methods Construction of SeMV clones in matchmaker Gal 4 two-hybrid system 3 vectors The gene sequences, corresponding to the encoded domains of SeMV, namely Pro, VPg, P10, P8, RdRp, CP and MP, were amplified with high fidelity phusion polymerase (New England Biolabs, 240 County Road, Ipswich, MA 01938-2723, USA), using sense and antisense primer corresponding to the 5′ and 3′ terminal region of the genes respectively ( Table 1 ) with full-length SeMV cDNA clone as the template (AY004291). An additional T was introduced at the frame shift site in the sense primer used for the amplification of RdRp to bring the nucleotide sequence in frame. For bait and prey construction, the PCR products were cloned at EcoR1 sites of pGAD T7 (Pro, P10, P8, RdRp, VPg and CP) vector having the Gal 4 activation domain with a N terminal hemagglutinin (HA) epitope tag and pGBK T7 (MP and MP deletion mutants) vector having Gal 4 DNA binding domain with a N terminal c-Myc epitope tag. All the recombinant clones were confirmed by PCR with T7 sense and gene specific antisense primers ( Table 1 ), followed by DNA sequencing. Y2H interaction assays The yeast strain AH109, along with the respective clones in bait and pray plasmid pGAD T7 and pGBK T7 from matchmaker Gal 4 two hybrid system (Clontech Laboratories, Inc. 1290 Terra Bella Ave. Mountain View, CA, 94043, U.S.A), providing HIS3, ADE2, MEL 1 and lac Z reporter constructs and allowing high stringency Y2H selection were transformed and colonies were grown according to the Yeast Protocols Handbook ((PT3024-1, Clontech, Heidelberg, 2001). The cells were grown in minimal medium (0.67% yeast nitrogen base, 2% glucose) with appropriate amino acid omission (-Leu and –Trp for yeast transformed with both bait and pray plasmid). Replica plating was performed under conditions of increasing stringency according to the manufacturer's suggestions and subsequently analysed for growth on –Leu/-Trp/-His, -Leu/-Trp/-His/-Ade synthetic drop out (SD) media. The replica plates were analysed with 5-Bromo-4-Chloro-3-indolyl α-D-galactopyranose (α-XGal) to monitor MEL1 expression. Images were captured after 4–6 days of growth at 30°C. To determine the strength of protein-protein interactions, β-galactosidase solution assay was performed using ortho-nitrophenyl-β- galactopyranoside as a substrate (ONPG; Sigma-Aldrich, 3050 Spruce St., St. Louis, MO 63103, USA). The transformed cells grown in 10 ml of -Leu –Trp –His medium or - Leu –Trp galactose medium (for transformants which failed to grow in the former medium) for 3 to 4 hrs at 30°C were lysed and assayed for β-galactosidase activity. The ß-galactosidase (ß-Gal) activity was quantitated using the following formula: 1,000×[OD 420 −(1.75)]/( T × V ×OD 600 ), where optical density at 420 nm (OD 420 ) is due to the product formed, OD 600 is the cell density of the culture, T is the reaction time in minutes, and V is the volume in ml. The assays were performed in triplicate. Proteins were also isolated from the transformed colonies to quantitate the expression of interacting proteins by ELISA. The proteins (Pro, P10, P8, RdRp, VPg and CP) expressed from pGAD T7 vector were quantified using HA polyclonal antibody as primary antibody, where as MP and the deletion mutants of MP expressed from pGBKT7 were quantified using cMyc monoclonal antibodies. For this purpose, the lysate of cells from 5ml liquid culture grown in appropriate SD medium (-Leu/-Trp/-His) was used for coating the wells. Construction and expression of GST-MP clones in E.coli SeMV MP gene was amplified from the full length SeMV cDNA clone as described in the previous section. In order to express MP with N terminal- GST-tag, the PCR product was also cloned at EcoRI site of pGEX 4T1 vector (GE Healthcare Bio-Sciences AB,SE-751 84 Uppsala, Sweden). The recombinant clones were confirmed by PCR with T 7 sense and MP antisense primers ( Table 1 ), followed by DNA sequencing. MP deletion mutant genes were amplified with high fidelity phusion polymerase using sense and antisense primers corresponding to the N and C termini of each of the mutant ( Table 1 ) and with pGEX-MP clone as the template. The PCR products were cloned at EcoRI site of pGEX 4T1 vector by site directed ligation. All the clones were confirmed by PCR with T7 sense and MP mutant specific antisense primers ( Table 1 ), followed by DNA sequencing. pGEX 4T1-MP clone was transformed into E.coli BL21(DE3) pLysS cells (Novagen, Merck KGaA, Darmstadt, Germany). A single colony was inoculated into 20 ml of Luria-Bertani (LB) medium containing 50 μg/ml ampicillin and allowed to grow overnight at 37°C. The overnight culture was inoculated into 500 ml of Terrific Broth (TB) containing 50μg/ml ampicillin and allowed to grow at 37°C till the optical density (OD) at 600 nm reached 0.6. The expression of MP was induced with 0.3 mM isopropyl-ß-D- thiogalactopyranoside (IPTG) (Sigma) and grown for 10 h at 15°C. The cells were harvested by centrifugation and resuspended in 30 ml of lysis buffer. The resuspended cells were sonicated for 15 min at an amplitude of 30 (Vibra cell, Sonics & Materials, Inc. 53 Church Hill Road,Newtown, CT 06470-1614 USA) and the lysate was centrifuged at 10,000 g (Avanti JE, Beckmen coulter) for 10 min at 4°C. The solubility of the expressed protein was checked using sodium dodecyl sulphate- polyacrylamide gel electrophoresis (SDS-PAGE) [27] . The protein bands were visualized by staining with coomassie blue (Sigma). GST-MP was purified from the soluble fraction obtained after sonication of E.coli BL21(pLysS) cells expressing GST-MP from the pGEX-MP clone by affinity chromatography using GST binding resin as described in the manufacturers protocol. The purified protein was extensively dialyzed against storage buffer [50 mM Tris HCl, 100 mM NaCl, 10 mM β-marcaptoethanol, (pH 8) and 10% glycerol] to remove the reduced glutathione and stored at −20°C. The same procedure was used for purification of GST-MP deletion mutants. The VPg over expressed in E. coli was purified as described earlier [22] . Electrophoretic Mobility Shift Assay (EMSA) SeMV was purified from infected Sesbania grandiflora leaves 21 days after inoculation and genomic RNA was isolated as described earlier [16] . Fixed concentration of nucleic acid (1 μg) (genomic RNA from SeMV, genomic RNA from Physalis mottle virus ( PhMV)) was purified as described earlier [28] , single stranded DNA from SV40 (Bangalore Genei, India), Double stranded DNA (PCR product of SeMV CP gene) were incubated with increasing concentrations of GST-MP or its mutants in incubation buffer [200 mM 3-(N-morpholino)propanesulfonic acid (MOPS), 20 mM sodium acetate pH 7] at 4°C for 30 min. After incubation, the samples were loaded on to 0.5% agarose gel in MOPS electrophoresis buffer [200 mM MOPS, 20 mM sodium acetate pH 7] with 1 μl RNA loading dye (0.25% bromophenol blue, 0.25% xylenecyanol, 30% glycerol in DEPC treated water) and 1 μl ethidium bromide (1 mg/ml) and run for 4 hours at 5 V/cm. After electrophoresis, image of the gel was captured under UV with a gel doc (Gel Doc 1000, Bio-Rad Laboratories, 1000, Alfred Nobel Drive, Hercules, CA 94547). ELISA for monitoring in vitro protein- protein interaction The interaction between GST-MP and VPg or P10 in vitro was tested by an enzyme-linked immunosorbent based assay (ELISA), as described previously with minor modifications [29] , [30] , [31] , [32] . The ELISA plate (F8 Nunc Maxisorp loose,Nunc, Kamstrupvej 90, Postbox 280, DK-4000, Roskilde, Denmark) wells were coated with 5.0 μg of purified VPg or P10 (100 μl /well) for 1h at 37°C. The protein was diluted in 1× Phosphate-buffered saline (PBS) (pH 7.4). The unadsorbed protein fraction was removed and wells were blocked with 10% skimmed milk in 1× PBS for 1 h at 37°C. The plates were then incubated with GST-MP for 2 h at 37°C. BSA was used as negative control. The wells were washed thrice; 3 min each with 1× PBS containing 0.05% Triton X 100 (PBST) and then thrice with 1× PBS. Polyclonal antibodies specific to the GST-MP (1∶5000) was added and incubated for 1 h. at room temperature. Washes were repeated as before and the wells were further incubated with goat anti rabbit secondary antibody conjugated to HRP (1∶10,000) in 5% skimmed milk in 1× PBS (100 μl /well) for 45 min followed by washing and addition of 1× substrate TMB/H 2 O 2 (diluted from 20× stock solution in distilled water) and monitored for the blue colour development. The reaction was stopped by the addition of 2 N H 2 SO 4 (50 μl /well). Interactions were quantified by reading the absorbance at 450 nm using a SpectraMax 340PC384 absorbance microplate reader (Molecular Devices, Inc., 1311. Orleans Drive, Sunnyvale, CA 94089-11361, USA,). All the experiments were done in triplicate and standard deviation was calculated. GST was used instead of GST-MP to rule out possibility of interaction between VPg and GST. Thin Layer Chromatography (TLC) and ATPase Assay To test for ATPase activity, GST MP was incubated with 0.1 μCi [γ 32 P]-ATP at 25°C in the reaction buffer (50 mM MOPS buffer, pH 7.0, 2 mM MgCl 2 , 2 mM dithiothreitol and 0.5 mg/ml of bovine serum albumin). Following incubation for 30 min, the reaction was stopped by adding EDTA (pH 8.0) to a final concentration of 10 mM. Buffer alone and calf intestine alkaline phosphatase (CIAP) were used as negative and positive controls, respectively. The Pi released was detected by TLC on polyethyleneimine-cellulose F plates (obtained from Merck, Germany). The solvent used was 1 M formic acid, 1mM EDTA (pH 8.0) and 0.5 M lithium chloride. The intensity of the spot was measured using phosphor imaging with image gauge software.

Results and Discussion Y2H Study for identification of ancillary proteins that interact with SeMV-MP The association of ancillary proteins with MPs presumably affect local and systemic spread of the virus within the host. In order to identify viral encoded ancillary proteins that could interact with SeMV MP and may assist in cell to cell movement, all the genes encoded by SeMV genome ( Figure 1A ) were cloned in matchmaker Gal 4 two hybrid system 3. The Pro, VPg, P10, P8, CP and RdRp PCR products were cloned into pGADT7 and wild type and deletion mutants of MP were cloned into pGBKT7 as described in the methods section. The recombinant pGADT7 and pGBKT7 clones were co-transformed into Saccharomyces cerevisiae AH109 strain in pairs as shown in figure 1B . pGBKT7-P53 (Murine p53 fused to GAL4 DNA-BD) and pGADT7-T Antigen (SV40 large T-antigen fused to GAL 4 DNA-AD) were used as positive control in these experiments, as p53 and T Antigen have been shown to interact in a yeast 2 hybrid assay earlier [33] . The growth was monitored on SD medium (SD) not containing Leu and Trp (–Leu –Trp) plates to ensure both the plasmids were co transformed into the yeast cells. The transformed colonies were replica plated on –Leu –Trp -His (medium stringency) plate to determine the interaction, if there is any, between the proteins. As shown in figure 1B , growth of AH109 cells was observed when p53-T Ag, CP-MP, P10-MP, VPg-MP were expressed together. On the other hand, the transformed yeast cells did not grow when P8-MP, RdRp-MP and Pro- MP were co expressed. The results confirmed earlier observations that MP interacts with CP and showed that MP might also interact with VPg and P10 but not with P8, RdRp or Pro domains. The transformed colonies which showed positive Y2H interaction were grown in liquid cultures (-Leu –Trp –His medium) and were assayed for β-galactosidase activity to validate and quantify the two-hybrid interactions. For testing the interactions in the pairs that did not grow in -Leu –Trp –His medium, the cells grown on -Leu –Trp medium were used. A single colony was picked from each of the selection medium plate and β Galactosidase assay was performed as described in the methods section. The results obtained are presented as percentage of arbitrary units of ß-galactosidase activity (the values are indicated on the top of each bar) obtained for the interaction between p53 and TAg (100%). The values represent the mean of at least three separate experiments ( Figure 2A ). The measurement of β-galactosidase activity in these transformants confirmed that MP interacts with CP, VPg and P10 but not with P8, RdRp or Pro. Activity measurements showed that the interaction between MP and VPg or P10 was nearly as strong as that of p53 and TAg (95.5 and 97.1% respectively); where as interaction of MP with CP was weaker with only 70% of the positive control. To ascertain whether the lack of interaction between MP and P8, RdRp or Pro domains was not due to the lack of expression of the proteins in the yeast cells, the expression of all the proteins, both from pGAD T7 and pGBK T7 clones was monitored using HA polyclonal and cMyc monoclonal antibodies respectively using the lysed transformed AH109 cells grown in -Leu -Trp -His medium. As can be seen from figure 2B , the expression level of all the HA epitope tagged SeMV proteins was comparable to the expression of cMyc tagged MP. MP-VPg Interaction VPg is covalently linked to the 5′ end of some viral genomes. VPg acts as a primer for replication by interacting with viral RdRp and also regulates translation by interacting with initiation factors [34] , [35] . However, the direct role of VPg in viral movement has not been clearly demonstrated thus far. Recently, SeMV, Potato virus Y (PVY, genus Potyvirus ) and Potato virus A (PVA, genus Potyvirus ) VPgs were reported to be “natively unfolded proteins” [22] , [36] , [37] . Such proteins possess a number of crucial biological functions including molecular recognition and regulation [38] , [39] , [40] . The functional diversity provided by disordered regions is believed to complement functions of ordered protein regions by protein-protein interactions [41] . Therefore we wanted to probe this unique MP-VPg interaction further and determine its possible functional significance. pGBK T7 MP and pGAD T7 VPg were retransformed into AH109 strain, and plated onto SD –Leu –Trp plates and incubated for 96 hrs. The transformants were marked and transferred to higher stringency plates as described earlier. As shown in Figure 3A , the transformants could grow at highest stringency level (–Leu –Trp –His –Ade +α X Gal plates) suggesting the activation of all the reporter genes to the same extent as was observed with p53 and TAg interaction. It was also ascertained that there was no auto activation of the reporter genes by the presence of proteins alone, as VPg or MP expressed independently along with the pGBK T7 or pGAD T7 vector alone respectively in AH109 cells showed no growth on the selection medium plates. These results clearly establish that MP interacts with VPg. Next, purified GST-MP and VPg were used for checking their interaction in vitro by ELISA, where, VPg (5μg) was immobilised on to the ELISA plates, and blocked with 10% skimmed milk in PBS followed by addition of GST-MP (5μg), which was used as the probe protein. The interaction between the two proteins was monitored by anti MP polyclonal antibodies. As shown in Figure 3B , a significant absorbance at 450 nm due to the interaction of MP with VPg was observed. In the experiment BSA and GST were used as negative controls. Further, cross reaction of the primary antibody to the immobilized protein, i.e. MP polyclonal antibody to VPg or the milk proteins used for blocking were also tested by ELISA in the absence of MP. From the results it can be concluded that MP interacts with VPg in vitro also. Nucleic acid binding studies of SeMV MP Many of the well characterised MPs bind to nucleic acids. Therefore, the nucleic acid binding property of purified GST-MP was investigated. Increasing concentration GST-MP was incubated with a fixed concentration of viral RNA purified from SeMV and the complex was analysed on 0.5% agarose gel and photographed under UV ( Figure 4A ). The mobility shift observed with increasing concentration of GST-MP confirms the ability of the protein to bind to genomic RNA. To rule out the possibility of GST- RNA interaction, EMSA was performed with genomic RNA and similarly expressed and purified GST ( Figure 4B ). No interaction was observed confirming that the mobility shift observed in the gels ( Figure 4A ) was a result of MP-RNA interaction. Interestingly, no interaction was observed when GST-MP was incubated with other nucleic acids such as PhMV genomic RNA ( Figure 4C ), M13 ssDNA ( Figure 4D ), and dsDNA in the form of SeMV CP gene PCR product ( Figure 4E ). These results demonstrate that SeMV MP can only recognize its own genomic RNA. It may be noted that only SeMV genomic RNA has covalently linked VPg. On the other hand, studies with MPs from different genera have shown that, they have diverse nucleic acid binding properties. For example, Red clover necrotic mosaic virus MP binds to both ssRNA and ssDNA, [42] , Cocksfoot mottle virus P1 protein can interact with ssRNA transcripts in a sequence-non-specific manner [43] , while Bean dwarf mosaic virus transports ssDNA and not ssRNA or dsDNA [44] . This unique feature of SeMV-MP with specificity towards its own genome was investigated further. Does MP recognize genomic RNA through interaction with VPg? The results presented thus far clearly demonstrated that SeMV MP interacts with VPg and it can also bind to genomic RNA specifically. Two possible explanations for the specific binding of SeMV MP to the genomic RNA are- i) sequence specific interaction of MP with the genome or, ii) direct interaction of MP with the VPg covalently linked to the 5′end of the RNA. To test these possibilities, in vitro transcripts of the CP gene, which lacked the covalently linked VPg at the 5′ end, were made and EMSA was performed with GST-MP. It was observed that MP was unable to bind to the in vitro transcribed RNA (data not shown). Next, the genomic RNA of SeMV was treated with Pronase {Protease from Streptomyces griseus , Sigma Aldrich, [45] } to remove the VPg and the RNA was purified by trizol method as described earlier [28] . When EMSA of the Pronase treated RNA was performed with GST-MP as described previously, it was observed that GST MP was unable to recognize genomic RNA without VPg ( Figure 5 ). To ascertain the presence of VPg covalently linked to the genomic RNA, ELISA was performed with 0.5 μg of genomic RNA samples with polyclonal antibodies to VPg or CP as primary antibody (data not shown). The well in which genomic RNA was coated gave signal at 450 nm when VPg antibodies were used and not when CP antibodies were used, thus ruling out the possibility of MP-RNA interaction via CP that might have co-purified with the genomic RNA. These results, demonstrate that SeMV MP interacts with genomic RNA via the covalently linked VPg. Such an interaction could facilitate specific viral RNA movement from one cell to another. In the case of potyviruses, which also have covalently linked VPg at the 5′ end of their genome, it has been demonstrated that VPg interacts with host encoded VPg interacting proteins (PVIP) and assists in cell to cell movement [46] . Generation of SeMV MP deletion mutants for the identification of domains necessary for interaction with the ancillary proteins Further experiments were carried out to delineate the regions in SeMV MP which may be involved in the interaction with the ancillary proteins. MP is primarily a α helical protein as predicted from the amino acid sequence. The predicted N-terminal three helices were deleted one at a time to obtain NΔ16, NΔ35 and NΔ49 mutant MPs. From the C terminus three serine residues were deleted to generate CΔ3 mutant. The stretch of three serine residues was predicted to have a high propensity for phosphorylation. Previously it was reported in literatures that some of the MPs are regulated by phosphorylation / dephosphorylation as has been shown for Tobacco mosaic virus (TMV), Tomato mosaic virus (ToMV) and Potato leafroll virus (PLRV), [47] . CΔ19 deletion mutant was created to remove a predicted nucleic acid binding domain. Deletion of 38 amino acids from the C-terminus resulted in the deletion of three of the cysteine residues and the nucleic acid binding domain. The deletion mutants were constructed by PCR amplification with appropriate sense and anti sense primers ( table 1 ). The PCR products were cloned into pGBK T7 vector at the EcoRI site to obtain bait constructs with Gal 4 DNA binding domain as described earlier for full length MP. The same mutants were also cloned into pGEX 4T1 vector at the EcoRI site as described earlier. These mutants were over expressed in E.coli BL21 pLys S cells. All the GST-MP deletion mutants were soluble and were of expected sizes. The proteins were purified using GSH affinity chromatography and used for in vitro experiments. Identification of domains of MP necessary for interaction with VPg MP mutants in pGBK T7 were transformed with pGAD T7 VPg into AH109 cells to map the domains in MP which could be important for interaction with VPg. Transformation was carried out as described earlier. As is evident from figure 6A , the interaction of MP with VPg was abolished at the highest stringency upon deletion of the N terminal 49 and C terminal 19 amino acids of MP. No colonies appeared in –Leu –Trp –His -Ade+X-Gal plates when these two pGBKT7 NΔ49 and CΔ 19 MP mutants were transformed with pGAD T7 VPg separately. Interaction between all the other mutants of MP with VPg produced blue colonies. Interestingly, even CΔ38 produced blue colonies. β–galactosidase activity of these transformants also corroborated with these results ( Figure 6B ). As evident, VPg- MP N Δ49 and VPg MP CΔ19 showed 27.7% and 18.5% activity when compared with that obtained with p53-TAg interaction. In conformity with the results presented in figure 6A , VPg- MP CΔ38 interaction resulted in 83.1% ß- galactosidase activity. It is possible that the deletion of C terminal 19 amino acids affects proper folding of MP, thereby preventing interaction with VPg. However deletion of C terminal 38 amino acids may restore the core structure of MP which might favour interaction with VPg. ELISA of the transformed AH109 cells showed that the level of proteins expressed from both the vectors remained nearly the same, confirming that the reduced level of activity was due to loss of interaction between MP NΔ49, MP CΔ19 with VPg. As explained earlier, the loss of interaction with MP CΔ19 could be due to improper folding of this mutant protein. These results demonstrate that the interaction of SeMV-MP with VPg is primarily through its N terminal 49 amino acids residues. Interaction of SeMV MP deletion mutants with genomic RNA Since the MP recognizes genomic RNA via interaction with VPg, the mutants which are unable to bind or recognize VPg may not be able to form RNP complex with the genomic RNA. To test this hypothesis, EMSA was performed with GST-MP and its deletion mutants with genomic RNA ( Figure 7A ) and the fractional binding was calculated by measuring the intensity of the band corresponding to genomic RNA and RNA protein complex and using the formula, Fractional binding = Protein bound RNA/ Total RNA (Bound+Free) as shown in Figure 7B . As can be observed MP NΔ49, MP CΔ 19 and MP CΔ 38 were unable to form complex with the genomic RNA. The lack of binding of viral RNA could be easily explained for the former two mutants as these mutants failed to interact with VPg in the Y2H study ( Figure 6 ). However, although MP CΔ 38 interacted with VPg ( Figure 6 ) it did not show the mobility shift with genomic RNA. It may be noted that deletion of C terminal 19 and 38 amino acids of MP results in the loss of predicted RNA binding domain. These results suggest that the initial interaction of MP with genomic RNA is mediated by interaction with VPg. This interaction might trigger conformational changes in MP that allows the RNA binding domain of MP to interact with RNA. MP- P10 interaction As shown earlier, P10 was another protein that interacted with SeMV MP in Y2H experiments. To further characterise the interaction in detail, pGBK T7 MP and pGAD T7 P10 were retransformed in AH109 strain and plated onto SD –Leu –Trp plates and incubated for 96 hrs. The transformed colonies were replica plated again on to SD –Leu-Trp plates to confirm the transformation. The colonies were marked and selected on higher stringency plates as described earlier and are shown in Figure 8A . MP and P10 were found to interact upto the highest level of stringency i.e., –Leu –Trp –His –Ade +X Gal as evident from the appearance of blue colonies on the plate similar to MP –VPg interaction, signifying a strong interaction between the two proteins. The interaction was specific, as pGBKT7 MP or pGADT7 P10 transformed individually along with pGADT7 or pGBKT7 respectively did not result in the appearance of colonies under any nutritional selection ( Figure 8A last two rows). Interaction of MP with P10 in vitro was also tested by ELISA. For this purpose, GST-MP and P10 over expressed in E.coli and purified were used in this experiment as described in the methods section. As apparent from figure 8B , plate bound P10 interacted with GST- MP specifically. Identification of domains in MP necessary for interaction with P10 MP mutants in pGBK T7 and pGAD T7 P10 were transformed into AH109 strain to map the regions of interaction between MP and P10. As evident from the figure 8C , the interaction between P10 and MP was reduced drastically by the deletion of first 49 amino acids from the N terminus of MP as only white colonies were produced in –Leu –Trp –His -Ade +X Gal plates, while interaction between all the other mutants of MP with P10 resulted in blue colonies. β –galactosidase assay also confirmed the same result ( Figure 8D ). As evident MP NΔ49- P10 Y2H interaction resulted in only 14% of the ß- galactosidase activity as compared to that observed with p53- T Ag interaction or the other MP deletion mutants with P10. This drastic reduction in the activity in MPNΔ49- P10 expressing cells was not due to the lack of expression of the proteins, since, the level of expression of the two interacting proteins in all the sets remained nearly the same (data not shown). These results show that MP and P10 interact with each other predominantly via the N terminal domain of MP. However, as shown in figure 8C all the transformants could grow on –Lue -Trp-His -Ade it is possible that P10 and MP might have other sites of interaction within the region N 49 to CΔ 38 of MP. It may be noted that all the three ancillary proteins CP (data not shown), VPg and P10 interact with MP via the N terminal domain primarily. It would be interesting to investigate whether or not these interactions are mutually exclusive. The specific interaction of MP with P10 could be of physiological relevance. The P10 domain of SeMV poly protein 2a was recently shown to exhibit ATPase function [23] . The cell-to-cell movement of viruses is an active process and some MPs or their ancillary proteins possess ATPase function. For example, HSP70h from Closterovirus which is involved in virus translocation from cell to cell, has a well characterized ATPase function [48] . Similarly, the TGBpl proteins encoded by distantly related viruses contain conserved sequence motifs of some ATPases and helicases [49] , [50] , [51] . The PVX 25K MP and its counterparts in Figwort mosaic virus and Barley stripe mosaic virus have been shown to exhibit ATPase activity [52] . To examine if SeMV MP also exhibited ATPase activity, ATPase assay was carried out at 25°C for 30 min in standard reaction mixtures as described in the methods section. GST- MP was unable to hydrolyse ATP where as calf intestine alkaline phosphatase (CIAP) used as positive control could hydrolyse ATP (data not shown). It is possible that P10 having the ATPase function assists the active movement process of M complex by interaction with MP. The results presented in this communication clearly demonstrate that SeMV MP interacts with two viral encoded proteins P10 and VPg in addition to CP. The recognition of cognate RNA by MP probably occurs via protein-protein interaction between MP and VPg. Interaction of SeMV MP with P10 might be crucial for the active transport of viral RNA complex, the energy for which could come from the hydrolysis of ATP by P10.

Results and Discussion Y2H Study for identification of ancillary proteins that interact with SeMV-MP The association of ancillary proteins with MPs presumably affect local and systemic spread of the virus within the host. In order to identify viral encoded ancillary proteins that could interact with SeMV MP and may assist in cell to cell movement, all the genes encoded by SeMV genome ( Figure 1A ) were cloned in matchmaker Gal 4 two hybrid system 3. The Pro, VPg, P10, P8, CP and RdRp PCR products were cloned into pGADT7 and wild type and deletion mutants of MP were cloned into pGBKT7 as described in the methods section. The recombinant pGADT7 and pGBKT7 clones were co-transformed into Saccharomyces cerevisiae AH109 strain in pairs as shown in figure 1B . pGBKT7-P53 (Murine p53 fused to GAL4 DNA-BD) and pGADT7-T Antigen (SV40 large T-antigen fused to GAL 4 DNA-AD) were used as positive control in these experiments, as p53 and T Antigen have been shown to interact in a yeast 2 hybrid assay earlier [33] . The growth was monitored on SD medium (SD) not containing Leu and Trp (–Leu –Trp) plates to ensure both the plasmids were co transformed into the yeast cells. The transformed colonies were replica plated on –Leu –Trp -His (medium stringency) plate to determine the interaction, if there is any, between the proteins. As shown in figure 1B , growth of AH109 cells was observed when p53-T Ag, CP-MP, P10-MP, VPg-MP were expressed together. On the other hand, the transformed yeast cells did not grow when P8-MP, RdRp-MP and Pro- MP were co expressed. The results confirmed earlier observations that MP interacts with CP and showed that MP might also interact with VPg and P10 but not with P8, RdRp or Pro domains. The transformed colonies which showed positive Y2H interaction were grown in liquid cultures (-Leu –Trp –His medium) and were assayed for β-galactosidase activity to validate and quantify the two-hybrid interactions. For testing the interactions in the pairs that did not grow in -Leu –Trp –His medium, the cells grown on -Leu –Trp medium were used. A single colony was picked from each of the selection medium plate and β Galactosidase assay was performed as described in the methods section. The results obtained are presented as percentage of arbitrary units of ß-galactosidase activity (the values are indicated on the top of each bar) obtained for the interaction between p53 and TAg (100%). The values represent the mean of at least three separate experiments ( Figure 2A ). The measurement of β-galactosidase activity in these transformants confirmed that MP interacts with CP, VPg and P10 but not with P8, RdRp or Pro. Activity measurements showed that the interaction between MP and VPg or P10 was nearly as strong as that of p53 and TAg (95.5 and 97.1% respectively); where as interaction of MP with CP was weaker with only 70% of the positive control. To ascertain whether the lack of interaction between MP and P8, RdRp or Pro domains was not due to the lack of expression of the proteins in the yeast cells, the expression of all the proteins, both from pGAD T7 and pGBK T7 clones was monitored using HA polyclonal and cMyc monoclonal antibodies respectively using the lysed transformed AH109 cells grown in -Leu -Trp -His medium. As can be seen from figure 2B , the expression level of all the HA epitope tagged SeMV proteins was comparable to the expression of cMyc tagged MP. MP-VPg Interaction VPg is covalently linked to the 5′ end of some viral genomes. VPg acts as a primer for replication by interacting with viral RdRp and also regulates translation by interacting with initiation factors [34] , [35] . However, the direct role of VPg in viral movement has not been clearly demonstrated thus far. Recently, SeMV, Potato virus Y (PVY, genus Potyvirus ) and Potato virus A (PVA, genus Potyvirus ) VPgs were reported to be “natively unfolded proteins” [22] , [36] , [37] . Such proteins possess a number of crucial biological functions including molecular recognition and regulation [38] , [39] , [40] . The functional diversity provided by disordered regions is believed to complement functions of ordered protein regions by protein-protein interactions [41] . Therefore we wanted to probe this unique MP-VPg interaction further and determine its possible functional significance. pGBK T7 MP and pGAD T7 VPg were retransformed into AH109 strain, and plated onto SD –Leu –Trp plates and incubated for 96 hrs. The transformants were marked and transferred to higher stringency plates as described earlier. As shown in Figure 3A , the transformants could grow at highest stringency level (–Leu –Trp –His –Ade +α X Gal plates) suggesting the activation of all the reporter genes to the same extent as was observed with p53 and TAg interaction. It was also ascertained that there was no auto activation of the reporter genes by the presence of proteins alone, as VPg or MP expressed independently along with the pGBK T7 or pGAD T7 vector alone respectively in AH109 cells showed no growth on the selection medium plates. These results clearly establish that MP interacts with VPg. Next, purified GST-MP and VPg were used for checking their interaction in vitro by ELISA, where, VPg (5μg) was immobilised on to the ELISA plates, and blocked with 10% skimmed milk in PBS followed by addition of GST-MP (5μg), which was used as the probe protein. The interaction between the two proteins was monitored by anti MP polyclonal antibodies. As shown in Figure 3B , a significant absorbance at 450 nm due to the interaction of MP with VPg was observed. In the experiment BSA and GST were used as negative controls. Further, cross reaction of the primary antibody to the immobilized protein, i.e. MP polyclonal antibody to VPg or the milk proteins used for blocking were also tested by ELISA in the absence of MP. From the results it can be concluded that MP interacts with VPg in vitro also. Nucleic acid binding studies of SeMV MP Many of the well characterised MPs bind to nucleic acids. Therefore, the nucleic acid binding property of purified GST-MP was investigated. Increasing concentration GST-MP was incubated with a fixed concentration of viral RNA purified from SeMV and the complex was analysed on 0.5% agarose gel and photographed under UV ( Figure 4A ). The mobility shift observed with increasing concentration of GST-MP confirms the ability of the protein to bind to genomic RNA. To rule out the possibility of GST- RNA interaction, EMSA was performed with genomic RNA and similarly expressed and purified GST ( Figure 4B ). No interaction was observed confirming that the mobility shift observed in the gels ( Figure 4A ) was a result of MP-RNA interaction. Interestingly, no interaction was observed when GST-MP was incubated with other nucleic acids such as PhMV genomic RNA ( Figure 4C ), M13 ssDNA ( Figure 4D ), and dsDNA in the form of SeMV CP gene PCR product ( Figure 4E ). These results demonstrate that SeMV MP can only recognize its own genomic RNA. It may be noted that only SeMV genomic RNA has covalently linked VPg. On the other hand, studies with MPs from different genera have shown that, they have diverse nucleic acid binding properties. For example, Red clover necrotic mosaic virus MP binds to both ssRNA and ssDNA, [42] , Cocksfoot mottle virus P1 protein can interact with ssRNA transcripts in a sequence-non-specific manner [43] , while Bean dwarf mosaic virus transports ssDNA and not ssRNA or dsDNA [44] . This unique feature of SeMV-MP with specificity towards its own genome was investigated further. Does MP recognize genomic RNA through interaction with VPg? The results presented thus far clearly demonstrated that SeMV MP interacts with VPg and it can also bind to genomic RNA specifically. Two possible explanations for the specific binding of SeMV MP to the genomic RNA are- i) sequence specific interaction of MP with the genome or, ii) direct interaction of MP with the VPg covalently linked to the 5′end of the RNA. To test these possibilities, in vitro transcripts of the CP gene, which lacked the covalently linked VPg at the 5′ end, were made and EMSA was performed with GST-MP. It was observed that MP was unable to bind to the in vitro transcribed RNA (data not shown). Next, the genomic RNA of SeMV was treated with Pronase {Protease from Streptomyces griseus , Sigma Aldrich, [45] } to remove the VPg and the RNA was purified by trizol method as described earlier [28] . When EMSA of the Pronase treated RNA was performed with GST-MP as described previously, it was observed that GST MP was unable to recognize genomic RNA without VPg ( Figure 5 ). To ascertain the presence of VPg covalently linked to the genomic RNA, ELISA was performed with 0.5 μg of genomic RNA samples with polyclonal antibodies to VPg or CP as primary antibody (data not shown). The well in which genomic RNA was coated gave signal at 450 nm when VPg antibodies were used and not when CP antibodies were used, thus ruling out the possibility of MP-RNA interaction via CP that might have co-purified with the genomic RNA. These results, demonstrate that SeMV MP interacts with genomic RNA via the covalently linked VPg. Such an interaction could facilitate specific viral RNA movement from one cell to another. In the case of potyviruses, which also have covalently linked VPg at the 5′ end of their genome, it has been demonstrated that VPg interacts with host encoded VPg interacting proteins (PVIP) and assists in cell to cell movement [46] . Generation of SeMV MP deletion mutants for the identification of domains necessary for interaction with the ancillary proteins Further experiments were carried out to delineate the regions in SeMV MP which may be involved in the interaction with the ancillary proteins. MP is primarily a α helical protein as predicted from the amino acid sequence. The predicted N-terminal three helices were deleted one at a time to obtain NΔ16, NΔ35 and NΔ49 mutant MPs. From the C terminus three serine residues were deleted to generate CΔ3 mutant. The stretch of three serine residues was predicted to have a high propensity for phosphorylation. Previously it was reported in literatures that some of the MPs are regulated by phosphorylation / dephosphorylation as has been shown for Tobacco mosaic virus (TMV), Tomato mosaic virus (ToMV) and Potato leafroll virus (PLRV), [47] . CΔ19 deletion mutant was created to remove a predicted nucleic acid binding domain. Deletion of 38 amino acids from the C-terminus resulted in the deletion of three of the cysteine residues and the nucleic acid binding domain. The deletion mutants were constructed by PCR amplification with appropriate sense and anti sense primers ( table 1 ). The PCR products were cloned into pGBK T7 vector at the EcoRI site to obtain bait constructs with Gal 4 DNA binding domain as described earlier for full length MP. The same mutants were also cloned into pGEX 4T1 vector at the EcoRI site as described earlier. These mutants were over expressed in E.coli BL21 pLys S cells. All the GST-MP deletion mutants were soluble and were of expected sizes. The proteins were purified using GSH affinity chromatography and used for in vitro experiments. Identification of domains of MP necessary for interaction with VPg MP mutants in pGBK T7 were transformed with pGAD T7 VPg into AH109 cells to map the domains in MP which could be important for interaction with VPg. Transformation was carried out as described earlier. As is evident from figure 6A , the interaction of MP with VPg was abolished at the highest stringency upon deletion of the N terminal 49 and C terminal 19 amino acids of MP. No colonies appeared in –Leu –Trp –His -Ade+X-Gal plates when these two pGBKT7 NΔ49 and CΔ 19 MP mutants were transformed with pGAD T7 VPg separately. Interaction between all the other mutants of MP with VPg produced blue colonies. Interestingly, even CΔ38 produced blue colonies. β–galactosidase activity of these transformants also corroborated with these results ( Figure 6B ). As evident, VPg- MP N Δ49 and VPg MP CΔ19 showed 27.7% and 18.5% activity when compared with that obtained with p53-TAg interaction. In conformity with the results presented in figure 6A , VPg- MP CΔ38 interaction resulted in 83.1% ß- galactosidase activity. It is possible that the deletion of C terminal 19 amino acids affects proper folding of MP, thereby preventing interaction with VPg. However deletion of C terminal 38 amino acids may restore the core structure of MP which might favour interaction with VPg. ELISA of the transformed AH109 cells showed that the level of proteins expressed from both the vectors remained nearly the same, confirming that the reduced level of activity was due to loss of interaction between MP NΔ49, MP CΔ19 with VPg. As explained earlier, the loss of interaction with MP CΔ19 could be due to improper folding of this mutant protein. These results demonstrate that the interaction of SeMV-MP with VPg is primarily through its N terminal 49 amino acids residues. Interaction of SeMV MP deletion mutants with genomic RNA Since the MP recognizes genomic RNA via interaction with VPg, the mutants which are unable to bind or recognize VPg may not be able to form RNP complex with the genomic RNA. To test this hypothesis, EMSA was performed with GST-MP and its deletion mutants with genomic RNA ( Figure 7A ) and the fractional binding was calculated by measuring the intensity of the band corresponding to genomic RNA and RNA protein complex and using the formula, Fractional binding = Protein bound RNA/ Total RNA (Bound+Free) as shown in Figure 7B . As can be observed MP NΔ49, MP CΔ 19 and MP CΔ 38 were unable to form complex with the genomic RNA. The lack of binding of viral RNA could be easily explained for the former two mutants as these mutants failed to interact with VPg in the Y2H study ( Figure 6 ). However, although MP CΔ 38 interacted with VPg ( Figure 6 ) it did not show the mobility shift with genomic RNA. It may be noted that deletion of C terminal 19 and 38 amino acids of MP results in the loss of predicted RNA binding domain. These results suggest that the initial interaction of MP with genomic RNA is mediated by interaction with VPg. This interaction might trigger conformational changes in MP that allows the RNA binding domain of MP to interact with RNA. MP- P10 interaction As shown earlier, P10 was another protein that interacted with SeMV MP in Y2H experiments. To further characterise the interaction in detail, pGBK T7 MP and pGAD T7 P10 were retransformed in AH109 strain and plated onto SD –Leu –Trp plates and incubated for 96 hrs. The transformed colonies were replica plated again on to SD –Leu-Trp plates to confirm the transformation. The colonies were marked and selected on higher stringency plates as described earlier and are shown in Figure 8A . MP and P10 were found to interact upto the highest level of stringency i.e., –Leu –Trp –His –Ade +X Gal as evident from the appearance of blue colonies on the plate similar to MP –VPg interaction, signifying a strong interaction between the two proteins. The interaction was specific, as pGBKT7 MP or pGADT7 P10 transformed individually along with pGADT7 or pGBKT7 respectively did not result in the appearance of colonies under any nutritional selection ( Figure 8A last two rows). Interaction of MP with P10 in vitro was also tested by ELISA. For this purpose, GST-MP and P10 over expressed in E.coli and purified were used in this experiment as described in the methods section. As apparent from figure 8B , plate bound P10 interacted with GST- MP specifically. Identification of domains in MP necessary for interaction with P10 MP mutants in pGBK T7 and pGAD T7 P10 were transformed into AH109 strain to map the regions of interaction between MP and P10. As evident from the figure 8C , the interaction between P10 and MP was reduced drastically by the deletion of first 49 amino acids from the N terminus of MP as only white colonies were produced in –Leu –Trp –His -Ade +X Gal plates, while interaction between all the other mutants of MP with P10 resulted in blue colonies. β –galactosidase assay also confirmed the same result ( Figure 8D ). As evident MP NΔ49- P10 Y2H interaction resulted in only 14% of the ß- galactosidase activity as compared to that observed with p53- T Ag interaction or the other MP deletion mutants with P10. This drastic reduction in the activity in MPNΔ49- P10 expressing cells was not due to the lack of expression of the proteins, since, the level of expression of the two interacting proteins in all the sets remained nearly the same (data not shown). These results show that MP and P10 interact with each other predominantly via the N terminal domain of MP. However, as shown in figure 8C all the transformants could grow on –Lue -Trp-His -Ade it is possible that P10 and MP might have other sites of interaction within the region N 49 to CΔ 38 of MP. It may be noted that all the three ancillary proteins CP (data not shown), VPg and P10 interact with MP via the N terminal domain primarily. It would be interesting to investigate whether or not these interactions are mutually exclusive. The specific interaction of MP with P10 could be of physiological relevance. The P10 domain of SeMV poly protein 2a was recently shown to exhibit ATPase function [23] . The cell-to-cell movement of viruses is an active process and some MPs or their ancillary proteins possess ATPase function. For example, HSP70h from Closterovirus which is involved in virus translocation from cell to cell, has a well characterized ATPase function [48] . Similarly, the TGBpl proteins encoded by distantly related viruses contain conserved sequence motifs of some ATPases and helicases [49] , [50] , [51] . The PVX 25K MP and its counterparts in Figwort mosaic virus and Barley stripe mosaic virus have been shown to exhibit ATPase activity [52] . To examine if SeMV MP also exhibited ATPase activity, ATPase assay was carried out at 25°C for 30 min in standard reaction mixtures as described in the methods section. GST- MP was unable to hydrolyse ATP where as calf intestine alkaline phosphatase (CIAP) used as positive control could hydrolyse ATP (data not shown). It is possible that P10 having the ATPase function assists the active movement process of M complex by interaction with MP. The results presented in this communication clearly demonstrate that SeMV MP interacts with two viral encoded proteins P10 and VPg in addition to CP. The recognition of cognate RNA by MP probably occurs via protein-protein interaction between MP and VPg. Interaction of SeMV MP with P10 might be crucial for the active transport of viral RNA complex, the energy for which could come from the hydrolysis of ATP by P10.

Conceived and designed the experiments: SRC. Performed the experiments: SRC. Analyzed the data: SRC HSS. Contributed reagents/materials/analysis tools: HSS. Wrote the paper: SRC HSS. Sesbania mosaic virus (SeMV) is a single strand positive-sense RNA plant virus that belongs to the genus Sobemovirus . The mechanism of cell-to-cell movement in sobemoviruses has not been well studied. With a view to identify the viral encoded ancillary proteins of SeMV that may assist in cell-to-cell movement of the virus, all the proteins encoded by SeMV genome were cloned into yeast Matchmaker system 3 and interaction studies were performed. Two proteins namely, viral protein genome linked (VPg) and a 10-kDa protein (P10) c v gft encoded by OFR 2a, were identified as possible interacting partners in addition to the viral coat protein (CP). Further characterization of these interactions revealed that the movement protein (MP) recognizes cognate RNA through interaction with VPg, which is covalently linked to the 5′ end of the RNA. Analysis of the deletion mutants delineated the domains of MP involved in the interaction with VPg and P10. This study implicates for the first time that VPg might play an important role in specific recognition of viral genome by MP in SeMV and shed light on the possible role of P10 in the viral movement.

We thank Prof. N. Appaji Rao and Prof. M.R.N. Murthy for valuable discussion. We thank Ms. Sreelatha for technical assistance. We thank Dr. Smita Nair for providing the P10 clone and Ms. Chhavi Mathur for her help in correction of the manuscript.

CC BY

no

2022-01-13 08:14:22

PLoS One. 2011 Jan 5; 6(1):e15609

oa_package/85/ae/PMC3016346.tar.gz

PMC3016347

21143988

Materials and methods Spinal slice preparation and whole-cell recording Transverse spinal cord slices (~600 μm, L4 or L5 segment) with an attached dorsal root from adult rats (male, 6-8 weeks old) were prepared with a vibrating microslicer and perfused in the oxygen-bubbled Krebs' solution (in mM: 117 NaCl, 3.6 KCl, 2.5 CaCl 2 , 1.2 MgCl 2 , 1.2 NaH 2 PO 4 , 25 NaHCO 3 , and 11 D-glucose) for a blind ruptured patch-clamp recording as our previous study [ 33 ]. Resistance of the patch electrodes was typically 4~10 MΩ. The internal eletrode solution contained (in mM: 135 K-gluconate, 0.5 CaCl 2 , 2 MgCl 2 , 5 KCl, 5 EGTA, 5 HEPES and 5 D-glucose). Currents were filtered at 2 kHz and digitized at 5 kHz (Axopatch 200B amplifier, Molecular Devices) and were analyzed by using pCLAMP8.5 program. The membrane potential was hold at -70 mV. To evoke Aδ- and C-fiber activation, the dorsal root stimulation was delivered with a suction electrode which was linked to a constant-current stimulator (Digitimer). Monosynaptic eEPSC was studied in the presence of 20 μM bicuculline and 2 μM strychnine. Frequency and amplitude of sEPSC were analyzed with Axograph (Molecular Devices). Afferent Aδ- or C-fibers were identified by the basis of the conduction velocity (CV) of afferent fibres (Aδ: 2~12 m/s; C: <1.2 m/s) calculated from the latency of EPSC from a stimulus artifact, the length of dorsal root, and the stimulus threshold (Aδ: 10~60 μA; C: 180~620 μA). The Aδ and C responses were considered as monosynaptic in origin when the latency remained constant and there was no failure during stimulation at 20 Hz for the Aδ-fiber evoked EPSCs, and at 2 Hz for the C-fiber evoked EPSCs. Drugs were applied through a superfusion exchange of the solutions in the recording chamber. The drugs used in the present studies included strychnine, bicuculline, MPP, 17β-estradiol (Sigma, USA), BIM and H89 (Calbiochem, USA). All drugs except for MPP, 17β-estradiol and H89 (where dimethyl sulphoxide was used as a solvent) were dissolved in distilled water at 1000 times the concentration in stock and kept at -20°C. On the experimental days, they were diluted to the desired concentration within Kreb's solution. Data analysis and statistics All numerical data were presented as the mean ± S.E.M. Statistical significance was determined as P < 0.05 using the student's paired t -test, n refers to the number of neurons studied.

Background It has been shown that estrogen is synthesized in the spinal dorsal horn and plays a role in modulating pain transmission. One of the estrogen receptor (ER) subtypes, estrogen receptor alpha (ERα), is expressed in the spinal laminae I-V, including substantia gelatinosa (SG, lamina II). However, it is unclear how ERs are involved in the modulation of nociceptive transmission. Results In the present study, a selective ERα antagonist, methyl-piperidino-pyrazole (MPP), was used to test the potential functional roles of spinal ERα in the nociceptive transmission. Using the whole-cell patch-clamp technique, we examined the effects of MPP on SG neurons in the dorsal root-attached spinal cord slice prepared from adult rats. We found that MPP increased glutamatergic excitatory postsynaptic currents (EPSCs) evoked by the stimulation of either Aδ- or C-afferent fibers. Further studies showed that MPP treatment dose-dependently increased spontaneous EPSCs frequency in SG neurons, while not affecting the amplitude. In addition, the PKC was involved in the MPP-induced enhancement of synaptic transmission. Conclusions These results suggest that the selective ERα antagonist MPP pre-synaptically facilitates the excitatory synaptic transmission to SG neurons. The nociceptive transmission evoked by Aδ- and C-fiber stimulation could be potentiated by blocking ERα in the spinal neurons. Thus, the spinal estrogen may negatively regulate the nociceptive transmission through the activation of ERα.

Findings Several studies suggest that estrogen plays an important role in the spectrum of neural functions, such as nociception [ 1 - 4 ]. Estrogen is synthesized in many neurons in laminae I-III of the spinal cord [ 5 - 8 ], and potentiates the pain behavior [ 8 ]. Estrogen may modulate nociceptive responses through the increase of glutamate-induced currents, the inhibition of γ-aminobutyric acid (GABA) and glycine (Gly) receptors, or the modulation of the opioid receptors in the spinal dorsal horn [ 9 - 11 ]. It is well known that the classical estrogen action in neurons is to activate nuclear estrogen receptor α and β (ERα/β), which cause long-term genomic effects [ 12 , 13 ], or to activate cytoplasmic signaling events at or near the plasma membrane [ 14 , 15 ] through either membrane-localized classical ERs [ 16 , 17 ] or novel ERs [ 18 ]. Recent studies showed that ERα is expressed in spinal laminae I-V, especially in laminae I-II, and is most abundant in the lower lumbar (L) and sacral segments [ 19 , 20 ]. However, whether the ERα is involved in estrogen-mediating pain behavior remains unclear. Considering that the superficial dorsal horn of the spinal cord, especially substantia gelatinosa (SG, lamina II), plays an important role in the modulation of synaptic transmission of fine myelinated A (Aδ)- and unmyelinated C-afferent fibers [ 21 , 22 ], we used a selective ERα antagonist, methyl-piperidino-pyrazole (MPP) [ 23 ], to examine the function of spinal ERα in nociceptive transmission in SG neurons. The dorsal root-attached spinal cord slices were prepared from adult rats and recorded with whole-cell patch-clamp technique. Whole-cell recordings were carried out in SG neurons. Stable recordings could be maintained in vitro for more than 8 hrs; and recordings could be made from a single SG neuron up to 2 hrs. The monosynaptic, Aδ-afferent evoked excitatory postsynaptic currents (eEPSCs) with a mean amplitude of 156 ± 25 pA (50~360 pA; V H = -70 mV) were found in ~70% of recorded neurons (18/25). In 8 out of these 18 neurons (~ 45%), superfusion of MPP (10 μM) increased the peak amplitude of the Aδ-eEPSC in a reversible manner (Figure 1A ). The enhancement was averaged at 130 ± 5% (n = 8) in magnitude. The monosynaptic C-afferent eEPSCs with a mean amplitude of 135 ± 31 pA (40~310 pA; V H = -70 mV) were found in ~60% of neurons (10/16). In 5 out of these 10 neurons, MPP (10 μM) treatment increased the peak amplitude of the C-eEPSC and normal Kreb's solution washed off the MPP-induced effect (Figure 1B ). The averaged magnitude of the enhancement was 150 ± 6% (n = 5). In other three neurons exhibiting both Aδ- and C-eEPSCs, MPP increased the amplitude of both types of eEPSCs (Figure 1C ). Further comparison of MPP-induced enhancement between Aδ- and C-eEPSCs showed that the increase in C-eEPSC amplitude during MPP application was more pronounced than that of Aδ-EPSC (Figure 1D ). In spite of the differences of their sensitivity to MPP, Aδ- and C-eEPSCs were responded with a similar time course following MPP superfusion. The current amplitudes had been changed maximally and measured at 3 min after MPP was applied. To examine whether MPP modulated the afferent synaptic transmission through pre- or post-synaptic action, the spontaneous EPSC (sEPSC) in SG neurons during MPP treatment were analyzed. We found that superfusion of MPP (10 μM) resulted in a reversible enhancement in sEPSC frequency (Figure 2A and 2B ; 234 ± 9% of control at 3 min following its application, n = 8; P < 0.001). Furthermore, the MPP-induced responses were dose-dependent. At a concentration of 1 and 5 μM, the increased sEPSC frequency was 124 ± 8% (n = 5) and 180 ± 6% (n = 5), respectively. However, the amplitude of sEPSC was not altered by the treatment with MPP (Figure 2B and 2C ). The modulation on sEPSC frequency, but not on amplitude, suggests that MPP regulates nociceptive transmission through a pre-synaptic action. To investigate whether exogenous estrogen could modulate glutamatergic excitatory synaptic transmission in SG neurons, we tested whether a treatment with 17β-estradiol could regulate the sEPSC. We found that the frequency of sEPSC was reduced by bath-applied 17β-estradiol (1 μM), and this effect could be reversed by MPP (Figure 3A , n = 6). Finally, we also found that a PKC inhibitor bisindolylmaleimide I hydrochloride (BIM) (1 μM), but not a PKA inhibitor H89 (5 μM), could reduce the MPP-induced enhancement of sEPSC frequency (Figure 3B , n = 5). The above results suggest that nociceptive transmission could be facilitated by blocking ERα, such as a selective antagonist MPP used in the current study. In addition, the endogenous estrogen may activate ERα in spinal dorsal horn to reduce glutamatergic excitatory transmission and inhibit the nociceptive responses. Previous studies showed that ERα is expressed in the small-diameter neurons in the dorsal root ganglion (DRG), a subset of nociceptive sensory neurons [ 24 - 26 ]. The ERα-mediated inhibition of ATP-induced Ca 2+ signaling in mouse DRG neurons [ 27 ] suggests that peripheral ERα negatively regulates nociceptive transmission. Moreover, ERα immunoreactivity has been found in the spinal cord. A larger numbers of ERα-immunoreactive neurons were found in the lower lumbar spinal cord segments. These ERα-containing neurons were mainly found in the spinal lamina II, and some were in laminae I, III, IV, V, and X. In the superficial layers of the medullary dorsal horn, ERα-immunoreactivity was mainly located in lamina II, which was also expressed noxious-induced Fos [ 19 , 20 , 28 ]. These findings provide an anatomical and neurochemical basis for the hypothesis that estrogen activates ERα directly to regulate pain transmission at the central level [ 28 ]. Consistent with early studies, our present study shows that in the spinal dorsal horn, ERα is involved in the modulation of nociceptive Aδ- and C-afferent transmission. Previous studies showed that the enzyme aromatase catalyzed the formation of estrogen from testosterone in the gonads and other tissues, such as many nociceptive neurons in the spinal laminae I-III. Moreover, the specific nonsteroidal aromatase inhibitor, vorozole, was found to inhibit the phosphorylation of aromatase in the spinal cord and induce an acute inhibition of the endogenous spinal estrogen synthesis, which could consequently lead to the inhibition of nociceptive responses [ 6 - 8 ]. Some studies showed that estrogen rapidly potentiated glutamate (kainate)-induced currents through a second-messenger cascade [ 9 ]. G-protein coupled to inwardly rectifying K channels could be inhibited by the estrogen-induced reduction of the potency of GABA and opioid receptor agonists [ 18 , 29 ]. Moreover, estrogen inhibited the GABA and Gly receptors or modulated the opioid receptor in the spinal dorsal horn [ 10 , 11 ]. The estrogen-induced potentiation of kainate currents and inhibition of GABA and Gly receptors may play a role in the activation of the central pain pathways [ 30 - 32 ]. Our present results show that activation of ERα by estrogen may inhibit nociceptive transmission through a PKC signaling pathway. Therefore, the ERα-mediated negative regulation of nociceptive transmission may be balanced by the effect of estrogen on other receptors in some certain extent. In conclusion, we propose that estrogen may inhibit the nociceptive transmission via the ERα in the spinal dorsal horn. Our results may help to understand the functions and mechanisms of estrogen in pain modulation, and suggest that ERα may be a potential target in relieving pain syndrome. Competing interests The authors declare that they have no competing interests. Authors' contributions XZ and KCL conceived and designed the study. YQZ performed the experiments. All authors read and approved the final manuscript.

Acknowledgements This work was supported by NNSFC 30621062, MOST 2006CB806600, CASKSCX2-YW-R-31. Dr. Kai-Cheng Li was supported by grant 06R214160 and 2007KIP402.

CC BY

no

2022-01-12 15:21:45

Mol Pain. 2010 Dec 11; 6:92

oa_package/e3/5e/PMC3016347.tar.gz

PMC3016348

21162731

Background Pain has a strong emotional component and is significantly associated with anxiety and depression. The amygdala plays a key role in emotional learning and memory as well as in affective disorders [ 1 - 4 ] and is also important for the emotional-affective dimension of pain and pain modulation [ 5 - 8 ]. Pharmacologic inhibition of amygdala hyperactivity has been shown to decrease nocifensive and affective responses in animal pain models [ 5 , 8 - 13 ]. Conversely, pharmacologic activation can produce pain behavior even in the absence of tissue injury [ 14 - 17 ]. The amygdala consists of several anatomically and functionally distinct nuclei [ 2 , 18 ]. The laterocapsular division of the central nucleus (CeLC) has been termed the "nociceptive amygdala" because it receives nociceptive-specific information from the spinal cord and brainstem (external parabrachial area, PB) and the vast majority of CeLC neurons respond exclusively or preferentially to noxious stimuli [ 5 , 8 , 19 ]. Synaptic plasticity of PB inputs to the CeLC has been shown in models of arthritic pain [ 20 - 23 ], visceral pain [ 24 ] and chronic neuropathic pain [ 25 ] and is associated with pain-related central sensitization of CeLC neurons [ 21 , 26 - 31 ]. Highly processed multimodal, including nociceptive, information reaches the CeLC from thalamus and cortex through the lateral-basolateral (LA-BLA) network [ 5 , 8 ]. The LA-BLA circuitry is critical for the emotional evaluation of sensory stimuli and for acquisition and consolidation of aversive associations [ 2 , 3 , 32 , 33 ]. Our previous studies showed pain-related synaptic plasticity of excitatory transmission at the LA-BLA and BLA-CeLC synapses [ 10 , 20 , 23 ]. The BLA can influence CeA processes via direct glutamatergic projections and through indirect disynaptic routes involving GABAergic neurons in the intercalated cell masses (ITC) that project to the CeA [ 2 , 32 , 34 ]. Activation of inhibitory ITC neurons and subsequent inhibition of CeA neurons has been suggested to play an important role in fear extinction [ 2 , 35 ]. However, the role of synaptic inhibition of CeLC neurons in pain-related plasticity remains to be determined and was addressed in this study. Another focus of this study was on the involvement of group I metabotropic glutamate receptors (mGluRs) in synaptic inhibition of CeLC neurons, because our previous studies showed that these receptors are important modulators of excitatory synaptic transmission in the CeLC [ 20 ]. Group I mGluRs comprise mGluR1 and mGluR5 subtypes and are involved in neuroplasticity associated with normal brain functions as well as in neurological and psychiatric disorders [ 36 - 39 ] and in pain mechanisms [ 40 - 42 ]. Group I mGluRs play a critical role in pain-related central sensitization of amygdala neurons [ 20 , 27 ] and in amygdala-mediated pain behaviors [ 9 , 15 , 43 ]. Using patch-clamp recordings in brain slices from arthritic rats (kaolin-carrageenan model) and from controls, we measured and compared pain-related changes in inhibitory and excitatory transmission from the BLA to the CeLC and the contribution of group I mGluRs to these changes.

Methods Animals Male Sprague Dawley rats (120-250 g) were individually housed in standard plastic cages (40 × 20 cm) in a temperature-controlled room and maintained on a 12 hr day/night cycle. Standard laboratory chow and tap water were available ad libitum. On the day of the experiment, rats were transferred from the animal facility and allowed to acclimate to the laboratory for at least 1 hr. Arthritis pain model In one group of rats ("arthritis"), arthritis was induced in one knee joint as described in detail previously [ 65 ]. A kaolin suspension (4%, 80-100 μl) was slowly injected into the joint cavity through the patellar ligament with a syringe and needle (1 ml,). After repetitive flexions and extensions of the knee for 15 min, a carrageenan solution (2%, 80-100 μl) was injected into the knee joint cavity, and the leg was flexed and extended for another 5 min. This treatment paradigm reliably leads to localized inflammation and swelling of the injected knee within 1-3 hr. The inflammation persists for up to 2 weeks. It does not spread systemically [ 65 ]. Another group of rats ("normal") did not receive any injections but was kept under the same conditions as the arthritis rats before brain slices were obtained for electrophysiological studies. Electrophysiology Amygdala slice preparation Brain slices containing the CeA were obtained from normal rats and from arthritic rats (4-6 h after arthritis induction) as described before [ 10 , 20 , 21 , 23 ]. Rats were decapitated, and the brains quickly were dissected out and blocked in cold (4°C) artificial cerebrospinal fluid (ACSF). ACSF contained the following (in mM): 117 NaCl, 4.7 KCl, 1.2 NaH 2 PO 4 , 2.5 CaCl 2 , 1.2 MgCl 2 , 25 NaHCO 3 , and 11 glucose. ACSF was oxygenated and equilibrated to pH 7.4 with a mixture of 95% O 2 /5% CO2. Coronal brain slices (500 μm) were prepared using a Vibroslice (Camden Instruments, London, UK). After incubation in ACSF at room temperature (21°C) for at least 1 h, a single brain slice was transferred to the recording chamber and submerged in ACSF (31 ± 1°C), which superfused the slice at 2 ml/min. Whole-cell patch-clamp recording Whole-cell voltage-clamp recordings were made from CeLC neurons (see Figure 1A and 1B ) in brain slices from normal and arthritic rats using the "blind" patch technique as in our previous studies [ 10 , 20 , 21 , 23 ]. One neuron was recorded in each slice and 1 or 2 slices were used per animal. Patch electrodes (4-6 MΩ tip resistance) were made from borosilicate glass capillaries (1.5 and 1.12 mm, outer and inner diameter, respectively; Drummond, Broomall, PA) pulled on a Flaming-Brown micropipette puller (P-97/PC; Sutter Instruments, Novato, CA). The internal solution of the recording electrodes contained (in mM): 122 K-gluconate, 5 NaCl, 0.3 CaCl 2 , 2 MgCl 2 , 1 EGTA, 10 HEPES, 5 Na 2 -ATP, and 0.4 Na 3 -GTP, pH was adjusted to 7.2-7.3 with KOH and osmolarity to 280 mOsm/kg with sucrose. Data acquisition of current signals was done using a dual four-pole Bessel filter (Warner Instruments), a low-noise Digidata 1322 interface (Molecular Devices), an Axoclamp-2B amplifier (Molecular Devices) and a Pentium personal computer. Evoked current data were acquired and analyzed using pCLAMP10 software (Axon Instruments). Head-stage voltage was monitored continuously on an oscilloscope to ensure precise performance of the amplifier. Neurons were voltage-clamped at -70 (chloride reversal potential) or 0 mV (reversal potential of EPSCs) for the study of excitatory and inhibitory transmission, respectively. High gigaohm seal and low series (20 MΩ) resistances were checked throughout the experiment (using pClamp9 membrane test function) to ensure high-quality recordings. Synaptic transmission Excitatory and inhibitory postsynaptic currents (EPSCs and IPSCs, respectively) were evoked in CeLC neurons (held at -70 mV or 0 mV) by electrical stimulation (150 μs square-wave pulses; S88 stimulator; Grass Instruments) of BLA afferents (see Figure 1B ) using a concentric bipolar stimulating electrode (David Kopf Instruments). The distance between stimulation and recording electrode was about 1 mm. Input-output relationships were obtained by increasing the stimulus intensity in 0.1 mA steps. For evaluation of a drug effect on synaptically evoked responses, the stimulus intensity was adjusted to 80% of the intensity required for the maximum response. Spontaneous and miniature (in 1 μM TTX) EPSCs and IPSCs were recorded at -70 and 0 mV, respectively [ 23 ]. A fixed length of traces (5 min) was analyzed for frequency and amplitude distributions using MiniAnalysis program 5.3 (Synaptosoft). The root mean square (RMS) of the background noise was computed for each set of data. The detection threshold for an event was set to 3-4 times the RMS value. Peaks were detected automatically, but each detected event was then visually inspected to avoid the inclusion of false data. Drugs The following drugs were used: 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7- sulfonamide disodium salt ( NBQX ; non-NMDA receptor antagonist); bicuculline (GABAA receptor antagonist); α-amino-4-carboxy-2-methylbenzeneacetic acid ( LY367385 ; selective mGluR1 antagonist) and 3-((2-Methyl-1,3-thiazol-4-yl)ethynyl)pyridine hydrochloride ( MTEP ; selective mGluR5 antagonist). Drugs were purchased from Tocris Cookson (Bristol, UK). All drugs were dissolved in ACSF to their final concentration on the day of the experiment. Selectivity and target concentrations have been established in the literature [ 40 , 42 , 52 , 53 ]. Drugs were applied to the brain slice by gravity-driven superfusion in the ACSF. ACSF contained (in mM): 117 NaCl, 4.7 KCl, 1.2 NaH 2 PO4, 2.5 CaCl 2 , 1.2 MgCl 2 , 25 NaHCO 3 , and 11 glucose. Solution flow into the recording chamber (1 ml volume) was controlled with a three-way stopcock. Drugs were applied for at least 8-10 min to establish equilibrium in the tissue. ACSF served as vehicle control in all experiments. Statistical analysis All averaged values are given as the mean ± SE. Statistical significance was accepted at the level P < 0.05. GraphPad Prism 3.0 software (Graph-Pad Software, San Diego, CA) was used for all statistical analysis. For multiple comparisons of I/O functions, one-way ANOVA or two-way ANOVA was used with appropriate posttests (Bonferroni to compare selected pairs of data; Dunnett's multiple comparison test to compare all data to a control value). Student's t test (paired or unpaired when appropriate) was used to compare two sets of data that have Gaussian distribution and similar variances. Kolmogorov-Smirnov test was used for cumulative distribution analysis of spontaneous and miniature synaptic events (MiniAnalysis program 5.3, Synaptosoft Inc., Decatur, GA).

Results Pain-related increase of excitatory transmission and decrease of glutamate-driven inhibitory transmission in CeLC neurons Excitatory and inhibitory postsynaptic currents (EPSCs and IPSCs, respectively) were evoked in CeLC neurons by electrical stimulation in the BLA (Figure 1A and 1B ). Monosynaptic EPSCs recorded in voltage-clamp at -70 mV were mediated by non-NMDA receptors because they were completely blocked by NBQX (10 μM, Figure 1C ). IPSCs recorded at 0 mV holding potential were blocked by a GABAA receptor antagonist (bicuculline, 10 μM, Figure 1C ). EPSCs, but not IPSCs, followed high-frequency (20 Hz) synaptic stimulation reliably (Figure 1D ). EPSCs had a fixed latency whereas the latencies of IPSCs showed larger variability (Figure 1E and 1F ). Average latency of EPSCs (from stimulus artifact to onset of synaptic current) was significantly shorter than that of IPSCs (see individual example in Figure 1F ; data are summarized in Figure 1G , n = 10 neurons; P < 0.01, paired t-test). Failure to follow high-frequency synaptic stimulation was significant in the sample of neurons (n = 15; P < 0.01, Dunnett's multiple comparison test, Figure 1H ). The data show that CeLC neurons receive monosynaptic excitatory and polysynaptic inhibitory inputs from the BLA. To determine pain-related changes of excitatory and inhibitory synaptic inputs, input-output (I/O) relationships were obtained by measuring peak amplitudes of EPSCs and IPSCs as a function of afferent fiber stimulus intensity for each neuron (see examples in Figure 2A and 2B ). CeLC neurons were recorded in slices from normal untreated rats and in slices from arthritic rats (4-6 hr after induction of arthritis in one knee joint; see Methods). Compared with controls (n = 11 neurons), I/O function of monosynaptic EPSCs at the BLA-CeLC synapse increased significantly in the arthritis pain model (n = 12 neurons, F 1,231 = 30.49, P < 0.0001, main effect of treatment, two-way ANOVA; Figure 2A ), whereas the I/O function of IPSCs decreased significantly (control, n = 11 neurons; arthritis, n = 12 neurons; F 1,231 = 22.15, P < 0.0001, main effect of treatment, two-way ANOVA; Figure 2B ). The increase of excitatory transmission relative to inhibitory transmission in the arthritis pain model is reflected in the significantly increased EPSC/IPSC ratio (P < 0.001, unpaired t test; Figure 2C ). Next we tested the hypothesis that polysynaptic inhibitory transmission is glutamate- driven. IPSCs were inhibited by a non-NMDA receptor antagonist (NBQX, 10 μM; Figure 3 ) in slices from normal animals (n = 5 neurons, F 1,88 = 24.11, P < 0.0001, main effect of drug, two-way ANOVA; Figure 3A ) and in slices from arthritic rats (n = 5 neurons, F 1,88 = 36.18, P < 0.0001, main effect of drug, two-way ANOVA; Figure 3B ). The effect of NBQX on I/O functions of inhibitory transmission was not significantly different in arthritis compared to normal conditions (P > 0.05, unpaired t-test; Figure 3C ). The pharmacological profile (blockade by NBQX) and synaptic characteristics (longer and more variable latencies and inability to follow high-frequency stimulation) indicate that IPSCs recorded in CeLC neurons are polysynaptic, involving a glutamatergic synapse. The results are consistent with morphological and functional evidence for glutamatergic projections from the BLA to GABAergic interneurons in the ITC [ 2 , 35 ] and suggest that CeLC neurons receive disynaptic feedforward inhibition. Pain-related loss of GABAergic inhibition of excitatory transmission We sought to determine if feedforward inhibition of CeLC neurons modulates excitatory synaptic transmission from the BLA and if that effect changes in the arthritis pain state. A GABA A receptor antagonist (bicuculline, 10 μM) significantly increased I/O function of excitatory transmission at the BLA-CeLC synapse under normal conditions (n = 5 neurons, F 1,88 = 8.80, P < 0.01, main effect of drug, two-way ANOVA; Figure 4A ), suggesting GABAergic control of excitatory inputs to the CeLC. In slices from arthritic rats, however, bicuculline had no significant effect on EPSCs evoked in CeLC neurons (n = 5 neurons, F 1,88 = 2.67, P > 0.05, main effect of drug, two-way ANOVA; Figure 4B ). The significantly decreased facilitatory effect of bicuculline in the arthritis model (P < 0.05, unpaired t-test; Figure 4C ) may suggest that loss of inhibition contributes at least in part to the pain-related increase of excitatory transmission (see Figure 2A ). mGluR1, but not mGluR5, can account for the pain-related changes of excitatory and inhibitory transmission in CeLC neurons Our previous studies showed that mGluR1 and mGluR5 modulate excitatory transmission in the CeLC and upregulation of mGluR1 is associated with pain-related synaptic plasticity [ 20 ]. Here we examined the modulation of inhibitory transmission by mGluR1 and mGluR5. Confirming the results of our previous study obtained with a different mGluR1 antagonist (CPCCOEt) [ 20 ], a selective mGluR1 antagonist (LY367385, 10 μM) inhibited excitatory synaptic transmission in CeLC neurons in slices from arthritic rats (n = 5 neurons, F 1,88 = 37.10, P < 0.0001, main effect of drug, two-way ANOVA; Figure 5B ) but had no significant effect under normal conditions (n = 3 neurons, F 1,44 = 1.03, P > 0.05, main effect of drug, two-way ANOVA; Figure 5A ). The inhibitory effect of LY367385 in the pain model was significantly different from that under normal conditions (P < 0.05, unpaired t-test; Figure 5C ). LY367385 increased inhibitory synaptic transmission in slices from arthritic rats significantly (n = 5 neurons, F 1,88 = 15.91, P < 0.0001, main effect of drug, two-way ANOVA; Figure 5E ) but had no significant effect under normal conditions (n = 3 neurons, F 1,44 = 0.56, P > 0.05, main effect of drug, two-way ANOVA; Figure 5D ). The facilitatory effect of LY367385 in the pain state was significantly different from that under normal conditions (P < 0.05, unpaired t-test; Figure 5F ). The results show that LY367385 can reverse pain-related changes of excitatory as well as inhibitory transmission in the CeLC. Like the mGluR1 antagonist, a selective mGluR5 antagonist (MTEP, 1 μM) significantly decreased the input-output functions of excitatory synaptic transmission in slices from arthritic rats (n = 5 neurons, F 1,88 = 16.12, P < 0.0001, main effect of drug, two-way ANOVA; Figure 6B ). Unlike LY367385, however, MTEP had inhibitory effects in slices from normal rats (n = 5 neurons, F 1,88 = 5.75, P < 0.01, main effect of drug, two-way ANOVA; Figure 6A ). The inhibitory effects on excitatory transmission were not significantly different between arthritis and normal conditions (P > 0.05, unpaired t-test; Figure 6C ). MTEP also decreased inhibitory synaptic transmission in slices from arthritic rats (n = 5 neurons, F 1,88 = 10.13, P < 0.01, main effect of drug, two-way ANOVA; Figure 6E ) and in slices from normal animals (n = 5 neurons, F 1,88 = 22.54, P < 0.0001, main effect of drug, two-way ANOVA; Figure 6D ). The inhibitory effects of MTEP on IPSCs were not significantly different between normal and arthritis conditions (P > 0.05, unpaired t-test; Figure 6F ). The results show that mGluR5 are involved in excitatory and inhibitory synaptic transmission under normal conditions and in the pain state, whereas mGluR1 contribute only to the pain-related changes of excitatory and inhibitory transmission in CeLC neurons. Presynaptic action potential-dependent action of mGluR1 and postsynaptic action of mGluR5 The analysis of spontaneous and miniature EPSCs and IPSCs is a well established electrophysiological approach to determine pre- versus postsynaptic mechanisms. Presynaptic changes at the transmitter release site affect frequency, whereas changes at the postsynaptic membrane would alter amplitude (quantal size) [ 44 ]. In slices from arthritic rats, LY367385 (10 μM) decreased frequency (Figure 7B ), but not amplitude (Figure 7C ), of spontaneous EPSCs (sEPSCs) significantly (cumulative frequency distribution, P < 0.05, Kolmogorov Smirnov test; mean frequency, P < 0.01, paired t-test; n = 6 neurons). Original recordings of sEPSCs are shown in Figure 7A . Recordings were made in slices from arthritic rats, because LY367385 had no significant effect under normal conditions (see Figure 5 ). LY367385 (10 μM) had no significant effect on frequency (Figure 7E ) and amplitude (Figure 7F ) of miniature EPSCs (mEPSCs) recorded in TTX (1 μM) in slices from arthritic rats (cumulative distribution, P > 0.05, Kolmogorov-Smirnov test; mean frequency and amplitude, n = 8 neurons, P > 0.05, paired t-test; original traces are shown in Figure 7D ). The lack of effect on mEPSCs suggests an action potential-dependent site of action. LY367385 (10 μM) increased frequency, but not amplitude, of sIPSCs (n = 7 neurons; Figure 7G-I ) and mIPSCs (n = 7 neurons; Figure 7J-L ) significantly (cumulative frequency distribution, P < 0.05, Kolmogorov-Smirnov test; mean frequency, P < 0.01, paired t-test). The data suggest that mGluR1 act presynaptically on GABAergic terminals to regulate glutamatergic transmission in the arthritis pain model. MTEP (10 μM) decreased the amplitude, but not frequency, of sEPSCs (n = 5 neurons; Figure 8A-C ) and mEPSCs (n = 5 neurons; Figure 8D-F ) recorded in CeLC neurons in slices from arthritic rats (cumulative amplitude distribution, P < 0.05, Kolmogorov-Smirnov test; mean amplitude, P < 0.01, paired t-test). MTEP also decreased the amplitude (n = 4 neurons; Figure 8G-I ), but not frequency (n = 4 neurons; Figure 8J-L ), of sIPSCs and mIPSCs in CeLC neurons significantly (cumulative amplitude distribution, P < 0.05, Kolmogorov-Smirnov test; mean amplitude, P < 0.001, paired t-test). The data suggest that mGluR5 regulate both excitatory and inhibitory synaptic transmission in CeLC neurons through a postsynaptic mechanism of action. Presynaptic modulation of excitatory transmission by mGluR1 involves GABAA receptors The data presented so far show that mGluR1 acts presynaptically to inhibit GABAergic transmission but also increases excitatory transmission through an action potential dependent "presynaptic" mechanism. We tested the hypothesis that the inhibitory action of mGluR1 on GABAergic terminals is a mechanism by which mGluR1 increase excitatory transmission. LY367385 (10 μM) decreased excitatory synaptic transmission (EPSCs, n = 5 neurons, F 1,88 = 42.06, P < 0.0001, main effect of drug, two-way ANOVA; Figure 9A ) and the number of synaptically evoked spikes (n = 5 neurons, P < 0.05, ANOVA with Bonferroni posttest; Figure 9B ) in CeLC neurons in slices from arthritic rats. The addition of bicuculline (10 μM) partially reversed the inhibitory effect of LY367385 on EPSCs (n = 5 neurons, F 1,88 = 14.85, P < 0.001, main effect of drug, two-way ANOVA; Figure 9A ) and on synaptically evoked spikes (n = 5 neurons, P < 0.05, ANOVA with Bonferroni posttest; Figure 9B ), whereas bicuculline alone had no effect in the arthritis pain state (see Figure 4B ). The data suggest that removal of the mGluR1-mediated blockade of inhibitory transmission with an mGluR1 antagonist restores inhibitory control of excitatory synaptic transmission as is evident from the facilitatory effect of a GABAA-receptor antagonist that was lost in the arthritis pain model. Therefore, activation of mGluR1 in arthritis may explain the loss of inhibitory control (disinhibition) of excitatory transmission in the CeLC. Monosynaptic IPSCs are not under control of mGluR1 in the arthritis pain model IPSCs evoked in some CeLC neurons showed little variability in latency and followed high-frequency stimulation, suggesting that they were monosynaptic (Figure 10A and 10B ). NBQX (10 μM) had no significant effect on monosynaptic IPSCs (n = 3 neurons, P > 0.05, paired t-test; Figure 10C ). In contrast to its facilitatory effect on polysynaptic IPSCs associated with feedforward inhibition of CeLC neurons (Figure 5E and 5F ), LY367385 (10 μM) had no significant effect on monosynaptic IPSCs in slices from arthritic rats (n = 3 neurons, P > 0.05, paired t-test; Figure 10D ). The origin of these monosynaptic inhibitory inputs remains to be determined.

Discussion The novel key findings of this study on amygdala function related to pain are as follows. 1) In contrast to the increase in excitatory synaptic transmission, inhibitory feedforward inhibition of CeLC neurons decreases in a model of arthritis pain, shifting the balance toward a dominance of excitatory inputs. 2) The differential change of excitatory and inhibitory transmission involves mGluR1 acting presynaptically on GABAergic terminals to decrease inhibitory and enhance excitatory transmission. 3) Postsynaptic mGluR5 contribute to both excitatory and inhibitory synaptic transmission under normal conditions and in the arthritis pain model; their effect on synaptic inputs does not change in the pain state. To conclude, the data confirm the results of our previous studies [ 20 , 27 ] that a change in the function of mGluR1 rather than mGluR5 contributes to enhanced excitatory transmission and increased activity of CeLC neurons; importantly, the present study provides novel insight into underlying mechanism by showing that mGluR1 exert their facilitatory effect through disinhibition, i.e., inhibition of inhibitory control of excitatory inputs to the CeLC (see Figure 11 ). The conclusion is supported by the following observations. We show for the first time a change in feedforward inhibition of CeLC neurons in a pain model. Pharmacological evidence (blockade by NBQX) and synaptic characteristics (longer and more variable latencies and inability to follow high-frequency stimulation reliably) indicate that synaptic inhibition of CeLC neurons is polysynaptic, involving a glutamatergic synapse. The results are consistent with reports in the literature that glutamatergic projections from the BLA do not only reach the CeA directly but also target a cluster of GABAergic interneurons in the ITC that are interposed between BLA and CeA [ 2 , 18 , 35 , 45 , 46 ]. The CeA serves as the output nucleus for major amygdala functions and regulates behavioral responses through projections to hypothalamic nuclei and brainstem areas [ 4 , 45 , 47 ]. The CeA receives affect-related information that is generated in the LA-BLA network through associative processes [ 3 , 4 , 18 , 32 ]. BLA axons projecting toward the CeA form excitatory synapses with ITC neurons that project to the CeA where they generate feed-forward inhibition [ 45 , 48 , 49 ]. Accumulating evidence suggests that direct glutamatergic projections from the BLA to the CeA are important for fear expression whereas activation of ITC neurons that inhibit CeA output neurons might account for the reduction of fear expression after extinction [ 2 , 35 , 50 , 51 ]. Our results show that feedforward inhibition controls excitatory inputs to the CeA under normal conditions but not in a pain state. The fact that bicuculline enhanced excitatory transmission while the CeLC neuron was held at -70 mV, the equilibrium potential for chloride ions, may suggest that GABAA-receptors modulate excitatory inputs indirectly through a "presynaptic" site of action or in the network rather than on the CeLC neuron itself. The significantly decreased facilitatory effect of bicuculline in the arthritis model suggests that a loss of inhibition that may contribute at least in part to the pain-related increase of excitatory transmission mediated by mGluR1. The results obtained with highly selective antagonists for mGluR1 and mGluR5 [ 40 , 42 , 52 , 53 ] show that presynaptic mGluR1, but not postsynaptic mGluR5, can account for the pain-related changes of excitatory and inhibitory transmission in CeLC neurons. The interaction between mGluR1 and GABAergic transmission further suggests that pain-related decrease of inhibitory transmission is an active process that involves activation of mGluR1. Based on the analysis of spontaneous and miniature synaptic events, mGluR1 act presynaptically on GABAergic inputs whereas their effect on excitatory transmission is indirect through a process that requires action potential dependent network activity. Excitatory transmission is under GABAergic inhibition that is lost in arthritis through a mechanism that involves mGluR1. Removal of the mGluR1-mediated blockade of inhibitory transmission with an mGluR1 antagonist restores inhibitory control of excitatory synaptic transmission. Therefore, activation of mGluR1 that is seen in arthritis but not under normal conditions may explain the loss of inhibitory control (disinhibition) of excitatory transmission in the CeLC. Group I mGluR subtypes mGluR1 and mGluR5 play important roles in physiological neuroplasticity as well as in neurological and psychiatric disorders [ 36 - 39 ] and in pain mechanisms [ 40 - 42 ]. Our previous studies showed that in the amygdala, activation of mGluR5, but not mGluR1, enhanced the excitatory responses of CeLC neurons to innocuous and noxious stimuli in naïve animals [ 54 , 55 ]. In the arthritis pain model, blockade of mGluR1 and mGluR5 decreased the enhanced activity of CeLC neurons to normal-like levels, suggesting a major change in the function of mGluR1 in pain. The underlying mechanism included presynaptic facilitation of excitatory transmission from the parabrachial area and the BLA to CeLC neurons by mGluR5 under normal conditions and by mGluR1 and mGluR5 in the arthritis pain model [ 20 ]. Largely based on agonist data we assumed that both mGluR1 and mGluR5 acted presynaptically in that model. However, the detailed analysis of miniature events in the present study suggests that mGluR5 are postsynaptic and mGluR1 have presynaptic effects. The new results further show that the action of mGluR1 on excitatory transmission involves the inhibition of disynaptic inhibitory inputs from the BLA (disinhibition). Inhibition of inhibitory synaptic transmission by group I mGluRs has been shown in the hippocampus (mGluR1 [ 56 , 57 ]), striatum (mGluR1 [ 58 ]), cerebellum (mGluR1 [ 59 ]), midbrain [ 60 , 61 ] and periaqueductal gray (mGluR5 [ 62 ]). The mechanism of inhibition was typically presynaptic, and some evidence suggests the involvement of retrograde endogenous cannabinoid signaling through CB1 receptors [ 62 - 64 ].

Conclusion Both increased excitatory transmission and decreased inhibitory transmission in the CeLC in a model of arthritis pain involve mGluR1. These receptors act presynaptically to decrease synaptic inhibition, thus dis-inhibiting excitatory inputs to the CeLC, which may explain the loss of inhibitory control and increase in excitatory transmission observed in the arthritis pain model. mGluR5 act postsynaptically to facilitate both excitatory and inhibitory inputs, but they cannot account for the differential pain-related changes that involve loss of presynaptic GABAergic control of excitatory transmission to the CeLC. The concept of disinhibition of amygdala function may provide important insights into emotional-affective pain mechanisms and potential therapeutic strategies.

Neuroplasticity in the central nucleus of the amygdala (CeA), particularly its latero-capsular division (CeLC), is an important contributor to the emotional-affective aspects of pain. Previous studies showed synaptic plasticity of excitatory transmission to the CeLC in different pain models, but pain-related changes of inhibitory transmission remain to be determined. The CeLC receives convergent excitatory inputs from the parabrachial nucleus in the brainstem and from the basolateral amygdala (BLA). In addition, feedforward inhibition of CeA neurons is driven by glutamatergic projections from the BLA area to a cluster of GABAergic neurons in the intercalated cell masses (ITC). Using patch-clamp in rat brain slices we measured monosynaptic excitatory postsynaptic currents (EPSCs) and polysynaptic inhibitory currents (IPSCs) that were evoked by electrical stimulation in the BLA. In brain slices from arthritic rats, input-output functions of excitatory synaptic transmission were enhanced whereas inhibitory synaptic transmission was decreased compared to control slices from normal untreated rats. A non-NMDA receptor antagonist (NBQX) blocked the EPSCs and reduced the IPSCs, suggesting that non-NMDA receptors mediate excitatory transmission and also contribute to glutamate-driven feed-forward inhibition of CeLC neurons. IPSCs were blocked by a GABAA receptor antagonist (bicuculline). Bicuculline increased EPSCs under normal conditions but not in slices from arthritic rats, which indicates a loss of GABAergic control of excitatory transmission. A metabotropic glutamate receptor subtype 1 (mGluR1) antagonist (LY367385) reversed both the increase of excitatory transmission and the decrease of inhibitory transmission in the arthritis pain model but had no effect on basal synaptic transmission in control slices from normal rats. The inhibitory effect of LY367385 on excitatory transmission was blocked by bicuculline suggesting the involvement of a GABAergic mechanism. An mGluR5 antagonist (MTEP) inhibited both excitatory and inhibitory transmission in slices from normal and from arthritic rats. The analysis of spontaneous and miniature EPSCs and IPSCs showed that mGluR1 acted presynaptically whereas mGluR5 had postsynaptic effects. In conclusion, mGluR1 rather than mGluR5 can account for the pain-related changes of excitatory and inhibitory synaptic transmission in the CeLC through a mechanism that involves inhibition of inhibitory transmission (disinhibition).

List of abbreviations BLA: basolateral nucleus of the amygdale; CeA: central nucleus of the amygdale; CeLC: latero-capsular division of the CeA; EPSC: excitatory postsynaptic current; IPSC: inhibitory postsynaptic current; ITC: intercalated cell masses; LA: lateral nucleus of the amygdale; mEPSC: miniature EPSC; mIPSC: miniature IPSC; sEPSC: spontaneous EPSC; sIPSC: spontaneous IPSC Competing interests The authors declare that they have no competing interests. Authors' contributions WR performed patch-clamp recordings, analyzed electrophysiology data, provided figures and wrote the first draft of the manuscript. VN conceptualized the hypothesis, designed and supervised the experiments, directed the data analysis, and finalized the manuscript. All authors read and approved the manuscript.

Acknowledgements This work was supported by National Institute of Neurological Disorders and Stroke Grants NS-38261 and NS-11255.

CC BY

no

2022-01-12 15:21:45

Mol Pain. 2010 Dec 16; 6:93

oa_package/8e/db/PMC3016348.tar.gz

PMC3016349

21129183

Background During general anaesthesia, pulmonary ventilation is secured with a tracheal tube or by a laryngeal mask and attention to the risk of complications related to a high intracuff pressure is important. When the cuff to tracheal wall pressure exceeds the tracheal capillary pressure (27-40 cm H 2 O) for approximately 15 min, the tracheal mucous membrane becomes ischemic [ 1 ]. The intracuff pressure approximates the cuff to tracheal wall pressures in high volume/low pressure cuffs [ 2 ] and a cuff pressure below 30 cm H 2 O is recommended to prevent ischemic injury [ 1 , 3 ]. Also recurrent laryngeal nerve palsy has been demonstrated in up to 5% of patients after intubation and a high cuff pressure is suspected to be important in that regard [ 4 , 5 ]. Similarly in patients provided with a laryngeal mask, a high cuff pressure may lead to palsy of the lingual, hypoglossal, and recurrent laryngeal nerves [ 6 - 8 ] but with the cuff pressure maintained below 60 cm H 2 O, the airway seal is optimized [ 9 , 10 ] and the incidence of a postoperative sore throat is low [ 11 - 13 ]. Here patients, provided with either a tracheal tube or a laryngeal mask during elective surgery requiring general anaesthesia, were assessed for the established cuff pressure. We aimed at evaluating the incidence of a cuff pressure that was outside the recommended level and hypothesized that especially overweight patients would be exposed to a high cuff pressure since a high body mass index caries a risk of gastroesophageal reflux [ 14 ] and that such patients require a high peak inspiratory pressure during mechanical ventilation [ 15 ].

Methods The study was evaluated by The Scientific Ethics Committee of the Capital Region of Denmark (Journal no. H-4-2010-075) but not considered to require ethical approval since it was directed to quality control and did not involve any experimental procedures. We determined the cuff pressure in consecutetively enrolled patients from two hospitals: 97 from Herlev Hospital and 104 from Rigshospitalet. We included adult patients planned for operation in general anaesthesia and who were provided either with a tracheal tube or a laryngeal mask, while we excluded patients who had been intubated prior to arriving at the operating room and those for whom the airway was kept patent with a double-lumen tracheal tube. When the patient arrived at the operating ward, the staff prepared the patient for anaesthesia and surgery according to local instructions and the anaesthesiologist in charge of the patient initiated the anaesthesia together with an anaesthesia nurse. For ventilatory and cardiovascular monitoring a Dräger CATO (type M32040, Lübeck, Germany) or Dräger Primus (G18155) anaesthesia machine was used together with a Phillips Intillivue MP70 monitor. Neuromuscular function was evaluated with a TOF monitor (Organon Dublin, Ireland). For induction of anaesthesia propofol or thiopental was administered guided by the weight of the patient and the administration was continued until the cilia reflex was eliminated. Anaesthesia was maintained with propofol, sevoflurane or desflurane and fentanyl or remifentanil was used for analgesia, while cisatracurium, rocuronium or suxamethonium facilitated tracheal intubation. The airway was kept patent with a tracheal tube (high volume/low pressure; Unomedical, Copenhagen, Denmark; n = 119) or a laryngeal mask (AuraOnce; Ambu A/S, Ballerup, Denmark; n = 82) for mamma (n = 49), gastrointestinal (n = 33), gynaecological (n = 11), orthopaedic (n = 5), plastic (n = 11), urological (n = 71), hepatic (n = 6), or vascular surgery (n = 15). Inflation of the cuff was not described in the local instructions for induction of anaesthesia and therefore carried out according to the disposition of the anaesthesiologist in charge of the patient and without the use of a manometer or a pressure release valve. In regard to this study there were no restrictions to the treatment of the patient before or after the cuff pressure was measured. According to local tradition, the airway cuff pressure is not adjusted during the surgical intervention and manipulation of the head does not regularly provoke a recheck of the cuff pressure. When anaesthesia was established and the tracheal tube or the laryngeal mask was in place, the cuff pressure was determined by a Universal cuff pressure gauge with an upper scale limit of 120 cm H 2 O (VBM Medizintechnik GmbH, Baden-Württemberg, Germany). The manometer was connected to the pilot balloon and the cuff pressure was read from the manometer and documented together with the height, weight, age, and sex of the patient and the type of surgery and the time from placement of the tube or mask to when the cuff pressure was measured. In case the cuff pressure was outside the recommended level [ 3 , 12 ], the pressure was adjusted. Considering that we should detect 10 patients provided with a too high laryngeal or tracheal tube pressure in each department and the incidence is about 25% [ 16 ], we evaluated the cuff pressure in 200 patients. Data are presented as median and range and a p-value < 0.05 was considered to be statistical significant. For correlations between cuff pressure and age, body mass index, type of surgery, and time from induction of anaesthesia to determination of the cuff pressure, Pearson's correlation was used.

Results We determined the cuff pressure in 110 female and 91 male patients, age 61 (18 - 93) years and body mass index 24.6 (14 - 48) kg m -2 and the cuff pressure was determined 58 (2 - 360) min after induction of anaesthesia. In the 119 patients provided with a tracheal tube during surgery, the cuff pressure was 30 (8 - 100) cm H 2 O and it exceeded 30 cm H 2 O for 54 patients, and the pressure was higher than 40 cm H 2 O for 33 patients (Figure 1 ) with no significant difference between values obtained in the two departments. For the 82 patients provided with a laryngeal mask, the cuff pressure was 95 (10 - 121) cm H 2 O reflecting that for 56 patients, the cuff pressure was above 60 cm H 2 O and it exceeded the upper gauge limit for 34 patients (Figure 2 ) and also for the established laryngeal mask cuff pressure, there were no significant difference between values obtained in the two departments. There was no significant relation between tracheal tube cuff pressure and age (r = 0.028), body mass index (r = 0.245), type of surgery (r = -0.001), or the time from induction of anaesthesia to determination of the cuff pressure (r = -0.168). Furthermore there was no significant relation between laryngeal mask cuff pressure and age (r = 0.129), body mass index (r = -0.015), type of surgery (r = -0.177), or the time from induction of anaesthesia to when the cuff pressure was determined (r = -0.074).

Discussion and Conclusion In contrast to what we expected, overweight patients were not more frequently exposed to a high cuff pressure than other patients but even without the use of nitrous oxide for maintained anaesthesia, the cuff pressure exceeded the recommended level for about half of the patients provided with a tracheal tube and for almost three quarters of those patients provided with a laryngeal mask. We expected a high cuff pressure to be predominant in obese patients since a high body mass index caries a risk of gastroesophageal reflux [ 14 ] and a high peak inspiratory pressure during mechanical ventilation [ 15 ]. A low laryngeal mask cuff pressure secures the airway seal [ 11 , 10 ] and reduces the incidence of postoperative sore throat [ 11 ]. It was therefore surprising to us that the majority of the patients were exposed to a too high cuff pressure and the most frequently measured laryngeal cuff pressure was in fact at a level that exceeded 120 cm H 2 O, arguing for that cuff pressure is an underestimated issue in general anaesthesia practice. Sengupta et al. [ 16 ] found that among 93 patients provided with a tracheal tube and undergoing general anaesthesia, 50% of the patients were having a cuff pressure above 30 cm H 2 O and 27% had a cuff pressure above 40 cm H 2 O. Similarly in a prehospital setting Galinski et al. [ 17 ] found that among 107 patients, the tracheal tube cuff pressure was larger than 27 cm H 2 O in 79% of the patients. Thus with inflation of air into the cuff until air sealing, the anaesthesiologist has a poor ability to estimate a correct cuff pressure by palpation of the pilot balloon [ 18 , 19 ]. When using inflation of a fixed volume of air into the cuff, a linear relationship between cuff volume and pressure is established [ 20 ], but the volume of air required (4.5 ml) to reach 50 cm H 2 O is only 50% larger than that required for establishing the safe tracheal tube cuff pressure of 30 cm H 2 O (3 ml), i.e. the safety margin is low. Focus on the cuff pressure is often directed to patients exposed to nitrous oxide during anaesthesia. In these patients cuff pressure monitoring with automatic tracheal tube cuff pressure control is both reliable and stable [ 21 ] and for paediatric patients intubated with a cuffed tube, the use of a pressure release valve prevents that high tube cuff pressures develop [ 22 , 23 ]. Use of a pressure release valve is relevant especially when a high compliance tracheal tube cuff is used, since it does not prevent a high cuff pressure to be transmitted to epithelia and the thin polyurethane membrane facilitates transmembrane diffusion of nitrous oxide with following rapid increase of the cuff pressure [ 24 ]. The window of application of a correct laryngeal mask cuff pressure is broader (< 60 cm H 2 O) and estimation of the cuff pressure by palpation of the pilot balloon is acceptable after some training [ 25 ]. Still the use of a cuff pressure gauge is favoured to confirm that a correct cuff pressure is established, both in regard to a tracheal tube and a laryngeal mask, considering the low cost of the device (~100 €). Among the limitations of this study is it that the cuff pressure was measured only once during anaesthesia and fluctuations in pressure by movement of the patient's head and change in the depth of anaesthesia or the level of neuromuscular blockade are not taken into account. It could be argued that the pressure reported is not representative for the whole period of anaesthesia, but measurements were spread over 358 min to obtain values both at induction of anaesthesia and during surgery. Also we did not register the technique used for inflation of air into the cuff, and we do not know whether the high incidence of cuff pressures above the recommended level is due to that the staff was unaware of the correct cuff pressure, or whether they did not inflate the cuff only to the level that secured the airway. To generalize the results of this study can also be argued: We only included two hospital departments, but a high incidence of cuff pressures above the recommended level is reported by others [ 16 , 17 , 26 ]. In conclusion this study demonstrates that more than 50% of patients are provided with a too high tracheal tube or laryngeal mask cuff pressure. We consider that to establish the correct cuff pressure is an important element in the anaesthetic procedures and that it involves the determination of the tracheal tube or laryngeal mask cuff pressure, not only at the induction of anaesthesia but also during its maintenance.

Discussion and Conclusion In contrast to what we expected, overweight patients were not more frequently exposed to a high cuff pressure than other patients but even without the use of nitrous oxide for maintained anaesthesia, the cuff pressure exceeded the recommended level for about half of the patients provided with a tracheal tube and for almost three quarters of those patients provided with a laryngeal mask. We expected a high cuff pressure to be predominant in obese patients since a high body mass index caries a risk of gastroesophageal reflux [ 14 ] and a high peak inspiratory pressure during mechanical ventilation [ 15 ]. A low laryngeal mask cuff pressure secures the airway seal [ 11 , 10 ] and reduces the incidence of postoperative sore throat [ 11 ]. It was therefore surprising to us that the majority of the patients were exposed to a too high cuff pressure and the most frequently measured laryngeal cuff pressure was in fact at a level that exceeded 120 cm H 2 O, arguing for that cuff pressure is an underestimated issue in general anaesthesia practice. Sengupta et al. [ 16 ] found that among 93 patients provided with a tracheal tube and undergoing general anaesthesia, 50% of the patients were having a cuff pressure above 30 cm H 2 O and 27% had a cuff pressure above 40 cm H 2 O. Similarly in a prehospital setting Galinski et al. [ 17 ] found that among 107 patients, the tracheal tube cuff pressure was larger than 27 cm H 2 O in 79% of the patients. Thus with inflation of air into the cuff until air sealing, the anaesthesiologist has a poor ability to estimate a correct cuff pressure by palpation of the pilot balloon [ 18 , 19 ]. When using inflation of a fixed volume of air into the cuff, a linear relationship between cuff volume and pressure is established [ 20 ], but the volume of air required (4.5 ml) to reach 50 cm H 2 O is only 50% larger than that required for establishing the safe tracheal tube cuff pressure of 30 cm H 2 O (3 ml), i.e. the safety margin is low. Focus on the cuff pressure is often directed to patients exposed to nitrous oxide during anaesthesia. In these patients cuff pressure monitoring with automatic tracheal tube cuff pressure control is both reliable and stable [ 21 ] and for paediatric patients intubated with a cuffed tube, the use of a pressure release valve prevents that high tube cuff pressures develop [ 22 , 23 ]. Use of a pressure release valve is relevant especially when a high compliance tracheal tube cuff is used, since it does not prevent a high cuff pressure to be transmitted to epithelia and the thin polyurethane membrane facilitates transmembrane diffusion of nitrous oxide with following rapid increase of the cuff pressure [ 24 ]. The window of application of a correct laryngeal mask cuff pressure is broader (< 60 cm H 2 O) and estimation of the cuff pressure by palpation of the pilot balloon is acceptable after some training [ 25 ]. Still the use of a cuff pressure gauge is favoured to confirm that a correct cuff pressure is established, both in regard to a tracheal tube and a laryngeal mask, considering the low cost of the device (~100 €). Among the limitations of this study is it that the cuff pressure was measured only once during anaesthesia and fluctuations in pressure by movement of the patient's head and change in the depth of anaesthesia or the level of neuromuscular blockade are not taken into account. It could be argued that the pressure reported is not representative for the whole period of anaesthesia, but measurements were spread over 358 min to obtain values both at induction of anaesthesia and during surgery. Also we did not register the technique used for inflation of air into the cuff, and we do not know whether the high incidence of cuff pressures above the recommended level is due to that the staff was unaware of the correct cuff pressure, or whether they did not inflate the cuff only to the level that secured the airway. To generalize the results of this study can also be argued: We only included two hospital departments, but a high incidence of cuff pressures above the recommended level is reported by others [ 16 , 17 , 26 ]. In conclusion this study demonstrates that more than 50% of patients are provided with a too high tracheal tube or laryngeal mask cuff pressure. We consider that to establish the correct cuff pressure is an important element in the anaesthetic procedures and that it involves the determination of the tracheal tube or laryngeal mask cuff pressure, not only at the induction of anaesthesia but also during its maintenance.

Background To prevent endothelium and nerve lesions, tracheal tube and laryngeal mask cuff pressure is to be maintained at a low level and yet be high enough to secure air sealing. Method In a prospective quality-control study, 201 patients undergoing surgery during anaesthesia (without the use of nitrous oxide) were included for determination of the cuff pressure of the tracheal tubes and laryngeal masks. Results In the 119 patients provided with a tracheal tube, the median cuff pressure was 30 (range 8 - 100) cm H 2 O and the pressure exceeded 30 cm H 2 O (upper recommended level) for 54 patients. In the 82 patients provided with a laryngeal mask, the cuff pressure was 95 (10 - 121) cm H 2 O and above 60 cm H 2 O (upper recommended level) for 56 patients and in 34 of these patients, the pressure exceeded the upper cuff gauge limit (120 cm H 2 O). There was no association between cuff pressure and age, body mass index, type of surgery, or time from induction of anaesthesia to the time the cuff pressure was measured. Conclusion For maintenance of epithelia flow and nerve function and at the same time secure air sealing, this evaluation indicates that the cuff pressure needs to be checked as part of the procedures involved in induction of anaesthesia and eventually checked during surgery.

Competing interests The authors declare that they have no competing interests. Authors' contributions KZR participated in development of the study design, collected the data, and wrote the first draft of the manuscript. NHS conceived of the study and contributed to protocol development and manuscript preparation. AMM developed the study design and contributed to manuscript preparation. HBN performed data analysis and contributed to manuscript preparation. All authors approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2253/10/20/prepub

CC BY

no

2022-01-12 15:21:45

BMC Anesthesiol. 2010 Dec 3; 10:20

oa_package/68/43/PMC3016349.tar.gz

PMC3016350

21143882

Background It is important to ensure an appropriate amount of pressure is exerted on the tracheal wall by an endotracheal tube cuff for two opposing reasons. Firstly, a higher pressure is desirable in order to form an effective seal, which reduces the pulmonary aspiration [ 1 ]. Secondly, a lower pressure is desirable in order to minimise pressure injury to the tracheal wall [ 2 - 4 ]. Consequently, it has been recommended that an endotracheal tube cuff should exert between 20 and 30 cmH2O of pressure on the tracheal wall [ 2 ]. This is thought to represent a balance between preventing pulmonary aspiration and protecting the tracheal wall from pressure injury. Traditional endotracheal tube cuffs are high-volume, low-pressure (HVLP) cuffs. These cuffs should not be fully inflated when used. This has two important consequences. Firstly, there are longitudinal folds in cuff wall because the cuff is not under tension. Secondly, the pressure exerted on the tracheal wall by the cuff is equal to the intracuff pressure. At an intracuff pressure of 30 cmH2O, the HVLP cuff exerts approximately 30 cmH2O of pressure on the tracheal wall. Therefore, it is recommended that HVLP cuffs are operated at an intracuff pressure of 30 cmH2O. In clinical practice, HVLP cuffs have been shown to allow pulmonary aspiration at an intracuff pressure of 30 cmH2O [ 5 , 6 ]. This occurs along the longitudinal folds, which develop in the cuff wall. It also appears that HVLP cuffs are routinely overinflated. Recent studies using HVLP cuffs have reported mean intracuff pressures between 35-62 cmH2O [ 7 - 9 ]. The desire to minimise pulmonary aspiration could explain why HVLP cuffs are routinely overinflated. The Lotrach endotracheal tube has a unique low-volume, low-pressure (LVLP) cuff, which has been designed to prevent pulmonary aspiration and avoid pressure injury to the tracheal wall [ 10 ]. The cuff is designed to be fully inflated when used. This has two important consequences. Firstly, the cuff does not develop longitudinal folds in cuff wall because the cuff is under tension. Secondly, the pressure exerted on the tracheal wall by the cuff is equal to the intracuff pressure minus the sum of elastic forces within the cuff. The sum of the elastic forces within the cuff is approximately 50 cmH2O. At an intracuff pressure of 80 cmH2O, the LVLP cuff exerts approximately 30 cmH2O of pressure on the tracheal wall. Therefore, it is recommended that the LVLP cuff is operated at an intracuff pressure of 80 cmH2O. The Lotrach also incorporates a cuff pressure controller to maintain an optimal intracuff pressure over time. The cuff has already been shown to prevent pulmonary aspiration in a pig model, in anaesthetised patients and in the critically ill [ 11 ]. The aims of this study were to estimate the pressure exerted on the tracheal wall by a HVLP cuff and the LVLP cuff in a bench-top, clinical and radiological study.

Method This prospective, observational study was conducted at the Queen Elizabeth Hospital Intensive Care Unit, Norfolk, UK. We used the Softseal [HVLP, Polyvinylchloride, internal diameter 8 mm, Portex, UK] endotracheal tube and the Lotrach [LVLP, Silicone, internal diameter 8 mm, Venner Medical, Singapore] endotracheal tube. All equipment was used according to the manufacturers' instructions. Bench-top study We placed the HVLP cuff in a rigid cylindrical tube with an internal diameter of 22 mm as an approximate model of normal adult human trachea and used a constant pressure device [Tracoe cuff pressure controller, Tracoe, Frankfurt, Germany] to inflate the HVLP cuff [ 12 ]. The HVLP cuff was overpressurised to an intracuff pressure of 50 cmH2O while a column of water was instilled above the cuff to a height of 40 cm. At an intracuff pressure of 50 cmH2O, the HVLP cuff exerts approximately 50 cmH2O of pressure on the tracheal wall. Next, the intracuff pressure was reduced to 30 cmH2O. At an intracuff pressure of 30 cmH2O, the HVLP cuff exerts approximately 30 cmH2O of pressure on the tracheal wall. The height of the column of water was re-measured once column of water had reached a plateau. Theoretically, the column of water would fall until it reached a height equal to the pressure exerted by the cuff on the tracheal wall, provided that the cuff did not continue to leak. If the cuff did continue to leak, it was assumed that the cuff was not fully inflated and the leakage occurred along the longitudinal folds in the cuff wall. In this instance, the intracuff pressure was assumed to be equal to the pressure exerted on the tracheal wall. The experiment was repeated 10 times so that any variation was averaged out. We did not attempt to assess amount or speed of fluid leakage past the cuff except to identify the pressure at which leakage stopped. The same protocol was used with the LVLP cuff. The LVLP cuff was overpressurised to an intracuff pressure of 100 cmH2O while a column of water was instilled above the cuff to a height of 40 cm. At an intracuff pressure of 100 cmH2O, the LVLP cuff exerts approximately 50 cmH2O of pressure on the tracheal wall. Next, the intracuff pressure was reduced to 80 cmH2O. At an intracuff pressure of 80 cmH2O, the LVLP cuff exerts approximately 30 cmH2O of pressure on the tracheal wall. The height of the column of water was re-measured once column of water reached a plateau. The experiment was repeated 10 times so that any variation was averaged out. Clinical study We use positive end expiratory pressure (PEEP) during recruitment manoeuvres in critically ill patients. We hypothesised that there would be gas leakage past the endotracheal tube cuff during a recruitment manoeuvre once the PEEP exceeded the pressure exerted on the tracheal wall by the cuff. All patients who required intubation over a 6 month period were included in our analysis. Patients were allocated to each group depending on their anticipated duration of intubation. Those patients who were anticipated to require more than 24 hours of intubation were intubated with a LVLP cuff. Forty eight patients were intubated with the HVLP cuff and 54 patients with the LVLP cuff. Both cuffs were operated at the recommended intracuff pressure: the HVLP cuff at 30 cmH2O and the LVLP cuff to 80 cmH2O. At these intracuff pressures, both cuffs exert approximately 30 cmH2O of pressure on the tracheal wall. Each patient underwent a staged recruitment manoeuvre while the intracuff pressure was maintained using a constant pressure device [Tracoe cuff pressure control, Tracoe, Frankfurt, Germany]. The PEEP was set to 15 cmH2O and then increased in 5 cmH2O increments every 5 seconds until 40 cmH2O was achieved. A second observer auscultated the anterior neck and the pressure at which air leak was heard was recorded. In practice, the point of air leakage was obvious and heard at the bedside along with the observation of the brisk bubbling of gas escape at the mouth. Theoretically, there would be no air leak up to a PEEP of 30 cmH2O. Above this, the PEEP would exceed the pressure exerted on the tracheal wall by the cuff and an air leak heard. Statistical analysis was performed using a two sided 95% confidence interval non-inferiority test. We hypothesised that the mean difference in the pressure exerted on the tracheal wall by the HVLP cuff and the LVLP cuff was within a clinically relevant magnitude. A figure 20% either side of 30 cmH2O was chosen to delineate the limits of a clinically relevant magnitude [ 13 ]. Radiological study We hypothesised that the anatomical distortion of the trachea caused by the inflation of the cuffs would be similar if the pressure exerted on the tracheal wall from both cuffs was similar. Both cuffs were operated at the recommended intracuff pressure: the HVLP cuff at 30 cmH2O and the LVLP cuff to 80 cmH2O. At these intracuff pressures, both cuffs exert approximately 30 cmH2O of pressure on the tracheal wall. All patients who required chest CT examinations as part of their ongoing medical care over a 6 month period were included in our analysis. In this part of the study, the intracuff pressure of the HVLP cuff was not controlled. In contrast, the intracuff pressure in the LVLP cuff was controlled because the device incorporates a cuff pressure controller. An independent radiologist examined the images of seven sequential patients who were intubated using the HVLP cuff and six sequential patients who were intubated using the LVLP cuff. The tracheal antero-posterior (AP) and transverse diameters were measured at the mid-cuff level, and 20 mm above the cuff and 20 mm below the cuff where there was no contact with the trachea (Figure 1 ). The degree of anatomical distortion of the trachea was recorded. Anatomical distortion was defined as the difference in diameter between the mid-cuff and the mean of the above and below cuff diameters. Statistical analysis was performed using a two sided 95% confidence interval non-inferiority test. We hypothesised that the mean difference in anatomical distortion of the trachea by the HVLP cuff and the LVLP cuff was within a clinically relevant magnitude. A figure 20% either side of the mean of the above and below cuff diameters was chosen to delineate the limits of a clinically relevant magnitude [ 13 ]. Ethical considerations This study was conducted with prospective approval from the Local Research and Ethics Committee. The requirement for patient consent or next of kin consent was waived because the patients were unconscious, the data was observational data from our routine practice, the data was collected for audit, service surveillance and improvement purposes and the data was anonymised. In the clinical study, each patient underwent a recruitment manoeuvre as part of his or her routine respiratory care. The data from this practice was recorded in the patients' electronic medical record. In the radiological study, each patient underwent a chest CT examination as part of as part of his or her medical care. The data from this practice was recorded in the patients' electronic medical record.

Results Bench top study The HVLP cuff failed to maintain any water above the cuff and a plateau could not be measured. Therefore, the pressure exerted on the tracheal wall was assumed to be equal to the intracuff pressure. The LVLP cuff achieved a plateau at a mean height of 25.2 cmH2O (SD 0.34). There was no leakage of water past the cuff once the column of water had reached a plateau. No statistical analysis was performed because the HVLP cuff did not achieve a plateau. Clinical study The mean pressure at which air leak occurred using the HVLP cuff was 32.4 +/- 0.7 cmH2O (SD 3.0). The mean pressure at which air leak occurred using the LVLP cuff was 30.0 +/- 0.8 cmH2O (SD 3.8). There was a statistically significant difference between the two groups: the 95% confidence intervals at which air leak occurred did not overlap. However, the limits of our clinically relevant magnitude were 24 cm and 36 cm. The confidence intervals for each cuff were enclosed within these limits and so difference was deemed not clinically relevant. Radiological study The mean duration of time between intubation and chest CT examination was 2.6 hours (SD 4.4) in the HVLP cuff group and 192.4 hours (SD 228.9) in the LVLP cuff group. The tracheal AP and transverse diameters at the mid-cuff level, and 20 mm above the cuff and 20 mm below the cuff are shown for the HVLP cuff and the LVLP cuff in tables 1 and 2 . The mean degree of anatomical distortion of the trachea in AP and transverse tracheal diameter using the HVLP cuff was 4.4 +/- 1.3 mm (SD 1.4) and 2.6 +/- 1.5 mm (SD 1.6) and using the LVLP cuff was 2.9 +/- 2.2 mm (SD 2.1) and 1.8 +/- 1.4 mm (SD 1.4). This represented a percentage change from baseline diameter of the trachea in AP and transverse tracheal diameter of 19.2% and 13.0% using the HVLP cuff and 14.1% and 9.8% using the LVLP cuff. There was not a statistically significant difference between the two groups: the 95% confidence intervals did overlap. However, the limits of our clinically relevant magnitude were 18.15 +/- 3.6 mm for the HVLP cuff and 16.9 +/- 3.4 mm for the LVLP cuff. The confidence intervals for the HVLP cuff exceeded these limits in both diameters and the LVLP cuff in the AP diameter and so the degree of anatomical distortion was deemed clinically relevant.

Discussion It has been recommended that the pressure exerted on the tracheal wall by an endotracheal tube cuff should be between 20 and 30 cmH2O [ 2 ]. This is thought to represent a balance between preventing pulmonary aspiration and protecting the trachea from pressure injury. Previously, the LVLP cuff has been shown to prevent pulmonary aspiration in a pig model, in anaesthetised patients and in the critically ill [ 11 ]. This is the first study to investigate the pressure exerted by the LVLP cuff on the tracheal wall in vitro and in vivo . Our bench top study estimated that the LVLP cuff transmitted approximately 30 cmH2O of pressure on the tracheal wall. This is demonstrated by the fact that there was only leakage of water past the cuff until the water column reached a height of 25.2 cm. This differs from 30 cmH2O because there is a range of acceptable values, which are allowed as part of the quality control during cuff manufacture. All LVLP cuffs operate within the range of 25-35 cmH2O. In contrast to the HVLP cuff, the LVLP cuff provided an effective seal at this pressure. The clinical study also estimated that the LVLP cuff exerted approximately 30 cmH2O of pressure on the tracheal wall, which was similar to that exerted by the HVLP cuff. These results are supported by our radiological study, which found that the degree of anatomical distortion of the trachea was similar in patients using the HVLP and LVLP cuff in both the AP and transverse diameters in normal clinical practice. Put together, these results suggest that the LVLP cuff exerts an acceptable amount of pressure on the tracheal wall, despite it having an intracuff pressure of 80 cmH2O. In the radiological study, the mean duration of time between intubation and chest CT examination was substantially longer in patients with LVLP cuffs (192.4 hours, SD 228.9) compared to patients with HVLP cuffs (2.6 hours, SD 4.4). This difference occurred because those patients who were anticipated to require more than 24 hours of intubation were intubated with a LVLP cuff and many of these patients went on to require a substantially longer period of intubation. Anatomical distortion of the trachea is known to occur after prolonged intubation [ 14 ]. Consequently, there is a bias in our study that favours anatomical distortion in patients with a LVLP cuff because they were intubated for a longer period time. This data further supports our assertion that the LVLP cuff exerts an acceptable amount of pressure on the tracheal wall when it is operated at the recommended intracuff pressure. The finding that the degree of anatomical distortion was deemed clinically relevant only in patients with HVLP cuffs was unexpected. A possible explanation is that the baseline tracheal diameters (as judged by the above and below cuff measurements) were lower in patients with HVLP cuffs. Therefore, the difference in the degree of the anatomical distortion with the HVLP cuffs can be attributed to the range of the patients' tracheal diameters and the geometrical configuration of the HVLP cuff. Interestingly, the degree of anatomical distortion was different in the AP and transverse tracheal diameters for both cuffs, which suggests that either the pressure from a cuff is not exerted homogenously on the trachea or that the compliance of the trachea is different in the two different axes. Previously, the pressure exerted on the tracheal wall by endotracheal tube cuffs has been estimated using calculations from balloons interposed between cuff and trachea, implantable transducers cuff, gas flowing across the cuff-trachea contact area through hollow sleeves and compliance curves [ 15 - 18 ]. There are common methodological problems with each technique, which may be categorised under the headings measuring probe error, pressure measurement volume displacement error and volume aliquot inflation error. Measuring probe error: interposing a transducer between the cuff and the trachea produces error because the transducer distorts the geometry of the trachea and/or the cuff [ 16 ]. In addition, if the transducer is fractionally on the luminal side of the trachea, the pressure may be overestimated. Alternatively, if the transducer is fractionally recessed into the trachea, the pressure may be underestimated. Furthermore, the position of the transducer in the trachea varies throughout inflation because the trachea is distensible while the transducer is not. Pressure measurement volume displacement error: gas within the measuring system produces error because it is compressible [ 19 ]. The greater the volume or pressure of the gas in the measuring system produces a greater underestimate of the true pressure. Volume aliquot inflation error: the addition of fixed aliquots of volume produces error because different cuffs have different compliance characteristics [ 17 ]. A small increase in volume for a cuff with low compliance will produce a greater increase in pressure relative to a cuff with high compliance. In this study, we estimated the pressure exerted on the tracheal wall by the cuff by measuring the ability of the cuff to support a column of water in vitro and a column of gas in vivo . Advantages of these techniques are that there is no transducer distorting the trachea/cuff, the measuring system does not contain compressible gas and the measurement is not affected by the compliance characteristics of the cuff itself. A disadvantage of these techniques is that measurement is not possible if there is a leak in the system as exemplified by our bench top study. A limitation of the radiological study was that it was observational data. This meant the intracuff pressure in patients with HVLP cuffs was not controlled. In contrast, the intracuff pressure in patients with LVLP cuffs was controlled because the device incorporates a cuff pressure controller. Therefore, any conclusions that are drawn between the two groups of patients are limited by the lack of control of the intracuff pressure in patients with HVLP cuffs. However, our results do reflect the usage of the two devices in normal clinical practice. The LVLP cuff is always used with a cuff pressure controller and so will always benefit from a second to second intracuff pressure adjustment back to normal. The HVLP cuffs commonly do not have a cuff pressure control system, notable exceptions being the Tracoe device described above and the Lanz endotracheal tube (Covidien, Mansfield, MA, USA). Furthermore, there were a limited number of patients in our radiological study because only patients who required chest CT examinations as part of their medical care were included in our analysis. However, these represented all of the available patients at the time of our analysis. Another limitation is that our study does not address the effect of the mucosal injury caused by shear forces on the trachea, which may occur with movement in a longitudinal or rotational basis between cuff and tracheal wall. This will require a further study.

Conclusions The bench-top study estimated that the LVLP cuff exerted approximately 30 cmH2O of pressure on the tracheal wall. The clinical study also estimated that the LVLP cuff exerted approximately 30 cmH2O of pressure on the tracheal wall, which was similar to that exerted by the HVLP cuff. These results are supported by our radiological study, which found that the degree of anatomical distortion of the trachea was similar using the HVLP and LVLP cuff in normal clinical practice. We conclude that the LVLP cuff exerts an acceptable amount of pressure on the tracheal wall when it is operated at the recommended intracuff pressure of 80 cmH2O.

Background The Lotrach endotracheal tube has a unique low-volume, low-pressure (LVLP) cuff, which has been designed to prevent pressure injury to the tracheal wall. We aimed to estimate the pressure exerted on the tracheal wall by the LVLP cuff and a conventional cuff in a bench-top, clinical and radiological study. Method In the bench-top study, a model trachea was intubated with the LVLP cuff and the conventional cuff. The cuff pressure was controlled using a constant pressure device. We assessed the pressure exerted on the tracheal wall by measuring the ability of the cuffs to support a column of water using a standard protocol. In the clinical study, we tested the ability of both cuffs to prevent air leak during a staged recruitment manoeuvre. In the radiological study, we recorded the degree of anatomical distortion of the trachea from both cuffs in the antero-posterior (AP) and transverse tracheal diameters. We performed statistical analysis using non-inferiority tests. Results In the bench-top study, the LVLP cuff achieved a plateau at a mean height of 25.2 cmH2O (SD 0.34). In contrast, the conventional cuff failed to maintain any water above the cuff and a plateau could not be measured. In the clinical study, the mean pressure at which air leak occurred was 30.0 +/- 0.8 cmH2O (SD 3.8) using the LVLP cuff and 32.4 +/- 0.7 cmH2O (SD 3.0) using the conventional cuff. In the radiological study, the mean degree of anatomical distortion of the trachea in AP and transverse tracheal diameter was 2.9 +/- 2.2 mm (SD 2.1) and 1.8 +/- 1.4 mm (SD 1.4) using the LVLP cuff and 4.4 +/- 1.3 mm (SD 1.4) and 2.6 +/- 1.5 mm (SD 1.6) using the conventional cuff. Conclusions The bench-top and clinical studies both demonstrated that the LVLP cuff exerted approximately 30 cmH2O of pressure on the tracheal wall. These results are supported by our radiological study. We conclude that the LVLP cuff exerts an acceptable amount of pressure on the tracheal wall when it is operated at the recommended intracuff pressure.

Competing interests Dr Peter Young is the inventor of the Lotrach endotracheal tube. In the past he has received funding and consultancy fees from Venner Medical. This work was not funded or financed by Venner Medical. Dr Peter Young is a minor shareholder in the intellectual property ownership of the Lotrach endotracheal tube and tracheal seal monitor (cuff pressure controller). The remaining authors do not declare any conflict of interest or competing interest. Authors' contributions AD carried out the bench-top and radiological study, drafted the write up for the bench-top study and is the lead author of the submitted manuscript. RS carried out the clinical study, drafted the write up for this section and gave final approval for the version to be published. MC interpreted the CT scans in the radiology study, drafted the write up for this section and gave final approval for the version to be published. MB devised the bench-top study, critically revised the manuscript and gave final approval for the version to be published. PY devised the clinical and radiological studies, critically revised the manuscript and gave final approval for the version to be published. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2253/10/21/prepub

Acknowledgements None declared.

CC BY

no

2022-01-12 15:21:45

BMC Anesthesiol. 2010 Dec 9; 10:21

oa_package/c1/2f/PMC3016350.tar.gz

PMC3016351

21092270

Background Asthma is a significant chronic health problem in the U.S. and other developed nations. It accounts for millions of hospitalizations and thousands of deaths per year. The incidence of this burdensome ailment has increased by 50% every ten years raising the prevalence in the U.S. to 26.7 million in 1997 [ 1 , 2 ]. A significant disparity in new asthma diagnoses has been noted between urban and suburban/rural children. Urban children suffer a significantly higher incidence of asthma than their non city-dwelling counterparts [ 1 ]. This effect has been attributed to differences in exposure to sensitizing allergens present in the environment. Namely, it is theorized that an allergen(s) present at higher concentration, or in a unique combination in urban environments is the principal aggravating factor. Suspected allergens include feline, canine, murine, dust mite and cockroach allergen and pollutants such as diesel particulates. A landmark paper published in 1997 identified a link between asthma incidence in urban children and antigens derived from the ubiquitous pest Blattella germanica , the German cockroach [ 3 ]. This link was further investigated in 1998 by the development of the first cockroach allergen induced model of asthma [ 4 ]. Publications from this lab have shown that aqueous house dust extract from the kitchens of severely asthmatic, urban children can induce allergic asthma-like symptoms in mice. Moreover, the most potent and abundant allergens identified in the dust extract were proteins of the German cockroach [ 5 ]. Thus, the study of cockroach allergen as an inducer of asthma in mice and humans is a rational first step in untangling the complex web of environmental exposure and allergic asthma. Despite the alarming increase in prevalence and incidence of asthma in recent times, treatments have evolved slowly. Current standards of care include a long lasting anti-inflammatory agent such as an inhaled glucocorticoid or long acting β2 adrenergic agonist, combined with a short acting β2 agonist rescue inhaler and or epinephrine injections. Other effective agents include anti-IgE monoclonal antibodies, leukotriene receptor antagonists, mast cell degranulation inhibitors, antihistamines and cholinergic antagonists [ 1 , 6 , 7 ]. These treatments focus on alleviating the symptoms of the disease rather than addressing the cause. Specifically, they manipulate events occurring downstream of the aggravating stimulus to diminish certain aspects of the response. While these drugs are very effective at decreasing exacerbations and halting attacks, they tend to lose effectiveness over time. Asthma progresses because chronic, inflammation-induced lung remodeling coupled with drug desensitization, allow the disease to become more severe and recalcitrant to treatment over time. Allergen desensitization is thus a highly attractive option for the treatment of allergic-type inflammatory diseases. This form of treatment has a significant advantage over standard therapeutic agents in that it addresses the fundamental cause of asthma rather than modifying downstream mediators. In addition allergen desensitization has the advantage of being tailored to the individual patient. A skin hypersensitivity panel can be performed for a wide array of potential allergens in order to identify the causative agent(s). Thus, the desensitization regimen can be narrowly focused on the most likely causative allergens, offering symptomatic relief based on rational and specific treatment targeted to the inciting allergens(s). Current desensitization procedures deliver increasing titers of the putative allergen subcutaneously over a period of weeks. The patient must travel to their physician's office and remain in the clinic for 30 minutes after the injection in case a life-threatening reaction occurs. In addition, the administering health professional must be specially trained and have the capacity to treat severe anaphylaxis [ 8 , 9 ]. These factors make current desensitization procedures relatively costly and inconvenient for both the patient and the care provider. Sublingual/oral allergen desensitization is starting to gain wider interest and acceptance[ 10 , 11 ] with significant advantages over subcutaneous desensitization. Chief among these is the lower potential for anaphylaxis. The patient is monitored for severe reactions only after the first dose; subsequently, he/she can self-administer treatment in their own home. This vastly increases compliance as there is very little disruption of the patient's daily schedule. This is especially true for asthmatic children in whom compliance is a significant issue. Therefore, oral desensitization is an ideal candidate for the treatment of allergic type asthma. Although oral tolerance is gaining wider acceptance, the mechanism of how this occurs has not been determined. We designed studies to examine whether oral tolerance would effectively relieve asthma-like pulmonary inflammation in response to cockroach allergens, and determine the mechanism(s) of why oral tolerance is effective.

Methods Experimental model We used exclusively female HSD-ICR mice at 18-20 grams. (Harlan Sprague Dawley Inc., Frederick, MD). All data represent the combination of 3 replicates except for direct airways resistance measurement which is the combination of 2 replicates. All experiments were reviewed and approved by the Boston University School of Medicine Institutional Animal Care and Use Committee. Allergen feeding Oral exposure to allergens was performed by gavage. Briefly, a solution containing 16 ug of combined Blag1 and Blag2 was prepared in 100 ul of PBS. Mice were lightly anesthetized with isoflurane and the solution delivered directly to the stomach by means of a metal gavage needle. Control mice received 100 ul of PBS by the same method. Feeding was performed daily for 4 days. The mice were given a 3 day rest period before the allergen sensitization protocol. For allergen specificity ovalbumin (OVA) was given on the same schedule and volume as those described above. The OVA solution was adjusted to provide the same total protein concentration as the CRA mixture, 7.35 mg/mL. Following the OVA tolerization period, the mice received CRA immunization and 2 CRA challenges along the same schedule as the CRA tolerized animals. Allergen sensitization Cockroach antigen (CRA) was purchased from Greer Laboratories (Lenoir, NC) as a lyophilized whole body extract of the German cockroach Blattella germanica . The CRA was reconstituted in sterile PBS and the concentration of components Blag1 and Blag2 were assayed by ELISA. The concentration of the solution was adjusted so that 50 ul contained 8 ug of combined Blag1 and Blag2. The immunization (day 0) and 2 challenges (days 14 and 21) were delivered intratracheally by direct pharyngeal delivery which is subsequently inhaled [ 12 ]. Briefly, the mouse was suspended by its front incisors on an incline board, its tongue was gently pulled forward and the CRA solution was placed at the back of the pharynx in two 25 ul aliquots for aspiration. The immunization dose was a 1:2 of the stock solution and the challenges were a 1:4 containing 4 ug and 2 ug combined Blag1 and Blag2 respectively. The naïve mice received no CRA challenges. The 0 Hr mice were administered 2 intratracheal challenges of CRA and were assayed and sacrificed at the time they would have received their final allergen challenge, i.e. they did not receive the third challenge. The 1.5 Hr and 24 Hr mice were given the full set of 3 challenges and were assayed and sacrificed at 1.5 and 24 hours post final challenge respectively. Respiratory measurements Mice were placed in unrestrained whole body plethysmograph chambers at the same time of day and exposed to a 2 minute aerosolization of PBS, 25 mg/mL or 50 mg/mL methacholine followed by a 5 minute recording period [ 13 , 14 ]. The mice were first allowed to explore the chambers with normal grooming behavior indicating that the mice had become acclimated. Direct resistance measurements were made on a flexivent instrument (ScireQ, Montreal, QC, Canada). Briefly, mice were anesthetized with pentobarbital, and the trachea was directly cannulated through a small incision. The mice were placed on the mechanical ventilator and then paralyzed with pancuronium bromide. A nebulizer attached to the instrument delivered PBS, and methacholine challenges at 25 and 50 mg/mL while airways resistance was measured. Sacrifice and Data Collection The mice were anesthetized with intrperitoneal ketamine/xylazine and then sacrificed by exsanguination and cervical dislocation. The trachea was opened and cannulated with a length of flexible tubing and the lungs were lavaged with 2 mL of warm HBSS in 250 ul aliquots. The left lung was removed and fixed in 70% ethanol for histology. The right lung was placed into ice cold Complete Protease Inhibitor Cocktail (Roche Chemicals, Switzerland). The right lungs are then homogenized, the homogenate centrifuged at 10,000 G for 15 minutes and the supernatant removed for cytokine analysis. The cellular components of the whole lung homogenate were then resuspended in 0.5% cetyltrimethylammonium chloride (CTAC) and sonicated to release the contents of the eosinophilic granules. The supernatant of this mixture was collected and assessed for peroxidase activity (EPO). The lavage fluid was centrifuged at 600 G for 5 minutes and the supernatant was removed. The cell pellet was resuspended in 200 ul of RPMI, the red cells lysed and counted on Coulter particle counter (Beckman Coulter, Fullerton CA). The cells were then adhered to a slide and counted at 100 × magnification. The absolute cell counts per BAL sample were calculated for total white cells, neutrophils, macrophages, eosinophils and lymphocytes. Blood samples were taken from each mouse at exsanguination for cell counting on a Hemavet (Drew Scientific, Dallas, TX). Cell counts were expressed as the absolute number of a particular cell per 20 ul blood sample. Cytokine and chemokine analysis Cytokines and chemokines were measured using sandwich ELISA [ 15 ]. Briefly, Nunc (Rochester, NY) plates were coated overnight at 4C with anti-cytokine antibodies, the plates were blocked, samples were incubated on plates for 2 hours at room temperature, a biotinylated secondary antibody was used to detect captured cytokines and chemokines following incubation with strepavidin conjugated horse radish peroxidase (SA-HRP) and a colorimetric reaction. BAL samples were diluted 1:2, lung homogenate samples were diluted 1:5 and standards contained an equal concentration of pooled naïve lung homogenate. Plates were read with a PowerWaveX plate reader (Bio-Tek Instruments, Hopkinton, MA). Statistics Statistical comparisons were performed by two-tailed t-test in Graphpad Prism 4.0 (La Jolla, CA). Power analysis was performed using freeware tools on http://www.biomath.info/. The coefficient of variance was calculated as the ratio of the standard deviation and the mean of each data set or CV = standard deviation/mean.

Results Respiratory Parameters Initially, we examined the respiratory parameters in CRA-fed (tolerized) and PBS-fed mice following allergen sensitization and 2 challenges. We compared respiratory data in response to 50 mg/mL of methacholine (Mch) because this dose provided a maximal response. The first parameter we analyzed was enhanced pause (Penh). Penh is a dimensionless composite parameter which can be used to screen experimental animals for airways hyperreactivity to Mch and obstruction [ 16 - 18 ]. Changes in Penh can be caused by any factor which alters the caliber of conducting airways including inflammation, smooth muscle constriction and luminal obstruction with mucus. The Penh values of the CRA Fed mice were found to be significantly lower than that of the PBS Fed animals (Figure 1A ). Because the reporting of plethysmograph data in isolation is controversial, this result was further verified by employing a forced oscillation device (FO) to directly assess airways resistance [ 19 - 21 ]. In these FO experiments, we compared the airways resistance of the CRA or PBS fed mice in response to 25 mg/mL of methacholine because this represented the plateau of airways hyper-responsiveness. The CRA fed mice confirmed the trend towards decreased airways resistance which was not significant (Figure 1A ). Power analysis revealed that 21 experimental animals per group would be needed to verify this difference in a statistically significant manner. This value is very close to the 19 animals per group which were needed to identify the difference using Penh. Finally, the directly measured resistance values in CRA and PBS fed mice were compared with the analogous Penh values using a ranked Pearson correlation. This analysis revealed a 93% correlation between resistance and Penh in the CRA fed mice and an 81% correlation in the PBS fed animals. Next we examined a host of other respiratory parameters which we have found to be correlated with the respiratory health of mice in our model of asthma. First we examined minute ventilation (MV). The minute ventilation of the CRA fed mice was significantly higher than that of the PBS fed animals (Figure 1B ). Maintenance of an elevated MV in times pulmonary distress is associated with superior respiratory health [ 22 - 24 ]. Namely, when lung disease is more severe, acute exacerbations cause involuntary decreases in MV. To determine the etiology of this difference we examined the components of MV, respiratory rate (RR) and tidal volume (TV). Respiratory rate was significantly higher in the CRA fed mice but there was no difference in TV (Figure 1C, D ). This indicates that the low MV in the PBS fed mice was due principally to a depressed RR. Next we compared the time of inspiration (Ti) and expiration (Te) of the CRA or PBS fed mice. The orally tolerized mice displayed significantly lower Ti and Te when compared to the PBS fed animals (Figure 2A, B ). Increased Ti and especially Te is associated with airways obstruction in inflammatory processes [ 25 - 28 ]. Finally we examined the peak inspiratory (PIF) and expiratory (PEF) flow rates which indicate the forcefulness of the inspiratory and expiratory cycles respectively. Serial measurements of PEF (peak flow) are often used to evaluate the effectiveness of asthma control regimens [ 25 ]. PIF was significantly depressed in the PBS fed animals as compared to the CRA fed mice while PEF showed no significant differences between the groups (Figure 2C, D ). These data, considered together suggest that the CRA Fed mice are in a superior state of respiratory health and were better able to compensate upon exposure to methacholine (by increasing RR, and PIF and maintaining low Ti and Te) in order to maintain an elevated minute ventilation. Pulmonary Inflammation To determine whether this improvement in respiratory health was due to inhibition of inflammatory cell recruitment we analyzed the inflammatory cells present in the bronchoalveolar lavage fluid (BAL) of the experimental mice. First we examined the levels of neutrophils, macrophages and lymphocytes. Neutrophils are potent inflammatory mediators in asthma and typically arrive to sites of inflammation in a rapid fashion [ 29 , 30 ]. Macrophages are the only inflammatory cells typically present in the lungs and serve as immune surveillance [ 31 ]. Finally, lymphocytes are key components of the adaptive immune system and T-regs are thought to be important in the execution of immune tolerance to ingested antigen [ 32 - 34 ]. However, the counts of neutrophils, lymphocytes and monocytes were not significantly different in the lavage fluid of the CRA fed mice as compared to the PBS fed mice (Figure 3A-C ). Lung eosinophilia is a hallmark of severe asthma. These cells respond to many chemotactic factors including the eotaxins (released from airways epithelial cells and macrophages among others) and the Th2 cytokines IL-4, 5 and 13 (released primarily from T cells) [ 35 - 37 ]. Eosinophil counts were significantly depressed in the CRA fed group (Figure 4A,C,D ). We verified the diminished presence of eosinophils by measuring the eosinophil specific peroxidase (EPO) activity of the lung homogenate and found that it was significantly decreased in the CRA fed animals (Figure 4B ). Finally, we measured the levels of circulating eosinophils in the blood using a Hemavet and found no statistically significant difference between the CRA-fed and PBS-fed groups of mice (Figure 4E ) Blood eosinophil numbers are displayed as the absolute number of cells per 20 ul blood sample. Each value is the mean ± SEM for n = 18. ** = p < 0.01 comparing CRA fed to PBS fed mice. We verified that the tolerization effect was antigen specific through OVA experiments. Mice were tolerized to OVA and then sensitized and challenged with CRA. There was no difference in the airways hyperreactivity as measured by Penh, bronchoalveolar lavage eosinophils or lung homogenate EPO levels (Figure 5A-C ). In order to determine a mechanism of depressed eosinophil recruitment in the CRA fed mice we examined a number of cytokines and chemokines in the bronchoalveolar lavage fluid (BAL) and lung homogenate supernatant (LH) (Table 1 ). The most obvious candidates, eotaxins 1 and 2 did not differ between the two experimental groups. Additionally, we found no significant differences in any of the Th2 cytokines (IL-4, IL-5 and IL-13) or in any of the mediators often associated with airways hyperreactivity such as TNF-α and IFN. To see if the depressed eosinophil recruitment could be related to antibody production, we measured serum IgG and IgE and found no significant difference between CRA and PBS fed mice. Additionally, histological analysis of whole lung sections with PAS stain revealed no difference in airways mucus levels between the experimental groups (Figure 6A-C ). Finally, we examined IL-10 levels in the BAL and LH as this cytokine is known to inhibit eosinophil recruitment to the lung [ 38 , 39 ]. The BAL showed equivalent levels of IL-10 between the CRA and PBS fed mice. However, the lung homogenate supernatant of the CRA fed mice contained significantly elevated levels of IL-10 compared to the PBS fed animals (Figure 7 ). Additionally, the serum levels of IL-10 were below detection limit in both groups of mice (data not shown).

Discussion A number of research groups have reported amelioration of experimental allergic and autoimmune diseases through the establishment of oral tolerance [ 40 - 42 ]. Most of these studies employed well defined solitary antigens such as OVA in order to isolate the desired response without excessive background interference. Our current work differed from previous approaches in that it employed a complex allergen mixture containing the defatted whole body extract of German cockroaches. This mixture contains the complete corporeal proteins of the cockroach, as well as a host of bioactive enzymes and the innate immune stimulants chitin and LPS. These experiments represent a broader approach to allergen desensitization. Namely, this study sought to tolerize experimental animals to the whole host of cockroach derived products which human subjects are likely to encounter in urban environments. It has previously been reported that oral exposure to OVA in the drinking water of experimental mice significantly decreased the airways hyperreactivity to methacholine and the production of Th2 cytokines such as IL-5 and IL-13 [ 43 ]. Our research obtained similar results to this previous study since the Penh of experimental mice fed with CRA was significantly lower than that of PBS fed mice. This indicated that the allergen fed mice were in a superior state of respiratory health as a result of the prior, gastric exposure to the putative antigen(s). As mentioned previously, other research groups, using isolated allergens have been able to induce significant declines in Th2 cytokines following antigen tolerization. Interestingly, we saw no differences in any of the cytokines and chemokines traditionally measured in allergic pulmonary inflammation. We therefore had to seek other explanations for the improved respiratory health we observed in our allergen tolerized mice. Other groups have shown decreases in inflammatory cell infiltrate following oral allergen desensitization [ 44 , 45 ]. In our studies we observed significantly diminished eosinophil recruitment to the lung airspaces with no differences in the BAL levels of other inflammatory cells. However, circulating blood eosinophils did not differ between the experimental groups. That an equivalent number of eosinophils were recruited into the blood in the CRA-fed and PBS-fed animals is not surprising considering that BAL and LH levels of chemotactic agents for these cells were equivalent between groups. This suggests that the mechanism for decreased eosinophil infiltration into the lung air spaces is regulated at the level of organ itself, potentially during adhesion/transmigration. In addition EPO activity was also significantly decreased in the CRA fed mice. Taken together, these data suggest that the recruitment of eosinophils has been inhibited to both the bronchoalveolar space and the lung parenchyma as a whole. Although the numbers of eosinophils in the lung were decreased in the CRA fed mice, the levels of chemotactic agents were not significantly different between the two groups. Eotaxin 1 and 2 are as their name suggests potent chemo attractants for eosinophils yet their concentrations were not significantly decreased in the CRA fed groups [ 46 , 47 ]. IL-4 and IL-5 are also known eosinophil chemotaxins and activators, but the levels of these cytokines were equivalent between the experimental groups [ 48 , 49 ]. In addition, IL-13 which has been identified as necessary for the entrance of eosinophils into the lung did not differ between the groups [ 50 , 51 ]. Having exhausted the traditional mediators of eosinophil recruitment we decided to probe IL-10 levels. This cytokine has been shown to inhibit the recruitment of eosinophils and alternately improve or aggravate airways hyperreactivity [ 38 , 39 , 52 ]. IL-10 levels were significantly elevated in the lung homogenate supernatant of the CRA fed mice with the means differing by approximately 1000 pg/mL. This represents a 20% increase in IL-10 production in the allergen fed mice. Whether this is a biologically significant difference is uncertain. IL-10 is known promote the development of oral tolerance, but the elevated levels seen after the final challenge are not directly related to the tolerization period which took place 24 days prior to the final challenge [ 53 , 54 ]. This supposition is supported in that the OVA tolerized mice did not differ in lung homogenate IL-10 levels as would be expected if a gastric tolerizing event was responsible for this late cytokine production (data not shown). Thus it seems likely that the post challenge increase in IL-10 is a separate event related to the pulmonary allergen exposure. Interestingly, the secretion of this immunomodulatory cytokine seems limited to the pulmonary environment as serum levels of IL-10 were below detection limit. In conclusion, oral exposure to cockroach allergen prior to pulmonary sensitization and challenge leads to significantly improved respiratory health in experimental mice. This improvement is due to reduced eosinophil recruitment into the air spaces and lung parenchyma. The inhibition of eosinophil recruitment may be related to increased production of IL-10 in the lung. Finally, this research suggests that oral tolerization to a complex environmental allergen is a viable option for desensitization in allergic airways disease.

Conclusion This research indicates that oral tolerization is a valid means to reduce pulmonary inflammation in a mouse model of allergic asthma. Oral tolerization presents an attractive therapeutic option for human asthmatics in that it addresses the primary cause of allergic asthma exacerbations rather than simply blunting the symptoms. In addition oral tolerization offers significant advantages over other desensitization procedures (epidermal injections) in that there is a much lower risk of anaphylaxis and treatments may be self-administered after an initial observation period.

Background Antigen desensitization through oral tolerance is becoming an increasingly attractive treatment option for allergic diseases. However, the mechanism(s) by which tolerization is achieved remain poorly defined. In this study we endeavored to induce oral tolerance to cockroach allergen (CRA: a complex mixture of insect components) in order to ameliorate asthma-like, allergic pulmonary inflammation. Methods We compared the pulmonary inflammation of mice which had received four CRA feedings prior to intratracheal allergen sensitization and challenge to mice fed PBS on the same time course. Respiratory parameters were assessed by whole body unrestrained plethysmography and mechanical ventilation with forced oscillation. Bronchoalveolar lavage fluid (BAL) and lung homogenate (LH) were assessed for cytokines and chemokines by ELISA. BAL inflammatory cells were also collected and examined by light microscopy. Results CRA feeding prior to allergen sensitization and challenge led to a significant improvement in respiratory health. Airways hyperreactivity measured indirectly via enhanced pause (Penh) was meaningfully reduced in the CRA-fed mice compared to the PBS fed mice (2.3 ± 0.4 vs 3.9 ± 0.6; p = 0.03). Directly measured airways resistance confirmed this trend when comparing the CRA-fed to the PBS-fed animals (2.97 ± 0.98 vs 4.95 ± 1.41). This effect was not due to reduced traditional inflammatory cell chemotactic factors, Th2 or other cytokines and chemokines. The mechanism of improved respiratory health in the tolerized mice was due to significantly reduced eosinophil numbers in the bronchoalveolar lavage fluid (43300 ± 11445 vs 158786 ± 38908; p = 0.007) and eosinophil specific peroxidase activity in the lung homogenate (0.59 ± 0.13 vs 1.19 ± 0.19; p = 0.017). The decreased eosinophilia was likely the result of increased IL-10 in the lung homogenate of the tolerized mice (6320 ± 354 ng/mL vs 5190 ± 404 ng/mL, p = 0.02). Conclusion Our results show that oral tolerization to CRA can improve the respiratory health of experimental mice in a CRA-induced model of asthma-like pulmonary inflammation by reducing pulmonary eosinophilia.

Competing interests The authors declare that they have no competing interests. Authors' contributions Vaickus, L.J. performed most of the data collection and analysis. Bouchard, J. provided data and made contributions to study design. Kim, J. provided data and made contributions to study design. Natarajan, S. provided data and made contributions to study design. Remick, D.G. is the principal investigator and mentor of the 1 st author and made contributions to study design. All authors have read and approved the final manuscript.

Acknowledgements The lead author would like to thank the Boston University School of Medicine Immunology Training Grant Program for instruction and funding. Research funded by NIEHS grant number: 5R01ES013538-04 and NIH training grant number: 2T32AI007309-21A1

CC BY

no

2022-01-12 15:21:45

Respir Res. 2010 Nov 23; 11(1):160

oa_package/67/c4/PMC3016351.tar.gz

PMC3016352

21134250

Introduction COPD in all stages of severity is a very prevalent disease and a great burden for patients and society [ 1 ]. In affluent countries COPD is related to smoking over a long period of time, whereas in many other countries it is also related to indoor and outdoor air pollution [ 1 ]. The pathology of chronic obstructive pulmonary disease include pulmonary inflammation, oxidants-antioxidants imbalance, protease-antiprotease imbalance, and both innate and adaptive immunity [ 2 , 3 ]. Smoking cessation has been proven to be effective in stopping further deterioration of pulmonary function, reducing symptoms and improving overall health [ 4 ]. Smoking cessation however, seems to have only limited influence on the inflammatory process that is associated with COPD. This inflammatory process is probably initiated by oxidative stress and forms the basis of the pathophysiology of COPD [ 5 - 8 ]. Thus the inflammatory process that is associated with COPD seems to be triggered by noxious gasses such as smoking and serious indoor or outdoor air pollution. Oxidative stress caused by these noxious gasses at the level of the epithelium of the bronchial tree might have play a key role in this inflammatory process. It is therefore possible that anti oxidant therapy or an intensive anti oxidant diet could have an influence on the inflammatory process and the progression of COPD. Over the last two decades a number of studies have suggested that COPD risk is associated with vitamins that all have antioxidant properties and with an anti oxidant diet. Low diet-intake of vitamins has been reported to reduce natural defenses and increase the possibility of airway inflammation [ 9 ]. Furthermore, a higher intake of fruits and vegetables was associated with a lower risk of COPD, lower mortality and an improvement of spirometric values [ 10 - 17 ]. When levels of vitamins were measured in the serum they were found to be significantly lower in COPD patients than in control subjects [ 18 ]. The association of vitamins with pulmonary diseases is further supported by a meta-analysis of 40 studies in patients with asthma. This meta-analysis revealed that relatively low dietary intake of vitamins A and C were associated with statistically significant increased odds of asthma and wheezing [ 19 ]. A large number of studies and reviews highlight an association of vitamins with lung function in healthy subjects and COPD patients [ 20 - 27 ]. Recently a randomized controlled trial suggested that a dietary shift to higher antioxidant food intake was associated with improvement in lung function [ 25 ]. Furthermore, several studies associate vitamins with a reduction in symptoms, respiratory infections and exacerbations [ 28 - 37 ]. Although for vitamin D the role in respiratory diseases has been clarified through its implication in immunity, for most other vitamins the mechanism of action is less clear [ 38 - 42 ]. Indeed we know that 1,25-dihydroxyvitamin D stimulates both innate and adaptive immunity, in addition to mineralization and calcium homeostasis. Further support on the important role of Vitamin D is given by the fact that it regulates genes that are implicated in apoptosis and cellular proliferation [ 39 ], known to be an important step in COPD pathogenesis [ 42 ]. Vitamin D has both immunomodulatory and antiiflammatory properties [ 43 ]. For vitamin A, B, C, and E, studies highlight their role in COPD risk as well as their connection with COPD outcomes such as symptoms and improvement in spirometric values, without a clear mechanism of action. This article aims to update the knowledge we have about the association between vitamin intake and COPD in outcome measures, as well as to assess the potential role of vitamin supplements.

Methods A systematic literature search was performed from 1989 until June 2010 in Pubmed, Embase and Cochrane Collaboration containing the following keywords: COPD in conjunction with smoking, gene polymorphisms, vitamins, FEV1, vitamin C, vitamin E, vitamin D, vitamin A, b-carotene, and vitamin supplements. Further articles were identified from the reference lists of the included articles. In order to be as accurate as possible we included in the present review only studies that have measured serum vitamin levels or used validated food frequency questionnaires to assess the role of vitamins. Other dietary factors -cured meats, fish, whole grains and alcohol- that have been reported to be associated with the risk of chronic obstructive pulmonary disease or with an increase in symptoms were not included in this review [ 44 - 51 ]. Further we did not examine the influence of caloric intake on COPD. Weight loss and muscle wasting, considered complications of COPD strongly associated with diet, are also not included in the review.

Results Methods of assessing vitamin status: which is best? The literature reports two essentially different ways to measure vitamins: Serum levels and Food Frequency Questionnaires (FFQ). Both measures have their advantages and disadvantages. Serum levels of vitamins The assessment of serum levels of vitamins can have the advantage of being more objective than patient's reported intake. However, serum level assessment of vitamins has the disadvantage that they represent the more recent intake, and for some vitamins such as vitamin C, the levels in peripheral blood are not representative of intake and do not change accordingly [ 19 , 52 ]. Food Frequency questionnaire Food frequency questionnaires on the other hand present a large heterogeneity with differences in assessing periods (from 1 day to two years), number of items on food questionnaires (ranging from 44 to 350) and use of portion size questions [ 53 ]. Further, it has been suggested that FFQs do not always detect weak associations [ 54 ]. Regarding vitamin C intake large differences were found between FFQs that used portion size questions instead of using standard portions [ 53 ]. Another problem regarding FFQ use is that it is difficult to determine which particular vitamin is associated with COPD and if it is the vitamins in fruits and vegetables that are associated with COPD, or another confounding nutrient. Resveratrol for example, a phenolic antioxidant is present in many fruits and is associated with anti-inflammatory activity [ 55 , 56 ]. Therefore it is not clear if in some cases, the vitamins have the beneficial effect, or other nutrients such as resveratrol with antioxidant and anti-inflammatory properties. Another important obstacle in FFQ is that an assessment of vitamin D from food intake would lead to an incorrect estimation in for example Mediterranean countries where there is a high skin synthesis because of the sunlight. We found 14 references [ 31 , 57 - 69 ] that assessed the relationship between a vitamin rich diet as assessed by a FFQ and the subsequent improvement of spirometric values and symptoms. Twelve studies [ 32 , 33 , 35 , 52 , 63 , 70 - 76 ] measured serum levels of vitamins. For the above mentioned reasons, and in order to have a more precise overview, we decided to include both FFQ and serum level studies. Food intake patterns: Effect of vitamins on the risk of developing COPD and associated mortality Varraso et al in a study of 72,043 women identified 754 cases of newly diagnosed COPD [ 44 ]. In this study a healthy diet (fruit, vegetables, fish, whole-grain products) was compared with a Western diet (refined grains, cured and red meats, desserts, French fries). The healthy diet was associated with a lower risk of COPD [ 44 ]. This could be considered to be due to the overall diet, or indicate a possible positive effect of vitamins on COPD risk, as fruits are considered sources rich with vitamins. From the same author another study comparing the same patterns of diet showed the same results in 111 self-reported cases of newly diagnosed COPD in men [ 46 ]. Celik et al used a food frequency questionnaire and found that the consumption of fruits and vegetables was significantly lower in COPD patients compared to the control group [ 77 ]. Fruit intake was related to a lower 25-year incidence of chronic bronchitis and emphysema [ 57 ] as well as spirometry improvement [ 78 ]. A recent randomized controlled trial has shown that a dietary shift to more anti-oxidant foods such as fruits and vegetables is associated with improvement in lung function [ 25 ]. These studies showed a protective role of vitamins against COPD but did not measure vitamins in serum or with a FFQ. Therefore conclusions about the special role of any vitamin in COPD could not be obtained. Vitamin D and COPD Vitamin D is extremely important for the human body. It has a significant role in bone mineralization, in calcium and phosphorous absorption, and is important in the immune system [ 39 , 40 ]. It's most important role however is in bone structure development and bone turn over, as a low vitamin D level is directly associated with osteoporosis. The main sources of vitamin D are skin synthesis and diet. The precursor form is 7-dehydrocholesterol which with UVB is transformed to vitamin D3. Vitamin D3 is transported via the D-binding protein (DBP) to the liver where with hydroxylation reactions transforms to 25(OH)D3 and which is transported again by DBP to the kidneys where it takes its active form of 1,25(OH)2D3. 25OHD3 can also be transformed in 1,25(OH)2D3 in the immune cells [ 42 ]. Some studies showed a DBP Gc-1F allele presence that was higher in COPD patients [ 79 , 80 ] and Schellenberg et al found that the Gc2 homozygous genotype was protective for COPD [ 81 ]. Other polymorphisms associated with vitamin D binding protein gene are related to clinical differences in families with alpha-1-antitrypsin deficiency [ 82 ]. COPD is characterized by inflammation induced by macrophages and neutrophils (innate immunity). COPD is considered a disease where proinflammattory cytokines are increased and has a Th2 response with a predominance of CD8 lymphocytes (adaptive immunity). 1,25-dihydroxyvitamin D stimulates innate immunity probably due to activation of cathelicidin (antimicrobial peptides) to enhance the bacterial killing via Toll-like receptors [ 83 , 84 ]. Vitamin D receptors (VDR) are present in various cells of both innate (ie.macrophages) and adaptive immunity (i.e.T and B cells). Vitamin D is able to modulate both types of immunity therefore minimizing inflammation [ 85 ]. Vitamin D in general is involved in modulating cellular proliferation, suppressing TH cells, [ 86 ], downregulating cytokines such as IL-2 [ 87 ], as well as in the inhibition of dendritic cells [ 88 ], all of which are known to be important in the COPD pathway. Regarding respiratory function, vitamin D plays a significant role in airway remodeling through the inhibition of TNFa and enhancement of Il-10 in immune cells [ 39 ]. Vitamin D also seems to play a role as an alternative treatment strategy to reverse glucocorticoid resistance through its ability to restore IL-10 response [ 89 ]. This is important since glucorticoid resistance is a pivotal barrier to the anti inflammatory treatment of COPD. Patients with COPD have an increased prevalence of osteoporosis (from 9-69%) and osteopenia (from 27-67%) [ 90 - 93 ]. Malnutririon and low vitamin D levels could be a cause of this higher prevalence [ 91 , 94 ]. The majority of COPD patients have vitamin D deficiency [ 39 , 41 , 95 - 97 ] therefore vitamin D supplementation in patients with COPD has been proposed [ 40 ]. Black et al reported that higher vitamin D levels were associated with better lung function [ 72 ]. In this study that used cross-sectional data from the Third National Health and Nutrition Examination Survey 14.091 people aged >20 years were included. The mean difference between the highest and the lowest quintile of 25-hydroxyvitamin D serum concentration was 126 ml in FEV 1, and 172 ml for FVC after adjustment for factors that affect lung function (age, gender, smoking, etc) [ 72 ] (Table 1 ). Vitamin D insufficiency has been reported to be associated with an increased incidence of chronic respiratory infections [ 29 , 33 - 35 ]. There are some studies that also suggest that low serum 25-hydroxy vitamin D levels are associated with upper and lower respiratory tract infection [ 33 - 35 ]. In one large cross sectional study with 18.883 participants, this association was stronger in COPD patients [ 33 ] (Table 1 ). Ebstein Barr virus infection, which is often found in COPD patients, is also associated with low levels of vitamin D [ 98 , 99 ]. Liou et al reported a relation between Toll-like receptors, external triggers and vitamin D-mediated innate immunity, and suggested that differences in the ability of human populations to produce vitamin D may contribute to susceptibility to microbial infections [ 100 ]. Finally, Vitamin D could play an important role as an antioxidant therapy, not only for the significant improvement in spirometric values, but also because it has been proposed as a novel treatment to cachexia and sarcopenia in COPD patients [ 101 ]. Vitamin C and E The role of vitamin C (also known as ascorbate or L-ascorbic acid) in the human body is essential. It has antioxidant properties, is involved in various metabolic reactions, and some studies report it also plays a role in the immune system [ 102 , 103 ]. It is considered important for the maintenance of the connective tissue and bone remodeling [ 102 ]. Vitamin E has antioxidant properties as well, and has been reported to have a protective role in the prevention of atherosclerosis and carcinogenesis [ 104 ]. In one study that included 3 European Countries a trend (P < 0.05) of lower COPD mortality was observed with vitamin E intake, while no trend was found with vitamin C after adjustment for age, smoking and country [ 16 ]. Higher levels of vitamin C and E in both serum and FFQ in healthy subjects were associated with an increase in FEV1 and FVC [ 31 , 32 , 58 - 66 , 69 , 70 ]. More details are depicted in Table 2 . In one study an increase of 20 micromol/Lt in plasma vitamin C concentration was associated with a 13% reduction in the risk of developing obstructive airway disease OR: 0.87 (CI:0.77-0.98) [ 75 ]. Studies regarding the role of vitamin C and E in respiratory symptoms showed that low levels were associated with more wheezing, phlegm production and dyspnea [ 28 , 31 , 32 , 36 , 37 ]. Tug et al found both vitamin E and vitamin A levels were significantly lower during exacerbations of COPD than in patients with stable COPD [ 30 ]. Takkouche et al, in 1667 cases of the common cold in the general population suggested that intake of vitamin C and zinc was not related to the occurrence of common cold [ 105 ]. Nevertheless vitamin C decreases the duration of common cold symptoms which might be important in patients with COPD [ 106 ]. Vitamin A and B Vitamin A (retinol and carotens) plays an important role in several functions of the human body including vision, bone and skin health, and further has an innate antioxidant activity. Vitamin B is involved in various steps of metabolism and enhances immunity. High levels of vitamin A, b-carotene and/or alpha-carotene were associated with increase in FEV1 and FVC in most of the studies [ 31 , 32 , 63 , 65 , 66 , 69 - 71 , 76 ] although there are some exceptions [ 57 , 61 , 68 ]. More details are presented in Table 2 . High serum beta carotene levels in a general population sample of 523 subjects were associated with the expression of a gene polymorphism that connected with a slower FEV1 decline [ 107 ]. Hirayama et al reported that the highest level of intake of vitamin A resulted in a 52% (p = 0.008) reduction in COPD risk [ 108 ] while in another study the risk for COPD was associated with lower levels of plasma vitamin A (p < 0.01)[ 108 ]. Fimognari et al reported lower levels of folate and vitamin B12 in COPD patients, resulted in an increased plasma level of total homocysteine, a known cardiovascular risk factor [ 109 ]. Regarding the role of b-carotene in respiratory symptoms, two studies showed a beneficial association with cough [ 32 , 36 ] and one study showed no correlation with symptoms, except for wheezing [ 31 ]. Vitamin supplementation Antioxidant supplementation has been proposed to be helpful in patients with COPD as a way to reduce oxidative stress and inflammation, and improve spirometric values [ 24 ]. Multivitamin supplementation has been reported to be popular for patients with COPD especially among older patients [ 110 ]. Seven studies reported the effect of vitamin supplements on several outcomes of COPD [ 18 , 36 , 111 - 115 ]. All these studies showed a large heterogeneity regarding: which vitamins had been supplemented, the dosage and the duration of the vitamin supplementation, ranging from 4 weeks to 5 years [ 111 - 115 ]. Secondly, different outcomes were measured such as spirometric values, symptoms and exercise capacity. A randomized controlled trial in high-risk individuals for cardiovascular events that received antioxidant vitamins (vitamin C, E and b-carotene) supplementation for 5 years failed to identify any improvement in 5-year mortality and in spirometric values or hospitalization due to COPD. However this study excluded patients with severe COPD [ 115 ]. More details are depicted in Table 3 . Also a cohort study of 77,719 participants using multi-vitamin supplements was not related to the total mortality [ 116 ]. Little is known regarding the prevention of upper respiratory tract infections after supplementation of Vitamin D although some studies report a trend for improvement [ 117 ] but some others do not confirm that (ranges of OR = 0.77-0.95) [ 118 , 119 ].

Conclusion The results of this review show that intake of various vitamins are associated with improvement in features of COPD such as symptoms, exacerbations and pulmonary function. Increased vitamin intake could probably reduce the annual decline of FEV1. Although the mechanisms behind these effects are often not clear, this might open possibilities to develop drugs that modify or prevent COPD. Dietary interventions directed towards high vitamin intake might be an additional approach towards COPD management. Although there are many studies that associate vitamins with improvement in lung function tests, there is no clear evidence of the benefit of vitamin supplements. Most studies regarding supplements showed no benefit of multivitamin supplemention in symptoms, spirometric function or hospitalization for COPD. This review suggests that future work is needed with prospective randomized controlled trials, that would explore the role of vitamins as well as the effectiveness of vitamin supplements on outcomes such as symptoms spirometric values, health status, risk of development of COPD and exacerbations rates.

Background Pulmonary inflammation, oxidants-antioxidants imbalance, as well as innate and adaptive immunity have been proposed as playing a key role in the development of COPD. The role of vitamins, as assessed either by food frequency questionnaires or measured in serum levels, have been reported to improve pulmonary function, reduce exacerbations and improve symptoms. Vitamin supplements have therefore been proposed to be a potentially useful additive to COPD therapy. Methods A systematic literature review was performed on the association of vitamins and COPD. The role of vitamin supplements in COPD was then evaluated. Conclusions The results of this review showed that various vitamins (vitamin C, D, E, A, beta and alpha carotene) are associated with improvement in features of COPD such as symptoms, exacerbations and pulmonary function. High vitamin intake would probably reduce the annual decline of FEV1. There were no studies that showed benefit from vitamin supplementation in improved symptoms, decreased hospitalization or pulmonary function.

Competing interests The authors declare that they have no competing interests. Authors' contributions Both authors (IGT, TvdM) wrote and revised the manuscript, and approved the final version.

CC BY

no

2022-01-12 15:21:45

Respir Res. 2010 Dec 6; 11(1):171

oa_package/30/71/PMC3016352.tar.gz

PMC3016353

21172020

Introduction Cancer cells are capable of evading regular immune responses for a number of reasons: they can secrete immunosuppressive factors [ 1 ], there can be down-regulation of antigen expression [ 2 , 3 ] or of major histocompatability complex (MHC) molecules [ 4 , 5 ] and also a lack of co-stimulation [ 6 , 7 ]. With the advent of gene therapy as a tool for cancer treatment, immunotherapy-related approaches to stimulate immune responses against cancer cells include the transfer of immune stimulatory genes such as cytokines or costimulatory genes into cancer cells, enhancing antigen presentation through the manipulation of antigen presenting cells (APCs) and genetic vaccination against cancer cell-specific antigens [ 8 , 9 ]. AAV has a number of properties that make it an ideal candidate as a gene delivery vector for the treatment of cancer. AAV elicits only mild host immune responses in vivo [ 10 ]; long term transgene expression can be achieved [ 11 , 12 ] and also many of the therapeutic genes for cancer treatment fall within the size limit dictated for rAAV. While vectors derived from AAV have shown great promise in the course of research into treatment of numerous indications ranging from cystic fibrosis to haemophilia B [ 13 , 14 ], only in recent years have they begun to be investigated in a cancer setting [ 15 - 18 ]. Granulocyte macrophage colony stimulating factor (GM-CSF) is a cytokine that acts as a critical factor for development and differentiation of macrophages and dendritic cells (DCs). Activation of T cells is enhanced by local GM-CSF mediated recruitment of DCs, allowing for the efficient uptake of antigens and presentation to T cells in the draining lymph node. Co-stimulatory molecules are essential for correct T cell activation and subsequent differentiation into effector T cells following their interaction with antigen presenting cells (APCs). The initial signal for activation is dependent on specific T cell receptor (TCR) recognition of the antigen presented by MHC molecules on APC. The second signal is delivered through the binding of co-stimulatory molecules expressed on the APC surface with their ligands on T cells. A lack of co-stimulatory signals allows tumour cells to induce antigen specific tolerance or anergy on the basis of MHC class I restricted presentation [ 19 , 20 ]. The CD28 receptor has been identified as one of the most important costimulatory receptors on T cells. The ligands for this receptor are members of the B7 family and include B7-1 (CD80) [ 21 , 22 ]. B7-1-transduced tumour cells are expected to present both the antigen and the co-stimulatory (CD28-mediated) signals to CD8 + CTL simultaneously, leading to efficient activation of CTLs without requiring the assistance of CD4 + helper T cells. Transfection/transduction with B7-1 has resulted in tumour cell rejection in several tumour models [ 19 , 23 - 26 ]. Studies have also demonstrated that cells modified to express GM-CSF or B7-1 can be used to induce protective, T cell-mediated immune responses. Different approaches have been taken for the modification of cells, including both ex vivo viral transduction of leukaemia cells [ 27 ] and non-viral delivery of the genes on plasmids to growing tumours [ 28 ]. For effective cytotoxic responses, in addition to effective education/priming of the immune system to tumour antigens, the local tumour environment must permit immune cell infiltration. Angiogenesis is the formation of new capillary blood vessels from existing microvessels which occurs in physiological and pathological states [ 29 ]. This process is controlled by numerous angiogenic factors that are able to attract endothelial cells from the surrounding tissues and represents a crucial stage in tumour growth and metastasis [ 29 , 30 ]. For cancer therapy, strategies based on the manipulation of angiogenesis are referred to as anti-angiogenic strategies and seek to prevent new vessel formation or to inactivate pre-existing vessels. Although angiogenesis is a discrete component of the tumour phenotype, it is often neglected by tumour immunologists. However, lymphocyte extravasation is tightly controlled by blood vessels and requires orchestration of multiple receptor-ligand interactions as well a favourable cytokine/chemokine micromilieu [ 31 ]. Moreover, ongoing angiogenesis induces profound morphological and molecular changes in tumour blood vessels and may thus contribute significantly to the tumour's intrinsic resistance to infiltration by immune cells. Therefore, effective tumour immune strategies require both fully armed effector cells and a tumour environment permissive for infiltration and destruction. The invasive and metastatic behaviour of tumour cells is regulated by extracellular growth factors like hepatocyte growth factor (HGF), which is a ligand for the c-Met receptor tyrosine kinase [ 32 , 33 ]. HGF is a heterodimeric molecule and functions of HGF include mitogenic, motogenic, morphogenic and anti-apoptotic activities [ 34 , 35 ]. In cancer, HGF stimulates malignant cell invasion behaviour through its binding to c-Met [ 32 , 33 , 36 ]. Nk4 (also known as IL32b) inhibits HGF-c-Met signalling and therefore tumour metastasis [ 36 , 37 ]. Nk4 also has an additional, independent function, promoting anti-angiogenic activities. This is achieved due to the make up of Nk4, which consists of the N-terminus of HGF, containing an N-terminal hairpin and four kringle domains (well described anti-angiogenic molecules) [ 38 - 41 ]. Nk4 augments anti-angiogenic activities through the competitive inhibition of binding of angiogenic growth factors such as VEGF, bFGF and HGF to endothelial cells by its N-terminus [ 36 , 42 , 43 ]. Angiogenesis-inhibitory as well as cancer-specific apoptosis inducing effects make the Nk4 gene an attractive candidate for gene therapy of cancer. The aim of this study was to assess AAV2 mediated delivery of the immune stimulating genes GM-CSF and B7-1 and Nk4 on different tumour models in vivo . Since immunotherapy has the potential to recruit a systemic immune response against tumour cells, and Nk4 treatment is known to inhibit angiogenesis and metastatic spread, a combination of these therapies may improve or replace traditional treatments currently available.

Materials and methods Vector constructs pAAV2-MCS (Stratagene) was used to generate reporter and therapeutic vectors and for the generation of AAV control particles. The mammalian expression vector pVivo1 was purchased from Invivogen (Cayla SAS, Toulouse, France). A version of this plasmid, designated pVivoGMCSF, B7-1, containing the murine GM-CSF and murine B7-1 genes transcriptionally controlled from two human glucose regulated protein (GRP) promoters GRP94 and hamster GRP78 promoters respectively was designed and cloning was performed on contract by Invivogen. An AAV plasmid encoding the GM-CSF, B7-1 expression cassette (pAAV2-GB) was constructed by excising the expression cassette from pVivoGMCSF, B7-1 using SspI and NheI and cloning the Klenow treated fragment into NcoI and XbaI sites of pAAV-MCS plasmid (Klenow treated). Inserts were confirmed by sequencing (MWG Biotech). The AAV2- Luc , AAV2- Nk4 and AAV-BB constructs have previously been described [ 44 ]. All constructs used in this study are illustrated in Figure 1 . Vector generation Recombinant AAV2 vectors (rAAV), AAV2-MCS, AAV2- GB , AAV2- Nk4 , AAV2-BB and AAV2- Luc were generated using the AAV Helper-Free System (Stratagene, Agilent, Dublin). rAAV particles were purified using the Virakit AAV Purification Kit (Virapur, San Diego, USA) per manufacturer's instructions. Purified AAV2- GB particles were used to transduce HT1080 cells and FACS analysis for B7-1 expression employed to determine the number of transducing units (TU). Purified AAV2-MCS, AAV2- Nk4 , AAV2-BB and AAV2- Luc preparations were titrated using real time PCR to determine the number of genome copies, using primers specific for the CMV promoter (forward: 5' aaatgggcggtaggcgtgta 3', reverse: 5' gatcggtcccggtgtcttct 3') and were synthesized by MWG Biotech, Germany. A fragment of length 124 bp is expected. Cell lines and tissue culture Murine JBS fibrosarcoma tumour cells [ 28 ] and murine Lewis Lung Carcinoma cells were maintained in culture at 37°C in a humidified atmosphere of 5% CO2, in Dulbecco's Modified Essential Medium (GIBCO, Invitrogen Corp., Paisley, Scotland) supplemented with 10% iron-supplemented donor calf serum (Sigma Aldrich Ireland, Ireland), 300 μg/ml L-glutamine. Cell densities were determined by visual count using a haemocytometer. Cell viability was confirmed by Trypan Blue Dye Exclusion (Sigma Aldrich Ireland, Ireland) to be > 95% for tumour induction. Human HT1080 fibrosarcoma cells were maintained in culture at 37°C in a humidified atmosphere of 5% CO2, in Eagle Minimum Essential Medium (GIBCO, Invitrogen Corp., Paisley, Scotland) supplemented with 10% iron-supplemented donor calf serum (Sigma Aldrich Ireland, Ireland), 300 μg/ml L-glutamine. In vitro transduction Cells were seeded in a 12-well plate (HT1080 at 2 × 10 5 , JBS at 5 × 10 4 cells per well, LLC at 1.5 × 10 5 cells per well) in complete medium 24 h before transduction. On the day of transduction, cells were 80% confluent. 9 × 10 8 genome copies (GC) of AAV2- Luc or 7 × 10 5 transducing units (TU) of AAV2- GB in a 0.5 ml volume of transduction medium (DMEM, 2% FBS) were added to individual wells. The plates were incubated for 2 h at 37°C, 5% CO 2 with gentle rocking at 30 min intervals during the incubation. 0.5 ml post infection medium (DMEM, 18% FBS) was added to each well and incubated at 37°C, 5% CO 2 for a further 24 h. Flow-cytometric analysis and ELISA of transduced cells Cell surface expression of B7-1 was detected by flow cytometry using a FACScan (Becton Dickinson, San Jose, CA) with CD80-specific antibody, clone L307.4 (BD Biosciences UK Ltd, Oxford, UK). Briefly AAV2- GB transduced and mock-infected cells were harvested 48 h post transduction. The cells were labelled with the CD80-specific antibody, an isotype control antibody F (ab') 2 Goat Anti Rat IgG: RPe Mouse ADS (Serotec) or unlabeled. 10,000 events were acquired and analyzed for PE fluorescence. PE was measured on the FL2-channel (short band pass 575 nm filter) and plotted against side scatter. Cells without a conjugated antibody and cells with an irrelevant antibody conjugated antibody were used as controls, thereby correcting for background fluorescence. Production of GM-CSF from JBS cells was quantified by enzyme-linked immunosorbent assay (ELISA) (Quantikine Mouse GM-CSF Immunoassay R&D Systems, Minneapolis, MN). For quantification of GM-CSF production in transduced cells, AAV2- GB transduced and untransduced cell supernatant was harvested 48 h post transduction and the assay was carried out as per the manufacturer's protocol. Animals and tumour induction Mice were obtained from Harlan Laboratories (Oxfordshire, England), and kept at a constant room temperature (22°C) with a natural day/night light cycle in a conventional animal colony. Standard laboratory food and water were provided ad libitum . Before experiments, mice were afforded an adaptation period of at least 14 days. Female Balb/C or C57Bl/6 mice in good condition, without fungal or other infections, weighing 16-22 g and of 6-8 weeks of age, were included in experiments. For routine tumour induction, 2 × 10 6 JBS cells or 5 × 10 5 LLC cells suspended in 100 μl of serum free DMEM or were injected subcutaneously (SC) into the flank. Following tumour establishment, tumours were allowed develop and monitored mostly by alternate day measurements in two dimensions using a Verniers Callipers. Tumour volume was calculated according to the formula V = ab 2 Π/6, where a, is the longest diameter of the tumour and b is the longest diameter perpendicular to diameter a. From these volumes, tumour growth curves were constructed. In cases of successful treatment, 100 days with no recurrence was considered a cure. In the case of recurrence, the animal was considered incurable and humanely euthanized when the tumour diameter was between 1.5 - 2 cm. Survival time extended from the time of first treatment to 100 days (successful treatments) or to sacrifice (recurrences). In vivo gene delivery All animal experiments were approved by the ethics committee of University College Cork. Mice were randomly divided into experimental groups and subjected to specific experimental protocols. For tumour experiments, mice were treated as soon as the tumour could be reliably injected (tumour diameter = 0.4 cm on average). For quadriceps muscle experiments, a single intramuscular injection was carried out into the right or left thigh of the animal. Mice were anaesthetized during all treatments by intraperitoneal (IP) administration of 200 μg xylazine and 2 mg ketamine. Viral vector particles were administered by direct intratumoural (IT) or intramuscular injection (IM) in a volume of 50 μl 2 × 10 8 - 2 × 10 9 GC of replication incompetent recombinant AAV2 particles. In vivo confirmation of Nk4 gene delivery and expression Muscle tissue from animals treated by IM injection of AAV2- Nk4 and untreated animals was excised at day 3. The muscle tissue was passed through a nylon membrane in order to disassociate the tissue and create a single cell suspension. The cells were precipitated by centrifugation, the DNA and RNA was simultaneously extracted from the cell pellet using the Qiagen Allprep DNA/RNA kit as per the manufacturers protocol. DNA and RNA concentration was determined using the nanodrop. AAV mediated delivery was confirmed by PCR and AAV mediated gene expression was confirmed by rtPCR. The primers were against the Nk4 sequence Forward: 5'CCTCTCTGATGACATGAAGAAG 3', Reverse: 5'TGTCACAAAAGCTCTCCCC 3'. PCR conditions were as follows HotstarTaq Activation 95°C-15 min, Denaturation 94°C-1 min, Annealing 59°C -1 min, Elongation 72°C-1 min. Nk4 DNA was detected by PCR in 50 ng of DNA using HotstarTaq Master Mix Kit (Qiagen) in a Mastercycler (Eppendorf,, UK) PCR machine. The PCR products were visualised on a 1% agarose gel. Nk4 expression of transduced muscle was confirmed by rtPCR. Extracted RNA was DNAse treated using Ambion DNAfree kit according to manufacturer's instructions. RNA concentration was determined using the nanodrop. Omniscript RT kit (Qiagen) was used to generate cDNA from 100 ng of total RNA in a 20 μl volume according to manufacturer's instructions. The cDNA was diluted to a final volume of 50 μl following cDNA synthesis using DNasefree H 2 O. 5 μl diluted cDNA was PCR amplified using HotstarTaq Master Mix Kit (Qiagen) in a Mastercycler (Eppendorf, UK) PCR machine. The PCR products were then visualised on a 1% agarose gel. Luminescence measurements For in vitro experiments, treated cells were analysed for luciferase activity 48 h post transduction using the Luciferase Assay System (Promega MSC, Dublin), as per manufacturer's instructions. Luminescence was measured using the IVIS Imaging System (Xenogen, UK). In vivo luciferase activity from tissues was analysed post- transduction as follows: 80 μl of 30 mg/ml firefly luciferin (Biosynth, Basil, Switzerland) was injected intraperitoneally (IP) and intratumourally (IT). Mice were anaesthetised as before. Ten minutes post-luciferin injection, live anaesthetised mice were imaged for 3 min at high sensitivity using the IVIS imaging system (Xenogen, UK). In vivo cytotoxicity assay The development of an immune-mediated anti-tumoural activity following treatment was tested by in vivo cytotoxicity assay [ 45 ]. The Winn assay was utilised as follows: mice (six/group) received injections of a mixture of JBS cells and splenocytes from either AAV2- GB cured mice or naive mice. Splenocytes were taken 100 days post tumour regression from 'cured' mice for use in Winn assays. Splenocytes were mixed with tumour cells and injected SC in a proportion of 50:1 (10 8 spleen cells to 2 × 10 6 JBS cells). Mice were then monitored on alternate days for tumour development. Statistical Analysis The primary outcome variable of the statistical analyses was the tumour volume in each mouse measured at each time point. The principal explanatory variables were the different treatment groups. Tumour volume was analyzed as continuous. Treatment groups were analyzed as categorical variables. At each time point, a two-sampled t-test was used to compare mean tumour volume within each treatment group depending on the number of groups being compared. Microsoft Excel 11.0 (Microsoft) and GraphPad Prism Version 4.0 (GraphPad Prism Software Inc, San Diego, CA, USA) were used to manage and analyze data. Statistical significance was defined at the standard 5% level. Survival was analysed using a two-sampled Student's t-test assuming equal variances to compare the average number of days survived per group.

Results Validation of vector constructs and gene expression Flow Cytometric analysis of cell surface expression of B7-1 and ELISA for GM-CSF confirmed the functionality of AAV2- GB particles in vitro . The human HT1080 fibrosarcoma cell line was used, being the standard cell line for AAV transduction assays. HT1080 cells were transduced with AAV2- GB or mock transduced with PBS. After 48 h, cells and supernatant were harvested for assays. Cells were labelled with anti-CD80 antibody, and the resulting overlay graph (Figure 2a ) demonstrated an increase of 38.2% in B7-1 expression in transduced cells (light grey overlay peak) in comparison cells labelled with an isotype (dark grey peak). GM-CSF protein was detected in cell culture supernatant in cells transduced with AAV2- GB at a level of 250 pg/ml and not in mock-infected cells (Figure 2b ). The efficiency of AAV2 mediated transduction of each of the test cell lines was determined in cells transduced with either AAV2- GB or AAV2- Luc particles. FACS analysis for cell surface expression of B7-1 confirmed transduction of both JBS (Figure 2c ) and LLC (Figure 2d ). These graphs demonstrate that JBS cells (13.4% increase) are more permissive to transduction with AAV2 than LLC cells (4.25% increase). Also evident from these data is that there is a low level of endogenous cell surface B7-1 expression in untreated LLC cells, which has also been reported by other groups [ 46 ]. A lower level of background B7-1 expression was observed in JBS cells. The efficiency of AAV2 mediated transduction of growing JBS and LLC tumours was also assessed. AAV2- Luc was administered IT to SC tumours and luciferase expression assessed using the IVIS system on day 7-post administration. Luminescence was detected in JBS tumours in Balb/C mice at 9.7 × 10 -1 p/sec/cm 2 /sr/gene copy administered (Figure 2g ) and in LLC tumours in C57Bl/6 mice at 1.64 × 10 -8 p/sec/cm 2 /sr/gene copy administered (Figure 2g ). In order to confirm transgene expression from AAV2- GB transduced tumours in vivo , LLC tumours were excised 7 days after IT delivery of AAV2- GB or PBS. Cell surface expression of B7-1 was detected by flow cytometry as previously described. Results indicated that administration of AAV2- GB resulted in an increase in cell surface B7-1 expression. A background level of B7-1 expression of approximately 10% was seen in PBS treated LLC cells while a 5.2% (+/- 1.48) increase in B7-1 positive cells was observed in AAV2- GB transduced LLC cells (Figure 2f ). AAV mediated immune gene therapy of tumours in vivo The AAV2- GB construct was used to deliver GM-CSF and B7-1 to JBS or LLC tumours. The JBS study consisted of three groups (n = 5): an AAV2- GB treated group, an AAV2 null vector treated group (AAV2- Luc vector), and an untreated group. The tumour growth curve (Figure 3a ) illustrates a significant decrease in tumour growth rate in those groups treated with AAV particles expressing GM-CSF and B7-1 genes in comparison with untreated and null vector treatment groups. There was a significant reduction in tumour growth on day 21 between the AAV2- GB treated group and the null vector treatment group (p = 0.028), confirming that tumour regression involved the therapeutic genes encoded by the particles and was not due to a response to the particle alone. The survival curve (Figure 3b ) illustrates a significant increase in survival for all mice treated with GM-CSF, B7-1 in comparison with the untreated controls (p < 0.0008). The treatment resulted in a cure for 60% of the treated animals. The LLC study consisted of three groups (n = 6): an untreated control group, an AAV null vector (AAV2-BB) group and an AAV2- GB treated group. The tumour growth curve (Figure 3c ) illustrates a marked decrease in tumour growth in the AAV2- GB treated group in comparison with the untreated and null vector controls. The reduction in tumour growth was significant on days 20 - 27 (p < 0.02) between AAV2- GB treated mice and both the untreated and null vector groups. There was no significant difference between the untreated and the null vector groups. The survival curve (Figure 3d ) illustrated a significant increase in survival in animals treated with GM-CSF, B7-1 in comparison with the untreated and null vector groups. Although the increase in survival was significant (p = 0.008), the therapy did not result in cure in any of the treated animals. Immunological memory following tumour treatment In cases where complete tumour regression occurred (60% JBS treated mice), 'cured' mice were rechallenged to assess for sustained anti-tumoural immunological responses. Mice were injected SC on the opposite flank to the original tumour challenge, with tumourigenic doses of the same tumour type (JBS) 30 days following tumour regression. AAV2- GB 'cured' mice remained tumour free to 100 days whilst all naïve mice developed tumours and were culled due to tumour burden by day 28 (Figure 3e ), indicating immunological memory to tumour antigens. In order to examine for a cell-mediated immune response as a result of AAV2- GB treatment, the cytotoxic activity of lymphocytes from treated mice was examined in vivo using a modified Winn assay [ 45 ]. Groups of mice received injections of a mixture of a tumourigenic dose of JBS cells and splenocytes from either AAV2- GB cured mice or naïve mice. All mice receiving splenocytes from 'cured' mice failed to grow tumours, while JBS tumours developed in all control animals receiving splenocytes from naïve mice (Figure 3f ), indicating adoptive transfer to naïve mice of anti-tumour lymphocytes conferring resistance to further tumour challenge. Nk4 therapy of growing subcutaneous LLC tumours AAV2 vector mediated transgene expression is known to be delayed initially before increasing over time [ 47 - 49 ]. Given the short therapeutic window available with the fast growing LLC tumour model, and since the Nk4 transgene product is secreted by transduced cells, we opted to administer AAV2- Nk4 prior to tumour induction, to quadriceps muscle, with the aim of producing systemic circulating Nk4 protein. The temporal pattern of AAV2-mediated expression from quadriceps was examined using AAV2- Luc (Figure 4a ). Expression was observed to increase up to day 14 and remain higher thereafter. For tumour experiments, quadriceps muscles were transduced with AAV2- Nk4 14 days prior to LLC tumour induction and tumour growth monitored on alternate days (Figure 4c ). Nk4 delivery and expression was confirmed by PCR (Figure 4b ). For animals treated with AAV2- Nk4 there was a significant reduction in tumour growth compared with the untreated group (p = 0.048 on day 20), although this significance was not seen when compared with the null vector treated group. The Kaplan-Meier curve (Figure 4d ) illustrates an increase in survival in animals treated with Nk4 , although not reaching statistical significance in comparison with the untreated (p = 0.06) or null vector groups (p = 0.26). Combination of immune and anti-angiogenic therapies In an effort to enhance both the immune and Nk4 protocols, a combined therapy was examined. The study consisted of 5 groups: an IT AAV2- GB /IM AAV2-BB treated group, an IT AAV2-MCS/IM AAV2- Nk4 treated group, a combined IT AAV2- GB /IM AAV2- Nk4 treated group, an IT AAV2-MCS/IM AAV2-BB treated group (null vector), and an untreated group. The tumour growth curve (Figure 4e ) again illustrates a marked decrease in tumour growth in groups treated with AAV expressing GM-CSF , B7-1 or Nk4 genes alone in comparison with the untreated and null vector groups. However, treatment with AAV2- Nk4 prior to immune therapy, eliminated the anti-tumour effects of AAV2- GB treatment (Figure 4e ), with tumour growth in this group similar to controls (p vs. untreated p = 0.52, p vs. null vector control p = 0.38).

Discussion Though certain viral vectors can elicit strong immune responses and systemic toxicity [ 50 ], gene transduction efficiency is extremely high. While non-viral delivery of plasmid DNA displays lower toxicity, a major obstacle that has prevented its widespread application is its relative inefficiency in gene transfection [ 51 ]. The construct used in this study encoded GM-CSF and B7-1 , and was designed such that GM-CSF would be secreted from the tumour cell with B7-1 expressed on the cell surface in an effort to elicit an anti-tumour immune response. We have previously demonstrated that non-viral delivery of these immune genes on a plasmid to JBS tumours leads to immune stimulation and consequent eradication of the treated tumour and associated metastases, when delivered by electroporation [ 28 , 52 ] or sonoporation [ 53 ]. Other work from our laboratory has shown that AAV2 mediated reporter gene expression in JBS tumours is significantly higher and more sustained than plasmid-mediated delivery. We decided to examine if the anti-tumour efficacy of GM-CSF, B7-1 could be improved by AAV2 delivery. The results achieved here were comparable with those observed with both non-viral strategies, suggesting that levels of immune gene expression may not be the limiting factor in recruitment of anti-tumour responses. However, it should be noted that a different temporal pattern of transgene expression is observed with AAV and plasmid vectors. Gene expression from plasmid vector is maximal 24 - 48 h post transfection, while AAV2 related transcription is delayed, taking 4 - 7 days to surpass plasmid levels. Given the relatively short window of therapeutic opportunity permitted by the murine tumour models used here, it is plausible that superior effects might be observed in clinical situations, where patient tumour growth is slower, thereby facilitating AAV vector transcription levels to reach full capacity [ 8 , 47 - 49 ]. Also, the range of patient tumour locations and sizes amenable to AAV vector administration is far wider than is practical for delivery involving electroporation or sonoporation equipment, which are currently useful only for accessible subcutaneous tumours. Furthermore, the potential for specific transduction of tumour cells following systemic administration of viruses has been validated, involving the inclusion of tumour specific targeting ligands to viral vector surfaces [ 54 , 55 ]. Our studies demonstrated that the observed tumour reductions were immune mediated, and that the immune response induced was, at least in part, T cell mediated. Adoptive transfer of the anti-tumour immune response to naïve animals prevented tumour establishment. The immune gene therapy was less effective in LLC than JBS and this might be attributed to a number of different reasons. Both of the tumour cell lines can be described as weakly or non-immunogenic. The JBS murine fibrosarcoma is derived from the 3T3 cell line and is described previously [ 56 ]. JBS grows in immunocompetent Balb/C at the same rate as in athymic mice, and vaccination strategies, using a diversity of approaches, schedules and cell treatments, failed to protect Balb/C mice from subsequent tumourogenic JBS challenges [ 28 , 56 ]. The well characterised LLC cell line has previously been shown to be weakly or non-immunogenic, with a number of approaches such as UV irradiation [ 57 ] or viral infection [ 58 ] being used to increase their immunogenicity in previous studies. As demonstrated both in vitro and in vivo , AAV2 transduces JBS cells more efficiently than LLC. The AAV2/2 serotype used in our studies has a reported 30% efficiency of transducing LLC in vitro [ 59 ]. We observed an even lower efficiency (data not shown). Different serotypes such as AAV2/5 have a 65% transduction efficiency in LLC [ 59 ]. Also, as demonstrated here, LLC cells have an endogenous B7-1 expression. It is possible that low level B7-1 expression following transduction with AAV-GB in LLC cells could result in preferential binding of B7-1 to CTLA-4 on Treg cells rather than CD80 on effector cells resulting in a subsequent tumour growth advantage, as demonstrated by Tirapu et al [ 46 ]. However, such an effect was not apparent in our studies, with the moderate increase in CD80 expression combined with GM-CSF through gene transduction sufficient to significantly reduce tumour growth. An increase in cell surface B7-1 expression breaks the immune tolerance allowing B7-1 to bind preferentially to its main receptor CD28 on antigen presenting cells (rather than Treg) creating an anti-tumour response. We have previously demonstrated that simultaneous depletion of Treg further improves immune therapy [ 60 ]. For both tumour types examined here, it is plausible that a minimal threshold of the percentage of tumour cells expressing GM-CSF and B7-1 is necessary for this system to effect complete tumour regression. Intratumoural distribution of transduced cells could also be important. In AAV2 treatment, it is likely that a large portion of the tumour is untransduced, as the transduction region is limited to the needle track. In an attempt to maximise transduction efficiency, we used multiple injections of AAV2 vector in an attempt to saturate the tumour with vector solution. This notwithstanding, it is unlikely that there is a requirement for transduction of every tumour cell, as the mechanism of tumour regression is immune mediated, whereby targeting of non-transduced tumour cells is achieved after the gene therapy induced immune sensitisation. Due to the nature of the AAV2- Nk4 construct, a null vector that encoded the blastocidin antibiotic resistance gene was used as a control. In contrast, the AAV2- GB construct includes no extra coding sequences. In this case, the control vector/s used included the AAV2-MCS or AAV2- Luc . Firefly luciferase is generally accepted to be non-immunogenic [ 61 ], and was used here for monitoring of vector-mediated expression in tumours during experiments. Specifically in relation to JBS tumours, growth curves also illustrate that there is a moderate, but statistically insignificant reduction in tumour growth for those tumours treated with the null vector (AAV2- Luc ) in comparison with the untreated group. No such tumour reduction in tumour growth was seen in LLC tumours treated with the null vector (AAV2-BB) in comparison with the untreated LLC group. In this study, we also assessed the effects on tumour growth of combining anti-angiogenic treatment with immune gene therapy. The process of angiogenesis provides an ideal target for treatment, has been studied extensively and anti-angiogenic agents are in clinical use [ 62 ]. Solid tumours create a unique microenvironment, featuring chaotic vasculature, resulting in zones of tumour ischaemia and poor intratumoural circulation, which can prevent immune cell access [ 31 , 63 - 65 ]. In recent years, evidence is accumulating that the immune and vascular environments are closely linked. Several studies have indicated that vascular components of the tumour stroma are targeted during immune-mediated tumour rejection [ 31 ]. Furthermore, recent reports indicate that certain anti-angiogenic therapies, rather than eliminating vasculature in the tumour and 'starving' cells, serve to normalise microcirculatory function, permitting access to the tumour for immune cells [ 63 ]. By developing a deeper understanding of the activity in vivo of various anti-angiogenic agents, new improved therapeutic regimes may be developed by addressing simultaneous anti-angiogenic activity and immune up-regulation, with more therapeutic potential than either therapeutic approach alone. We have previously shown that IT administered AAV2- Nk4 did not significantly reduce SC LLC tumour growth [ 44 ]. In the current study, systemic production of Nk4 induced prior to tumour establishment provided superior anti-tumour immune responses to IT delivery, with a reduction in tumour burden and an increase in survival in AAV- Nk4 treated animals, although differences were not statistically significant. However, the combination studies described here showed that Nk4 therapy eliminated the efficacy of the immune therapy. In our experiments involving prior expression of anti-angiogenic agent, it is possible that Nk4 mediated reduction in tumour neovasculature may act as an obstacle to immune therapy by preventing migration of immune effector cells into established tumour parenchyma [ 31 , 64 , 65 ]. The schedule and timing of combined treatment may be key in achieving improved responses with such an approach. In addition, since Nk4 is a multifunctional molecule, we cannot rule out the possibility that its activity other than anti-angiogenic, might play a role in the observed responses. The design of the experiment addressed the possibility that pre-exposure to the first AAV vector IM precluded gene transfer with the IT vector one week later due to any potential immune response to the vector. The experimental group that received AAV2- GB IT also received AAV-BB IM one week previously, and a significant reduction in tumour growth was still observed, ruling out an immune reaction to AAV preventing expression of the immune genes. Furthermore, as indicated in Figure 4a , IM delivery of AAV2- Luc resulted in long-term gene expression indicating the absence of immune responses to AAV-transduced muscle cells. Much study is required with respect to elucidating the complex cascade of events that is needed to reduce angiogenesis while permitting effector cell entry into tumours. Researching these mechanisms will enable cancer biologists to specifically target the tumour environment and further improve therapeutic efficacies. Together with overcoming the obstacle of inefficient effector cell generation in cancer patients, 'angio-immuno' therapy may provide new opportunities to permit tumour infiltration and destruction. Exploiting the benefits of gene therapy, especially utilising viral vectors such as AAV, in terms of local activity and duration of therapeutic activity, may overcome current obstacles to successful cancer treatment with systemically administered drugs.

Background Many strategies have been adopted to unleash the potential of gene therapy for cancer, involving a wide range of therapeutic genes delivered by various methods. Immune therapy has become one of the major strategies adopted for cancer gene therapy and seeks to stimulate the immune system to target tumour antigens. In this study, the feasibility of AAV2 mediated immunotherapy of growing tumours was examined, in isolation and combined with anti-angiogenic therapy. Methods Immune-competent Balb/C or C57 mice bearing subcutaneous JBS fibrosarcoma or Lewis Lung Carcinoma (LLC) tumour xenografts respectively were treated by intra-tumoural administration of AAV2 vector encoding the immune up-regulating cytokine granulocyte macrophage-colony stimulating factor ( GM-CSF ) and the co-stimulatory molecule B7-1 to subcutaneous tumours, either alone or in combination with intra-muscular (IM) delivery of AAV2 vector encoding Nk4 14 days prior to tumour induction. Tumour growth and survival was monitored for all animals. Cured animals were re-challenged with tumourigenic doses of the original tumour type. In vivo cytotoxicity assays were used to investigate establishment of cell-mediated responses in treated animals. Results AAV2-mediated GM-CSF, B7-1 treatment resulted in a significant reduction in tumour growth and an increase in survival in both tumour models. Cured animals were resistant to re-challenge, and induction of T cell mediated anti-tumour responses were demonstrated. Adoptive transfer of splenocytes to naïve animals prevented tumour establishment. Systemic production of Nk4 induced by intra-muscular (IM) delivery of Nk4 significantly reduced subcutaneous tumour growth. However, combination of Nk4 treatment with GM-CSF, B7-1 therapy reduced the efficacy of the immune therapy. Conclusions Overall, this study demonstrates the potential for in vivo AAV2 mediated immune gene therapy, and provides data on the inter-relationship between tumour vasculature and immune cell recruitment.

Competing interests The authors declare that they have no competing interests. Authors' contributions SAC performed the in vitro and in vivo experiments, and contributed to drafting the manuscript. AB constructed AAVNk4, AAVBB. MFS, PTH, GCO'S, DMO'H and MT were the coordinators of the project. MT designed the studies and drafted the manuscript. All authors read and approved the final manuscript.

Acknowledgements This work was supported through grant funding from Science Foundation Ireland (06/RF/BIC055) and Cancer Research Ireland (CRI07TAN).

CC BY

no

2022-01-12 15:21:45

Genet Vaccines Ther. 2010 Dec 20; 8:8

oa_package/57/08/PMC3016353.tar.gz

PMC3016354

21129195

Background Clinical practice guidelines (CPGs) are designed to support decision-making processes in patient care and are classically defined as "systematically developed statements to assist practitioners' and patients' decisions about appropriate healthcare for specific circumstances" [ 1 ]. While there is some evidence that CPGs improve outcomes when they are appropriately implemented, [ 2 , 3 ] there are numerous factors which influence their acceptability and use by healthcare providers [ 4 ]. Some of the major barriers to guideline adherence can be related to practitioner, patient, organisational or guideline factors, [ 5 - 7 ] although practitioners appear to be a key factor. Aspects such as knowledge, perceptions of and attitudes toward CPGs in general, are professional-related. Other factors that affect the success of guideline implementation are content or format related, like clarity and presentation of the recommendations or the existence of a quick reference guide or algorithm [ 8 ]. The grading system used to classify the quality of the evidence and the strength of recommendations is an important characteristic of guideline format. However, the growing number of different grading systems employed across different institutions [ 9 ] could possibly be a source of confusion, impeding the correct comprehension of recommendations and their application in clinical practice. The evidence on clinicians' knowledge, perceptions and attitudes toward CPGs is growing. The majority of published studies on this topic used quantitative design (surveys or record audits), investigating the association between barriers or facilitators and CPG use. Alternative approaches used by some investigators were qualitative designs, where concepts are collected in focus groups, interviews or through open-ended questions in questionnaires. Finally, some studies considered a mixed-method design including both qualitative and quantitative techniques. Survey results were systematically reviewed in two publications, [ 4 , 10 ] concluding that most clinicians were supportive of CPGs, finding them to be useful, educational and likely to improve quality of care. Less frequent findings were that CPGs were impractical, unable to be used for individual patients, limited clinician autonomy, increased the likelihood of litigation or disciplinary action, and were used to cut costs [ 10 ]. A recent systematic review and meta-synthesis of qualitative studies of general practitioners' (GPs') attitudes to and experience with the use of CPGs [ 11 ] offers a conceptual framework for their interpretation. Six broad themes were identified in the analysis: questioning the guidelines; GPs' experience; preserving the doctor-patient relationship; professional responsibility; practical issues and guideline format. No systematic pattern in the distribution of these themes was found. However, comparative analysis and synthesis suggested that GPs' reasons for not following the CPGs differed according to whether the guidelines encouraged (prescriptive type) or discouraged (proscriptive type) particular interventions or behaviours [ 11 ]. Some authors suggest that clinicians find guideline contents too complex and confusing, and perceive difficulties in understanding the CPG development process. More specifically, there is very little research on clinicians' perceptions and understanding of the grading systems used to rate the evidence, quality, and strength of recommendations [ 12 , 13 ]. In our context, a qualitative study about the GRADE system showed important difficulties in understanding it and an important level of disagreement among the participating clinicians, with and without previous experience in CPG development [ 14 ]. However, the participants had hardly any experience with the use of this grading system, which requires substantial methodological expertise and practice. Given this context there is a need for systematically reviewing the evidence on this topic and an assessment of the situation in our context. Additionally, the scarce information about the perceptions and understanding of the grading systems by healthcare professionals in the international literature provide the impetus for an investigation, both qualitative and quantitative, about this important aspect of clinical practice guidelines.

Methods and design Goal and objectives The main goal of this study is to explore the perceptions and attitudes of Spanish practitioners (GPs and hospital-based specialists, both with or without previous experience in guideline development) towards clinical practice guidelines and grading systems. The specific objectives are to: - Categorize, synthesize, and compare the perceptions of these different groups of healthcare professionals toward CPGs; - Identify and describe the adoption/non-adoption factors specific to each group; - Investigate whether different grading systems are better understood, and the influence of the wording of recommendations on understanding, acceptability and use of CPGs; - Investigate whether different formats of CPGs influence acceptability and use of CPGs; - Explore and compare the level of knowledge and understanding of different grading systems in these groups. The design of this two-phase project will include qualitative and quantitative research techniques. During Phase I, with a qualitative design, group discussions will be carried out with the participation of GPs and hospital-based specialists, to explore in depth their perceptions and attitudes towards CPG and grading systems. In a posterior quantitative stage, Phase II, we will complement this information by means of a survey. The study has been approved by the Ethics Board of the Hospital de la Santa Creu i Sant Pau, Barcelona, Spain. Phase I: Qualitative study Qualitative research methods provide in-depth knowledge concerning perceptions, beliefs and values of the persons and groups involved and have proved to be extremely useful in the healthcare field [ 15 , 16 ]. These methods provide researchers with the opportunity to explore potential hypotheses on why participants hold the views they do. We thus considered this approach, and more specifically group discussions, as the most appropriate technique to gather information for our study objectives. The added value of group discussions [ 17 , 18 ] is that they permit us to explore dynamic interactions among group members in the light of certain cultural characteristics, concrete values, and references that may underlie individual preferences and shape the general position of the participants. The information gathered from group discussions will be analyzed following the sociological analysis disclosure model [ 19 , 20 ]. In this technique the reading and organization of data does not rely on the fragmentation of texts but on the interpretation of the different discursive positions of the participants and the main explanatory axes of their interventions [ 21 ]. 1. Group discussions structuring and participants The basic topics to explore during the group discussions will be pre-selected by a group of experts (medical doctors, psychologists and sociologists with wide experience in CPGs development and implementation). These topics will be based on some of the most relevant and controversial issues around the development and implementation of CPGs in clinical practice, such as confidence, utility, knowledge and understanding of CPGs. Emphasis will be given to the importance of grading systems. Once these areas have been agreed upon, a semi-structured manual will be developed to ensure similarity in the moderation of the sessions. A total of six groups are expected to be run: three groups of GPs (one group with, and two groups without previous experience in CPG development) and three groups of other hospital-based specialists (same distribution as GPs regarding previous experience). By "previous experience" in CPG development we will consider having participated in a guideline development group. Each group will consist of 8-9 clinicians, as generally recommended, [ 22 ] but not less than six. In these cases the sessions will be canceled and rescheduled. Participants will be selected taking into account their structural inter-group representation: medical specialty and years of experience, place of work, age and sex [ 18 ]. We will gather a purposefully selected sample, choosing participants according to their representativeness and capacity to provide rich and varied information in relation to the study objectives. On the other hand, the sample size will be considered sufficient when the information obtained has reached saturation, i.e., when information is repeated, redundant and provides no new aspects. The group discussions will be carried out in four Spanish regions: Andalusia, Aragon, Catalonia and Madrid. We will recruit potential participants by means of a formal letter of invitation, or e-mail, and their travel costs will be reimbursed. Each group discussion will last approximately 60-90 minutes and will be recorded, with the consent of the participants. 2. Moderation of group discussions All investigators involved in group discussions will receive training in qualitative research methods, and on focus groups technique in particular. In order to achieve optimal standardization of the sessions, the previously developed semi-structured manual will be used. Still, its content might be a subject of constant adjustments depending on the findings that arise during the development of the group discussions [ 23 ]. An interviewer and a collaborator will be involved in moderating each group discussion. The role of the interviewer will be to lead the discussion by introducing the subjects of interest and managing the group dynamics so as to maintain focus and cover the programmed issues. We will take great care to avoid sharing opinions on the topic. The role of the collaborator will be to observe group dynamics and annotate any elements of the discussion, such as non-verbal language, responses to the interviewer's interventions, etc. The interviewer will introduce the study objective indirectly, with the objective of analyzing the group's approach to the subject, and explain the need to record the session. Assurance of confidentiality will be provided to all group members, as well as of the independent nature of the study, explaining clearly to the potential participants that knowledge and quality of clinical practice will not be evaluated. Some basic demographic information will be collected before the beginning of each discussion group. Group sessions will be designed to be flexible and open to possible new topics and issues raised by participants. The two investigators will meet together after finishing the group meetings. This information will be analyzed in conjunction with the data obtained in the group discussions [ 23 ]. 3. Data organization, validation and analysis All information generated will be audio-taped and stored digitally, thus representing the 'information corpus'. Audio-tapes will be transcribed literally and completed by the information collected by the interviewers and observers. To guarantee accuracy, transcriptions will be verified by the moderators of each group. For final data validation, the participants will be invited to read and comment on transcription contents [ 24 ]. Themes and patterns will then be identified and coded by two independent investigators, keeping in mind the texts as a whole and the contexts in which they have been produced. Each step in the configuration of the potential explicative axes requires new purposeful and iterative reading of texts to legitimate and corroborate the interpretation of findings [ 25 ]. Thus, to assure credibility, the data of the first group interviews will be re-assessed in the light of new findings. In the final stage of results interpretation the main explanatory axes will be organized and contrasted with the empirical reference of texts, in order to select the most representative verbatim statements. The investigators will develop a list of the identified concepts. The process and provisional analysis will be triangulated among investigators and, in case of divergence; consensus will be obtained through discussion. Phase II: Survey 1. Questionnaire design We will elaborate a single questionnaire divided into three general sections. The first part will collect information on the demographic characteristics and professional profile of participants (such as sex, age, qualifications and work experience). The second and third parts of the questionnaire will evaluate practitioners' knowledge, perceptions of and attitudes toward CPGs and grading systems. The information obtained in the group discussions (phase I) will be used as a starting point in the development of the questionnaire, providing insight into the social and symbolic representation of CPGs in our local circumstances. Previous to its application, the questionnaire will be piloted with a group of practitioners in order to modify or eliminate confusing and contradictory questions. 2. Study sample We have estimated a sample size of 600 respondents needed to answer one of the specific objectives of the survey (adequate comprehension of grading systems) in the GPs group, with a precision of 4%. To obtain this precision we dichotomized one item of the survey (adequate comprehension: yes/no) assuming maximum variability (50% of affirmative responses). A confidence interval of 95% will be applied to the percentage of affirmative responses. 3. Recruitment of participants Approximately 1200 clinicians working in Primary Care or Hospital Departments of the four regions participating in the study will be contacted. The questionnaire will be sent via e-mail to all potential participants in the two subgroups of interest (GPs and specialists) with a letter inviting them to collaborate and a link to an electronic version of the questionnaire. Three reminders will be sent to those who have not responded within two weeks. An investigator's contact details will be provided for participants to raise questions or doubts about the questionnaire or the project. To potentiate practitioners' participation in the survey and to guarantee a high response rate, we will apply the following strategies which have proved to be successful: [ 26 ] the questionnaire will be concise and easy to complete (approximate duration 10-15 minutes); we will assure confidentiality of all personal data and will offer the participants the possibility to receive the final results of the survey. 4. Analysis For qualitative outcomes, between-group comparison will be described by means of contingency tables and inferences with the Chi Square Test or Fisher's Exact Test. Quantitative outcomes will be described using means and their corresponding standard deviations. Inferences will be made using an analysis of variance (ANOVA). If we encounter clearly defined application problems (abnormality and heteroscedasticity) the corresponding non-parametric test will be used. All analyses will be conducted using a bilateral approach. We will use the usual 5% significance level (α = 0.05) and the SPSS statistical package, version 11.5, to run the proposed analyses.

Discussion Clinicians' knowledge, perceptions and attitudes toward CPGs are important aspects towards the necessary implementation of these tools and subsequent change in clinical practice. To date, these aspects have been systematically reviewed in three main papers [ 4 , 10 , 11 ]. A number of later studies surveyed clinicians' knowledge about and attitudes towards CPG in general in different clinical settings: among general practitioners (GPs), [ 27 - 30 ] specialists, [ 28 , 31 - 34 ] and healthcare providers in Intensive Care Units [ 35 - 37 ]. Generally, most findings reported in these studies confirm that higher familiarity with the guidelines is related to more positive attitudes towards them and to more frequent reported use of CPGs. Guidelines concerning both diagnosis and treatment of common diseases, developed on a national level and adapted locally, presented in summarized format and transmitted directly by the practitioners of the department, have the best chance of being used, as Riou et al. observed in a described setting [ 32 ]. On the other hand, the predictive role of characteristics, such as age, personality traits and professional qualifications, in professionals' attitudes towards CPGs is still controversial [ 31 , 36 - 38 ]. Additionally, several studies have focused on the factors which influence the use and acceptance of a specific guideline in a local context, [ 39 - 44 ] making the generalization of their results difficult. Carlsen et al. [ 45 ] carried out focus groups amongst Norwegian GPs and add two important findings. First, the participants were concerned that guideline recommendations may be more heavily influenced by economic considerations than by clinical aspects. Second, in contrast to earlier findings, changes in recommendations due to evidence-based updates were mostly viewed positively. This latter point should be viewed in light of the Norwegian context, which has a relatively small number of published guidelines. In addition to these findings Boivin et al., using semi-structured focus group interviews, observed that in circumstances where clinicians judge patient participation in decision-making to be important, they perceive tension between the need to respect patients' preferences and the pressure to apply guidelines [ 46 ]. The authors conclude that CPGs should include information that supports shared decision-making in addition to their current focus on influencing prescription patterns, costs and health outcomes. In Spain, to the best of our knowledge, this study will be the first to explore the perceptions, beliefs and attitudes of clinicians (GPs and other specialists, with or without previous experience in guideline development) towards CPGs. In relation to grading systems used to rate the evidence, quality and strength of recommendations, it will be one of the first to explore this issue qualitatively in more detail. Limitations and difficulties which we expect to face during the study originate from the natural complexity of qualitative designs and more specifically, group discussions. The researcher, or moderator, in these types of studies has less control over the data produced [ 47 ] than in either quantitative studies or in-depth interviews. The moderator has to allow participants to talk to each other, ask questions and express doubts and opinions, while having very little control over the interaction other than generally keeping participants focused on the topic. By its nature, the group discussion method is open-ended and cannot be entirely predetermined. Additionally, it should not be assumed that the individuals in a group discussion are expressing their own definitive individual view. They are speaking in a specific context, within a specific culture, and thus sometimes it may be difficult for the researcher to clearly identify an individual message. It is to be anticipated that dominant group members influence the statements and opinions of others, as they do in "real" group meetings and other social settings [ 45 ]. Nevertheless, there are also important advantages to carrying out group discussions. They allow for gathering a lot of relevant information in a short period of time, offering insight on the existing contradictions and differences in the individual opinions of the participants. This social interaction is produced in a real context and helps getting a broader conceptual perspective of the issues discussed. Another potential disadvantage of corpus analysis is that the findings are not necessarily valid in other settings and cannot be extended to wider populations. However, we will undertake our study in a wide and diverse population of clinicians, assuring wide applicability of our results in Spain and probably elsewhere. The limitations would be potentially neutralized by completing initial results from group discussions with survey findings. On the other hand, the most apparent limitation of the survey could be a possible low response rate. However, in survey research, no single response rate is considered a standard [ 48 , 49 ]. For some surveys, a response rate of 80 percent is desired; in others, 60 percent is deemed adequate. Mail surveys typically have lower response rates than other types of surveys, and because non-response may introduce error, researchers should take steps designed to promote responses. Some of these steps include personally contacting potential respondents and asking them to participate, sending a reminder to non-respondents, assuring respondents of confidentiality, and making the survey short and easy to complete. In our study, all these measures will be adopted. Finally, caution will be taken when interpreting information about the connection between attitudes and actions. Research indicates that positive attitudes to guidelines do not necessarily mean that doctors follow their recommendations [ 10 , 50 ]. Based on the results from both phases of our study, we expect to deepen our understanding about the perceived difficulties and barriers to the comprehension and application of CPGs and to answer whether different grading systems, recommendation phrasing and guideline format presentation influence these processes. We expect to identify and describe the CPGs' adoption factors specific to each professional group participating and thus inform health professionals and decision makers about the needs and challenges in this field.

Background Clinical practice guidelines (CPGs) have become a very popular tool for decision making in healthcare. While there is some evidence that CPGs improve outcomes, there are numerous factors that influence their acceptability and use by healthcare providers. While evidence of clinicians' knowledge, perceptions and attitudes toward CPGs is extensive, results are still disperse and not conclusive. Our study will evaluate these issues in a large and representative sample of clinicians in Spain. Methods/Design A mixed-method design combining qualitative and quantitative research techniques will evaluate general practitioners (GPs) and hospital-based specialists in Spain with the objective of exploring attitudes and perceptions about CPGs and evidence grading systems. The project will consist of two phases: during the first phase, group discussions will be carried out to gain insight into perceptions and attitudes of the participants, and during the second phase, this information will be completed by means of a survey, reaching a greater number of clinicians. We will explore these issues in GPs and hospital-based practitioners, with or without previous experience in guideline development. Discussion Our study will identify and gain insight into the perceived problems and barriers of Spanish practitioners in relation to guideline knowledge and use. The study will also explore beliefs and attitudes of clinicians towards CPGs and evidence grading systems used to rate the quality of the evidence and the strength of recommendations. Our results will provide guidance to healthcare researchers and healthcare decision makers to improve the use of guidelines in Spain and elsewhere.

Abbreviations CPGs: Clinical practice guidelines; GPs: general practitioners; GRADE: Grading of Recommendations Assessment, Development and Evaluation; ANOVA: analysis of variance; SPSS: Statistical Package for the Social Sciences. Competing interests The authors declare that they have no competing interests. Authors' contributions PAC and IS conceptualized the study. AK and PAC drafted a first version of the protocol. All authors participated in revising it critically for important intellectual content and have given final approval of the version to be published. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1472-6963/10/328/prepub

Acknowledgements Funding This project has been funded by the Instituto de Salud Carlos III, co-financed by the European Regional Development Fund (PI08 90647). The publication of this document has been funded within the framework of collaboration designed for the Quality Plan of the Spanish National Health System, under the agreement signed by the Carlos III Health Institute and the Aragon Health Science Institute as GuiaSalud secretariat. Pablo Alonso-Coello is funded by a Miguel Servet contract by the Instituto de Salud Carlos III (CP09/00137).

CC BY

no

2022-01-12 15:21:45

BMC Health Serv Res. 2010 Dec 3; 10:328

oa_package/0c/07/PMC3016354.tar.gz

PMC3016355

21138576

Background "Managed care" is a global term for health care systems that integrate the delivery and financing of health care. Managed care contrasts with liberal medical practice, which allows doctors to make clinical decisions and bill for their services without interference from managers or payers. Traditional forms of managed care include the staff-model health maintenance organization (HMO) and the office-based independent provider association [ 1 , 2 ]. However, many variants exist. Luft notes that "in reality, each HMO is a highly complex combination of economic incentives, bureaucratic structures, and personalities" [ 3 ]. Another definition characterizes managed care programs by their use of a variety of interventions, including economic incentives for doctors and patients, review of the medical necessity of specific services, increased cost sharing, controls on inpatient admissions and lengths of stay, and selective contracting with health care providers [ 4 ]. Thus managed care is akin to a tool-box. The tools include the use of practice guidelines [ 5 ], gate-keeping [ 6 ], health care networks [ 7 ], second opinion requirements [ 8 ], and pre-approval requirements for expensive treatments or hospitalisation [ 9 , 10 ] (or utilization review). Other tools act through financial incentives: doctor payment for performance or by salary instead of fee-for-service payment [ 11 , 12 ], as well as the possibility for insurers to choose which doctors to reimburse (selective contracting) [ 13 ]. While managed care originated in North America, its tools have spread internationally. For instance, generalists in several Northern European countries regulate access to specialists, and have responsibility over a per capita annual budget. France has created a governmental agency to develop clinical practice guidelines. In Switzerland, the Health Insurance Law of 1996 has authorised managed care plans that apply various cost containment measures [ 14 ]. Therefore managed care tools have become a concern for doctors in many countries. How doctors perceive these tools is not well known. Several studies have explored doctors' perceptions of practice guidelines and managed care in general [ 7 , 15 - 22 ], but evidence is scant concerning other tools [ 23 - 26 ]. Because the methods and survey instruments vary across studies, direct comparisons of various tools are impossible. Furthermore, global opinions such as overall acceptability may hide differences in specific impacts. For example, doctors may believe that utilization review will reduce total health care expenditures but that quality of care and access to treatments may suffer. In addition to quality and costs, managed care tools may also have an impact on doctors' autonomy and on their relationships with patients. We are not aware of studies comparing various managed care tools on their various perceived impacts. Understanding doctors' perceptions of managed care tools is important, because the doctors' acceptance of such tools is required for their successful implementation. The aim of this study was to explore doctors' opinions about the impact of eight managed care tools (use of guidelines, gate-keeping, health care networks, second opinion requirement, doctor payment by performance or salary, selective contracting, and pre-approval of expensive treatments) on four aspects of medical care (quality of care, cost control, professional autonomy and relations with patients).

Method Setting This study was conducted in canton Geneva, Switzerland. The Swiss health care system mixes liberal medical practice in ambulatory care with a publicly run hospital sector. Managed care, as implemented in the United States, is not much developed; about <10% of the population is enrolled in managed care plans. Reinhardt describes Swiss health care as a "consumer directed health care" system with tight government rules [ 14 ]. The majority of hospitals are public and most hospital doctors are paid by salary (only a few senior doctors are allowed to have a part-time private practice at the hospital). Doctors in private practice are reimbursed on the basis of a fee-for-service schedule negotiated between caregivers and insurers. Swiss residents are covered by mandatory health insurance; most have access to any doctor of their choice, and only a fraction are enrolled in managed care networks. Health insurance premiums vary across regions (cantons) and insurance carriers but are identical within an insurance company and canton for all adults aged 25 or older (children and young adults pay lower premiums), regardless of health status or pre-existing conditions. Premiums can be lowered by choosing a higher annual deductible (up to 2500 Swiss francs) or managed care plans. The government subsidizes low-income residents, and also determines which drugs and services are reimbursed under the basic health insurance plan. In addition to basic insurance, some people contract private insurance policies to reimburse additional services, mostly hospitalization in private clinics. Health care expenditures represented 10.6% of the gross domestic product in 2007 and were shared between households (31.7%), private insurance (9.2%), the state (16.2%), and social insurance (42.9%) [ 27 ]. Managed care is a relevant topic for Swiss doctors. In addition to the possibility to join existing managed care plans, two measures are currently discussed: selective contracting is promoted by the health insurers, and managed care networks, coordinated by primary care doctors, are promoted in federal policy circles. Study design and sample We conducted a mail survey of all doctors active in clinical care in Geneva, Switzerland. We included doctors who worked at public hospitals or in private practice, and who were board certified or still in training. We used two databases: the Geneva Medical Association membership register and the University Hospitals of Geneva human resources file. After exclusion of duplicate records, invalid addresses and doctors who did not work with patients (e.g., pathologists, public health specialists, etc), 2746 doctors remained eligible. Our sample matched the number of doctors in Geneva according to the Swiss Medical Association [ 28 ]. We sent to each doctor the initial survey package and up to two reminder packages between November 2007 and February 2008. This study was approved by the Research Ethics Committee of the University Hospitals of Geneva. Attitudes toward management tools This section of the questionnaire started with the following statement: "In recent times, new measures have changed the liberal practice of medicine, and several others are being discussed. Please check for each of the following measures if you judge that it has or would have a positive or negative impact on professional autonomy, cost control, quality of care, and relations with patients." This was followed by a list of eight managed care tools: (a) systematic use of guidelines for clinical practice (abbreviated as "guidelines"), (b) patients' access to specialists guided by a primary care doctor (gatekeeping), (c) necessity to ask the insurance carrier for an authorization before any hospitalisation (utilization review), (d) necessity to ask for a second medical opinion for any costly investigation or treatment (second opinion), (e) doctors' participation in health care networks which coordinate care for affiliated patients (network), (f) possibility for the insurer to choose which doctor he will reimburse (selective contracting), (g) doctor's pay for performance based on cost control, patient satisfaction, or quality of care (pay for performance), and (h) payment by salary (salary). Each managed care tool was followed by a list of four impacts - professional autonomy, control of health care costs, quality of care, and relations with patients. Possible answers to each item were: 1 = very negative, 2 = rather negative, 3 = neutral, 4 = rather positive and 5 = very positive. There were 32 assessments in total (4 impacts for 8 tools). The questions were developed by the investigators, and were pretested in face to face interviews with 15 doctors. Other variables The questionnaire included questions on the respondent's sex, year of birth, specialty (either completed or planned) and practice setting (private practice in solo, in group or in a clinic or salaried practice in a clinic or a public hospital). Respondents were also asked where and when they graduated, if and when they received their specialty certification, and whether they were part of any managed care network. The questionnaire also evaluated work satisfaction and opinions on insurance policies and quality of care; these topics are not discussed in this article. Statistical analysis We described the response rate and the characteristics of participants, comparing respondents and non-respondents. As negative opinions globally predominated, we reported proportions of negative opinions regarding each tool's impact on each aspect (autonomy, costs, quality, and relations with patients), grouping "very negative" and "rather negative" opinions. The four impact ratings turned out to be correlated, for all of the managed care tools. Therefore we constructed global opinion scores for each of the 8 tools as the average of the 4 expected impacts. For each scale, we reported the internal consistency coefficient (Cronbach alpha). However, absolute expected impacts (i.e., negative or positive) differed across the four domains for most managed care tools, so descriptive statistics are shown separately for each of the 32 ratings (4 impacts for 8 tools). Scores were compared across subgroups of respondents (sex, age, specialty, practice setting and membership in a managed care network). We classified specialties into five groups: primary care doctors (generalists and general internists), internal medicine specialists (including neurologists), paediatricians, psychiatrists and technical specialists (surgeons, anaesthetists, ear-nose-throat specialists, ophthalmologists, dermatologists, gynaecologists-obstetricians and radiologists). We distinguished between three categories of practice settings: independent private practice, public hospital practice in training or public hospital practice as senior. For each tool impact score we built a multivariate model using doctor characteristics as predictors. We used SPSS 15.0 for all analyses.

Results Sample characteristics The survey response rate was of 56.3% (1546/2746). Participation was not related to age, setting of practice and source data base. However men responded more frequently than women (58.0% vs. 53.7%, p = 0.027) and participation varied according to specialty, from 52.6% in technical specialists to 62.2% in primary care doctors (p = 0.003). The sample included more men than women (Table 1 ). More than half of all respondents worked as independent doctors, either in solo or in group practice and one out of five belonged to a managed care network. Primary care doctors and technical specialists represented each about a third of the sample, the remaining third consisting of psychiatrists, internal medicine specialists and paediatricians. On average, doctors were 47.3 years old (SD 11.6) and graduated 20.1 years ago (SD 11.1). Specialists had obtained their certification 14.0 years ago (SD 9.4) on average, and those in private practice established their practice 15.2 years ago (SD 9.0). Most doctors who worked in managed care belonged to one or two of 4 organizations: A (N = 88), B (N = 87), and C (N = 25), and D (N = 18). Organisations A and B were independent provider organisations that included generalists and internal medicine specialists; both had a program of quality circles, practiced gate-keeping (gynaecology check-ups and paediatric consultations were not subjected to gate-keeping), and used fee-for-service payments. Neither used direct financial incentives or imposed a global budget, and both had contracts with several health insurance carriers. Organisation C was similar, but was restricted to care for migrants and asylum seekers, and worked more closely with the public hospital in providing care to this population. Organization D was a group practice set up by a single insurer, where doctors were paid by salary. None of the participants had been exposed to pay for performance, utilisation review, or selective contracting. Proportions of negative opinions Use of guidelines received the smallest proportion of negative opinions (18.0% to 35.6%) followed by gate-keeping and health care networks (Table 2 ; full item distributions of the items are available from the last author on request). In contrast, pay for performance, utilization review and selective contracting gathered the highest percentages of negative judgments. Payment by salary and second opinion stood in between. Respondents reported more often negative impacts on professional autonomy and relations with patients than on quality of care and on the control of health care costs. Global impact scores Cronbach alpha coefficients for impacts of the four aspects of care within each tool ranged from 0.75 to 0.89 (Table 3 ). Mean scores ranged from 1.56 (selective contracting) to 3.18 (guidelines), and standard deviations were all less than 1. Use of guidelines was the only tool with an average score above 3, slightly better than neutral. Gate-keeping, health care networks, second opinion and payment by salary were rated between 2 and 3 on average (between neutral and rather negative). Pay for performance, utilization review and selective contracting, had mean scores below 2 (between rather and very negative). Differences across subgroups Mean global impact scores for each tool varied across subgroups of respondents in univariate (Table 4 ) and multivariate analysis (Table 5 ). Men had a more positive attitude than women concerning second opinion and pay for performance, but rated payment by salary more negatively. These differences decreased in multivariate analysis. Older respondents reported more negative opinions about all tools except about second opinion. In multivariate analysis, age differences remained significant only for guidelines, gate-keeping and second opinion. Most tools were rated lower by private practitioners than by doctors working in public hospitals. Practice differences persisted for almost every measure in multivariate analysis, except for second opinion and the utilization review. Primary care doctors had a more positive attitude not only toward guidelines but also toward gate-keeping. Along with paediatricians, they had the highest scores for all tools except pay for performance, utilization review and selective contracting. For the latter, technical specialists held the most positive opinions. Psychiatrists appeared to be the most reluctant specialists, particularly concerning gate-keeping. Internal medicine specialists usually gave intermediate ratings. In multivariate analysis, differences between specialists remained strongest for ratings of pay for performance, utilization review, and selective contracting. Among independent practitioners, those involved in a medical network rated second opinion better than other doctors and also had a positive attitude toward gate-keeping and networks.

Discussion Overall, the Swiss doctors who participated in this study expressed negative opinions about the expected impact of most managed care tools. Only the use of guidelines received a lukewarm endorsement from the respondents. Perhaps predictably, managed care tools that remain under the control of the medical profession - guidelines, gate-keeping and health care networks - were rated more favourably than tools that put payers in charge, such as pay for performance, utilization review and selective contracting. These findings are compatible with previous observations from the United States that managed care reduces career satisfaction through its impact on doctor's autonomy [ 29 ], and that managed care constraints may hamper the doctor's ability to provide high quality care [ 30 ]. However, as we did not obtain the doctors' explanations of why they gave a positive or negative impact rating, the specific reasons for these opinions of Geneva doctors remain unknown. That doctor's attitudes are context-specific is illustrated in qualitative study comparing doctor's perceptions of practice guidelines in Norway and Denmark [ 31 ]. Subgroup comparisons Differences between specialties were rather small, with few exceptions. One was the more positive view of gatekeeping and networks by primary care doctors. It is understandable that generalists should support tools that give them an active role in the health care system. Other studies have suggested likewise [ 7 , 32 ]. In addition, general practitioners may be less motivated by financial incentives than other specialties [ 33 ], and therefore may be more willing to accept payers' intrusion into medical practice. Primary care doctors had opinions of guidelines that were similar to other specialities (excepting psychiatrists). Some previous studies have observed that generalists were less prone to using guidelines than specialists [ 34 ], but that was not the case in our data. The higher than average survey participation rate of primary care doctors' suggests that this group may be particularly interested in managed care issues. For most tools, psychiatrists held more negative opinions than other doctors. Psychiatrists may feel particularly threatened by managed care [ 35 ], because traditional psychotherapy requires multiple doctor visits that may not be approved under many cost cutting policies. Older doctors and those in private practice had more negative opinions of the various managed care tools. These are subgroups that have the greatest experience of a traditional "liberal" model of medical practice, grounded in professional autonomy and fee-for-service compensation, which is under pressure from managed care. Similar findings have been reported in previous studies of managed care [ 7 ] and guideline acceptability [ 15 , 17 , 21 , 22 , 33 , 36 ]. Private practitioners who were members of a managed care network had a more positive opinion of gate-keeping and health care networks, and salaried doctors rated payment by salary higher than other respondents. Whether this reflects a selection bias or the effect of greater familiarity with these measures is unclear. Previous research supports the notion that familiarity with some managed care tools is associated with more positive opinions [ 7 , 33 , 37 ]. If indeed familiarity led to positive opinions, implementation of managed care tools despite initial reluctance on the doctors' part may be a viable strategy. However, familiarity may improve perceptions only if a priori apprehensions are disproved by experience. Impact on different aspects of care We have conducted most analyses using a summary score, the global impact score, for each managed care tool. This was justified by the high internal consistency coefficients of the summary scales, and allowed a synthetic presentation of the results. However, as others have reported [ 7 , 23 ], doctors differentiate the impact on costs from impacts on other aspects of care. Many doctors consider that these tools help rein in health expenditures, but few believe that they promote professional autonomy, quality of care or satisfactory relations with patients, as seen in Table 2 . As health care expenditures correspond in part to doctors' incomes, this discrepant assessment of impacts is understandable. Because the summary score was not planned initially, we did not ask what importance respondents would attribute to each of the four aspects and, thus, how much each aspect would influence their own global opinion on each tool. The summary score we computed assigns equal importance to each aspect, which is arbitrary. Other weighting schemes may lead to different results, but none would capture the inter-individual differences. Furthermore, the correlations between impacts may be partly due to a halo effect, whereby the general attitude toward a managed care tool influences the specific ratings. Importantly, correlation between the four impacts does not preclude differences between ratings in absolute terms. Limitations By asking closed-format questions, we have constrained the respondents' ability to express their detailed views about the managed-care tools and to explain the reasons for their ratings. Another limitation is that we did not provide definitions of the terms employed. E.g., "health care costs" probably meant medical expenditures to most respondents, who may not have thought about indirect costs or non-medical costs. This, and similar differences in other definitions, may have added unexplained variance to the opinions. The moderate response rate, although typical for doctor surveys, raises the issue of selection bias. Respondents may have been more critical toward managed care than non-respondents, but have no data to substantiate this belief. We conducted this survey among a population of doctors who work within a given health system. Our results may be safer to generalize to contexts where the ethos of independent liberal practice is still present. However, more general findings may be valid regardless of context: e.g., familiar procedures may be seen as having a more positive impact than unfamiliar procedures, and tools that are imposed from the outside may be less acceptable that those that allow a measure of control by the doctors themselves. But generalisability is an empirical question, and defining the validity of our results in other contexts would require further surveys. Regarding the issue of familiarity, we did not document each doctor's experience with each of the tools, beyond participation in a managed care organisation. In some cases, therefore, the rating reflected expectations, in others, experience. These situations are not equivalent and averaging opinions of these types of respondents may result in loss of information. Finally, this study concerned itself only with the opinions of doctors. Other stakeholders in the health care system may have valid, yet different, opinions about the same managed care tools.

Conclusion Doctors in Geneva, Switzerland, expressed a positive attitude only toward the use of guidelines and otherwise held predominantly negative opinions about managed care tools. They were particularly severe concerning selective contracting, utilization review, and pay for performance. While they agreed that several measures can help control health care costs, they were particularly concerned about loss of autonomy, worsening of relations with patients, and reduced quality of care. While we did not query the respondents about their preferences, our results suggest that managed care tools and incentives that remain at least partially under control of the medical profession and that interfere the least with the current payment mechanisms may have the greatest acceptability. From a policy perspective these generally negative attitudes are informative, as they may influence attempts at implementing managed care tools on a larger scale. Further research in this area should address the following: qualitative analysis of doctors' opinions, assessment of various combinations of managed care tools from the doctors' perspective, and assessment of the opinions of the general public regarding these policies.

Background How doctors perceive managed care tools and incentives is not well known. We assessed doctors' opinions about the expected impact of eight managed care tools on quality of care, control of health care costs, professional autonomy and relations with patients. Methods Mail survey of doctors (N = 1546) in Geneva, Switzerland. Respondents were asked to rate the impact of 8 managed care tools on 4 aspects of care on a 5-level scale (1 very negative, 2 rather negative, 3 neutral, 4 rather positive, 5 very positive). For each tool, we obtained a mean score from the 4 separate impacts. Results Doctors had predominantly negative opinions of the impact of managed care tools: use of guidelines (mean score 3.18), gate-keeping (2.76), managed care networks (2.77), second opinion requirement (2.65), pay for performance (1.90), pay by salary (2.24), selective contracting (1.56), and pre-approval of expensive treatments (1.77). Estimated impacts on cost control were positive or neutral for most tools, but impacts on professional autonomy were predominantly negative. Primary care doctors held more positive opinions than doctors in other specialties, and psychiatrists were in general the most critical. Older doctors had more negative opinions, as well as those in private practice. Conclusions Doctors perceived most managed care tools to have a positive impact on the control of health care costs but a negative impact on medical practice. Tools that are controlled by the profession were better accepted than those that are imposed by payers.

Competing interests The authors declare that they have no competing interests. Authors' contributions TP and PB proposed the study; MD, PB and TP wrote the protocol; MD conducted the survey; MD and TA analysed the data; all authors interpreted the results; MD wrote the paper and TA, PB and TP revised the paper. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1472-6963/10/331/prepub

Acknowledgements Funding was provided by the Research and Development Program of University Hospitals of Geneva.

CC BY

no

2022-01-12 15:21:45

BMC Health Serv Res. 2010 Dec 7; 10:331

oa_package/84/c4/PMC3016355.tar.gz

PMC3016356

21138584

Background Unintentional (accidental) injuries are the leading cause of death for New Zealand children aged one to fourteen years, accounting for 35% to 43% of deaths in this age group [ 1 ]. The associated annual mortality rate is among the highest in high-income countries [ 2 ]. Serious non-fatal injuries account for almost 14,000 hospitalisations each year [ 1 ] imposing a substantial burden on the children involved and their families. Overseas research indicates that many children whose injuries are severe enough to warrant hospitalisation experience long-term impairments in physical and cognitive function, behavioural difficulties, and psychological distress [ 3 - 9 ]. The psychological impact on parents and siblings can also be considerable [ 7 - 11 ] with many caregivers experiencing employment and financial hardships [ 12 , 13 ]. The period of acute care of seriously injured children in hospital is of particular significance in this context. High quality hospital care is estimated to reduce child injury mortality rates by eight percent [ 14 ]. This phase also provides families the opportunity to gain valuable knowledge regarding the treatment their children required and services and support systems that can enable optimal recovery following discharge from hospital. Unmet needs in these areas are therefore important concerns relating to the quality of trauma care. A study from the United States found that parents of hospitalised injured children experienced problems regarding the content and timing of communication during acute care of their children and difficulties in accessing and understanding the system of health care [ 15 ]. These concerns can be compounded by racial and ethnic disparities in the quality of trauma care that can occur in the context of broader social and economic inequities [ 16 ]. This is particularly important in New Zealand where significant socioeconomic and ethnic disparities exist in the burden of childhood injury with disproportionately higher representation of Māori and Pacific children among injury hospitalisations [ 17 - 19 ]. Māori (the indigenous people) comprise 15% and Pacific peoples (a range of groups who have migrated over decades from the Pacific region) 7% of the over four million people residing in New Zealand in 2006 [ 20 ] and both groups experience significantly poorer outcomes in most health, social and economic indices [ 21 ]. Although New Zealand studies have explored the concepts and perceptions relating to unintentional injuries amongst Māori and Samoan families [ 22 , 23 ], there is a gap in research knowledge regarding the perspectives of those who provide health and support services to injured children. In order to address this gap, the present study explored the perspectives of 21 service providers from a range of professional backgrounds regarding the key issues facing injured children and their families accessing their services as well as other services with which the families interacted. This research was part of a larger study investigating the health and social impact of childhood injuries, and complements an exploration of the perspectives of families whose children were admitted to hospital, reported elsewhere [ 24 ].

Methods To ensure an appropriate research design and effective participant engagement, the multi-ethnic research team was guided by a steering group comprising a range of professional and cultural experts. The study was approved by the Auckland Ethics Committee, New Zealand. Design, setting and participants A qualitative research design involving in-depth individual interviews or small focus groups was used as this was considered the most appropriate means of obtaining in-depth, descriptive detail about the roles, experiences and views of study key informants [ 25 ]. For the purposes of our study key informants were defined as key people from a range of professional disciplines and organisations involved in the care and support of injured children and their families in Auckland, New Zealand. They were selected purposefully using maximum variation sampling to ensure a wide range of expertise and experience was represented [ 26 ] and were identified through the networks of the study investigators and steering group members. They included representatives of emergency medical, trauma, general and specialist surgical services, Māori and Pacific Family Support services, therapy services, hospital school educators, rehabilitation service providers (public and private), the disability support sector and the Accident Compensation Corporation (ACC). Potential participants were approached in the first instance by phone, email or in person by one of the study investigators. As the Maori Family Support service, the Pacific Family Support service and the hospital school educator service each comprised a small team, all their team members were invited to participate. All those approached agreed to participate in the study. Eleven individual interviews and three focus groups involving ten participants (in total) were conducted face to face using a semi-structured discussion guide. Individual interviews were used in situations where the participant's role was distinctive from other participants. Three distinct focus groups were used for the Māori Family Support team (Kaitiaki), the Pacific Family Support team and the hospital school educators as these participants worked in teams that had similar roles and responsibilities. Interview process The interviews and focus groups were undertaken during March to May 2002 by six researchers. Nine of the eleven interviews were undertaken by the first two authors (SA, SA), the remaining two interviews and the Kaitiaki service focus group by two Māori researchers with medical backgrounds, the Pacific Family Support team focus group by the Pacific family researcher (LA), and the school focus group by the first author (SA) and a study investigator with strong school links. Interviews took place at a location chosen by the participants and in all cases this was their workplace. The purpose and process of the research were reviewed at the commencement of the interview and key informant(s) were given time to ask questions and clarify points. Written consent was then obtained. Interviews/focus groups took approximately 30-60 minutes and were audio-taped where participants gave permission. Those in one focus group did not wish to be audio-taped so comprehensive notes were taken. The key informants' discussion guide was developed by the investigator team in conjunction with the advisory group. It comprised a small set of common questions that were then fashioned to suit the individual's or group's particular area of expertise and were broad, generic, open-ended and exploratory in nature. After establishing an understanding about the services provided by participants, these questions explored their views regarding how injured children and their families' needs were met by their services, and a deeper exploration of the key concerns they had about their services or other services with which these families interacted. Data Analysis The taped interviews were transcribed and, where requested, a copy of the transcript was sent back to the participant to review and, if desired, add comment. No amendments were requested. Thematic analysis [ 26 ] was then undertaken whereby the data were analysed inductively by identification of fundamental ideas or meaning units in the transcripts which were then coded into sub-themes and further amalgamated into a number of themes by the second author (SA). A first draft of key themes was circulated amongst the five other interviewing researchers and revisions made based on feedback. A later draft was also discussed in depth with the whānau/family researchers, comparing findings with those from the whānau/family interviews, published previously [ 24 ].

Results In total, the views of 21 key informants were elicited, representing a range of professional groups. These were: paediatricians (3), hospital nurses (2), occupational therapists (2), a physiotherapist, a play therapist, a trauma coordinator, an Accident Compensation Corporation case worker, Maori whānau/family support workers (4), Pacific family support workers (3) and hospital and community-based teachers (3). All Maori whānau/family support workers and two other participants were Māori (n = 6) and all Pacific family support workers were of Pacific ethnicity (n = 3). All other participants were Pākehā (n = 12). Three of the participants were based in the community and all others were based either at the Starship Children's Hospital or KidzFirst Hospital, the major children's hospitals in the Auckland region. A number of themes arose out of the key informant interviews and focus group discussions. Since only one key informant spoke about events prior to admission to hospital and did not mention any concerns, the key issues and concerns discussed in this paper are focused on the hospital and community environments. As the key informants were from a range of services and professions and they engaged with injured children and their whanau/families in different ways and at different points along the continuum of care, they each made relevant contributions in certain areas more than others. It was this breadth of key informants' experience and views that the study sought to capture. The theme descriptions incorporate the views and experiences of those participants with the relevant experience and quotes have been selected that best articulate and illustrate the theme components. Difficulty meeting the range of needs of injured children A dominant theme was that the hospital system in particular and the health system in general did not adequately meet the needs of children with mild injuries. These children had less contact with all staff, particularly those of the allied professions, physiotherapy, occupational therapy and social workers. This was largely attributed to their relatively rapid transit and discharge from hospital and the obvious and justifiable prioritising of more serious cases in a situation of limited staffing, resources and bed space. 'The ward that they come to is a very high turn-over ward... obviously staffing levels are an issue. You are chasing your tail to look after the unstable patients, so the ones that are perhaps in for a minor type injury, these parents are there, the reality is they are not top of your list.' (Hospital nurse) Even if a child's immediate physical injuries were reasonably well assessed and dealt with in the hospital setting, it was acknowledged that their emotional needs and concerns were usually not. Sometimes the child's physical injury might be relatively minor but if they were considered responsible for injuring others or if they had witnessed a serious motor vehicle crash, participants felt their emotional trauma could be extensive. 'They may have been in a car accident and although they [injuries] are actually of a minor nature, they are part of an event that may have resulted in death or injury. The child themselves may have caused an incident that may have severely injured another child and they themselves, their injuries were of a minor nature. So they [injuries] don't necessarily happen in isolation.... And I don't know how well they get picked up or followed up.' (Hospital nurse) Some informants perceived the lack of paediatric-trained staff in the emergency department and the poor understanding of developmental stages, even amongst some staff who had paediatric training, important issues in a context where the needs of children would differ according to their developmental stage. '[We need to] strengthen people's knowledge about the developmental issues that happen around children, then we provide a much more supportive environment for children ... An adult nurse has quite different skills to a paediatric nurse and many of the emergency departments around the country will have predominantly emergency nurses who are adult trained, so you look through different eyes.'(Play therapist) Accommodating the particular needs of adolescents was also considered sometimes difficult as these varied greatly from those of younger children and even varied between early and late adolescence. Dealing with the psychosocial needs of families Hospital-based informants identified influences that placed substantial emotional and psychological stresses on families when their child was hospitalised. These included: the family's previous experience of hospital, whether they were involved or had been hurt in the accident, whether they felt responsible for the accident, fear about how to cope with other children at home, fear about money or the need for time off work because of their child's injury and other issues going on in their lives. These multiple demands were perceived to place enormous pressures on individuals and relationships within families. 'I think [staff] under-estimate the impact of these sorts of things on family dynamics. Mothers have farmed off other children and they work. So for any acute hospitalisation there's quite widespread impact on the family, even if a child is only in for 24 hours. And I think sometimes we are a bit unsympathetic.'(Hospital nurse) There was strong empathy for the onus of responsibility felt by sole parents. 'I think it's very tough on single-parent families. They have to take it all on board and make all the decisions...Having to make the decisions and being responsible for when to do things and what to do and how to plan it and organise it all with the rest of the [extended] family and things like that.' (Physiotherapist) Practical considerations, such as the difficulty and cost of parking and access to meals, became important issues for families dealing with these stresses. 'I think dishing out the old taxi voucher to get families home or to get other family members in, all those sorts of things that are quite small, really help... And food. Kids coming in acutely... their mothers haven't had anything to eat and there's no food for them and if they haven't brought money with them... All those sorts of things add stress for the family'. (Hospital nurse) Services were perceived to deal very poorly with the emotional and psychological impact of trauma on the families. Key informants acknowledged that many parents experienced considerable guilt and anger because they had not prevented their child's injury or were afraid that it might appear that they were careless or abusive. Several informants mentioned that, while they and other staff members were often aware of parents' difficult emotions, there was no systematic process for acknowledging and assisting parents to deal with them. Insinuations about intentional injury were seen to put additional stresses particularly on Māori and Pacific families. In the context of intensified media coverage of intentional injury amongst Māori at the time, a few reported that it was more likely that reference would be made to the hospital based child protection services ( Whakaruruhau) when injured Māori and Pacific children were admitted to hospital than was the case with Pākehā children. 'Lots of times, because they are Māori or Pacific Island [children], Whakaruruhau tends to be talked about more. Sometimes that's quite a common trend... I don't know what sort of thing it has on the outcome, except that, if they do think of the family, that puts the whānau through a lot of stress.' (Kaitiaki focus group participant) Such insinuations, even when shown to be unfounded, were perceived to have a lingering impact on staff attitudes. 'As a staff member it [the suggestion of abuse] definitely does cloud your thinking. Even if someone says "No it's not, it's been cleared", you can't help but have that logged into your brain. Is it or isn't it, type thing'. (Hospital nurse) Issues for Māori and Pacific family support services A key role of the Kaitiaki and Pacific Family Support teams was providing practical and emotional support to Māori and Pacific families and resolving misunderstandings between families and staff. They identified a number of important issues for families and assisted where they could; however, they felt constrained at times in their ability to resource families in need. They did not have a budget of their own to draw from and, instead, had to get approval from the social worker or, if after hours, the duty manager. 'I don't believe [we have] the full resources that we need if we have a wh ā nau that is stuck. OK, they've come in on an ambulance and they really don't have a vehicle to get home, they don't have a phone to get someone. Lots of those things are issues for our families. No telephone, no car. We've still got to go and ask for a taxi voucher to get them home.' (Kaitiaki focus group participant) One of the difficulties they experienced was that families often confused them with social workers, about whom they were very wary, and therefore were reluctant to engage their assistance. '[Families] often mistake us for social workers and are scared we'll take away their benefit or take away their kids.' (Kaitiaki focus group participant) Another perceived reason for families being wary of accessing their service was concern about confidentiality. This was an important consideration for Pacific families. For example, Pacific key informants reported that some families were hesitant about using an official interpreter for fear that their confidentiality would be compromised. Information needs Key informants were very aware of the importance of good information as a means to empower children and families in the recovery and rehabilitative process. However, some felt that there was often not enough time to ensure parents and children did in fact understand what was told to them. They reported that sometimes parents received conflicting information from different professionals and that this was confusing. 'We give them a booklet [but] we need to go over the booklet to be sure they've got it... Sometimes there's too much, you can get too much information, but having it written down [helps]... And I think we forget that we just have to make it as simple as possible. Not too much information.' (Play therapist) Moreover, there was a perceived lack of comprehensive written information for parents to take home. '[We] constantly talk about giving information to families but our written information is not so good...It is written by people who are maybe nurses and doctors, they are good at being nurses and doctors, but are not necessarily good educators. So those are real gaps.' (Play therapist) Where children were from other cultures or there were language barriers the importance of ensuring information was appropriately offered and understood was considered paramount. 'The other thing we don't do well at the moment, [but] we are actually investigating getting it done, is translating [information] into different languages other than English. Even if it's translated into Samoan, Tongan and Cantonese, for example, then at least we are covering a big area.' (Occupational therapist) Staff continuity and coordination issues Poor staff continuity and coordination was a common theme. Each phase of the injured child's care from the emergency department to community rehabilitation had a different set of staff, and different methods of administration and data collection. Apart from there being no continuity of personnel for these phases, hand-over from one to the other was often not seamless. Consequently, key informants felt that parents experienced frustration when they had to repeat the same information to numerous staff. 'They arrive, are asked by the ambulance [officer], asked by the triage nurse, they are asked by the nurse that looks after them in ED, by the house surgeon, by the registrar, by the consultant and the same thing happens when they get to the ward. Everybody writes the same thing in the notes and, having been a parent who's stayed in hospital, it does become a little tedious'. (Hospital nurse) The high staff turnover, especially of medical and nursing staff, created several problems to do with continuity and coordination of care. Medical House Surgeons changed three monthly and registrars six monthly. New staff members were often inexperienced or unfamiliar with 'unwritten' procedures and consequently certain processes, such as referrals, follow-up and coordinating with other health professionals in the ward setting, were not carried out as required. These issues were exacerbated during the winter months when the wards were very busy and many permanent staff members were away sick. Another aspect of staff coordination related to professional boundaries. Key informants were respectful of the roles played by other professional groups and saw their roles as complementary. There was, however, recognition that staff needed to be constantly vigilant to ensure these relationships remained productive and that the children's needs were paramount. 'There is a tension between guarding your own profession - and that's a noble and a useful and a good thing - and also focusing on the needs of the child and seeking the best way to do that. So at times there are tensions. 'Is this something that a nurse should do or a physio should do?' But I think there are ways to dialogue those issues through and you keep them on the table.' (Hospital nurse) Issues after discharge from hospital Issues of poor continuity between the hospital and community settings were also identified. Key informants felt that initial post-discharge supports were scarce. The brevity of the hospital stay was considered a contributing factor here. Rapid discharge and busy staff meant that parents were at times poorly equipped practically and emotionally to deal with caring for the child at home. Some parents were perceived to be still shocked about their child having had an accident and dealing with a range of emotions, including guilt and fear, that affected their decision-making and coping ability once their child was home. A rapid discharge without clear, well planned follow up and support was seen to place added strain on these families. 'Children used to come in with a broken arm and they would be in here for a week. So you would have time to assist the parents through the shock. The shock of one minute your child is playing in the street and everything's fine and the next minute your life is so fragile. And not only is it fragile, this could have been your child dying. I think a lot of children are going home with parents who are still in that traumatic time themselves... I think we're not assessing. We are not asking these questions: how is it going to be for you in terms of managing? We ask the physical [ones], like, can you manage this child getting to the toilet and da-da-da? But we don't do such a good emotional stocktake .' ( Play therapist) In recognition that families were discharged rapidly and might find the first days very difficult, several informants suggested that a follow-up phone call by hospital staff after the first 24 hours in order to monitor how the parents were coping and to offer advice and support would be helpful. Exacerbating the early discharge issues was the difficulty of maintaining a strong and effective link between hospital and community services. Rotating hospital staff and the transience or short funding terms for some community agencies meant that any coordination that was established was often difficult to maintain. The well-recognised problems of non-attendance or difficulties in attending GP or out patients' clinics (e.g., transport, parking and long waiting time) were mentioned as ongoing issues of concern.

Discussion To our knowledge, this is the first study to provide in-depth, descriptive detail about the experiences and needs of children hospitalised for injury and their families from the perspective of health and support service providers. Strengths of the study included the wide range of key informant providers who willingly participated and the qualitative approach which enabled them to speak freely about their experiences and views. However, the findings must be interpreted in light of several limitations and the scope of this study. The research was undertaken in the context of two large urban children's hospitals. The information gleaned may not be applicable to all other hospitals, particularly those where resources are even less accessible. Smaller hospitals in provincial or rural settings may, however, have closer engagement with their communities which could overcome some of the challenges identified in this study. Despite some changes to New Zealand's health and hospital services since the collection of these data, feedback from health professionals at presentations of the study findings at several national meetings in 2008 and 2009 suggest that many issues identified by key informants have persisted and are not unique to the hospitals involved in this research, or indeed, trauma services alone. The findings of the present study as well as feedback at presentations since indicate a high level of awareness among service providers regarding the many shortcomings of a busy and overloaded health service and empathy and insight into the innumerable issues faced by children and their families. Most felt they and their health service colleagues were doing the best they could within the constraints of an inadequately resourced system, whose services were stretched, particularly in clinical settings where relatively short-stay admissions are common. It is possible that given the specific research questions explored in this study, participants were prompted to reflect particularly on service gaps and limitations rather than the areas in which the needs of children and families were adequately met. However, there was considerable agreement between the service issues and concerns identified by the key informants in this study and those mentioned by the whānau/families in the parallel strand of research in the larger study [ 24 ]. Both participant groups identified as issues; problems around information and communication needs, difficulties managing the multiple stresses on families when children are injured and hospitalised, the stressful impact of cultural stereotyping and the need for more appropriate resourcing for Māori and Pacific families. This high level of accord adds weight to their significance. The study findings have direct implications for services that provide care for injured children and their families during and after hospitalisation. In general, the hospital system does not seem to adequately meet the needs of children with mild injuries. Evidences suggest that an injury which may be 'minor' with respect to threat to life, may have a considerable impact on the family involved as well as wider society [ 27 ]. Depending on the circumstances surrounding the event, the emotional consequences experienced by these children and their families can be extensive. A high prevalence of post traumatic stress disorder symptoms has been found in children with mild to moderate injury up to 18 months after the event [ 28 ]. Other studies, while not exclusively focusing on children, have reported that a considerable proportion of patients with minor traffic injuries had a slow recovery and long-term adverse psychological and social consequences. Psycho-social outcomes were found at times to linger for much longer than physical outcomes and were poorly predicted by the severity of the physical injury [ 29 - 31 ]. Several key informants were aware of this issue and perceived it as a function of time and resource constraints, rather than assuming that these children did not require much attention. Key informants' concerns about the inadequacies of the hospital system in allaying parental distress appear well founded as one study found that parental distress played a more important role in the persistence of post traumatic stress disorder symptoms in children than the extent of the injury or the course of hospitalisation [ 28 ]. A number of factors contribute to parental distress during and after their child's hospitalisation. Important amongst these is inadequate information about the child's condition, care requirements and prognosis. Consistent with concerns expressed by families participating in the complementary arm of this study [ 32 ], key informants highlighted several problematic areas relating to communication, such as parents having to repeat the child's history often, language barriers and the need for better written information. Not alluded to by key informants but considered important by whānau/families was the need for appropriate and effective oral communications. Effective communication through a range of means is the cornerstone of ensuring patient health literacy, particularly where cultural miscommunications can arise [ 33 ]. Several studies have found that linguistic and cultural barriers adversely influence family involvement in the care of their children while in hospital and contribute to healthcare disparities for many minority groups [ 34 - 37 ]. The key informant Māori and Pacific support workers in the present study were aware of the significant stresses experienced by families of Māori and Pacific children who are known to experience a disproportionately high rate of unintentional injury [ 16 - 18 ]. However, they were concerned about resource constraints that limited the assistance they could offer these families, and the reality that their service was not accessible to families of most children with 'minor' injuries admitted for short periods This indicates that while the existence of such support services is crucial, these must be well resourced and extend to effective service delivery that is supported by an organisational culture that is responsive to the needs of these populations. Problems with the coordination of services both within the hospital setting and between the hospital and the community were noted. The movement of patients between services is typically a difficult area to streamline but, in the case of an injury, the involvement of a number of professionals coupled with rapid discharge appears to exacerbate this difficulty. The key informants recognised the first 24 hours at home following discharge as a crucial transition phase. During this period, the 'second crisis of injury' may occur when responsibility shifts from professional care givers to family members [ 38 ]. Many informants felt that there was more that could be done of a relatively minor nature that could reduce the stress of this transition for families. For example, a simple follow-up phone call to check on how the child was, how the family was coping and to answer any questions was a frequent recommendation. The communication and coordination between hospital and community or educational services were also recognised as areas that needed more attention. These concerns suggest the need for carefully developed discharge plans that provide continuity of care and effectively work across the boundaries between hospital, home and school, encouraging and reinforcing well developed relations between hospital and community based care services. For all of the reasons noted above, it is critical that information about hospital processes, issues that would become problematic following discharge to the community, and the resources available to support children and their families to negotiate these transitions, is given in a systematic and proactive manner rather than on an ad hoc basis. This would require the use of a variety of mechanisms to ensure that families under stress understand and retain necessary information. These are likely to include the need for culturally appropriate communication both orally and using written material, informative discharge plans, and where relevant, referrals to primary care and social services. It would be particularly important to ensure that all injured children and families who require further follow up by case managers in New Zealand's Accident Compensation Corporation (ACC), have access to this service. While the resources of this government-funded insurance scheme can serve as a potentially useful safety net for all New Zealanders requiring post-injury health care and rehabilitation, inequities in access to services by some ethnic groups, particularly including Māori, are well documented [ 39 ]. The accord between issues identified by the key informants in this study and those mentioned by the whānau/families in the parallel strand of research [ 24 ] suggests that many staff working with injured children and their families have a keen awareness of what constitutes optimal care but because of structural, time and other constraints are not always able to deliver it. The tension between providers' perceptions of optimum care and their ability to provide it is clearly an area for ongoing exploration and development. Systemic factors can provide significant barriers to effective care, however, health service providers' attitudes and behaviours also remain important influences on families' experience of health services and on the quality of care received [ 33 , 40 ]. In our study staff members' judgements around the cause of injury were a case in point. Barriers to quality care for children hospitalised for injury are complex and multi factorial and a clear understanding and analysis of these barriers is needed if service improvements are to occur. Systemic, health provider and patient factors have been identified as key barriers to quality services [ 41 ] and interventions at these three levels have recently been proposed as a means to reduce ethnic disparities in the quality of trauma care [ 42 ]. Strategies that would appear to have particular relevance for our suggested improvements to services for children hospitalised for injury and their families include cultural competency training with a focus on patient health literacy, and clinical audits with feedback loops that support continuous quality improvement. Research designed to quantify the importance and predisposing factors that underlie the issues identified in this study could provide useful guidance to prioritise areas of action.

Conclusions Key informants provided valuable insights into key issues and concerns for children hospitalised for unintentional injury and their families within their own service and within those services with which they interacted. Given the significant impact of unintentional injury on childhood mortality and morbidity, improvement in the quality of health care and follow-up services is required to meet the needs of injured children and their families. Such initiatives might include focusing more attention on children with minor injuries, dealing with stresses of injury hospitalisations on families, ensuring more effective and appropriate communication of information, ensuring provider cultural competency and more effective support services for Māori and Pacific whānau/families and other ethnic groups and improving coordination of services both within the hospital setting and between the hospital and the community. These recommendations for service quality improvements for injured children include interventions at systemic, provider and patient levels, recognising that barriers to service quality operate at these levels and that a multi pronged approach to address them is likely to be most effective.

Background Unintentional injuries are the leading cause of death and hospitalisation among New Zealand children, with indigenous Māori and ethnic minority Pacific children significantly over represented in these statistics. International research has shown that many children hospitalised for injury, as well as their families experience high levels of stress, and ethnic disparities in the quality of trauma care are not uncommon. The research on which this paper is based sought to identify key issues and concerns for New Zealand's multi-ethnic community following hospitalisation for childhood injury in order to inform efforts to improve the quality of trauma services. This paper reports on service providers' perspectives complementing previously published research on the experiences of families of injured children. Methods A qualitative research design involving eleven in-depth individual interviews and three focus groups was used to elicit the views of 21 purposefully selected service provider key informants from a range of professional backgrounds involved in the care and support of injured children and their families in Auckland, New Zealand. Interviews were transcribed and data were analysed using thematic analysis. Results Key issues identified by service providers included limited ability to meet the needs of children with mild injuries, particularly their emotional needs; lack of psychological support for families; some issues related to Māori and Pacific family support services; lack of accessible and comprehensive information for children and families; poor staff continuity and coordination; and poor coordination of hospital and community services, including inadequacies in follow-up plans. There was considerable agreement between these issues and those identified by the participant families. Conclusions The identified issues and barriers indicate the need for interventions for service improvement at systemic, provider and patient levels. Of particular relevance are strategies that enable families to have better access to information, including culturally appropriate oral and written sources; improve communication amongst staff and between staff and families; and carefully developed discharge plans that provide care continuity across boundaries between hospital and community settings. Māori and Pacific family support services are important and need better resourcing and support from an organisational culture responsive to the needs of these populations.

Competing interests The authors declare that they have no competing interests. Authors' contributions SAm, SAb and SC conceived of and designed the study and obtained funding support. SAm and SAb analysed the data and drafted the manuscript as joint lead authors. STT assisted with the data analysis and drafting of the manuscript. LA and SM participated in the conduct of the study and data collection. All authors critically reviewed the draft manuscript, and read and approved the final manuscript. Authors' Information SAm is based at the Section of Epidemiology & Biostatistics, School of Population Health, University of Auckland. All other authors were based at the Department of Maori & Pacific Health of the Faculty of Medical & Health Sciences, University of Auckland, during the conduct of this study. LA is now based at the Centre for Social and Health Outcomes Research & Evaluation, Massey University. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1472-6963/10/333/prepub

Acknowledgements We gratefully acknowledge the contribution of interviewers, investigators and advisors involved in the Child Injury Project, in particular, Brooke Arlidge, Robyn Dixon, Julie Park, Nicola Taylor, Teuila Percival, Rhys Jones and Tania Riddell. The funding for all researchers was provided by a project grant awarded by the Health Research Council of New Zealand.

CC BY

no

2022-01-12 15:21:45

BMC Health Serv Res. 2010 Dec 7; 10:333

oa_package/c1/8e/PMC3016356.tar.gz

PMC3016357

21143853

Background Acute Myocardial Infarction (AMI) is a health problem of paramount importance in terms of frequency, seriousness, social and economic costs, amenability to medical intervention and priority-ranking by policy makers and the community [ 1 , 2 ]. Although guidelines that effectively reduce AMI in-hospital case-fatality rates (AMI-CFRs) exist, they are not uniformly applied [ 3 - 7 ]. Conversely, reports of AMI-CFRs were thought to favorably affect quality initiatives [ 8 ]. The relationship between better care and lower AMI-CFRs led several countries, national and international agencies, and consumer's organizations to select AMI-CFR as a quality indicator and to publish individual providers' rates [ 3 , 9 - 11 ]. To organize the care of AMI patients in Belgium, a three-level structure has been set up. It consists of community hospitals (labeled A) having no catheterization facility, intermediary hospitals (labeled B1) providing coronary angiography but no Percutaneous Coronary Intervention (PCI), and tertiary hospitals (labeled B2-B3) offering PCI and/or Coronary Artery Bypass Graft. A recent study, however, showed unable to establish better outcomes for patients treated in services with catheterization facilities [ 12 ]. Current professional knowledge at the time of our study (2002-5) included the need to distinguish between ST-segment elevation myocardial infarction (STEMI) and non-ST-segment elevation myocardial infarction (NSTEMI), because of their different prognosis; the need for reperfusion as quick as possible, by PCI where possible or by thrombolysis otherwise; and, the safety of transport of patients from community to tertiary hospitals [ 13 - 15 ]. Changes in definition and the diversity of the various cardiac troponin assays may have heavily affected AMI incidence rates and AMI-CFRs [ 16 - 20 ]. At the time the study data were registered and coded by the hospitals, the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) did not distinguish between STEMI and NSTEMI, but as from October 2005 new ICD-9-CM regulations provide guidance to do so [ 21 ]. Herein STEMI is coded as 410.0-6, 410.8; NSTEMI as 410.7; and AMI "Not Otherwise Specified" (AMI-NOS) as 410.9. It became then possible for one of the authors (NT) to recode a posteriori the data in terms of STEMI or NSTEMI. According to the OECD "Case-fatality rates measure the proportion of patients with a given diagnosis, here acute myocardial infarction (AMI), who die within a specified time period, here 30 days [ 11 ]. Ideally, the case-fatality rates would be based on each individual patient who would be tracked for at least 30 days. However, as most countries do not have unique patient identifiers and lack the ability to track patients after hospital discharge, the indicator is based on unique hospital admissions and restricted to mortality within the hospital. Thus, differences in practices in discharging and transferring patients may influence the findings [ 11 ]". Due to the specifics of our administrative data and in alignment with the Agency for Healthcare Research and Quality (AHRQ) definition of the Acute Myocardial Infarction Mortality Rate, our definition of AMI-CFR is exclusively based on hospitalized cases and fatalities within the hospital regardless of any time constraint [ 9 , 22 ]. A fair inter-hospital comparison requires high-quality data [ 23 , 24 ]. Indeed, they have to be sufficiently detailed to allow fair comparison between a hospitals' practices and the then professional knowledge but also to assess the hospitals' organizational performance, especially its transfer policy and the symptom-onset-to-balloon time [ 25 ]. Risk-adjustment constitutes a constant threat to the fairness of the comparisons and widely used proprietary risk-adjustment systems may lead to erroneous conclusions [ 26 , 27 ]. Accountability of caregivers and health authorities to the community is internationally considered of paramount importance [ 28 ]. In spite of known limitations, public reporting of comparative information about the quality of health care, often derived from administrative data, is frequently put forward as an important quality improvement tool, which attempts to stimulate caregivers to grade up the provision of services and to reassure patients by demonstrating accountability [ 29 - 31 ]. In this context, ensuring data quality is a continuous challenge especially if the same data are used for reimbursement and for measuring quality [ 31 , 32 ]. In an effort to encourage the hospital system to assume responsibility, the Belgian Ministry of Public Health decided to stimulate initiatives of quality improvement. Hereto a limited set of indicators was selected from the AHRQ Inpatient Quality Indicators, including the AMI-CFR [ 33 ]. These indicators were to be routinely analyzed, using routinely collected, Belgian hospital discharge records that are known to differ across institutions in quality of data [ 34 ] and recognizing that the above-mentioned differences in practices in discharging and transferring patients may influence the findings. In this study we explore the AMI-CFRs' potential as a quality-improving tool. More precisely we aimed, by determining the existence of inter-hospital differences in AMI-CFR, (1) to evaluate to which extent Belgian discharge records allow the assessment of quality of care in the field of AMI, and (2) to identify starting points for quality improvement.

Methods Data sources Belgian hospitals are required to register discharge data on each sojourn, which are processed and stored in a dataset called the Minimal Clinical Data (MCD). The MCD contains patient data (among which year of birth, gender, residence, and anonymous hospital and patient identifiers) and stay data (among which year, month and day of the week but, due to privacy restrictions, not the precise date of admission and discharge; length of stay; transfer to another hospital with specification of the type of hospital). It further includes an unbounded number of ICD-9-CM coded diagnoses and procedures but neither results of laboratory investigations, such as cardiac enzymes, nor of technical examinations such as electrocardiograms. Due to privacy regulations a unique identifier, allowing for follow-up of a transferred patient, is lacking. Nonetheless, these data allow for the identification of a cardiogenic shock, the most feared complication of AMI, as well as for the computation of the Charlson co-morbidity index (CCI) (see additional file 1 : Charlson), particularly well suited to study AMI-CFRs [ 2 , 20 , 35 - 37 ]. Incompleteness and inaccuracy are well-known drawbacks of administrative data; therefore, we compared the MCD data with those from the Ghent and Bruges registries for acute coronary events, which are collected according to the protocol that was originally developed in the WHO Multinational Moni toring of Trends and Determinants in Ca rdiovascular Diseases (MONICA) project [ 16 , 38 ] (see additional file 2 : MONICA definitions). In this project the record form is intentionally kept as simple as possible [ 38 ]. It consists of (1) items characterizing the person (date of registration, gender, date of birth and data of onset of the acute attack) and of (2) medical and diagnostic data (whether it was an hospital case or managed elsewhere; whether it was a first or a recurrent event; whether the patient survived at 28 days or not; determination of the diagnostic category based on symptoms, electrocardiograms, serum enzymes and necropsy findings). In addition, the registries of Ghent (covering the city of Ghent (about 250,000 inhabitants)) and Bruges (covering the district of Bruges (about 260,000 inhabitants)) record whether or not a revascularization procedure (PTCA or thrombolysis) was carried out. Coding of all items is checked by both external and internal quality control [ 38 ]. In the light of these uniform registration practices and the quality control carried out, we consider the MONICA registries the standard against which we evaluated the MCD. For the years 2002-2004 and for the age groups between 25 and 74 years, we compared the MONICA hospitalized fatalities (fatal definite (F1) + fatal possible (F2)), the numerator of the AMI-CFR, and MONICA hospitalized cases (non-fatal "definite" (NFl) + "possible" (NF2) + fatal "definite" (F1) + "possible" (F2)), the denominator of the AMI-CFR, with respectively the AMI fatalities and cases in the same age groups of the MCD. Unlike the AHRQ definition, a case fatality in MONICA is defined as death within 28 days after the occurrence of the first symptoms [ 38 ]. In the framework of routine analyses of the MCD, for reasons of monitoring over the years, and to comply with the OECD requirements, we preferred to stay in line with the AHRQ model, adopted by the OECD, recognizing the difference in definition of in hospital AMI-CFR between MCD and MONICA. However in Belgium, this difference leads to the very small 0.5% difference in AMI-CFR [ 39 ]. Due to privacy regulations and according to the agreement with the hospitals not to perform any analysis on hospital level, we were unable to compare both databases neither on a hospital nor on an individual level. However, to formally assess the completeness of the data we intended to compare on register level the number of cases as well as the case-fatality rates of the Ghent and Bruges registries with the corresponding MCD. To do so log-linear models were to be fitted with the number of cases as the dependent variable, and gender, place and registry as covariates. The need to include registry as a covariate was to be examined by leaving registry out from the explanatory variables. Fitting log-binomial regression models [ 40 ] would allow us to observe a possible difference in risk ratio between both registries and MCD, implying either an underestimation or an overestimation of case-fatality rates by the MCD. Notice that, according to the MONICA protocol, fatalities occurring within the first hour after hospitalization were recorded by the registries of Ghent and Bruges. Definition of the study population Building on the work of the Agency for Healthcare Research and Quality (AHRQ) [ 9 ] we selected, for the years 2002-2005, from the MCD all stays having AMI as principal diagnosis (ICD-9-CM code 410.*1, the fifth digit indicating an initial episode of care). Cases with no information regarding vital status at discharge or aged less than 18 years or, to avoid double counting of patients when computing rates, transferring-out to another short-term hospital were excluded. We also excluded short-term hospitals registering less than 20 cases a year. Applying these criteria we obtained our study population consisting of 46,287 cases (30,271 males and 16,016 females) and 7,099 fatalities (3,841 males and 3,285 females), registered in 109 short-term hospitals. Design Our aim was to identify, at the same time, hospitals with higher quality of care, for benchmark and exemplary function purposes, and hospitals with lower quality, to help them improve their performance. To this effect, the AMI-CFR of each individual hospital was compared with the corresponding rate of the whole of the other Belgian hospitals. To assess hospital-specific rates of AMI-CFR a cohort study of all hospitalized AMI cases of the years 2002 to 2005 was carried out. Hereby, two types of analyses were performed: a global one focusing on the AMI-CFRs of the entire time span of the study (the "period") and a temporal one (the "trend"), focusing on its per-semester evolution and intended to refine the initial, global assessment by taking temporal evolution into account. It has been suggested that in analyses, founded on administrative databases, confounding effects of unmeasured or mismeasured variables can be equivalent in magnitude to the effect of the association of interest and cannot be ruled out as an explanation of rather small, yet statistically significant effect sizes, such as a relative risk (RR) of, say, 0.75 to 1.35, which are roughly each other's inverses [ 41 ]. To take this caution against over-interpretation into account, we defined an inconclusive zone, where the AMI-CF rate or AMI-CFR trend of a hospital, compared with these of the other hospitals, should not be interpreted as being "higher" or "lower," but rather be considered "inconclusive." First, we computed a hospital's relative risk (RR) of having a higher/lower period or trend AMI-CFR than those of the other hospitals. Subsequently, this RR allowed us to calculate a "departure," equaling (RR-1) × 100, which we used to define an "inconclusive zone." Regarding the period, we fixed the lower and upper boundaries of the inconclusive zone at a departure of - 25% and of + 35%, respectively, in line with the aforementioned relative risks. In the absence of literature regarding important departures of the trend, we arbitrarily fixed these boundaries at - 5% and + 5%, allowing for a maximal increase of + 41% or for a maximal decrease of -30%, respectively, of the AMI-CFR during the entire time span of the study. The technical details of our approach are described in the paragraph devoted to the statistical analysis. The results of the analyses regarding hospitals outside the inconclusive zone, were interpreted according to the degree of statistical evidence. We labeled this evidence as: 1) "strong," if the probability of finding a departure, as important or bigger than that of the hospital under consideration, was smaller than or equal to 0.05 divided by the number of hospitals to be compared (the so-called Bonferroni correction for multiple comparisons [ 42 ]); 2) "moderate," if this probability was smaller than or equal to 0.05 but greater than 0.05 divided by the number of hospitals to be compared; 3) "weak," otherwise [ 34 ]. For both the period and the trend analyses, two AMI-CFRs were computed: a first one in which transferring-out cases are excluded, so as not to compute national rates with duplicate observations [ 9 ], and a second one in which all transferred cases are excluded, to allow for a comparison between hospitals in the subset of cases they treated from uptake to discharge. We also aimed at identifying transferred patients using composite keys consisting of a patient's year of birth; gender; residence; year and month of admission and discharge; and day of the week of discharge ("discharge" facility's data) or admission ("intake" hospital's data). In these patients, we intended to compare the CCI distributions before and after transfer to evaluate the comparability across hospitals regarding the coding of co-morbidity, a component of the risk-adjustment in our regression analyses. From the viewpoint of quality of care, we also compared in these patients the time span between admission and transfer with the then prevailing guidelines. Statistical methods For our analyses, we used so-called fixed-effects models, motivated by the fact that the entire population of Belgian hospitals is considered, rather than sampling from them. We performed multivariable logistic regressions [ 43 ], aiming to identify outlying hospitals, i.e. characterized by an important and statistically significant, Bonferroni-corrected [ 24 ] departure from the other hospitals' AMI-CF rate or AMI-CFR evolution over time. Given that we cover the short time span of only eight semesters, we merely fitted models with a linear time trend, which for brevity we called "trend". By incorporating an interaction term in the logistic regression between a linear time trend, expressed in semesters, and individual hospitals, those hospitals with an abnormal evolution in time were identified. Precisely, we compared the slope of each hospital's time trend with that of the other hospitals using linear contrasts. When the outcome of interest, such as the percentage of fatal cases in the AMI case fatality rate, exceeds 10%, or if the odds ratio is greater than 2.5 or less than 0.5, the estimation of the relative risk by the adjusted odds ratios, derived from the logistic regression no longer adequately approximates the relative risk, which may become heavily biased. Indeed, the more frequently that outcome occurs, the more the odds ratio overestimates the relative risk when it is greater than 1 or underestimates the relative risk when the odds ratio is less than one [ 44 ]. To reduce this bias, we used the approximation of the RR by Zhang [ 44 ], which has been used to compute the aforementioned departure. The relation between RR Z and the odds ratio is given by RR Z = OR/((1-P0) + (P0 *OR), where P0 indicates the incidence of the outcome of interest in the non-exposed group [ 44 ]. Following the AHRQ model [ 22 ], we carried out two types of main analyses: a first one wherein transferred-out cases were excluded and a second one with exclusion of all transferred cases. In each analysis, adjustment was made for five-year age groups, gender, per-semester evolution of the AMI-CFRs, and shock. To account for correlation within the data, rescaling techniques were used [ 45 ]. To study a possible national upward trend we used so-called Generalized Estimating Equations (GEE), a refinement of the logistic regression that corrects for within-hospital, generally within-unit, correlation [ 46 ]. Feedbacks The results of the analysis serves two purposes: feedback to the hospitals to enable improvement of care, and feedback to the Belgian College of Physicians to identify hospitals with "strong" evidence of either superior or inferior quality. The feedback to the hospitals consists mainly of (1) a graphical display of the "departure" of each of the hospitals from the rate and trend of the other hospitals, and (2) an anonymous and tabular representation of these departures as well as of an indication of the level of statistical evidence. An aid in the interpretation, combining the information of both the period and trend analyses, is provided alongside. Its decision tree is given in the Appendix of the Supplementary Materials (see additional file 3 : decision tree). In the feedback to the Belgian College of Physicians, we present an "Average" and two "Outlying" categories of hospitals. For the period analysis, a first, outlying category, the 'high AMI-CFR' group, consists of those hospitals with a departure of >+35% and statistically significant (Bonferroni-corrected). A second, outlying category, the 'low AMI-CFR' group, consists of those hospitals with a departure of <-25% and statistically significant (Bonferroni-corrected). The other hospitals are grouped into the 'average AMI-CFR' group. A similar approach is followed regarding the trend, with boundaries at +5% and -5%. The decision tree is identical to that, used in the feedback to the hospitals, except that hospitals recommended for an external audit are now divided in "high AMI-CFR" and "low AMI-CFR" groups and that the other hospitals are regrouped in an "average AMI-CFR" group. Sensitivity analyses Due to the prognostic differences, related to the type of AMI, its complications and treatment modalities, several sensitivity analyses were carried out. After an initial analysis, modeling AMI-CFR adjusted for age, gender, CCI and time (period as well as trend), subsequent analyses were carried out, consecutively introducing a variable mentioning whether shock was present at admission or not, and a variable indicating whether or not an angioplasty had been performed. We repeated the subsequent analyses, this time replacing shock by a variable describing the type of AMI i.e. STEMI, NSTEMI or AMI NOS. We performed these analyses twice: firstly after exclusion of the transferred-out cases and subsequently after exclusion of all transferred cases. In the trend analyses, we additionally modeled an interaction term between trend and individual hospitals. To summarize the effect of including additional explanatory variables into the model, we classified for each model the hospitals into seven categories, ranging from strong, then moderate, to weak evidence of finding themselves above or below the inconclusive zone. A last category consisted of hospitals showing no interpretable departure. In so doing, we were able to compare between the hospitals their changes of category due to the modeling process. Because differences in hospitals' discharge policy may influence their AMI-CFRs we also compared the hospitals' cumulative AMI-CFR at five points in time, namely at the 7th, 14th, 21st, and 28th day of hospitalization, and at the latest day of discharge of any AMI-case [ 11 ]. Given the changing numerators of the rate over time, we based our comparison on the hospitals' change in deciles of AMI-CFR. To this end, we computed a maximum change in relation to the initial AMI-CFR. We further determined for each hospital the maximum length of stay (LOS) of any patient whether deceased or not, and the maximum LOS of any deceased patient, both after exclusion of the transfer-out cases and after exclusion of all transferred cases. To provide an idea of the variability of some major covariates we determined across hospitals and by level of care, averages of the lower quartile, mean, median and upper quartile of the CCI, LOS and rate of shock. In the same way we computed the average 95th CCI-percentile and the ranges of the LOS. The study being (1) of a retrospective, non-interventional type and (2) anonymous with respect to both hospitals and patients, no approval by an ethics committee is required under the Belgian law.

Results Patient and hospital characteristics by level of care The MCD dataset consisted of 46,287 cases and 7,099 fatalities [AMI-CFR: 15.34 (95% CI: 15.01;15.67) from 109 hospitals. The majority of cases (65,4%) and of the fatalities (54.1%) were males. In females the AMI-CFR was higher than in males [age-adjusted OR: 1.13 (95% CI: 1.04; 1.22)]. In tertiary hospitals, younger age groups (Pearson chi-square with 10 df = 794.7; p < 0.0001), males (Pearson chi-square with 2 df = 214.5; p < 0.0001) and shock (Pearson chi-square with 2 df = 6.29; p = 0.043) are over-represented (Table 1 ). The age-adjusted mortality was higher in Type A [OR:1.17; 95%CI(1.09;1.26)] hospitals than in Type B2-B3 hospitals. However, we were unable to establish a significant difference in age-adjusted mortality between Type A en Type B1, and between Type B1 and Type B2-B3 hospitals (Table 1 ). Males had lower mortality figures than females [OR:0.58; 95%CI(0.55;0.61)]. The sex-adjusted mortality was higher in Type A [OR:1.37; 95%CI(1.30;1.45)].and Type B1[OR:1.18; 95%CI(1.10;1.27)] hospitals than in Type B2-B3 hospitals. Type A hospitals also displayed a higher sex-adjusted odds ratio [OR:1.16; 95%CI(1.08;1.26)] than Type B1 (Table 1 ). Type B2-B3 hospitals had a lower CFRs in case of cardiogenic shock {(OR A vs B2-B3 :1.14 [95%CI(1.23;1.67)] and OR B1 vs B2-B3 :1.89 [95%CI(1.51;2.21)])}, and Type A hospitals also had a lower CFR than Type B1 (OR A vs B1 :0.76 [95%CI(0.60;0.96)]). Type B2-B3 hospitals had also a lower CFRs in case of absence of cardiogenic shock {OR A vs B2-B3 :1.70 [95%CI(1.58;1.82)] and OR B1 vs B2-B3 :1.32 [95%CI(1.20;1.45)] }, but now Type A hospitals had a higher CFR than Type B1 (OR A vs B1 : 1.29 [95%CI(1.17;1.41)]) (Table 1 ). With respect to STEMI cases we observed higher AMI-CFRs of Type A and B1 versus Type B2-B3 hospitals {(OR A vs B2-B3 :1.36 [95%CI (1.27;1.47)] and OR B1 vs B2-B3 :1.22 [95%CI (1.12;1.34)])} and of Type A versus Type B1 hospitals (OR A vs B1 : 1.11 [95%CI (1.01;1.22)]). In NSTEMI cases we observed substantially lower AMI-CFRs in Type B2-B3 hospitals {(OR A vs B2-B3 :2.01[95%CI (1.79;2.26)] and OR B1 vs B2-B3 :1.84 [95%CI (1.55;2.18)])} (Table 1 ). In contrast we were unable to demonstrate either a difference in AMI-CFR between Type A and Type B1 hospitals or evidence of any deviating AMI-CFR in cases of AMI-NOS. The proportions of cardiogenic shock were very high, viz 9.9% in STEMI, 4.2% in non-STEMI infarctions and even 12.7% in the AMI NOS group. On the hospital level we noticed huge variability across institutions of the same type regarding CCI, length of stay and to a lesser extent shock (expressed in %). However the distribution of these variables was very similar in the three types of hospitals (data not shown). We also observe a huge variability in volume, i.e the number of cases, within and between types of hospital (Table 1 ). Based on the volumes of the lower quartile, the median and the upper quartile of the Type B1 hospitals in the data, transfers out excluded, we grouped the hospitals in four classes. We observed a huge variability in volume between and within Types of hospital (Pearson chi-square of 300019 with 6 df;p < 0.0001) with excesses of low volume hospitals in Type A hospitals (41 out of the 61 of these hospitals found themselves in the lowest class) and of high volumes in Type B2-B3 hospitals (26 out of the 29 of these hospitals found themselves in the highest class). The Type B1 hospitals were almost equally distributed over the four classes. Similar findings were encountered in the analysis on the data with exclusion of all transfers. Comparison between MCD and MONICA The comparison of the MCD with MONICA (Ghent and Bruges) shows important differences between both datasets (Table 2 ), characterized by more fatalities and significant higher case-fatality rates in the MONICA registry. Notice that in the MONICA registry 54 out of the 190 fatal cases occurred within the first hour after hospitalization. At that moment these patients probably found themselves in the emergency services, where in Belgium as yet no specific ICD codes are systematically registered. Regarding the number of cases, we were unable to determine a significant difference between both datasets, the difference in the Pearson chi-squared statistics of our log-linear models being not significant (p = 0.19). Conversely, we observed a significant difference in risk ratio implying an important underestimation of case-fatality rates by the MCD (RR:0.39[95%CI:0.31;0.51]). However in the absence of a significant interaction term between registry and place (p = 0.82) we were unable to establish a differential underestimation according to place. Leaving out the fatalities occurring within one hour after admission led to similar results. Notice that the number of (mortality) cases registered in the MCD alternately was larger or smaller than those of MONICA. We were also able to compare PTCA figures between MONICA and MCD. For the years 2002-4 MONICA-Ghent registered 83 PTCAs in males and 19 in females versus respectively 102 and 20 PTCAs in the MCD. For Bruges these numbers amounted to 87 and 20 in MONICA versus 149 and 30 in the MCD. We paired the data and computed their proportions as well as the corresponding 95% confidence interval [ 47 ]. Apart from the numbers for females in Ghent, the MCD numbers of PTCA exceeded significantly (the lower bound of the CI being ≥1) those of MONICA. For the same years we also compared the recurrent events registered in MONICA with the AMI cases in the MCD mentioning in addition an "old myocardial infarction" (ICD-9-CM code: 412) as a secondary diagnosis. This exercise also showed substantial differences between MONICA and MCD. In Ghent MONICA totaled 53 recurrent events in males and 11 in females, versus 22 and 3 respectively in the MCD. For Bruges these numbers amounted to 76 and 14 in MONICA versus 49 and 6 in Bruges. Each time the MONICA figures significantly outnumbered (the lower bound of the CI being ≥1) those of the MCD. Transfers From the MCD data, before the exclusion of transfer cases, we identified 6,555 stays mentioning only a transfer-in from another hospital, 12,409 stays mentioning only a transfer-out to another hospital, and 4592 mentioning both a transfer-in and a transfer-out. We only succeeded to pair respectively 2,524, 4592 and 2657 transfers. In the subset of 2,524 paired transfer-in stays, we also assessed the accuracy of our co-morbidity data. The first quartile, median and third quartile of the distribution of the CCI amounted to 0, 0 and 2 in the "discharging" and to 0,1 and 2 in the "intake" facilities respectively. We also determined the time of referral, which in over 50% of the cases exceeded 24 hours. Sensitivity analyses In 27 hospitals neither the adjustments used nor the stays excluded according to type of transfer, altered the category of evidence to which they belonged. In 45 hospitals the change did not exceed 2 categories; in 31 hospitals the change equaled 3 categories, without crossing the overall rate; and, in 5 tertiary hospitals the change amounted at least to 3 categories and crossed the overall rate. In the latter cases poorer AMI-CFRs were found when adjustment was made for shock or STEMI and for carrying out an angioplasty. The range of the point estimators of the departures was considerable and varied substantially across the fitted models (Table 3 ). In a first series of analyses, wherein transferred-out cases, were excluded, the cumulative AMI-CFR, adjusted for age, gender, CCI and shock, showed a huge inter-hospital variability in AMI-CFR over time. The maximum change in deciles amounted to 0, 1, 2, 3 and 4 or 5 deciles in 21, 49, 25, 12 and 2 hospitals respectively. Analogously, after exclusion of all transferred cases, a maximum change in deciles of 0, 1, 2, 3, 4 or 5 was observed in 24, 58, 15, 11 and 1 hospitals, respectively. Similarly, excluding the transferred-out cases, we observed across hospitals ranges of the maximum LOS of any patient from 36 to 337 days and from 12 to 233 days in deceased patients. Very similar ranges from 36 to 243 days and from 12 to 232, respectively, were obtained after exclusion of all transferred cases. Determinants and evolution of the national AMI-CFR Shock (adjusted ORs of 23.0[95%CI: 20.9;25.2] and of 21.0 [95%CI: 19.0;23.2]) proved to be by far the strongest determinant of AMI-CFR in all the models, followed by age (adjusted ORs between 1.35 and 1.37), and severe co-morbidity (for example in the Initial model 1 a CCI of four has an adjusted OR of 1.09 4 = 1.41. Indeed, CCI being modeled as an interval variable, an adjusted odds ratio between successive levels of CCI equals 1.09.) (Table 4 ). The adjusted odds ratios of gender and trend, although statistically significant, found themselves in the inconclusive zone. Inter-hospital comparison We observed substantial and statistically significant inter-hospital differences in AMI-CFRs, both in the period and the temporal analysis, based on a model with age, gender, CCI and shock as explanatory variables. These differences arose in both the models with exclusion of transferred-out cases and those wherein all transferred cases were excluded. From the model wherein all transferred cases were excluded, we represented in Figure 1 the Bonferroni-corrected, outlying hospitals by green diamonds ("high") or blue circles ("low") and the other hospitals by red squares ("average"). Departures ranged from -65% up to +196% in the period, and from -47% up to +39% in the trend analysis, resulting in seven "high AMI-CFR" and nine "low AMI-CFR" outlying hospitals in the period and one "high AMI-CFR" and three "low AMI-CFR" outlying hospital in the trend analysis. For the model with exclusion of transferred-out cases, we observed departures ranging from -60% to +185% in the period and from -45% to +30% in the trend analysis, resulting in four "high AMI-CFR" and eight "low AMI-CFR" outlying hospitals in the period and one "high AMI-CFR" and one "low AMI-CFR" outlying hospital in the trend analysis. The same analyses, performed on the subset of tertiary level hospitals, also showed marked and statistically significant inter-hospital differences in AMI-CFRs in both models. In the model with transfer-out cases excluded, the departures in the period analysis ranged from - 36% to +70%, resulting in two "high AMI-CFR" and one "low AMI-CFR" outlying hospitals. In the trend analysis no significant departures were observed. In the model with all transfers excluded we observed one "high AMI-CFR" outlying hospital in the period analysis (departure +60%) and one "low AMI-CFR" outlying hospital in the trend analysis (departure - 14%). According to type of cardiac care provided, we found important, i.e. beyond the inconclusive zone, and statistically significant differences between community and tertiary level hospitals, except for "Initial 1 model" (Table 4 ) where, in spite of a statistical significant difference, no firm conclusion could be made, the excess being situated in the inconclusive region. The differences between intermediary and tertiary level hospitals fell outside the inconclusive zone for the models wherein all transfers are excluded { model "Initial 2": OR B1 vs B2-B3 :1.38 [95%CI (1.25;1.52)]) and model "Shock 2" OR B1 vs B2-B3 :1.48 [95%CI (1.33;1.65)])} (Table 4 ). The fitted models showed very good discrimination properties reflected by areas under the receiver operating characteristic (ROC) curves between 0.74 and 0.85 [ 48 ].

Discussion and Conclusions Data quality of the MCD Our results suggested numerous quality-of-data-related problems and prompted a very cautious interpretation of the quality-of-care-related findings. At first, AMI is characterized by diagnostic uncertainty, manifesting itself in marked differences in sensitivity and specificity of the measurement of myocardial proteins, ECG recordings and imaging modalities [ 18 , 49 ]. This may be reflected by the denominator differences of the AMI-CFR we found between the MCD and MONICA data. Part of these differences is probably due to the absence of national guidelines leaving the choice of the diagnostic criteria to the individual clinicians, whereas in MONICA univocal diagnostic criteria were used. In a German study [ 50 ], similar to ours, the hospital's AMI-CFR amounted to 13.5% versus the 28% derived from the German MONICA data, whereas in ours these rates amounted to 7.5% and 19,0%, respectively. The comparison between MONICA and the MCD, unfortunately very limited, due to privacy and agreement regulations, shed light on other quality-of-data problems. The outnumbering in PTCA of the MCD may illustrate the need of univocal definitions of AMI and coding practices and may reflect the propensity of administrative data for maximizing coding. By contrast the deficient registering of old myocardial infarctions, which is financially not rewarding in our prospective payment based reimbursement of the hospitals, comes as no surprise. We also tried to assess as much as possible the quality-of-data through internal comparisons. Thus and contrary to the findings in a recent publication stressing the sensitivity of the CCI to coding practices [ 27 ], our comparison of the CCI distribution in transferred patients did not suggest too large before-after transfer differences. However, our sensitivity analyses seemed to indicate a sizable inter-hospital variability in the accuracy of the data. The completeness of the data also was a cause for concern. Since about 50% of the fatal STEMI cases occur in the first two hours after the onset of the symptoms, the under-registration of fatalities may be due to the lack of ICD-coding of the pathologies during a stay in an emergency department [ 4 ]. Important information such as symptom-onset to needle-time, the time lag between the onset of symptoms and the initiation of the treatment, is lacking in our data, as well as pharmacological treatment details, which are essential with respect to quality assessment, especially in primary hospitals. Also, although low socio-economic status (SES) is related with lesser AMI-outcomes, we have no information regarding SES [ 51 ]. Further, one has to take the basic tension into account that exists between using the same data for reimbursement of the hospitals and for quality improvement purposes. Indeed, in the former case the purpose is to maximize the coding of complications and co-morbidities (e.g., CCI), while in the latter it is to restrict it to conditions really affecting care [ 32 ]. Our shock data may well illustrate this phenomenon. For example, we observed rates of 9.9% and 4,2% in STEMI and non-STEMI infarctions, respectively, which are substantially higher than the 5 to 8.6% and 2.5% in the literature, respectively, and to which, in addition, the 12.73% of the AMI-NOS are to be added [ 20 , 37 ]. The doubling of the reimbursement rate in these cases reflects the possible financial gains through maximization, but the high CFRs in case of cardiogenic shock seem not to indicate important efforts of maximization. Whereas in Belgium it is impossible to routinely obtain 30-day mortality data of discharged patients, they are very well comparable to the in-hospital mortality data [ 39 ]. By contrast, the comparison with MONICA revealed a substantial proportion of probably early fatalities not registered in the MCD. As a result, the in-hospital AMI-CFR may chiefly be biased in the initial phases of the fatal process, preventing the in-hospital AMI-CFR to be used as a reliable estimator of the population AMI-CFR. In that respect population-based, acute coronary syndrome registries remain irreplaceable. Since we were not able to assess whether the proportion of unrecognized, early AMI deaths is equally distributed across hospitals, the inter-hospital comparison may, apart from unevenly distributed, dubious coding practices [ 34 ], be biased as well. Statistical modeling To achieve the fairest possible inter-hospital comparison we performed two types of multivariable logistic regressions leaving out respectively transferred-out cases, and all transferred cases. In both types of analysis a risk-adjustment for age, gender, CCI and cardiogenic shock was performed. All the models had a C-index, a measure of the discriminative performance of the models, of 0.832 to 0.844, that are comparable to those obtained in a study evaluating five risk scores (PURSUIT, GUSTO-1, GRACE, SRI and EMMACE) for risk stratification of acute coronary syndromes [ 52 ]. The authors of this study conclude that simpler risk models had comparable performance to the more complex ones [ 52 ]. The Simple Risk Index (SRI) for instance, which consists of age, systolic blood pressure and heart rate, is very similar to our here-mentioned model. In our models, we kept away from inserting either angioplasty - not to model away the effect of angioplasty - or STEMI, because of both its correlation with cardiogenic shock and its difficult-to-handle class of AMI-NOS (about 13% of our study population and a case-fatality rate of 86.2%), prone to biased analyses [ 53 ]. To maximally avoid the inclusion of iatrogenic, in-hospital complications into the risk-adjustment, we also preferred the CCI, based exclusively on chronic conditions to widely used discharge abstract-based software, which fail to distinguish co-morbidities from complications [ 26 , 27 ]. The choice of including cardiogenic shock into the risk-adjustment is an arguable point; being much more often a non-iatrogenic phenomenon than a complication, we decided to incorporate it into our risk-adjustment. Indeed a recent article devoted to cardiogenic shock stated that MI with LV failure remains the most common cause of CS but also that approximately three fourths of patients with CS complicating MI develop shock after hospital presentation of which in some, medication use contributes to the development of shock [ 20 ]. Hierarchical models, usually taking the form of so-called random-effects models, would have been an alternative to the logistic regression approach we used. However, outlier detection based on such models is currently methodologically underdeveloped. Indeed, the theory dealing with outliers still has to be further developed for linear mixed models, with even less development for non-linear mixed models [ 46 , 54 ]. Finally, in the random-effects models the hospitals in the set of data are considered a random sample from the larger population of all hospitals, contrary to fact in our study wherein the entire population of Belgian hospitals is considered, rather than sampling from them. Inter-hospital comparison Two methods of calculating AMI lethality are included in the AHRQ's Inpatient Quality Indicators. The first one (excluding transferred-out cases) ensures the inclusion of all AMI patients. The second method (excluding all transferred cases) reflected the desire of users to have an alternative method of measuring AMI mortality that excluded patients transferred from another hospital. However, this approach results in the loss of transferred AMI patients from any quality measurement. Therefore, in order to allow both types of interpretation we presented the results obtained from both methods. While important, the above-mentioned quality-of-data limitations led inescapably to the question whether the MCD were "good enough" to carry out an inter-hospital comparison [ 55 ]. Contextual reasons including the accountability of both the hospitals and the authorities, as guarantors of the quality of health care, brought us to do so as well as the conviction "that what cannot be measured cannot be changed" [ 56 ]. Also one had to keep in mind that administrative data, such as the MCD, not only are the most accessible comparative data source for examining all patients admitted to a hospital, but also the only ones allowing a nationwide inter-hospital comparison with at least a minimal risk-adjustment [ 56 - 58 ]. Further, the magnitude of the observed inter-hospital differences in AMI-CFRs - the highest AMI-CFR is six times higher than the lowest - and the influence of type of hospital warrant further inquiry. Indeed, the principal interest of our study in this regard rests on its ability of screening substandard or above-standard care, to be completed by a formal assessment, and its stimulating effect on initiatives of quality improvement. From the transfer exercise for instance we learned that, going against the prevailing guidelines [ 14 ], a considerable number of transfers was realized more than 24 hours after intake, suggesting the carrying out of an elective rather than a rescue angioplasty. The case of the five tertiary level hospitals, which in the sensitivity analysis showed the greatest AMI-CFR variability, may be related to this phenomenon. Indeed, it may that these centers frequently carry out this type of angioplasty, as it was precisely the inclusion in the modeling process of shock, underrepresented in this type of patients, and angioplasty that caused the change in AMI-CFR. Since this result appeared during the modeling process, one may assume it not to be an artifact but rather an expression of sub-optimal care. Another intriguing finding, which requires elucidation, was the apparent heterogeneity regarding AMI-CFR in the group of tertiary level hospitals. We further think to have gathered some evidence in favor of PCI over other treatments. Indeed, comparing our results with those of a previous study [ 12 ] and in line with a Swedish study, its practice seemed to have improved over time [ 12 , 59 ]. Finally, our analyses revealed important inter-hospital differences in medical practices but do not seem to indicate a systematic, early discharge practice of "patients about to die" in any hospital, intended to diminish its AMI-CFR. Also, the AMI-CFR of the hospital, admitting transferred out patients, seems to be protected to a certain extent by the modeling of shock, age group, gender and co-morbidity of this type of patients. Nowadays, discharge records are used to compare AMI-CFRs between countries. It may well that these data suffer from similar limitations regarding the symptom-onset to needle-time, type of AMI, and that they are used both for reimbursement of the hospitals and for quality improvement purposes. Furthermore, in case of a hospital stay, taking place in more than one department, some countries collect discharge records of these partial stays, leading to an inflation of the denominator of the rate. As a consequence, one should very cautiously interpret differences in AMI-CFRs between countries. The OECD for instance suggests that this indicator should be considered in conjunction with length-of-stay and transfer rates and recommends risk adjustment for clinical factors [ 11 ]. Quality improvement In a perspective of quality improvement, implementation of evidence-based diagnostic and therapeutic practices and outcome monitoring have been shown to gradually improve outcomes in all types of hospitals and to decrease between-hospital variation [ 60 , 61 ]. More specifically, the measuring and tracking of performance is considered relevant to physicians, hospital managers, scientific bodies and policy-makers [ 62 ] and fit in the shift of focus, observed in recent years, from the "no blame" paradigm [ 63 ] to a more aggressive approach to poorly performing caregivers [ 64 ]. Our approach consisted mainly in a cautious ordering of findings in degrees of evidence by examining a hospital's departure both over the whole time period and over time in a perspective of improvement. Conscious of the data limitations, we refer to "screening" rather than "assessing" quality of care [ 65 ]. Therefore we avoided to establish a ranking of the hospitals, in itself a dubious technique [ 66 ], and to make our results available to the public by so-called report cards. The latter not only have been shown not to significantly improve composite process-of-care indicators for AMI [ 67 ] but also to lead to a rising post-discharge mortality rate, conceivably due to discharging patients in unstable conditions [ 68 ]. Further, taking the organizational nature of adverse events into account, we provided an anonymous feedback to the clinicians, the hospital management, the Belgian College of Cardiologists and the policymakers. As an input for the College of Cardiologists, we aimed at identifying the few, biggest outliers, which constitute an operationally manageable group to scrutinize with respect to the putative superior or substandard quality of care provided, and to propose corrective measures if necessary. Not outlying hospitals, displaying substantial departures, are suggested to proceed to an internal audit. Conversely, in our opinion it is more efficient to advise the vast majority of hospitals, finding themselves in the inconclusive zone, to comply with updated evidence-based guidelines. To conclude, we are of the opinion that administrative data may provide hospitals and policy makers with enough evidence to encourage quality improvement initiatives. However, to measure progress it will be necessary to (1) routinely assess and assure the completeness and accuracy of the data; (2) to have univocal case definitions; and (3) to be able to trace patients across hospitals.

Discussion and Conclusions Data quality of the MCD Our results suggested numerous quality-of-data-related problems and prompted a very cautious interpretation of the quality-of-care-related findings. At first, AMI is characterized by diagnostic uncertainty, manifesting itself in marked differences in sensitivity and specificity of the measurement of myocardial proteins, ECG recordings and imaging modalities [ 18 , 49 ]. This may be reflected by the denominator differences of the AMI-CFR we found between the MCD and MONICA data. Part of these differences is probably due to the absence of national guidelines leaving the choice of the diagnostic criteria to the individual clinicians, whereas in MONICA univocal diagnostic criteria were used. In a German study [ 50 ], similar to ours, the hospital's AMI-CFR amounted to 13.5% versus the 28% derived from the German MONICA data, whereas in ours these rates amounted to 7.5% and 19,0%, respectively. The comparison between MONICA and the MCD, unfortunately very limited, due to privacy and agreement regulations, shed light on other quality-of-data problems. The outnumbering in PTCA of the MCD may illustrate the need of univocal definitions of AMI and coding practices and may reflect the propensity of administrative data for maximizing coding. By contrast the deficient registering of old myocardial infarctions, which is financially not rewarding in our prospective payment based reimbursement of the hospitals, comes as no surprise. We also tried to assess as much as possible the quality-of-data through internal comparisons. Thus and contrary to the findings in a recent publication stressing the sensitivity of the CCI to coding practices [ 27 ], our comparison of the CCI distribution in transferred patients did not suggest too large before-after transfer differences. However, our sensitivity analyses seemed to indicate a sizable inter-hospital variability in the accuracy of the data. The completeness of the data also was a cause for concern. Since about 50% of the fatal STEMI cases occur in the first two hours after the onset of the symptoms, the under-registration of fatalities may be due to the lack of ICD-coding of the pathologies during a stay in an emergency department [ 4 ]. Important information such as symptom-onset to needle-time, the time lag between the onset of symptoms and the initiation of the treatment, is lacking in our data, as well as pharmacological treatment details, which are essential with respect to quality assessment, especially in primary hospitals. Also, although low socio-economic status (SES) is related with lesser AMI-outcomes, we have no information regarding SES [ 51 ]. Further, one has to take the basic tension into account that exists between using the same data for reimbursement of the hospitals and for quality improvement purposes. Indeed, in the former case the purpose is to maximize the coding of complications and co-morbidities (e.g., CCI), while in the latter it is to restrict it to conditions really affecting care [ 32 ]. Our shock data may well illustrate this phenomenon. For example, we observed rates of 9.9% and 4,2% in STEMI and non-STEMI infarctions, respectively, which are substantially higher than the 5 to 8.6% and 2.5% in the literature, respectively, and to which, in addition, the 12.73% of the AMI-NOS are to be added [ 20 , 37 ]. The doubling of the reimbursement rate in these cases reflects the possible financial gains through maximization, but the high CFRs in case of cardiogenic shock seem not to indicate important efforts of maximization. Whereas in Belgium it is impossible to routinely obtain 30-day mortality data of discharged patients, they are very well comparable to the in-hospital mortality data [ 39 ]. By contrast, the comparison with MONICA revealed a substantial proportion of probably early fatalities not registered in the MCD. As a result, the in-hospital AMI-CFR may chiefly be biased in the initial phases of the fatal process, preventing the in-hospital AMI-CFR to be used as a reliable estimator of the population AMI-CFR. In that respect population-based, acute coronary syndrome registries remain irreplaceable. Since we were not able to assess whether the proportion of unrecognized, early AMI deaths is equally distributed across hospitals, the inter-hospital comparison may, apart from unevenly distributed, dubious coding practices [ 34 ], be biased as well. Statistical modeling To achieve the fairest possible inter-hospital comparison we performed two types of multivariable logistic regressions leaving out respectively transferred-out cases, and all transferred cases. In both types of analysis a risk-adjustment for age, gender, CCI and cardiogenic shock was performed. All the models had a C-index, a measure of the discriminative performance of the models, of 0.832 to 0.844, that are comparable to those obtained in a study evaluating five risk scores (PURSUIT, GUSTO-1, GRACE, SRI and EMMACE) for risk stratification of acute coronary syndromes [ 52 ]. The authors of this study conclude that simpler risk models had comparable performance to the more complex ones [ 52 ]. The Simple Risk Index (SRI) for instance, which consists of age, systolic blood pressure and heart rate, is very similar to our here-mentioned model. In our models, we kept away from inserting either angioplasty - not to model away the effect of angioplasty - or STEMI, because of both its correlation with cardiogenic shock and its difficult-to-handle class of AMI-NOS (about 13% of our study population and a case-fatality rate of 86.2%), prone to biased analyses [ 53 ]. To maximally avoid the inclusion of iatrogenic, in-hospital complications into the risk-adjustment, we also preferred the CCI, based exclusively on chronic conditions to widely used discharge abstract-based software, which fail to distinguish co-morbidities from complications [ 26 , 27 ]. The choice of including cardiogenic shock into the risk-adjustment is an arguable point; being much more often a non-iatrogenic phenomenon than a complication, we decided to incorporate it into our risk-adjustment. Indeed a recent article devoted to cardiogenic shock stated that MI with LV failure remains the most common cause of CS but also that approximately three fourths of patients with CS complicating MI develop shock after hospital presentation of which in some, medication use contributes to the development of shock [ 20 ]. Hierarchical models, usually taking the form of so-called random-effects models, would have been an alternative to the logistic regression approach we used. However, outlier detection based on such models is currently methodologically underdeveloped. Indeed, the theory dealing with outliers still has to be further developed for linear mixed models, with even less development for non-linear mixed models [ 46 , 54 ]. Finally, in the random-effects models the hospitals in the set of data are considered a random sample from the larger population of all hospitals, contrary to fact in our study wherein the entire population of Belgian hospitals is considered, rather than sampling from them. Inter-hospital comparison Two methods of calculating AMI lethality are included in the AHRQ's Inpatient Quality Indicators. The first one (excluding transferred-out cases) ensures the inclusion of all AMI patients. The second method (excluding all transferred cases) reflected the desire of users to have an alternative method of measuring AMI mortality that excluded patients transferred from another hospital. However, this approach results in the loss of transferred AMI patients from any quality measurement. Therefore, in order to allow both types of interpretation we presented the results obtained from both methods. While important, the above-mentioned quality-of-data limitations led inescapably to the question whether the MCD were "good enough" to carry out an inter-hospital comparison [ 55 ]. Contextual reasons including the accountability of both the hospitals and the authorities, as guarantors of the quality of health care, brought us to do so as well as the conviction "that what cannot be measured cannot be changed" [ 56 ]. Also one had to keep in mind that administrative data, such as the MCD, not only are the most accessible comparative data source for examining all patients admitted to a hospital, but also the only ones allowing a nationwide inter-hospital comparison with at least a minimal risk-adjustment [ 56 - 58 ]. Further, the magnitude of the observed inter-hospital differences in AMI-CFRs - the highest AMI-CFR is six times higher than the lowest - and the influence of type of hospital warrant further inquiry. Indeed, the principal interest of our study in this regard rests on its ability of screening substandard or above-standard care, to be completed by a formal assessment, and its stimulating effect on initiatives of quality improvement. From the transfer exercise for instance we learned that, going against the prevailing guidelines [ 14 ], a considerable number of transfers was realized more than 24 hours after intake, suggesting the carrying out of an elective rather than a rescue angioplasty. The case of the five tertiary level hospitals, which in the sensitivity analysis showed the greatest AMI-CFR variability, may be related to this phenomenon. Indeed, it may that these centers frequently carry out this type of angioplasty, as it was precisely the inclusion in the modeling process of shock, underrepresented in this type of patients, and angioplasty that caused the change in AMI-CFR. Since this result appeared during the modeling process, one may assume it not to be an artifact but rather an expression of sub-optimal care. Another intriguing finding, which requires elucidation, was the apparent heterogeneity regarding AMI-CFR in the group of tertiary level hospitals. We further think to have gathered some evidence in favor of PCI over other treatments. Indeed, comparing our results with those of a previous study [ 12 ] and in line with a Swedish study, its practice seemed to have improved over time [ 12 , 59 ]. Finally, our analyses revealed important inter-hospital differences in medical practices but do not seem to indicate a systematic, early discharge practice of "patients about to die" in any hospital, intended to diminish its AMI-CFR. Also, the AMI-CFR of the hospital, admitting transferred out patients, seems to be protected to a certain extent by the modeling of shock, age group, gender and co-morbidity of this type of patients. Nowadays, discharge records are used to compare AMI-CFRs between countries. It may well that these data suffer from similar limitations regarding the symptom-onset to needle-time, type of AMI, and that they are used both for reimbursement of the hospitals and for quality improvement purposes. Furthermore, in case of a hospital stay, taking place in more than one department, some countries collect discharge records of these partial stays, leading to an inflation of the denominator of the rate. As a consequence, one should very cautiously interpret differences in AMI-CFRs between countries. The OECD for instance suggests that this indicator should be considered in conjunction with length-of-stay and transfer rates and recommends risk adjustment for clinical factors [ 11 ]. Quality improvement In a perspective of quality improvement, implementation of evidence-based diagnostic and therapeutic practices and outcome monitoring have been shown to gradually improve outcomes in all types of hospitals and to decrease between-hospital variation [ 60 , 61 ]. More specifically, the measuring and tracking of performance is considered relevant to physicians, hospital managers, scientific bodies and policy-makers [ 62 ] and fit in the shift of focus, observed in recent years, from the "no blame" paradigm [ 63 ] to a more aggressive approach to poorly performing caregivers [ 64 ]. Our approach consisted mainly in a cautious ordering of findings in degrees of evidence by examining a hospital's departure both over the whole time period and over time in a perspective of improvement. Conscious of the data limitations, we refer to "screening" rather than "assessing" quality of care [ 65 ]. Therefore we avoided to establish a ranking of the hospitals, in itself a dubious technique [ 66 ], and to make our results available to the public by so-called report cards. The latter not only have been shown not to significantly improve composite process-of-care indicators for AMI [ 67 ] but also to lead to a rising post-discharge mortality rate, conceivably due to discharging patients in unstable conditions [ 68 ]. Further, taking the organizational nature of adverse events into account, we provided an anonymous feedback to the clinicians, the hospital management, the Belgian College of Cardiologists and the policymakers. As an input for the College of Cardiologists, we aimed at identifying the few, biggest outliers, which constitute an operationally manageable group to scrutinize with respect to the putative superior or substandard quality of care provided, and to propose corrective measures if necessary. Not outlying hospitals, displaying substantial departures, are suggested to proceed to an internal audit. Conversely, in our opinion it is more efficient to advise the vast majority of hospitals, finding themselves in the inconclusive zone, to comply with updated evidence-based guidelines. To conclude, we are of the opinion that administrative data may provide hospitals and policy makers with enough evidence to encourage quality improvement initiatives. However, to measure progress it will be necessary to (1) routinely assess and assure the completeness and accuracy of the data; (2) to have univocal case definitions; and (3) to be able to trace patients across hospitals.

Background In-hospital case-fatality rates in patients, admitted for acute myocardial infarction (AMI-CFRs), are internationally used as a quality indicator. Attempting to encourage the hospitals to assume responsibility, the Belgian Ministry of Health decided to stimulate initiatives of quality improvement by means of a limited set of indicators, among which AMI-CFR, to be routinely analyzed. In this study we aimed, by determining the existence of inter-hospital differences in AMI-CFR, (1) to evaluate to which extent Belgian discharge records allow the assessment of quality of care in the field of AMI, and (2) to identify starting points for quality improvement. Methods Hospital discharge records from all the Belgian short-term general hospitals in the period 2002-2005. The study population (N = 46,287) included patients aged 18 years and older, hospitalized for AMI. No unique patient identifier being present, we tried to track transferred patients. We assessed data quality through a comparison of MCD with data from two registers for acute coronary events and through transfer and sensitivity analyses. We compared AMI-CFRs across hospitals, using multivariable logistic regression models. In the main model hospitals, Charlson's co-morbidity index, age, gender and shock constituted the covariates. We carried out two types of analyses: a first one wherein transferred-out cases were excluded, to avoid double counting of patients when computing rates, and a second one with exclusion of all transferred cases, to allow the study of patients admitted into, treated in and discharged from the same hospital. Results We identified problems regarding both the CFR's numerator and denominator. Sensitivity analyses revealed differential coding and/or case management practices. In the model with exclusion of transfer-out cases, the main determinants of AMI-CFR were cardiogenic shock (OR adj 23.0; 95% CI [20.9;25.2]), and five-year age groups OR adj 1.23; 95% CI [1.11;1.36]). Sizable inter-hospital and inter-type of hospital differences {(OR comunity vs tertiary hospitals 1.36; 95% CI [1.34;1.39]) and (OR intermediary vs tertiary hospitals 1.36; 95% CI [1.34;1.39])}, and nonconformities to guidelines for treatment were observed. Conclusions Despite established data quality shortcomings, the magnitude of the observed differences and the nonconformities constitute leads to quality improvement. However, to measure progress, ways to improve and routinely monitor data quality should be developed.

Competing interests The authors declare that they have no competing interests. Authors' contributions WA conceived of the study, drafted the manuscript, and participated in the design of the study and in the statistical analysis. NT was responsible for the data management and participated in the statistical analysis. GM supervised the statistical analysis and participated in the design of the study. GDB participated in the design of the study (comparison MONICA). CV participated in the design of the study (cardiological aspects). MvS participated in the design of the study (epidemiological aspects). All authors made important contributions to the interpretation of data and the redaction of the final manuscript, and approved it. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1472-6963/10/334/prepub Supplementary Material

Acknowledgements Disclaimer The authors of this article are responsible for its content. No statement in this article should be construed as an official position of the Belgian Federal Service of Health, Food Chain Safety and Environment, Directorate-General for the Organization of Health Care Establishments.

CC BY

no

2022-01-12 15:21:45

BMC Health Serv Res. 2010 Dec 8; 10:334

oa_package/0a/06/PMC3016357.tar.gz

PMC3016358

21143949

Background Patient safety has become a priority in the past decade, and many Western countries have examined the rate of adverse events in hospital care [ 1 - 8 ]. A systematic review reported that in 9.2% of all hospital admissions one or more adverse events (incidents with patient harm) occurred, while nearly half (43.5%) of these could have been prevented and 7.4% contributed to death [ 9 ]. Up to now, the focus in patient safety has mainly been on hospital care. Most patients, however, receive their healthcare in primary care settings, particularly in countries with a strong primary care system [ 10 , 11 ]. Relatively little is known about patient safety in these settings. In a previous review, the rate of incidents in primary care has been estimated as ranging between 5 and 80 in 100,000 consultations [ 12 ]. However, the majority of the included studies used event reporting by professionals as their research method. Reporting systems considerably under-report patient safety events and are not appropriate for estimating incidence rates [ 13 , 14 ]. A recent study found evidence to suggest that the rate is much higher. This record review study reported that an incident with harm occurred every 48 consultations [ 15 ]. The literature on incidents in primary care so far has not included out-of-hours primary care. Many patient contacts with primary care occur out-of-hours. In the Netherlands, out-of-hours primary care is organised by general practitioner (GP) cooperatives involving 40 to 250 GPs. GPs working in such cooperatives are registered GPs who work in primary care practices in daytime. The GP cooperative is intended to deal with urgent requests for medical attendance that cannot wait until the next day; it is available daily from 5 pm to 8 am on weekdays and the entire weekend (see Table 1 for more characteristics). Patient safety is of particular importance in GP cooperatives because of the high patient throughput and the diversity of clinical conditions presented, which are more likely to be urgent than in daytime primary care. Identification of medical urgency during telephone contacts with GP cooperatives proved to be suboptimal [ 16 - 18 ]. GPs and nurses working in cooperatives treat patients they do not know and have to decide on advice and medical treatment with limited knowledge of the patients' medical history. In addition, GPs work in shifts and have to collaborate with other healthcare providers, which increases the risk of errors caused by discontinuity in information transfer [ 16 , 17 ]. The aim of our study was to gain an understanding of the incidence, types, causes, and consequences of patient safety incidents in out-of-hours GP cooperatives. A secondary aim was to examine factors that are associated with the occurrence of patient safety incidents.

Methods Study design and setting In 2009, a retrospective patient record review study was performed to examine patient safety incidents in GP cooperatives providing out-of-hours primary care in the Netherlands. Data were collected in a sample of general practices associated with selected GP cooperatives, as general practices keep complete documentation of the healthcare their patients received, including their contacts with GP cooperatives and subsequent healthcare use. Four GP cooperatives were selected. In this selection, we aimed to provide a good reflection of the national situation by taking into account location in the country, degree of urbanisation, and size of the cooperatives. A convenience sample of seventeen general practices linked to these GP cooperatives were invited to participate in the study, sixteen of which agreed to participate. For each of the four GP cooperatives, we selected a sample of at least 250 patients who had contacted the GP cooperative in April or May 2009. For each GP cooperative, the first 250 contacts that were eligible for review were included in the study (consecutive sampling). A 'contact' was a patient who visited the GP cooperative, received telephone advice from the cooperative, or received a home visit from a GP working for the cooperative. For patients with multiple contacts with a GP cooperative within the study period, only the first contact was included in the sample as the index contact. We excluded administrative reports without a patient contact with the GP cooperative (for example, a note from a hospital reporting a patient's demise). Several measures were taken to ensure the confidentiality of the information we collected. No patient or physician names were included in the database, and reviewers and researchers (study staff) signed a confidentiality agreement to maintain the confidentiality of the information. The Arnhem-Nijmegen ethical committee waived approval for this study. Patient record review procedure The record review procedure consisted of three phases (Figure 1 ). In the first phase, reviewers assessed the medical records of all sampled patients. The reviewers were an experienced GP and a medical student in his final year. They used a review form that had been developed for the study and was based on a form used for incident reporting in general practice and checked for face validity by an expert panel. The review procedure was pre-tested on 100 patient records in one general practice, resulting in a few small alterations to the review form. The pre-tested records were reviewed once more, using the adapted review form, and were included in the study. The reviewers assessed the patient contacts in the general practices. The general practices supplied the reviewers with the electronic medical records of sampled patients, including information about the contact with the GP cooperative, test results, and specialists' letters. Reviewers assessed whether one or more potential patient safety incidents had occurred during the patient's contact with the GP cooperative. A patient safety incident was defined as an unintended event during the care process that resulted, could have resulted or still might result in harm to the patient [ 19 ]. Firstly, the reviewers registered some basic patient characteristics: age, gender, whether the patient had contacted his GP about the same health problem within one week before contacting the GP cooperative, and whether the patient was a high-risk patient. Patients with cardiac and vascular disease, diabetes mellitus, asthma/COPD, polypharmacy (> 5 medications), immune system disease, malignancy (active), pregnancy, or a combination of these conditions, were defined as high-risk patients. Secondly, the reviewers registered measures concerning the patients' contact with the GP cooperative: contact type, contact time, documented degree of urgency, reason for contact, diagnosis, and medical treatment. For a period of at least four months after the contact, moreover, reviewers registered whether patients had follow-up contacts (with a GP, GP cooperative, or hospital casualty department) relating to the index health problem, whether patients were admitted to a hospital, or whether patients died. Finally, the reviewers assessed whether the healthcare provided at the GP cooperative was potentially unsafe [ 20 ]. If the reviewers had doubts about a particular case, the patient's GP was interviewed to clarify what had happened. If the reviewers signalled potentially unsafe healthcare, the patient's medical record proceeded to phase two. In phase two, the medical records of patients who had received potentially unsafe healthcare were discussed by a panel of physicians to determine if a patient safety incident had indeed occurred. This panel discussed the potential incidents until consensus was achieved. Besides the two reviewers, the panel consisted of two experienced physicians. In the third and final phase, the physician panel tentatively classified the incidents according to type, causes, and consequences (Table 2 ). Six types of incidents were distinguished: 'organisation', 'communication', 'diagnosis', 'treatment', 'prevention', and 'triage'. The first four types were derived from a classification model that had been developed for and is commonly used in primary care [ 21 ]. Causes of the incidents were analysed using the Eindhoven Classification Model (ECM) of the PRISMA-method,[ 22 , 23 ] which has proved to be a reliable tool [ 24 ] and has been used as a foundational component in the conceptual framework for the International Classification for Patient Safety (ICPS) of the World Health Organization (WHO) and its World Alliance for Patient Safety programme [ 25 , 26 ]. The consequences of the incidents were classified, using the 'severity of outcome' dimension of the International Taxonomy of Medical Errors in Primary Care [ 27 ]. In addition, the probability of (severe) harm in the future was assessed (very likely, likely, not likely). Inter-rater reliability The first ten patient records from five general practices (N = 50) were independently assessed by the two reviewers to determine their agreement on the presence or absence of potentially unsafe healthcare. Statistical analysis Study results were first described using descriptive statistics and frequency tables. To test which patient and contact factors predicted the occurrence of patient safety incidents, univariate multilevel logistic regression analyses and a forward stepwise multilevel logistic regression analysis were performed, using GP cooperative as random factor in the model (PROC GLIMMIX). Results were considered statistically significant at p < 0.05. Data were analysed using SPSS 15.0 and SAS 9.2.

Results Description of patient safety incidents Incidence and types A total of 1,145 patient records were reviewed (for practical reasons, 248, 328, and 319 records were reviewed in three GP cooperatives and 250 records in the fourth cooperative). Agreement between the two reviewers was 98.0% and Cohen's kappa was 0.79 (95% CI: 0.39 to 1.00). The quality of the patient records was predominantly judged as good (94%). In 1,145 patient records, reviewers identified 56 potential patient safety incidents. The physician panel judged 27 of these to be patient safety incidents, which is an incident rate of 2.4% (95% CI: 1.5% to 3.2%). Three incidents were related to more than one incident type. The most frequent incident type was treatment: patients receiving inadequate or suboptimal treatment (N = 15; 56%). Nine incidents (33%) were related to triage, meaning the urgency or care of patients' complaints had not been correctly assessed either by the triage nurse or by the supervising physician. Six incidents (22%) were related to diagnosis: misguided diagnostic reasoning or wrong diagnoses (see Table 3 for examples). Causes For 27 patient safety incidents, a total of 30 causes could be identified. The causal factors fell into three different categories: clinical reasoning, protocols, and patient-related factors. All incidents had at least partly been caused by failures in clinical reasoning: the inability of individuals to apply their existing knowledge to a novel situation. In two cases, a patient-related factor was relevant, that is, a failure related to patient characteristics or conditions that are beyond the control of staff. In one case, the absence of an adequate protocol contributed to the incident. No technical causal factors were identified. Actual and potential consequences The majority of patient safety incidents did not result in actual patient harm (N = 19; 70%). Eight incidents had consequences for patients: an extra intervention was needed in six cases, and two patients had to be admitted to a hospital. No incidents resulted in permanent harm or death. Most incidents were not likely to result in patient harm in the long term (N = 24; 89%). In three cases (11%), future consequences were possible but not likely. Factors associated with the occurrence of patient safety incidents Table 4 presents an overview of contact and patient characteristics for the total sample of contacts and the subsample of contacts with patient safety incidents. Univariate analyses showed that age and being a high-risk patient were positively related to incident rate. The patients' gender, preceding contacts with patients' own GP, type and time of contact, and the urgency of the contact were not related to the incident rate. The mean age in the total sample was 36.6 years (SD = 24.8). The mean age was 52.2 years (SD = 23.5) for patients with incidents, compared to 36.2 years (SD = 24.8) for patients without incidents. Multilevel analyses showed that the likelihood of an incident increased with 1.03 for each year of advancing age (95% CI: 1.01 to 1.04). The likelihood of an incident was 3.05 times higher for high-risk patients than for normal-risk patients (95% CI: 1.42 to 6.58). In the stepwise multilevel logistic regression, only the effect of age was significant.

Discussion Main findings Patient safety incidents occurred in 2.4% of all patient contacts with GP cooperatives (95% CI: 1.5% to 3.2%). The incidents identified in our study were related to triage, diagnosis, or treatment. They were all (at least partly) caused by failures in clinical reasoning. The majority of the incidents did not result in (permanent) harm to patients, but a few incidents were associated with temporary harm. With increasing patient age, an incident's probability of occurrence increased significantly. Strengths and limitations Our study was based on a large sample of patient contacts. We performed our study in four GP cooperatives that were spread across the Netherlands and varied in size and degree of urbanisation because we pursued a nationally representative study. GP practices were a convenience sample. Although our unit of analysis was patient contacts with multiple GPs taking shifts at four GP cooperatives, we cannot exclude a risk of selection bias. Data were collected in general practices associated with the GP cooperatives. This enabled reviewers to study not only the contact information of the GP cooperative, but also the medical records of the patients' own GPs, which include information about patients' healthcare use after they contacted the GP cooperative. Descriptive retrospective analyses depend on data quality. Some details of patient contacts are not written down, especially in case of telephone contacts [ 28 , 29 ]. The patient record reviewing method was carefully developed and tested, and proved to have high inter-rater reliability. However, no method for identifying incidents is perfect,[ 13 , 14 ] and a combination of methods has been recommended [ 30 , 31 ]. The study design we used will likely have enabled us to discover severe incidents but might have led us to underestimate the number of minor incidents without consequences and incidents in particular domains, such as accessibility issues. The number of causes identified per incident was small. It was difficult to obtain information on all contributing factors just by reading patient records. Interviewing people who were involved in the incidents would have produced a more complete view of the causal factors involved. However, confidentiality clauses made it impossible for us to do so in our study. Furthermore, the majority of causes were human. Healthcare providers may be less inclined to note down technical or organisational factors in the medical records of individual patients, which might explain the small numbers of technical and organisational factors identified. Finally, the study period was two months (April and May 2009). As the quantity and variety of health problems presented by patients vary with time of year, a study covering an entire year might have provided a different perspective. Interpretation of the results The results show that patient safety incidents do occur in out-of-hours primary care but that most do not result in harm to patients. However, as there are large numbers of patient contacts, the accumulated effect of the incidents could be significant. No earlier studies have examined incidents in GP cooperatives or similar out-of-hours healthcare settings. In a review of patient safety in daytime primary care, the incidence rate was estimated between 0.005% and 0.8% [ 12 ], which is much lower than the rate found in our study (2.4%). The research methods used may have contributed to this discrepancy: most studies in the review used event reporting, whereas we used patient record review. Furthermore, our definition of patient safety incidents was broad and included more than just events that resulted in patient harm. Our findings are comparable to patient record review studies into incidents in daytime primary care reporting incidence rates of 2.1% [ 15 ] and 2.5% (Gaal, personal communication, 2010). These findings suggest that patient safety in out-of-hours primary care is comparable to that in daytime primary care, even though the symptoms presented and medical treatments provided differ substantially between those settings. Failures in clinical reasoning played a part in all incidents in our study. These failures might be related to the fact that GPs in GP cooperatives do not know patients and must decide on medical treatment without having full knowledge of patients' medical history. In addition, they have virtually no possibilities for diagnostic examinations such as X-rays and laboratory tests without referral to a specialist. We found that the probability of experiencing an incident was higher for older patients (OR = 1.03) but that this did not apply to high-risk patients. This result was unexpected because co-morbidity was believed to be the explanation for higher age-related probability. Deafness or cognitive disorders in older patients might impede communication, which makes clinical reasoning, diagnosis, and treatment more difficult for GPs. A study by Sari et al. (2008) in an NHS hospital also found that, with increasing age, there was a raised risk of experiencing an adverse event (OR = 1.03) [ 32 ]. Finally, nurse telephone triage has been adopted to reduce GP workload, but its safety is under debate [ 33 - 36 ]. Our study did not find an increased risk of incidents in patients that received telephone advice as compared to consultation at the clinic or home visits. Recommendations for practice Information about incidents can be used to develop interventions to improve patient safety. It is recommended for out-of-hours healthcare services to examine incidents by way of periodic patient contact audits, including follow-up information from the patients' own GPs, and to combine these with incident reporting by professionals. Incidents should be discussed with professionals to improve clinical reasoning and to reduce the likelihood of reoccurrence. GPs should adhere to guidelines and document their reasons for deviating from these guidelines. Moreover, system-wide interventions are suggested to enhance patient safety in primary care, such as developing leadership and culture supporting safety behaviour; constructing barriers in systems to prevent human error; and using sensible technology, such as meaningful warnings, communication, and knowledge coupling [ 37 ]. Finally, existing national clinical guidelines may be less applicable to out-of-hours primary care. Relevant topics related to acute medical problems need to be elaborated [ 38 ]. Recommendations for future research We need to improve our understanding of the reasons why GPs in out-of-hours healthcare make errors in clinical reasoning, as these are the most frequent causes of patient safety incidents. Lack of information on patients' medical history, insufficient medical knowledge, and high workload could all play a part. Further research into the underlying mechanisms of failures in clinical reasoning is recommended. Because the number of patients actually suffering harm from incidents is small, actual harm is not a good outcome measure for evaluating the effectiveness of patient safety interventions, such as evaluating the implementation effectiveness of clinical guidelines. Outcome measures such as unnecessary prolongation or aggravation of symptoms will be more valuable for evaluating prevention strategies. We recommend, therefore, that future research should include this broader view of patient safety.

Conclusions In this study, we found that patient safety incidents occurred in 2.4% of all patient contacts with GP cooperatives. The incidents identified were related to triage, diagnosis, or treatment and were all (at least partly) caused by failures in clinical reasoning. Most incidents did not result in harm to patients. With increasing patient age, an incident's probability of occurrence increased significantly. Our findings report a relatively low incident rate at GP cooperatives in the Netherlands. As there are large numbers of contacts and patients, however, the accumulated effect of the incidents on patient well-being and the healthcare delivery system could be significant. Moreover, as clinical reasoning played an important part in these incidents, further understanding of clinical reasoning and guideline adherence at GP cooperatives could contribute to healthcare safety.

Background Most patients receive healthcare in primary care settings, but relatively little is known about patient safety. Out-of-hours contacts are of particular importance to patient safety. Our aim was to examine the incidence, types, causes, and consequences of patient safety incidents at general practice cooperatives for out-of-hours primary care and to examine which factors were associated with the occurrence of patient safety incidents. Methods A retrospective study of 1,145 medical records concerning patient contacts with four general practice cooperatives. Reviewers identified records with evidence of a potential patient safety incident; a physician panel determined whether a patient safety incident had indeed occurred. In addition, the panel determined the type, causes, and consequences of the incidents. Factors associated with incidents were examined in a random coefficient logistic regression analysis. Results In 1,145 patient records, 27 patient safety incidents were identified, an incident rate of 2.4% (95% CI: 1.5% to 3.2%). The most frequent incident type was treatment (56%). All incidents had at least partly been caused by failures in clinical reasoning. The majority of incidents did not result in patient harm (70%). Eight incidents had consequences for the patient, such as additional interventions or hospitalisation. The panel assessed that most incidents were unlikely to result in patient harm in the long term (89%). Logistic regression analysis showed that age was significantly related to incident occurrence: the likelihood of an incident increased with 1.03 for each year increase in age (95% CI: 1.01 to 1.04). Conclusion Patient safety incidents occur in out-of-hours primary care, but most do not result in harm to patients. As clinical reasoning played an important part in these incidents, a better understanding of clinical reasoning and guideline adherence at GP cooperatives could contribute to patient safety.

Competing interests The authors declare that they have no competing interests. Authors' contributions MS contributed to data analysis and wrote the manuscript. LH contributed to the data collection process and the writing of the manuscript. BK collected and analysed data and contributed to the writing of the manuscript. EdF collected data and critically read the manuscript. MW and PG designed the study, supervised data collection, contributed to data interpretation, and critically revised the manuscript for intellectual content. MW arranged funding. All authors approved the final version of the manuscript. Funding The study presented in this paper was part of a large study on patient safety in primary care in the Netherlands. The Dutch Ministry of Health, Welfare and Sport (VWS) initiated and funded the project (without restrictions on the scientific work; grant number 313741). Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1472-6963/10/335/prepub

Acknowledgements The authors would like to thank all participating primary care practices and their staff for providing data. We confirm that all patient/personal identifiers have been removed or disguised, so the patients/person(s) described are not identifiable and cannot be identified through the details of the story. We thank our co-workers Vera Renaud and Maaike Gieben for their contribution to data collection. We thank Mirjam Harmsen for coordinating the larger project and Reinier Akkermans for his support in the multilevel analyses.

CC BY

no

2022-01-12 15:21:45

BMC Health Serv Res. 2010 Dec 10; 10:335

oa_package/14/e7/PMC3016358.tar.gz

PMC3016359

21162747

Background Professional burnout is an important issue in health and human service professions, due to its strong association with outcomes such as increased medical errors [ 1 ], decreased quality of patient care [ 2 ] and increased turnover and absenteeism [ 3 ]. The high prevalence of burnout in cancer care workers is a particularly salient issue due to the highly vulnerable patient groups involved and the importance of retaining skilled and experienced personnel in an occupational field marked by existing staff shortages. International research has clearly established high levels of burnout and psychological distress in oncology staff [ 4 - 6 ]. More than half (56%) of oncologists in a US sample reported experiencing an episode of burnout at some stage during their career, with incidence of burnout rising with increasing time spent in direct patient contact [ 7 ]. Similarly, high levels of occupational distress are reported by Australian cancer care workers, with 33% of those whose work involves direct patient contact and 27% of those without patient contact exhibiting high levels of emotional exhaustion [ 8 ]. Historically, professional burnout in health care providers has been assessed via the Maslach Burnout Inventory - Human Services Survey (MBI-HSS), which is a well-validated 22-item self-report questionnaire with strong psychometric properties [ 9 ]. Its three sub-scales measure distinct, but interrelated, aspects of burnout: emotional exhaustion (EE), depersonalisation, and personal accomplishment. However, with a completion time of 10-15 minutes, the MBI-HSS does not easily lend itself to routine administration for screening purposes in most health care settings, where time allocated to staff welfare and occupational health and safety often is at a premium. Of the three MBI sub-scales, EE has been widely regarded as the core component of the multidimensional construct of burnout [ 10 , 11 ], with fatigue and exhaustion reported as central features of burnout [ 12 ]. The important role of EE is also evidenced by previous findings that self-diagnosis of burnout is based on EE [ 13 , 14 ]; and only the EE component of burnout was predictive of cancer care workers' intention to leave the profession in a Canadian study [ 4 ]. Previous research indicates that the single-item measure of self-defined burnout developed initially for the Physician Worklife Study [ 15 ] may be a satisfactory screening tool for burnout, due to its demonstrated positive association with the EE sub-scale of the MBI-HSS in a US sample of physicians [ 14 ]. However, the potential usefulness of this brief screening measure being administered routinely in Australian clinical settings is unknown, given lack of Australian data on its validity as a burnout measure. This paper reports on the predictive validity of self-defined burnout assessed using a single item by measuring its association with the EE sub-scale of the Maslach Burnout Inventory in a sample of Australian cancer care workers.

Methods Sample & Procedure The Clinical Oncological Society of Australia (COSA) is the peak national body representing health professionals from a range of multidisciplinary groups, whose main work is in the area of cancer control. COSA members (N = 1,322 at May 2007) received a letter from the COSA secretariat with initial information about the study. Contact details of members who had not declined further contact (n = 1,157) were sent to the researchers for all further communication regarding the study. Members received study information (n = 1,059 by email; n = 98 by post), including a URL for accessing the web-based survey and a personal log-in and password. Non-responders received reminders two, three and six weeks after the initial invitation date. Completion of the survey was taken as consent to participate. No further contact was made with non-responders. The University of Newcastle Human Research Ethics Committee approved the study. Instruments Maslach Burnout Inventory The 22-item Human Services version (MBI-HSS) [ 9 ] was administered to respondents who reported having direct patient contact as part of their work. The MBI-HSS consists of three sub-scales: Emotional Exhaustion, Depersonalisation, and Personal Accomplishment. The cut-off scores recommended by the MBI scale developers were applied to indicate low, average or high levels of burnout on each sub-scale separately [ 9 ]. The sub-scales were scored as recommended by the developers of the scale; and only the EE sub-scale data are reported in this paper. Self-defined burnout A single item developed by Schmoldt et al [ 15 ], was included to assess self-defined burnout, with five response options: (i) I enjoy my work. I have no symptoms of burnout; (ii) Occasionally, I am under stress, and I don't always have as much energy as I once did, but I don't feel burned out; (iii) I am definitely burning out and have one or more symptoms of burnout, such as physical and emotional exhaustion; (iv) The symptoms of burnout that I'm experiencing won't go away. I think about frustration at work a lot; (v) I feel completely burned out and often wonder if I can go on. I am at the point where I may need some changes or may need to seek some sort of help. Statistical analyses Analyses were conducted using SPSS software. Mean score and prevalence of high burnout on the MBI-EE were calculated. An ANOVA analysis compared the MBI-EE sub-scale to the categorical responses to the self-defined burnout item. ANOVA was considered appropriate, as the data displayed normally distributed residuals and homogeneity of variance. A Pearson correlation coefficient was also calculated as a measure of association. A false negative score on the self-defined burnout item was defined as a score of 1 on the single item and either average or high on EE, or 2 on the single item and high on EE. A false positive score was defined as a score of 5 on the single item and either low or average on EE, or 4 on the single item and low on EE.

Results Sample Of the 1157/1322 financial COSA members willing to receive the initial survey invitation, 9 were ineligible and 165 declined contact from the research team. A total of 740 surveys were completed, representing a response rate of 56% of the known eligible COSA membership and a consent rate of 64.5% of eligible members who received the study information. The secondary analyses reported here are of the sub-sample of 638 participants who reported that their work involved direct patient contact. Participant demographic and occupational characteristics are in Table 1 . Burnout - EE versus single item The EE mean score of 21.3 (SD = 11.7) in this Australian sample is similar to that from published norms (Mean = 22.19, SD = 9.53, p = 0.07 ) from 1104 physicians and nurses in the USA [ 9 ]. Almost one-third of our sample (32%, n = 204) was identified as having high levels of burnout on the EE sub-scale, compared to 28.2% (n = 180) classifying themselves as definitely burning out (20.7%), having persistent symptoms of burnout (4.7%), or being completely burned out (2.8%) on the self-defined burnout item (Mean = 2.3, SD = 0.8). As shown in Table 2 , results of the ANOVA procedures comparing self-defined burnout and EE sub-scale scores indicate that the Emotional Exhaustion sub-scale of the MBI and the self-defined burnout measure indeed tap into a similar construct, with R 2 = 0.5 (p < .0001). This association is further evidenced by a significant moderate to high positive correlation between the two measures (r = 0.68, p < .0001). In order to further assess how well the self-defined burnout measure performed against the EE sub-scale, the proportion of false negatives and positives detected by the self-defined burnout measure was calculated. The self-defined burnout measure detected one false positive, and 76 (12%) false negatives. The majority of these false negatives (11%) were respondents who had high MBI EE scores but defined themselves as only occasionally under stress and sometimes lacking energy, but not feeling burned out . This finding highlights the fact that fatigue and exhaustion are the central features underpinning the EE sub-scale, but suggests that some respondents indeed may experience "emotional exhaustion" without feeling burned out.

Discussion The significant positive correlation between self-defined burnout assessed by a single item and the emotional exhaustion component of burnout in this sample of Australian clinical cancer workers is consistent with previous US findings [ 14 ]. The finding of a minor proportion of false negatives on the self-defined burnout measure relative to the emotional exhaustion sub-scale appears to be an artifact of the inclusion of concepts relating to stress and fatigue in one of the low-burnout response options of the single-item measure. Further evaluation of this measure should therefore ideally incorporate assessment of the items' construct validity through inclusion of related objective measures of burnout, such as absenteeism. Given the brevity and ease of administration of this measure, it has significant potential to be routinely used to effectively screen for burnout in health care settings which are time-poor for detecting symptoms related to fatigue and emotional exhaustion more comprehensively. Caution should be exercised in drawing inferences about the usefulness of the single item burnout measure in other health care settings, due to some under-represented professional groups in the current sample and the inherent self-selection bias introduced by the survey methodology and membership of COSA. However, the finding that the current sample reports levels of emotional exhaustion comparable to those in a large US normative sample of health professionals lend support to the validity of these findings across other health care settings.

Conclusions The current study indicates that a single-item measure of self-defined burnout can be effectively used to screen for burnout in an Australian oncology setting. Given the importance of preventing and addressing burnout in oncology settings, this easily administered burnout measure has got potential for being routinely administered as part of staff health and welfare procedures.

Background Burnout has important clinical and professional implications among health care workers, with high levels of burnout documented in oncology staff. The aim of this study was to ascertain how well a brief single-item measure could be used to screen for burnout in the Australian oncology workforce. Methods During 2007, 1322 members of the Clinical Oncological Society of Australia were invited to participate in a cross-sectional nationwide survey; 740 (56%) of eligible members consented and completed the survey. Data from the 638 consenting members who reported that their work involved direct patient contact were included in the secondary analyses reported in this paper. Burnout was assessed using the MBI Human Services Survey Emotional Exhaustion sub-scale and a single-item self-defined burnout scale. Results Emotional exhaustion was "high" in 33% of the sample when assessed by the psychometrically validated MBI. The single-item burnout measure identified 28% of the sample who classified themselves as "definitely burning out", "having persistent symptoms of burnout", or "completely burned out". MBI Emotional Exhaustion was significantly correlated with the single-item burnout measure (r = 0.68, p < 0.0001) and an ANOVA yielded an R 2 of 0.5 (p < 0.0001). Conclusions The moderate to high correlation between the single-item self-defined burnout measure and the emotional exhaustion component of burnout suggest that this single item can effectively screen for burnout in health care settings which are time-poor for assessing burnout more comprehensively.

Competing interests The authors declare that they have no competing interests. Authors' contributions Both VH and AG were involved in the conception and design of the study, and development of study measures. VH was responsible for data collection and analysis, and for drafting the manuscript. AG participated in data collection and analysis, and critically revised the manuscript. Both authors approved the final manuscript. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1472-6963/10/341/prepub

Acknowledgements This secondary analysis was undertaken on data collected in the larger study commissioned by the Clinical Oncological Society of Australia and funded by Cancer Australia. The assistance of both organisations in facilitating the completion of the research is gratefully acknowledged. We thank Ms Margaret McJannett and her team for facilitating the contact with COSA members; to COSA Council members for assistance in reviewing the survey; to Christophe Lecathelinais for statistical assistance, the University of Newcastle Corporate Information Unit for online survey administration; and most importantly, the members of COSA who took the time to complete the survey.

CC BY

no

2022-01-12 15:21:45

BMC Health Serv Res. 2010 Dec 16; 10:341

oa_package/b3/a7/PMC3016359.tar.gz

PMC3016360

21129198

Background This paper is a second in a series of three contributions discussing the geographic distribution and bionomics of the dominant vector species (DVS) of human malaria [ 1 , 2 ]. It deals specifically with the DVS of Africa, Europe and the Middle East. Despite highly variable levels of transmission across Africa [ 3 , 4 ], the global public heath impact of P. falciparum malaria is overwhelmingly felt on this continent [ 5 , 6 ]. Africa contains areas with the highest entomological inoculation rates [ 3 , 7 ] and prevalence levels [ 8 ] globally, and thus the highest morbidity and mortality [ 5 ]. This situation arises partly because Africa has the most effective and efficient DVS of human malaria [ 9 , 10 ]: An. gambiae ( sensu stricto - herein, referred to as ' An. gambiae '; it is not necessary to use ' sensu stricto ' (or the abbreviation ' s.s. ') when there is no doubt that the biological species being referred to is the one that bears the name An. gambiae ) [ 5 , 10 ], with its sibling, An. arabiensis , also of major importance [ 11 ]. The DVS members of the An. gambiae complex also include the salt water tolerant, coastal species An. melas and An. merus [ 12 ] and these, whilst not being as efficient at transmitting malaria as An. gambiae or An. arabiensis , are often found in such high densities that they achieve DVS status [ 13 - 15 ]. Other members of the An. gambiae complex are either highly restricted in their distribution ( e.g . An. bwambae , only currently known to occur in geothermal springs in western Uganda [ 11 , 16 ]) or are zoophilic in behaviour and not considered vectors of human malaria ( e.g. An. quadriannulatus and An. quadriannulatus B) [ 17 ]. In addition to the four DVS within the An. gambiae complex, large parts of Africa are also home to other DVS, including An. funestus , An. nili and An. moucheti , with An. funestus , in some cases, having a greater impact on malaria transmission even than An. gambiae [ 10 , 11 , 18 ]. The anthropophilic habits of these DVS are a major contributing factor to their public health impact, indeed An. funestus is considered to be one of the first species to have adapted to human hosts [ 19 ]. The vast majority of current malaria control efforts use interventions aimed at limiting human-vector contact [ 20 , 21 ]. Foremost among these interventions has been the rapid scale-up of insecticide treated bednets (ITNs) [ 22 ], followed by the scale-up of indoor residual spraying (IRS) in Africa [ 23 ]. These interventions are often deployed without a detailed understanding of the distribution, species composition and behaviour of local vectors. This complicates impact monitoring [ 24 ], the appraisal of arguments for more holistic integrated vector control [ 25 ] and evaluation of the potential of novel vector control methods [ 26 - 28 ]. Distribution maps can also be applied to gauge the importance of emerging insecticide resistance among the DVS of Africa [ 29 - 37 ]. In contrast to Africa, the European and the Middle Eastern region contain areas with low to no malaria transmission [ 8 ]. Despite this, the existence of Anopheles species with the capacity to transmit malaria is often highlighted as providing the potential for the re-introduction of malaria [ 38 - 43 ]. A number of vector species modelling and mapping strategies have been applied on a country ( e.g. [ 44 - 50 ]) and regional scale [ 51 ] and across the African continent [ 24 , 52 - 55 ], with fewer attempts directed at the European and Middle Eastern species [ 56 - 58 ]. No previous mapping efforts formally incorporate expert opinion (EO) distributions and the methods used range in complexity, from simply plotting presence or abundance on a map [ 24 , 44 , 48 , 57 , 58 ], to the application of more sophisticated predictive models [ 45 - 47 , 49 , 50 , 52 - 56 ]. This makes comparison between the maps difficult. Further difficulties also arise in the interpretation of existing maps as many previous studies include all historical occurrence records to compensate for poor data coverage. This can introduce taxonomic ambiguity; the An. gambiae complex, for example, was only fully categorised in 1998, with the addition of the provisionally designated An. quadriannulatus species B [ 12 , 59 ] and, even now, the status of An. funestus is under question [ 60 - 63 ]. Moreover, the morphological similarity that hides members of a species complex adds a level of uncertainty to the identity of species data recorded before the advent of cytological or molecular identification techniques. This current work attempts to overcome many of these problems. The same Boosted Regression Tree (BRT) methodology is applied to all DVS making comparison between predicted maps possible. Despite only using data collected after 31 December 1984 the assimilated DVS occurrence records together comprise the largest contemporary dataset for prediction, with this evidence base to be made available in the public domain. Significant efforts were also expended to update the EO maps for all species [ 1 ] and these were used to inform the predictions. The outcome of these efforts and that of a comprehensive bionomics review are presented here for the DVS of Africa, Europe and the Middle East.

Methods The data assembly and mapping methods, climatic and environmental variable grid pre- and post-processing methods and the modelling protocol summarised here are described in detail in Sinka et al. [ 2 ]. The selection of the DVS is detailed in Hay et al. [ 1 ]. In brief, 13 DVS from a final list of 41 species and species complexes worldwide were considered, seven of which are found solely in Africa (Table 1 ) [ 1 ] with a further six distributed across Europe, the Middle East and in limited areas of northern Africa (Table 2 ). Data assembly, data checks and expert opinion maps Building on the existing Malaria Atlas Project (MAP [ 64 ]) library of parasite rate surveys, a systematic search of the published, peer-reviewed literature using online scientific bibliographic databases was performed and augmented with a range of other information previously described [ 2 ]. Literature searches were concluded on 31 October 2009 and all citations meeting our search criteria [ 2 ] were reviewed. Occurrence data extracted from these sources (a detailed protocol is given in Hay et al. [ 1 ]) were subjected to a series of rigorous checks before being migrated from Excel into a web-based PostgreSQL database where a final series of checks were conducted (see Sinka et al. [ 2 ]). Globally, the literature search resulted in 3857 publications or reports containing potential data to be reviewed. Of these publications, 2276 fulfilled the inclusion criteria, providing data for 147 countries. A total of 727 sources detailed surveys conducted across 46 countries in Africa with 45 sources found for 49 countries in Europe and the Middle East. Using EO map overlays (Additional file 1 : Expert opinion distribution maps for the seven DVS of Africa and the six DVS of the Europe and Middle Eastern region (Raster prediction files are available on request)), initially digitised from published, authoritative sources (Table 1 , 2 ) and further refined by a Technical Advisory Group (TAG) of Anopheles experts (see acknowledgements), preliminary maps were produced displaying the occurrence data for each species. These maps were examined and points that fell outside the EO range were checked and either corrected or the EO maps adjusted to include all confirmed areas of occurrence. Boosted Regression Trees, climatic/environmental variables and model protocol The BRT method [ 65 , 66 ] was chosen to generate the predictive maps of each DVS distribution. In a review comparing 16 species modelling methodologies, BRT consistently performed well [ 67 ] and benefits from being flexible (accommodating both categorical and continuous data), using freely available, reliable and well documented R code [ 68 ] and producing maps that are simple to interpret and include a ranked list of environmental or climatic predictors [ 2 ]. The method is described in full by Elith et al . [ 66 ] and its implementation for DVS mapping summarised by Sinka et al . [ 2 ]. The BRT also produces a number of evaluation statistics including Deviance, Correlation, Discrimination (Area Under the operating characteristic Curve: AUC) and Kappa (κ) which are used here as a guide to the predictive performance of each map. The BRT model was provided with a suite of open access, environmental and climatic variable 5 × 5 km resolution grids, relevant to the ecology and bionomics of the DVS in the African, European and Middle Eastern regions. Each grid has undergone a series of processing steps to ensure all land and sea pixels exactly correspond, and, using nearest neighbour interpolation, to fill in any small gaps in the data due to, for example, cloud cover (see Sinka et al. [ 2 ]). Where the remotely sensed imagery was available as multi-temporal data, temporal Fourier analysis (TFA) was applied to ordinate the data, generating seven products for each temporal variable: the overall mean, maximum and minimum of the data cycles; the amplitude (maximum variation of the cycle around the mean) and the phase (the timing of the cycle) of the annual and bi-annual cycles [ 69 ]. The environmental/climatic variables applied to the BRT model included a digital elevation model (DEM) [ 70 - 72 ], precipitation and temperature [ 73 , 74 ], land surface temperature (LST), middle infrared radiation (MIR) and the normalized difference vegetation index (NDVI) (Advanced Very High Resolution Radiometer (AVHRR) [ 75 - 78 ]), and 22 individual categories of land cover plus a further three grouped classes that encompassed flooded areas, forested areas and dry areas (Globcover [ 79 ]). The AVHRR grids (LST, MIR and NDVI) were applied to all DVS except the European species An. messeae and An. atroparvus . These two species have the most northerly distribution of all the DVS, with An. messeae ranging up to 65° north. At these latitudes, the AVHRR satellite data can be problematic. Instead MODIS (MODerate Resolution Imaging Spectroradiometer) [ 70 ] data were used because it provides better coverage and fewer data gaps for these northern distributions. The MODIS grids include the Enhanced Vegetation Index (EVI) and LST [ 70 ]. Following the same protocol described in Sinka et al. [ 2 ], numerous model iterations were run to assess the 'optimal' mapping outputs, including assessing the buffer size surrounding the EO range from where pseudo-absences would be drawn, the number of pseudo-absences to apply to the model and the effects of including half weighted pseudo-presence data, allocated at random from within the EO boundary, alongside the occurrence data. As each of these categories required the use of different data inputs to the BRT, statistical comparison using the evaluation metrics was not strictly possible. Therefore the 'optimal' settings chosen are inherently subjective and based on visual examination and comparison of the various maps guided by, but not relying on, the evaluation statistics. Bionomics A full protocol describing the methodology used to extract species-specific bionomic data from the available literature (Table 3 , 4 ) is given in the supplemental information accompanying Sinka et al. [ 2 ]. The bionomics summary of each species is included to accompany the predictive maps as the success of interventions and control methods, such as ITNs or IRS, in reducing malaria transmission is closely related to the behavioural characteristics of the local DVS. This review does not, however, include detailed information relating to insecticide resistance. This was a purposeful omission as it would not be possible to do full justice to this highly dynamic and important aspect of the DVS within the space confines of the current work. Moreover, insecticide resistance is being addressed in detail by other groups, including those at the Liverpool School of Tropical Medicine and the Innovative Vector Control Consortium (IVCC) [ 80 ]. Furthermore, there are a number of comprehensive reviews that have been recently produced that detail insecticide resistance amongst Afrotropical species which should be considered alongside this current work ( e.g. [ 31 , 35 , 81 , 82 ]).

Results African DVS A total of 4581 independent sites, of which 4234 were successfully geo-referenced, reported the presence of one or more African DVS, relating to 9300 (8646 geo-referenced) occurrences (i.e. including one or more temporal sample conducted at one independent site) (Table 5 ). The following results refer only to geo-referenced data, and of these 3951 sites were at a resolution (points and wide areas, <10 km 2 and between 10 and 25 km 2 respectively) suitable to be applied to the BRT model (from here on, for simplicity, referred to as points). Data were recorded from a total of 46 countries, 44 of which reported points. The largest number of data were reported from Kenya, with a total of 757 sites (all area types), 686 points and 1599 occurrence data (all area types). In contrast, only one data point was reported from Togo (Kantindi) where An. gambiae was found [ 83 ] and studies from Mauritius only provided DVS location information, at a polygon level, for two sites. African DVS data were reported from Egypt, but only in the form of a polygon location that could not be successfully geo-referenced. Anopheles gambiae was reported from the largest number of countries (34) and from the highest number of point locations (1443), however occurrence data (from point locations only) were greater for both An. funestus and An. arabiensis (2692 and 2301, respectively) than for An. gambiae (2291). The least prevalent species was An. moucheti reported from only 66 point locations (Table 6 ) and Cameroon had the highest diversity of DVS with three sites (Nkoteng, Tibati and Mayo Mbocki) showing the presence of five DVS ( An. arabiensis , An. funestus , An. gambiae , An. nili and An. moucheti ) [ 84 - 86 ]. Adult resting collections were the most popular sampling method, with 424 studies collecting females resting inside houses compared to 178 studies that collected females biting indoors. Outdoor resting sampling was comparably rare with 56 studies collecting from outdoor shelters, 22 studies searching inside animal sheds and 21 studies where the details of the outdoor location sampled were not recorded. Outdoor landing catches were conducted in 132 studies and 181 studies collected larvae, relating to 675 point locations. Molecular techniques examining nucleic acids, which have only been applied for identification on a regular basis since the 1990s [ 87 ], were well represented, with 338 studies reporting the use of Polymerase Chain Reaction (PCR) methods. Morphological methods were used in 363 studies, often in conjunction with PCR techniques. At the other end of the scale, salinity tolerance tests were only attempted in four studies and cross-mating experiments only in five. European and Middle Eastern DVS Across the European and Middle Eastern region, 49 countries reported the presence of one or more DVS from 2820 point locations (all locations: 2891), of which 2784 were successfully geo-referenced (all geo-referenced locations: 2848) (Table 7 ). Relatively few polygon data were reported (all: 71/2891, georeferenced only: 64/2848) and longitudinal studies were also rare, with only 18 studies reporting sampling on more than one occasion at the same site. A total of 3020 geo-referenced occurrence data across all area types, with 2946 from point locations, were compiled. Considering only the geo-referenced data, DVS presence was reported from the most sites in Italy (all sites: 423, point only: 409). Anopheles atroparvus was the species reported most often across the region, found at 1051 geo-referenced locations, of which 1044 were available to be used in the analyses. Anopheles sergentii was only present at 35 point locations, and within 11 polygon areas, but these related to a total of 113 occurrence data (102 points, 11 polygons) (Table 8 ). In the European and Middle Eastern region, larval collections were the most common sampling method, with 23 studies sampling at 86 sites. Sampling methods were unknown for a large proportion of the data (1553 sites), of which 1488 related to a single data source [ 56 ]. Possibly due to the zoophilic nature of the majority of the European and Middle Eastern species (see below), resting adult females were collected from animal sheds at 85 locations compared to only 31 where resting collections were conducted inside human dwellings. Human landing collections were conducted indoors in only two studies, relating to only three sites, with three studies collecting by outdoor human landing at only eight sites. Identification methods, amongst those studies that reported them, mainly relied on morphological characteristics and were conducted on specimens from 175 locations. Only 10 studies reported using PCR identification techniques but due to a large number of unknown or unreported methods, this ranked as the second most popular method, and was applied to specimens collected from 67 sites. Mapping trials The results for each mapping trial are given in Additional file 2 (Additional file 2 : Summary tables showing evaluation statistics for all mapping trials and final BRT environmental and climatic variable selections for the final, optimal predictive maps). Optimal mapping categories were evaluated visually and using the deviance and AUC statistics, with the caveat that these could only be used as a guide rather than a definitive indication of predictive performance. The EO mapping test indicated that where random pseudo-presences were created within the EO range, and no real occurrence data were included, the model would predict a high probability of presence within the whole EO range and calculate a high deviance value for all species, indicating an overall poor predictive performance. This was the case for the African species and those from the European and Middle Eastern region, and consistent with the results for the nine DVS in the Americas [ 2 ]. Where the hybrid method was used that incorporated both real occurrence data plus 500 half-weighted pseudo-presence points randomly assigned within the EO range, the mapping performance was greatly improved. Maps created using only the real presence data produced a low deviance value, but visually, predictive performance was judged to be poor, possibly due to a paucity of data for some species. It was therefore considered that the hybrid maps performed better overall and are presented here. The optimal buffer width for the African DVS was judged to be 1500 km, producing the lowest deviance value for five out of the seven species. For the European and Middle Eastern species maps, all buffer widths other than 1000 km had high deviance values for all species. The 1000 km buffer therefore was judged to perform better for all six species and applied consistently to all final maps. For both the African and the European and Middle Eastern species, a ratio of 10:1 pseudo-absences to presence data (not taking into account the 500, half weighted pseudo-presence created in the hybrid maps) was judged to perform better overall, but for both regions, the number of pseudo-absences appeared to have little effect on the predictive maps. Predictive maps The BRT maps for all seven African DVS and for the six European and Middle Eastern species are given in Additional file 3 (Additional file 3 : Predictive species distribution maps for the seven DVS of Africa and the six DVS of the Europe and Middle Eastern region). Spatial constraints prevent all species being discussed in detail here, however, Anopheles gambiae (Figure 1 ) is the iconic and possibly the most important vector of malaria [ 88 ], and therefore is discussed further below. There have been a number of attempts to model the distribution of An. gambiae but the majority tend to focus on single countries and often just map presence points or abundance without further analysis ( e.g. [ 44 - 49 ]). Continent-wide predictive maps for An. gambiae (plus other members of the An. gambiae complex) have also been attempted [ 1 , 89 ], making use of satellite-derived environmental or climatic variables [ 24 , 52 - 55 ] (Table 9 ). The methods range from simply overlaying presence and absence points over rainfall maps [ 24 ] to the application of more complex, spatial ecological niche models [ 53 , 55 ]. Precipitation, in one form or another, is identified repeatedly in previous models (where these data are presented, Table 9 ) as an influential variable in predicting the range of An. gambiae . Within the top five contributing covariates from the suite applied to the BRT model, precipitation was identified three times, with mean precipitation as the highest contributor with a relative influence of over 37%. Maximum precipitation was placed second (19.42%) with the amplitude of the bi-annual cycle of precipitation ranked forth (8.85%). In common with the Maxent niche model presented by Moffett et al. [ 55 ], elevation (altitude) and minimum land surface temperature were also identified by the BRT model within the top five influencing climatic/environmental variables (relative influence of 12.36% and 5.68%, respectively). Anopheles gambiae larvae are commonly found in temporary, shallow, small bodies of water, such as puddles in hoof prints, wheel ruts and small ground pools (see below), sites which are only present after rainfall. Hence the high influence of precipitation on the distribution of this species identified by the BRT model corresponds closely with the known bionomics of An. gambiae . The predictive map of An. gambiae (Figure 1 ) loosely follows the boundary and distribution indicated by the EO map (Figure 1 , inset) with one clear exception: the large gap in the range over southern Kenya and a large proportion of northern and central Tanzania. This gap may be driven by the presence of savannah-type vegetation [ 89 ] more commonly associated with An. arabiensis , or the increasing altitude of this region, and may be causal to the identification of elevation as an influencing factor to the distribution of An. gambiae. Similar gaps are also seen in the maps produced by Moffett et al. [ 55 ] and, to a slightly lesser extent, in the map of Levine et al. [ 53 ]. In Madagascar, Léong Pock Tsy et al. [ 44 ] identified altitude as a limiting factor for An. gambiae with numbers diminishing as altitude increased until, other than two specimens found at 1300 m, it was considered essentially absent over 1000 m. However, in the Kenyan highlands An. gambiae is commonly identified up to 2000 m [ 89 - 92 ] and specimens have been confirmed at sites up to 1800 m in Uganda [ 93 ]. Sampling across Africa, as stated by Coetzee [ 88 ], reflects the distribution of entomologists and not necessarily the distribution of the mosquitoes, and the area within this predicted gap, along with a great swath through central Africa, is clearly lacking in empirical occurrence data. Acknowledging these caveats, and similar ones in parts of the range of many of the DVS, it is obvious that samples from these poorly known areas would help improve substantially our predictive mapping. Bionomics of the African DVS Anopheles arabiensis Anopheles arabiensis , when compared to An. gambiae , is described as a zoophilic, exophagic and exophilic species [ 94 ]. However, it is also known to have a wide range of feeding and resting patterns, depending on geographical location [ 11 , 95 , 96 ]. This behavioural plasticity allows An. arabiensis to adapt quickly to counter indoor IRS control, where suitable genotypes occur [ 97 ], showing behavioural 'avoidance' (deterrence from a sprayed surface) depending on the type of insecticide used [ 95 , 98 ]. Anopheles arabiensis is considered a species of dry, savannah environments and sparse woodland [ 11 , 24 , 97 , 99 ], yet it is known to occur in forested areas, but only where there is a history of recent land disturbance or clearance [ 24 ]. Its larval habitats are similar to those of An. gambiae (see below): generally small, temporary, sunlit, clear and shallow fresh water pools [ 100 - 103 ] (Table 10 ), although An. arabiensis is able to utilize a greater variety of locations than An. gambiae , including slow flowing, partially shaded streams [ 103 - 106 ] and a variety of large and small natural and man-made habitats (Tables 11 , 12 ). It has been found in turbid waters [ 100 , 107 , 108 ] and, on occasion, in brackish habitats [ 109 ] (Harbach, unpub. obs.). It readily makes use of irrigated rice fields (Table 11 ), where larval densities are related to the height of the rice, peaking when the plants are still relatively short and then dropping off substantially as the rice plants mature [ 110 - 113 ]. Such density fluctuations are also reflected in the adult population, which also peak when rice stalks are small and decline as the plants mature [ 114 - 116 ]. These patterns may be due to a preference for sunlit areas of water with relatively limited emergent vegetation (Table 10 ), with densities decreasing as shade from the growing plants increases. Moreover, there is evidence that An. arabiensis may be attracted by the application of fertilisers or by the amount of dissolved oxygen within the paddy water [ 111 - 113 , 117 , 118 ]. However, with fertiliser application occurring at the start of plant cultivation, and dissolved oxygen content related to sunlight exposure ( e.g. via increasing photosynthesis), the primary oviposition attractant in rice fields is uncertain. The behavioural variability of An. arabiensis is clearly evident (Table 13 ), with similar numbers of studies reporting either anthropophilic or zoophilic behaviour. Bøgh et al. [ 119 ] stated: 'There is... great variation in the feeding preference depending on the local variation in host availability and composition of the local genotypes of the vector' [ 95 , 96 , 120 ]. Tirados et al. [ 121 ] suggested the existence of an east-west behavioural cline. They proposed that those populations found in western Africa display higher levels of anthropophily, and preferentially feed and rest indoors, whereas those in the east exhibit greater zoophily and rest outdoors. Overall, however, biting patterns tend to be exophagic [ 121 - 124 ], but such behaviour is often reported in comparison with highly endophagic species such as An. gambiae. For example, Fontenille et al. [ 125 ] reported An. arabiensis as 'more exophagic than An. gambiae and An. funestus ' with 65.4% of vectors found biting outdoors identified as An. arabiensis , yet 59% of those found biting indoors were also identified as An. arabiensis . Blood feeding times also vary in frequency but biting generally occurs during the night. Peak evening biting times can begin in the early evening (19:00) or early morning (03:00) [ 121 , 123 , 126 - 131 ] (Table 13 ). This species does, however, demonstrate a predisposition to exophilic (or partial exophilic) behaviour regardless of where it has blood fed or the source of its meal [ 121 , 125 , 130 , 132 - 135 ], a behavioural trait considered to be related to polymorphic chromosomal inversions, to a greater or lesser extent, depending on location [ 97 , 132 , 136 , 137 ]. Anopheles funestus Anopheles funestus is a member of the Funestus Subgroup [ 138 ] (often mistakenly referred to as An. funestus complex), which includes: An. aruni , An. confusus , An. funestus , An. parensis and An. vaneedeni. The members of this subgroup exhibit important variation in their biology and behaviour, especially in regard to malaria vectorial capacity and are only morphologically distinguishable during certain stages in their development [ 10 , 11 , 18 , 139 ]. Only An. funestus is regarded as an important vector of malaria in this subgroup [ 18 ]. A typical An. funestus larval habitat is a large, permanent or semi-permanent body of fresh water with emergent vegetation, such as swamps, large ponds and lake edges. Larvae have been found in shaded and sunlit environments (Table 10 ) and Gillies & de Meillon [ 10 ] concluded that An. funestus uses emergent vegetation as refuge against predation while the shading it casts, or the presence of shade from overhanging plants, is of lesser importance. In some areas, An. funestus larvae, as with An. arabiensis , are associated with rice cultivation ( e.g. Madagascar, Mali) [ 140 - 144 ] (Table 11 ). Where they are found, their favoured environmental conditions are very different to those of An. arabiensis . Anopheles funestus replaces An. arabiensis in a successive temporal process during rice plant growth, exhibiting higher densities in older, maturing fields compared to the preceding open conditions preferred by An. arabiensis [ 115 , 143 , 144 ]. Anopheles funestus is considered to be highly anthropophilic [ 10 , 86 , 122 , 145 - 151 ] (Table 13 ) (but see below), which led Charlwood et al. [ 19 ] to propose that An. funestus may have been the first anopheline species to specialise on biting humans, surmising that its preferred larval sites (permanent water bodies in savannah-like environments) are likely to have been areas where humans first settled. Behaviourally, its late-night biting patterns would also allow ready access to human blood without incurring undue density-dependant host avoidance. This late-night biting preference is clearly evident throughout its range, with all studies reviewed reporting a peak biting period occurring after 22:00, and most commonly between midnight and the early hours of the morning [ 123 , 124 , 128 , 131 , 145 , 152 - 157 ] (Table 13 ). Endophilic resting behaviour is also commonly reported [ 84 , 86 , 114 , 124 , 125 , 145 , 146 , 149 , 152 , 156 , 158 , 159 ], and combined with a relatively high longevity, makes it as good a vector, or better in some areas, as An. gambiae [ 10 , 11 , 18 , 160 ]. These characteristics are also responsible for promoting the success of vector control using IRS and ITNs. However, this exposure has resulted in selection pressure and rapid development of insecticide resistance to pyrethroids, now well established in some populations and implicated as the primary reason for a major resurgence of epidemic malaria reported in Kwazulu-Natal, South Africa in the late 1990s [ 18 , 161 ]. Compared to other DVS in Africa, An. funestus shows fairly consistent behaviour (generally anthropophilic and endophilic) throughout its range; however, it is a highly adaptable species, allowing it to occupy and maintain its wide distribution and utilise and conform to the many habitat types and climatic conditions contained therein. Behavioural differences between chromosomal forms have been identified, for example, Lochouarn et al. [ 162 ] reported anthropophilic behaviour in western Senegal and zoophilic behaviour in the east of the country, behaviours which correspond to chromosomal polymorphisms that also follow this east-west cline. Costantini et al. [ 60 ] identified two chromosomal forms in Burkina Faso associated with different resting and biting behaviour. This, coupled with a lack of heterokaryotypes in areas where the two forms co-exist, prompted these authors to suggest that the two forms were incipient species, and hence of the concept of an An. funestus complex. More recently, An. funestus populations from 12 countries have been divided into three molecular types: M, W, and MW, correlating to geographical locations, whereby M is essentially found in eastern Africa, W from western and central Africa, and MW from southern Africa [ 61 ]. Further investigations showed a more complicated situation with specimens from Malawi showing all three types, specimens from Tanzania showed the M- and MW-types, whereas specimens from Kenya showed M- and W-types. In addition, two more types were described, type Y from Malawi, and type Z from four localities of Angola, Malawi, Ghana and Zambia [ 62 ]. Finally, adding further to the complexity surrounding this species, recent studies in Malawi have revealed a new species of the subgroup, named An. funestus -like [ 63 ] that is identical to An. funestus but appears to have a different biology and role in malaria transmission, although this needs confirmation. Anopheles gambiae Anopheles gambiae is considered to be one of the most efficient vectors of malaria in the world and is one of the most well studied [ 88 ]. Like An. funestus , the variable ecological conditions present within the large geographical range of An. gambiae indicate a highly plastic species with corresponding chromosomal diversity currently separated into five chromosomal forms: Forest, Bamako, Savanna, Mopti and Bissau [ 163 ]. There is suggestion of reproductive isolation among the sympatric forms, and hence, of incipient speciation between them [ 163 - 165 ]. Independent of these chromosomal categories, two molecular forms, 'M' and 'S', have also been described [ 165 ], and are the forms more commonly referred to in the recent literature. These different forms exhibit ecological adaptations which further indicate possible speciation, for example the Mopti and M forms are associated with semi-permanent, often man-made, larval habitats such as rice fields or flooded areas, whereas the Savanna/Bamako and S forms are seen more commonly in temporary, rain-dependent sites such as ground puddles [ 166 - 171 ]. There appear to be no definitive studies that explicitly describe variability in adult biting or resting behaviour or role in malaria transmission between the two molecular forms. Despite its wide range and variable ecology, a combination of traits allows An. gambiae to maintain its position as one of the most efficient vectors in sub-Saharan Africa. It is a relatively long-lived species (although not as long as An. funestus [ 160 ]) [ 172 , 173 ], with a short larval development period and is often found in larval habitats associated with human activity ( e.g. water in hoof prints, wheel ruts or areas of rice cultivation) (Tables 11 , 12 ). It is considered to be highly anthropophilic, with 11 of 15 studies that examined biting behaviour (Table 13 ) reporting a marked preference for human hosts [ 131 , 145 , 149 , 150 , 157 , 159 , 174 - 177 ]. However, there are a number of studies that indicate An. gambiae is less discriminant and more opportunistic in its host selection and that host choice is, as with the majority of African DVS, highly influenced by location, host availability and the genetic make-up of the mosquito population. Moreover, many studies that report host preference using blood meal analysis are often conducted on resting, blood-fed specimens collected inside houses, thus introducing a potential study design or sampling bias favouring the likelihood that the blood meal will be from a human host [ 178 ]. Of the studies that report some level of zoophily, Diatta et al. [ 178 ] specifically examined the host preference of An. gambiae and An. arabiensis by comparing the number of females of each species captured either in a calf-baited or a human-baited net trap. There was no statistical difference between the host preferences of the two species, both expressing greater zoophily ( e.g. 31% of An. gambiae were found in the human-baited trap and 69% in the calf-baited trap). Duchemin et al. [ 122 ] also reported zoophilic behaviour, yet highlighted this as unusual, suggesting that the high density of cattle in the sampling area may have influenced the propensity for zoophily in the population. Bøgh et al. [ 119 ] reported no specific preference for either human or animal hosts but that An. gambiae would feed readily on cattle. As with An. arabiensis , An. gambiae larvae typically inhabit sunlit, shallow, temporary bodies of fresh water such as ground depressions, puddles, pools and hoof prints (although see above) [ 91 , 101 , 175 , 179 - 183 ] (Table 10 , 12 ). Gillies & de Meillon [ 10 ] suggested that this aspect of their bionomics allow members of the An. gambiae complex to avoid most predators, and the larvae are able to develop very quickly (~six days from egg to adult under optimal conditions and temperatures), possibly in response to the ephemeral nature of their larval habitats. Water in these larval sites can appear clear, turbid or polluted [ 101 , 180 , 184 - 186 ] (Table 10 ). Typically An. gambiae larval habitats are described as containing no (or very sparse) vegetation (Mbogo, unpub. obs.) due to their temporary nature. Gillies & de Meillon [ 10 ] summarised the great diversity of habitats utilised by An. gambiae , and as described before, different molecular or chromosomal forms are associated with either vegetated ( e.g. rice fields) or temporary and non-vegetated ( e.g. hoof prints) larval sites [ 101 ]. The studies reviewed here report An. gambiae from habitats containing floating and submerged algae, emergent grass, rice, or 'short plants' in roadside ditches and from sites devoid of any vegetation [ 91 , 101 , 109 , 180 , 181 , 183 ] (Table 10 ). Females of An. gambiae typically feed late at night, a characteristic shared with An. funestus that may increase their ability to effectively transmit malaria parasites (see above) [ 19 , 123 , 127 , 145 , 153 , 154 , 157 , 175 , 177 , 185 , 187 - 190 ] (Table 13 ). Anopheles gambiae is often described as an endophagic and endophilic species, both biting and resting indoors, however, the majority of studies listed herein (nine of 11), that compared indoor and outdoor human-landing catches reported no difference in the numbers of females collected at either location [ 123 , 127 , 145 , 149 , 153 , 157 , 175 , 190 , 191 ] and an equal number of studies recorded post-feeding exophilic resting [ 122 , 131 , 154 , 175 ] as resting indoors [ 145 , 149 , 159 , 178 ]. Bockarie et al. [ 175 ] linked differences in the exo-or endophilic behaviour of An. gambiae to their chromosomal forms, suggesting the Forest form (with no inversion) demonstrated stronger exophily in southern Sierra Leone whereas the Savannah form, with a 2La inversion, was mostly endophilic. Odiere et al. [ 192 ] used clay pots to sample outdoor resting females in western Kenya and found no clear preference for indoor or outdoor resting. They suggested that the designation of An. gambiae as a predominantly endophilic species may have been based on poor sampling comparisons. As with host preference, this species appears to exhibit greater phenotypic plasticity and opportunism in blood feeding and resting locations than commonly thought. Anopheles melas There is relatively little contemporary information about the behaviour of An. melas , perhaps because it is generally considered to be a vector of lesser importance, specifically where it occurs in sympatry with An. gambiae or An. arabiensis . Anopheles melas has a comparably lower sporozoite rate than either An. arabiensis or An. gambiae ( e.g. 0.35% compared to 3.5% for An. gambiae in The Gambia) [ 13 , 95 , 193 ], yet in coastal areas where it can occur in very high densities it is still a problematic vector of malaria [ 13 ]. With the dearth of available contemporary data, those studies conducted prior to 1985 that closely examined the behaviour of this species have been included here. Anopheles melas is commonly associated with brackish water and can utilise saline environments that other species, for example, An. gambiae , cannot tolerate [ 109 , 171 ], yet does not appear to require brackish water for larval stage development [ 194 - 196 ]. It is generally restricted to coastal areas [ 194 - 197 ] but has been found up to 150 km inland along the Gambia River, where salt water can intrude great distances (up to 180 km) upriver [ 109 , 171 , 193 ]. Unlike other African DVS, the density fluctuations of An. melas are closely associated with tidal changes rather than seasons, for example, Gelfand [ 194 ] identified a peak in adult numbers 11 days after spring tides. The larvae of this species are associated with salt marsh grass ( Paspalum spp.) and mangroves, but only trees of the genus Avicenna , which include white, grey and black mangrove, and not those from the genus Rhizophora ('true' or red mangrove spp.) [ 109 , 194 , 195 , 197 ]. These positive and negative associations with mangroves are thought to be strongly influenced by the predominant soil type associated with the different tree genera. Anopheles melas preferentially oviposits on damp ground at low tide, rather than in open water, where the eggs are able to survive some degree of desiccation [ 196 ] until the tides rise again, and appears to prefer the poorly drained, peaty-like soil common to Avicenna forests compared to the sandy, gravelly or smooth, fibrous peat soils common to the Rhizophora stands [ 195 , 198 ]. Giglioli [ 198 ] surmised, that this behaviour guarantees the larvae will have sufficient time to complete their larval development and pupate in the less saline, relatively permanent waters of the new tide before it begins to recede and the water either becomes too salty, or dries out completely. Adult biting behaviour appears to be opportunistic. Anopheles melas has been described as both highly anthropophilic and a zoophilic species [ 193 , 194 , 197 , 199 , 200 ]. In a choice experiment, Muirhead-Thomson [ 197 ] varied the numbers of animal and human baits in traps to attempt to describe host preference. He found An. melas to be fairly indiscriminate: where there were more animal baits, An. melas would feed more often on animals, but still feed on humans. On the contrary, where there was an increase in the number of human hosts, a sharp decrease in the number of females feeding on animals occurred. Sampling bias towards anthropophily may be reported when blood fed females collected resting inside houses are tested for host blood type because An. melas generally appears to rest outdoors after feeding [ 193 , 194 , 197 ], although there has been limited success in locating and collecting from such natural outdoor resting sites. As previously described for An. gambiae , those females that bite and rest indoors are more likely to have fed on humans, and those biting or resting outdoors (or in animal sheds) are more likely to have bitten animals. Blood feeding activity appears to be fairly continuous throughout the night [ 194 , 197 , 200 ]. Gefland [ 194 ] observed continual biting from 19:00 to dawn, although Muirhead-Thomson [ 197 ] saw two peaks of biting activity: the first, and slightly smaller peak, between midnight and 02:00 and a second, larger peak, between 04:00 and dawn. Anopheles merus Anopheles merus has previously been considered as only a minor, or even an unimportant vector, potentially unable to sustain malaria transmission alone [ 95 ]. However, is has been identified as playing an 'unexpectedly important role' along the Tanzanian coast [ 14 ] and more recently in Mozambique [ 15 ]. It is also a species for which there is limited contemporary information. The differences in egg and larval morphology that distinguish An. melas from An. gambiae do not occur in An. merus and identification, before the advent of molecular techniques, was based on physiological characteristics involving larval salinity tolerance tests [ 201 ]. Originally, An. merus was referred to as a 'salt water An. gambiae ' variant or subspecies. Indeed, Jepson et al. [ 202 ] had a number of specimens collected in the 1940s from saline, coastal swamps in Mauritius examined for distinguishing features, and found no obvious morphological distinguishing characters and stated 'All the specimens proved to be typical forms [of An. gambiae ] and there was no evidence of the presence of An. gambiae var. melas '. They continued to regard ' An. gambiae ' as a species with 'a considerable tolerance for pollution and salinity and is therefore to be found in domestic wastes and in crab holes and pools near the sea side, in addition to a host of natural breeding places such as marshes, rock pools and casual rainwater pools'. This Mauritian species was finally designated a subspecies of An. gambiae by Halcrow [ 203 ], who provisionally named it An. gambiae litoralis based on larvae found in '...water of high salinity in crab holes, depressions in coralline rocks, small tidal lagoons, pools close to tidal zone and [interior] salt pans, and are not associated with mangroves...' [ 203 , 204 ]. Paterson [ 205 ] provided definitive proof of the specific status of An. merus and the validity of the name [ 206 ]. Halcrow's [ 203 ] description highlights a specific difference between An. merus and An. melas . Anopheles merus is rarely found in the mangrove forests on the east coast, however this may be due to the composition of the trees and soil type under of the stands of mangrove in this zone rather than inherent behavioural differences between the two species [ 10 ]. Anopheles merus is, instead, found in high numbers in shallow brackish pools and marsh or swamp areas along the coast. As a consequence, this species does not exhibit density changes in response to the tidal fluctuations as seen with An. melas , nor does it appear to tolerate the same high levels of salinity [ 201 , 207 ]. Anopheles merus is also known to occur further inland, using salt pans and saline pools larval habitats [ 11 , 208 - 211 ], and cross-mating experiments between inland and coastal populations have produced viable offspring indicating they are conspecific [ 212 ]. The biting behaviour of An. merus is similar to that of An. melas : generally opportunistic in host selection, depending on host availability [ 203 , 213 ] and with a tendency to bite [ 207 , 214 ] and rest outdoors [ 201 , 206 , 213 , 214 ]. Gillies & de Meillon [ 10 ] suggested that An. merus shows a preference for animal hosts, referring to a laboratory test where, given a choice, females consistently fed on calf versus human bait. Two of the studies reviewed here reported anthropophily [ 150 , 214 ], one indicated zoophily [ 203 ] and another concluded that no obvious preference was detected [ 213 ]. In the latter study, blood meal analysis was conducted on mosquitoes collected resting indoors (59.2% had fed on humans), and those collected resting outdoors (71.4% had fed on cattle and only 1.6% contained human blood) [ 213 ], highlighting the bias in drawing conclusions on host preference if only indoor or outdoor resting specimens are tested. Only one study, conducted on the Kenyan coast, examined the biting times of An. merus [ 214 ], which reported the number of bites gradually rising from early evening (18:00) peaking between midnight and 01:00 and then declining to 06:00 which corresponds to the accepted biting pattern for this species across its range (Bangs and Mbogo, unpub. obs.). Anopheles moucheti Anopheles moucheti is a species with two morphological forms: An. moucheti moucheti , and An. m. nigeriensis which are distinguishable by morphological features of the adult and larval stages [ 10 ]. Anopheles m. bervoetsi , previously considered a third morphological form, has recently been raised to full species status: An. bervoetsi by Antonio-Nkondjio et al. [ 215 ]. However, these authors do assert a level of caution in this new status as they point out that An. bervoetsi has only ever been reported from its type locality (Tsakalakuku, DRC) and has never been found in sympatry with An. moucheti . They do cite unpublished data that detected P. falciparum infection in An. bervoetsi specimens, and thus raises the possibility that this species could be transmitting malaria in central Africa [ 215 ]. The bionomic information detailed here is, in the most part, taken from sources that present data for ' An. moucheti '. Of these, the majority of studies have been conducted in Cameroon by Antonio-Nkondjio and colleagues or in Nigeria, so based on current knowledge the assumption is that these data refer to An. moucheti and not An. bervoetsi. Despite its status as a DVS, An. moucheti is a poorly studied species. It is the only DVS with its range entirely restricted to forested areas [ 216 ], specifically where the canopy is broken allowing sunlight to penetrate to the ground, such as is found where large rivers flow through the forest [ 10 ]. Human activity, such as road building, settlements or cultivation, can therefore be beneficial to this species by breaking up the forest canopy, although larger areas of deforestation may decrease the density of An. moucheti and allow replacement by An. gambiae [ 217 , 218 ]. Anopheles moucheti larvae are found at the edges of large, slow flowing or lentic rivers, often with turbid waters, and are associated with Pistia spp (water lettuce/water cabbage) [ 89 , 217 , 219 ]. Antonio-Nkondjio et al. [ 217 ] studied the larval habitats along the river networks of southern Cameroon and found the greatest numbers of An. moucheti larvae along the margins of rivers within deep, evergreen forest, substantially fewer in the degraded forest and none in the savannah areas. Where they were found, larvae were abundant near to areas of human habitation. Although the range of An. moucheti is relatively restricted within the equatorial forests, it derives its status as a DVS from its highly anthropophilic and endophilic behaviour (Table 13 ) [ 86 , 145 , 174 , 219 , 220 ]. Gillies & de Meillon [ 10 ] suggested such behaviour is unsurprising due to the lack of domestic animals found within forested environments. Anopheles moucheti is also described as highly endophagic, however this characteristic appears to be less than clear cut. For example, Antonio-Nkondjio et al. [ 220 ] found that in urbanised, forested environments (where An. moucheti was less abundant and replaced by An. gambiae ) compared to rural localities (where An. moucheti was dominant), only 43% of females were found biting indoors, whereas in the rural areas 66% were found biting indoors. In a study conducted in a village only 2 km from Yaounde, Cameroon, Antonio-Nkondjio et al. [ 145 ] reported 51% biting indoors and described the sampled populations as 'mainly endophagic'. Overall, An. moucheti appears endophilic [ 86 , 145 ] (Table 13 ). In a countrywide survey of Cameroon, of all females found resting, 1234 were located indoors, whereas only 12 were captured in outdoor shelters [ 86 ]. Only two studies examined the biting cycle of An. moucheti , with both reporting biting gradually increasing towards the second half of the night to dawn [ 145 , 221 ]; Mattingly [ 221 ] reported peak biting activity in the early morning between 03:15 and 06:15. Anopheles nili complex The An. nili complex includes An. carnevalei , An. nili , An. ovengensis and An. somalicus [ 12 ]. As with An. moucheti , species of this complex have been generally overlooked in African vector studies despite being described as highly efficient vectors [ 6 , 89 , 222 , 223 ]. Amongst members of the complex, An. nili is considered the most important vector, although An. carnevalei and An. ovengensis are implicated as secondary vectors of P. falciparum in Cameroon [ 86 , 224 ]. Anopheles somalicus is considered zoo- and exophilic [ 6 , 10 ]: it was not found to bite humans in Somalia [ 10 ] and no females were found in houses in Cameroon despite an abundance of larvae in the area [ 6 ]. Larvae of all members of the An. nili complex are found in vegetation at the edges of fast flowing streams and rivers [ 10 , 89 , 195 , 217 ]. However, An. ovengensis and An. carnevalei appear to be restricted to areas of deep forest, whereas An. nili is more abundant along rivers in degraded forest and savannah [ 217 ]. A comprehensive survey of the river systems across Cameroon found An. nili larvae associated with sunlit sites whereas An. carnevalei larvae were more commonly found in shaded areas [ 217 ]. Anopheles nili is considered to be strongly anthropophilic [ 10 , 86 , 145 , 148 , 225 - 227 ], and will readily bite both indoors and out [ 145 , 149 , 226 , 228 ] (Table 13 ). Carnevale & Zoulani [ 226 ] described biting patterns that exploited the behaviour of their human hosts, biting outdoors in the early evening when people are socialising, and then continuing to bite indoors once people move inside, with peak feeding occurring after midnight [ 145 ]. The resting habits of An. nili are also described as 'variable' [ 10 ]. Krafsur [ 227 ], in a lowland region of western Ethiopia, rarely found An. nili resting indoors despite the high densities found biting indoors, indicative of exophilic behaviour. Conversely, Antonio-Nkondjio et al. [ 86 ] examined populations across Cameroon and reported An. nili overwhelmingly resting indoors (466 females), with only one female captured in an outdoor shelter. In the same study they found no An. carnevalei females resting indoors or in outdoor shelters whereas all resting An. ovengensis captured were found indoors. Conversely, Awono-Ambene et al. [ 224 ] stated that An. ovengensis was rarely found resting indoors and concluded it had 'exophilic habits'. Bionomics of the European and Middle Eastern DVS Anopheles atroparvus Anopheles atroparvus is a member of the Maculipennis Subgroup, which also includes An. ( Ano. ) daciae , An. ( Ano. ) labranchiae , An. ( Ano. ) maculipennis , An. ( Ano. ) martinius , An. ( Ano. ) melanoon , An. ( Ano. ) messeae , An. ( Ano. ) persiensis and An. ( Ano. ) sacharovi [ 12 ]. Of these, An. labranchiae , An. messeae and An. sacharovi are also designated as DVS (see below). Anopheles atroparvus is described as a species with a preference for brackish larval habitats [ 229 - 232 ]. Hackett & Missiroli [ 231 ] summarised: 'In general it may be said that over its extensive range [ An. ] atroparvus is found in water of moderate salinity not exceeding 10 parts per 1000. It prefers relatively cool water, and its range does not overlap that of [ An. ] labranchiae , a warm water breeder'. However, the larval sites listed in the literature still include a number of predominantly fresh water habitats, for example canals, ditches, river margins, pools in river beds and rice fields [ 230 ], and Cambournac [ 233 ] defines An. atroparvus as a 'fresh water breeder'. Hackett [ 234 ] also stated that, in southern Europe, An. atroparvus 'inclines to breed in fresh water'. Of the few studies reporting primary data (Tables 14 - 17 ), larvae were identified in marshes and ditches/ground flood pools [ 235 ], pools in river beds, river margins and streams, rock pools, cement tanks, rice fields, wells and ground pools [ 229 ] and in small collections of water in used tyres [ 236 ] (Tables 15 , 16 ). Becker et al. [ 230 ] described sites to be 'usually sun exposed' and to contain 'a considerable amount of filamentous green algae and other floating submerged vegetation'. Pires et al. [ 229 ], in a study that sampled comprehensively across Portugal, reported An. atroparvus larvae to be found more frequently in sun-exposed habitats, although 'some shade was provided by grasses and aquatic vegetation'. They also reported filamentous algae present in 48 of 93 sites positive for An. atroparvus (Table 14 ). Anopheles atroparvus is generally considered zoophilic [ 229 , 230 , 237 ], and described as 'very zoophilic' by Cambournac [ 233 ], who also stated that its hosts, in order of preference, are rabbit, horse, cow, pig and sheep, and suggested that a long association between rabbit and An. atroparvus (since approx. 1000 BC) may be responsible for this hierarchy of preference. Indeed, An. atroparvus has been implicated as an effective vector of the myxomatosis virus to domestic rabbits in the UK [ 238 , 239 ]. Elsewhere, however, An. atroparvus is described as anthropophilic [ 89 ], which perhaps indicates the opportunistic nature of this species. Four studies identify An. atroparvus as zoophilic [ 229 , 237 , 240 , 241 ] and one study, that did not distinguish a preference, reported the collection of An. atroparvus during night catches on horse bait, from indoor resting sites and during day- or night-time catches on humans [ 235 ] (Table 17 ). There is no clear evidence or information among any of the published studies, nor within the general literature, that identifies this species as preferentially biting indoors or outdoors. The opportunistic nature of its feeding habits and zoophilic proclivity in host choice, however, would suggest it is probably exophagic but that biting location could also depend upon the setting and accessibility of the host. Anopheles atroparvus rests and hibernates in animal sheds and stables [ 229 , 230 , 235 , 237 , 238 , 240 , 241 ]. It hibernates as an adult female and is known to periodically feed, specifically if she has taken refuge in a relatively warm locality, but these meals do not result in egg production (i.e. gonotrophic disassociation) [ 230 - 232 ]. A number of investigators have discussed the inability of An. atroparvus to transmit tropical strains of P. falciparum , with most referring to studies conducted by Shute [ 242 ]. Unfortunately this reference could not be found, but in a study testing the susceptibility of Russian anopheline species to imported P. falciparum [ 40 ], no infection was detected in An. atroparvus females. Curtis & White [ 243 ] concluded (also referring to Shute [ 242 ]) that An. atroparvus is refractory to both Asian and African P. falciparum but competent in supporting a European strain, a conclusion reiterated by de Zulueta et al . [ 39 ] with Cambournac [ 233 ] stating that refractoriness of An. atroparvus to African and eastern strains of P. falciparum is an 'uncontroversial fact'. However, Capinha et al . [ 244 ] claimed the existence of local An. atroparvus in Portugal that could be infected with 'exotic strains of plasmodia', with reference to a comprehensive study by Souza [ 245 ]. However, on closer examination of these findings, even though Sousa did indeed infect An. atroparvus with P. falciparum , this was only after numerous attempts that resulted in formation of oocysts in only five out of 736 females. It would seem, therefore, that although An. atroparvus can be infected by tropical P. falciparum strains, it is very unlikely to happen under natural conditions and there is currently no conclusive evidence that such infection would result in salivary gland invasion by sporozoites. Anopheles labranchiae Despite similarity in larval site characteristics, An. labranchiae and An. atroparvus do not, or only have limited, overlap in their distributions [ 231 ]. This lack of sympatry may be simply a factor of temperature, with An. labranchiae making use of warmer waters than typical of An. atroparvus [ 230 , 231 ]. However, when Capinha et al. [ 244 ] modelled the habitat suitability of An. atroparvus across Portugal, they concluded that the most suitable locations include drier areas with higher temperatures (i.e. conditions where An. labranchiae typically dominate), whereas wetter areas with milder temperatures, where An. atroparvus are mostly found, were unsuitable. They concluded that An. atroparvus is not found in many other 'suitable' Mediterranean areas due to competitive exclusion. Conversely, de Zulueta [ 246 ] suggested that the absence of An. atroparvus in Sardinia allowed the wide distribution of An. labranchiae on the island, where, despite a five-year eradication campaign instigated in 1946, An. labranchiae still occurs [ 247 , 248 ]. Both species utilise brackish water marshes and lagoons along the coast [ 231 ], although in contrast to An. atroparvus , An. labranchiae will preferentially oviposit in fresh water [ 89 , 247 , 249 - 251 ]. Marchi & Munstermann [ 247 ], in a survey conducted across Sardinia, only identified An. labranchiae in fresh water sites, including rock holes, pits, ditches, drains or canals, streams/rivers, flooded ground pools and ponds, lakes or reservoirs. Despite an ability to tolerate some salinity, An. labranchiae larvae are not generally found at sites with significant levels of organic or mineral pollutants ([ 252 ], Mouchet, pers. com.). Larval sites are typically described as sunlit [ 89 , 230 , 249 , 253 ], although in Sardinia Aitken [ 251 ] found larvae in 'almost every type of habitat except the very densely shaded', and Macdonald [ 250 ] also associated this species with habitats that have some level of shade. In general, An. labranchiae larvae are found in stagnant or slow moving waters [ 230 , 249 ] and can make use of, and become very abundant in, rice fields [ 89 , 253 - 256 ]. Indeed, Bettini et al. [ 254 ] described a survey in central Italy that identified high numbers of larvae in newly established (two years old) rice fields with correspondingly high numbers of adults found resting in animal shelters near these fields. Female An. labranchiae can aggressively attack human hosts [ 230 , 255 ], and are described as 'persistent' in their attempt to enter bedrooms during the night [ 230 ]. Nonetheless, this species is also described as zoophilic in some of the general literature, but overall, An. labranchiae appears opportunistic in its host choice, readily biting either humans or animals (Table 17 ) [ 89 , 249 , 250 , 253 , 255 - 257 ]. Romi et al. [ 256 ] found high percentages (86% and 90.7%) of females engorged with human blood resting inside houses whereas they also found that almost all specimens collected resting in animal shelters had fed on animals. Anopheles labranchiae rests inside houses, animal shelters, and, to some degree, in natural shelters, depending on the location of its blood source [ 249 , 253 - 259 ]. D'Alessandro et al. [ 249 ] described An. labranchiae as both endo- and exophilic, using whatever shelters are available. Females hibernate in stables/animal shelters and in natural sites such as crevices and tree cavities. Both incomplete (with occasional blood feeding but without ovipositioning) and complete (with fat bodies, without feeding and non-gonoactive) hibernation have been noted for this species [ 89 , 230 , 231 , 249 ]. As with An. atroparvus , An. labranchiae has been found to be refractory to exotic strains of P. falciparum , with de Zulueta et al. [ 39 ] failing to infect An. labranchiae , albeit a small sample, with a Kenyan strain of P. falciparum . However, Toty et al . [ 58 ] reported historical evidence of naturally infected An. labranchiae , plus the results of a contemporary study conducted by the Centre de Production et d'Infection d'Anophèles (CEPIA) in Paris where 14% (13/99 specimens) of Corsican An. labranchiae were experimentally infected with the African NF54 laboratory-cultured strain of P. falciparum . This study also detected sporozoites in the salivary glands of three specimens, indicating that An. labranchiae is not only susceptible but also potentially able to transmit at least some strains of African P. falciparum [ 58 ]. However, this conclusion must only be considered alongside the knowledge that the NF54 P. falciparum strain is a highly attenuated, long-standing laboratory culture which may no longer reflect its origins (Bangs, unpub. obs.). Anopheles messeae Anopheles messeae is the third member of the Maculipennis Subgroup [ 12 ] to be designated as a DVS. It is the most widespread species of the subgroup [ 230 ], with a distribution extending from Ireland across Europe and Asia and into China and Russia [ 260 ]. A great deal of work on this species has been conducted in Russia and China. This review is therefore presented with the caveat that there may be details and data reported in the Chinese or Russian literature that are not included here due to access difficulties. Di Luca et al. [ 261 ] identified a number of genetic polymorphisms within An. messeae and defined five separate haplotypes associated with different geographical areas across its distribution. However, they could not confirm whether these polymorphisms were indicative of altered behaviour at these different locations, although the large range of this species combined with such genetic variability would suggest that some area-specific biological or behavioural adaptations are likely to have occurred. The larvae of An. messeae are typically found in shaded, clear, very slow flowing or stagnant, fresh water sites [ 230 , 262 - 264 ] such as lake margins and marshes [ 263 - 265 ]. Despite only sampling resting females, Adamovic, in Serbia and Montenegro [ 266 - 268 ] and Adamovic & Paulus [ 269 ], in surveys of Slovenia and Croatia, continually associated the presence of adult An. messeae with stagnant, fresh water oxbow swamps and marshes within alluvial plains or valleys of large river systems and at sites near large lakes. Localities along rivers with saline or alkaline soil did not provide the same association [ 266 , 268 ], however they did report the presence of An. messeae at sites near a marshy plain with brackish water [ 267 ]. Takken et al. [ 263 ] also inferred the presence of An. messeae in more brackish habitats, presenting a photograph in their paper of a drainage ditch labelled as containing brackish water and vegetation which 'supports Anopheles messeae '. However, they also indicated that engineering works in the Netherlands allowed the transition of brackish sites to fresh water, so whether or not they did find this species in brackish water is still unclear. Nonetheless, Takken et al. [ 263 ] did identify locations where An. messeae larvae were collected, including sites containing reeds, and those containing floating aquatic weeds and algae, relatively open ditches inside forests and clear water in small lakes within dunes. Only one study could be found that sampled An. messeae females inside human habitations, animal shelters and in natural outdoor shelters [ 270 ]. All other studies only searched in animals shelters [ 240 , 263 , 266 - 269 , 271 - 273 ]. Where comparisons were made, no An. messeae were found resting outdoors in urban areas ( e.g. in vegetation surrounding buildings) but were found indoors such as in entryways, staircases and basements, although not in large numbers. In rural areas, An. messeae dominated the collections made from cattle sheds, with few specimens collected from natural outdoor sites (hollows, ground cavities, amongst vegetation surrounding marshes, ponds and streams) [ 270 ]. Takken et al. [ 263 ] argued that in the Netherlands An. messeae has never been considered as a malaria vector despite it being as susceptible to P. vivax infection as An. atroparvus [ 40 ]. They stated that its high degree of zoophily and outdoor feeding behaviour makes the likelihood of it being involved in local malaria transmission very remote. They supported their argument reporting that all resting An. messeae females collected in their study had fed on animals; however, all their samples were collected from animal shelters. Bates [ 265 ] also mentioned that in Albania, An. messeae (along with other members of the Maculipennis Group) is not considered a malaria vector using the same reasoning: '[ An. maculipennis , An. messeae and An. melanoon (as An. subalpinus )] are generally supposed not to be malaria vectors because of their non-anthropophilous [ sic ] food habits'. Fyodorova et al. [ 270 ] found that 40% of the An. messeae females collected in urban areas contained human blood, with the remaining 60% having fed on cats (40%) and chickens (20%). However, in rural areas, no human blood meal was detected. Becker et al. [ 230 ] summed up the biting preferences of An. messeae somewhat ambiguously, stating that 'Blood-meals are taken from humans only when the density of An. messeae is very high and there is a shortage of livestock, but they also may attack humans in houses'. No studies were found that examined the feeding cycle of An. messeae . Anopheles messeae , like An. atroparvus and An. labranchiae , hibernates as an adult female. However, unlike these other two species, An. messeae chooses hibernation sites in abandoned buildings, in the absence of animals [ 230 , 271 ]. They enter full diapause, and do not feed during the winter, but instead, gain energy from fat reserves [ 271 ]. There is some evidence to suggest that, along with An. atroparvus , An. messeae may also be refractory (or essentially refractory) to tropical P. falciparum strains. In their study, testing the susceptibility of Russian anophelines to imported P. falciparum , Daškova & Rasnicyn [ 40 ] were unable to infect An. messeae. Indeed, the vector status of An. messeae has come into question, specifically since the discovery of a new species in 2004, formally named An. daciae , in Romania [ 273 ], which has since been recorded from south-western England [ 274 ]. Anopheles daciae can only be distinguished from An. messeae using egg morphology or by sequencing the internal transcribed spacer 2 (ITS2) of ribosomal DNA and the cytochrome oxidase 1 (COI) of mitochondrial DNA. It has been suggested that the presence of An. daciae , potentially sympatric across the full range of An. messeae , may be responsible for the high polymorphism previously reported for An. messeae [ 261 , 274 ]. Combined with an ongoing debate about the capacity of An. messeae to transmit malaria ( e.g. it is not considered a vector in northwestern Europe [ 275 ]), it is feasible that An. daciae , and not An. messeae , could be involved in malaria transmission, which will only be confirmed with further investigation into the epidemiological importance of each respective species (Harbach, unpub. obs.). Anopheles sacharovi Anopheles sacharovi is the final member of the Maculipennis Subgroup defined as a DVS and has been the target of a number of focussed, anti-vector campaigns across its range including Israel, Greece and Turkey [ 262 , 276 - 279 ], yet this species still persists in all areas. Anopheles sacharovi is highly plastic in both adult behaviour and its choice of larval habitats. Zahar [ 262 ] states simply: '[ An. sacharovi ] breeds in all small water collections containing aquatic vegetation'. It makes use of fresh water habitats but is also described as more tolerant of salinity (up to 20%) than any other member of the Maculipennis Subgroup [ 230 , 262 ]. It can survive in waters up to 38-40°C ([ 280 ] references within), and although it is generally considered to breed in stagnant waters, it can also cope with some, albeit weak, current [ 281 , 282 ]. Throughout the literature there is general agreement that this species prefers sunlit sites with plenty of emergent and/or floating vegetation [ 89 , 230 , 262 , 283 - 285 ]. A typical habitat would be an area of swamp or marsh [ 265 , 279 , 282 ], but larvae are also found at margins of rivers, streams and springs [ 281 , 282 ], seepages [ 281 ], wadis [ 286 ], pools and ditches [ 265 , 287 ]. It is associated with rice cultivation and other irrigated areas, specifically where irrigation channels are poorly constructed causing leakage, creating boggy areas or standing water [ 89 , 230 , 277 , 279 , 282 , 284 , 288 , 289 ]. Despite its apparent adaptability, An. sacharovi cannot tolerate organic pollutants [ 262 , 285 ]. Indeed, Saliternik [ 285 ] lists the organic pollution of streambed habitats, previously densely populated with An. sacharovi larvae, of greater impact than the wide-scale IRS application of DDT as causal to the near elimination of this species in Israel in the 1960s. Anopheles sacharovi females feed opportunistically, despite being generally considered as anthropophilic [ 89 , 230 ]. Only one study reviewed specifically tested host preference. Demirhan & Kasap [ 290 ], using baited feeding rooms, concluded that in the presence of other, equally available hosts (human, cow, sheep, chicken, horse and donkey), An. sacharovi preferentially fed on donkeys, and had a negative preference for humans. They also analysed the blood meals of engorged females from human habitations, animal shelters and abandoned or ruined buildings and reported the 'feeding preference' of females captured in the human dwellings to be cow, human, sheep, horse and chicken. Other studies reported similar results. Yaghoobi-Ershadi et al. [ 291 ] found high numbers of females collected from cow sheds or chicken coops had fed on animals (85.6 - 92.5%), whereas of those collected from bedrooms, only 38.5% had fed on humans, 38.5% on other animals and 23% on both. Boreham & Garrett-Jones [ 292 ] reported predominantly human blood in specimens collected from houses, predominantly animal blood (sheep or goat) from animal shelters, and those collected from outdoor pit shelters generally contained blood of mixed animal origin (sheep, goat, horse, dog or cow). Hadjinicolaou & Betzios [ 276 ] reported a high percentage of females containing human blood from human habitations, whereas females taken in pit shelters and animal sheds had mostly fed on domesticated animals. They concluded that An. sacharovi still exhibited significant levels of anthropophily despite a high ratio of animals to people (between 9:1 and 7.2:1) in the study area. Boreham & Garrett-Jones [ 292 ] suggested that An. sacharovi had increased tendencies towards zoophilic behaviour due to previous DDT spraying campaigns, but was reverting back to anthropophily. Anopheles sacharovi , contrary to the accepted night-time biting habits of most anophelines, can 'in deeply shaded situations... attack viciously throughout the day' [ 289 ]. However, Djadid et al. [ 293 ], indicated that An. sacharovi (plus other members of the Maculipennis Subgroup) generally start biting in the early evening, peaking between 20:00 or 21:00 (refers to Djadid MSc thesis), with Alten et al. [ 294 ] noting higher densities of An. sacharovi between 20:00 and 22:00, although they did not specifically examine biting behaviour. Hadjinicolaou & Betzios [ 276 ] observed An. sacharovi to bite indoors and outdoors. Biting location is likely to be driven by host behaviour, for example, in the hotter parts of Turkey where both people and animals spend the night outdoors, biting would tend towards exophagy [ 289 , 294 ]. Anopheles sacharovi is principally described as endophilic [ 89 , 230 , 284 ]. Its choice of resting location (and, arguably, for all species) is most likely driven by the need to find the most suitable microclimate for increased survival [ 289 ]. Demirhan & Kasap [ 290 ] observed An. sacharovi feeding on cows outside, and then entering houses or abandoned shelters to rest. Yaghoobi-Ershadi et al. [ 291 ] found An. sacharovi in cow sheds, chicken coops and bedrooms, but were unable to find any females resting outdoors. Boreham & Garrett-Jones [ 292 ] searched two artificial pit shelters and found 10 and 42 specimens compared to 377 and 333 from two cattle sheds and 260 in a village house. Abdel-Malek [ 282 ] failed to find An. sacharovi resting outdoors, but again, repeatedly found them resting in animal stables and human habitations. However, insecticide residual spraying in many areas has apparently affected endophilic behaviour [ 283 , 295 , 296 ], summed up by Gokberk [ 283 ]: 'Following the last ten years of DDT spraying, An. sacharovi recently began to show a tendency to be less domestic in habits'. Yet, there is evidence that once these IRS programmes ceased, An. sacharovi began to revert to more typical endophilic behavioural patterns [ 276 ]. As with other European or Middle East DVS that occur in warmer climates, hibernation is incomplete, with intermittent feeding during winter, but without oviposition [ 89 , 297 ], often making use of the same localities chosen for resting in the summer months [ 288 ]. Anopheles sergentii There is some confusion as to the taxonomic status of An. sergentii . It has previously been considered to have two geographical forms: An. sergentii sergentii and An. sergentii macmahoni , but in accordance with the published literature, the Walter Reed Biosystematics Unit (WRBU) online catalogue of the Culicidae [ 298 ] and the Mosquito Taxonomic Inventory [ 299 ], An. macmahoni is currently considered a subspecies of An. sergentii . This subspecies has never been found biting humans and is of no known medical importance [ 10 , 300 ]. Anopheles sergentii is known as the 'oasis vector' or the 'desert malaria vector' due to its distribution within oases across the Saharan belt in northern Africa into the Middle East, and its ability to cope with the extreme climate across this region [ 301 , 302 ]. It may be able to survive in such harsh conditions due to its adaptability. It makes use of a range of larval habitats, including streams, seepages, canals, irrigation channels, springs, rice fields [ 103 , 230 , 262 , 286 , 300 , 303 - 308 ], and most other non-polluted, shallow sites that contain fresh water with a slow current, slight shade and emergent vegetation or algae (Table 14 ) [ 89 , 103 , 230 , 262 , 302 , 306 , 308 ]. However, larvae have also been found in moderately brackish habitats, areas of stagnant water, light to moderately polluted locations or in sites in full sunlight [ 303 - 305 , 307 ]. In general, the presence of vegetation or algae seems to be the only characteristic common to all larval habitats of this species [ 103 , 230 , 302 , 303 , 305 - 308 ]. Farid [ 302 ] described An. sergentii as '...an indiscriminate biter of both humans and animals, both indoors and out.', however, no study could be found that specifically tested host preference. Six studies have reported blood meal analyses of resting mosquitoes, taken from both human and animal shelters (Table 17 ). Of these, five described An. sergentii as principally or even highly zoophilic [ 103 , 259 , 306 , 308 , 309 ], with Kenawy et al . [ 308 ] stating the key factor that limits oasis malaria transmission in Egypt is the zoophilic feeding behaviour of An. sergentii . Faraj et al. [ 259 ], in Morocco, also described 'a marked preference for zoophily' in this species. Kenawy et al . [ 309 ] related human biting to local animal stabling practices. They found that in villages where animals were housed in rooms within human habitations, a lower proportion of the An. sergentii females collected resting in the houses contained human blood. In an earlier study, however, the proportion of females with human blood was higher in those taken from houses containing animal rooms (86%), although the absolute numbers (calculated here from percentages reported in the paper) of 18/21 versus 26/54 (48%) from houses with no animal rooms also indicate that An. sergentii were diverted away from human hosts and towards the animals when in close proximity of one another. In this latter study, the animal rooms within the houses had the highest number of resting mosquitoes (209), of which 13% (equivalent to 27 mosquitoes) had fed on humans compared to those in isolated animal sheds (126) with only 7% (equivalent to nine mosquitoes) containing human blood. No studies were found that specifically tested biting location, yet Abdoon & Alshahrani [ 310 ] reported high numbers biting outdoors and concluded An. sergentii was both exophagic and exophilic after finding few females resting inside houses. However, with no comparable indoor biting data and no sampling of resting mosquitoes from animal shelters, this conclusion may not indicate a true blood feeding preference. Indeed, Saliternik [ 285 ] described An. sergentii as feeding and resting both indoors and outdoors, but referred to 'exophilic habits'. Barkai & Saliternik [ 304 ] suggested that an 'exophilic strain' of An. sergentii had developed because of indoor spraying with DDT in Israel, finding fewer adults at indoor resting places where they had been common in the past, despite the local abundance of larvae. Anopheles sergentii can overwinter as both adult females or larvae [ 230 , 302 ], although no details regarding hibernation, blood feeding and oviposition could be found. Anopheles superpictus Preliminary data on An. superpictus populations sampled across Iran recently identified three genotypes (designated X, Y and Z) and raised the possibility of An. superpictus as a species complex [ 311 ]. These data have yet to be confirmed, but the wide distribution of this species across a number of diverse climatic regions (Mediterranean across to central and southwestern Asia) and the existence of eight junior synonyms, suggests the realistic possibility of An. superpictus being a complex of species and therefore warrants further investigation (Harbach, unpub. obs.). In the published literature, An. superpictus larvae are continually associated with gravel or pebble river and stream beds in shallow, slow-flowing clear water in full sunlight [ 230 , 250 , 265 , 282 , 285 , 289 , 304 , 312 - 314 ]. Typical, natural sites are small pools within or next to drying river beds, conditions which are closely related to seasonal fluctuations in precipitation [ 230 , 265 , 289 , 314 , 315 ]. At such sites, larval abundance increases only in late summer when pools are created as the river levels decline and, once water levels rise with the increasing rain during the onset of winter, these locations again become unsuitable as aquatic habitats [ 230 , 289 , 315 ]. Such natural limiting conditions could restrict both the distribution, abundance and period of adult activity of this species, however An. superpictus has easily adapted to human-influenced habitats, making use of irrigation channels and storage tanks and pools formed from their leakage, rice fields, ditches, borrow pits and hoof prints, amongst others [ 230 , 282 , 313 , 315 , 316 ]. Anopheles superpictus larvae have also been found in brackish water habitats [ 314 ] and in stagnant water [ 304 , 313 ]. Jetten & Takken [ 275 ] state that it can occur in polluted sites, although here, no primary data were found to confirm this statement, which is also contradicted by other observations. For example, Berberian [ 314 ], stated that ' A. superpictus is never found in polluted or filthy water...' and the decline of An. superpictus (and An. sergentii ) in Israel has been closely associated with sewage pollution of many of the natural streams it previously inhabited [ 285 ]. Anopheles superpictus survives at relatively high altitudes, up to 2800 m [ 264 ], replacing An. sacharovi [ 316 ] that may dominate at lower altitudes. No publication could be found that reports any definitive host preference for An. superpictus , but it is generally given to be a zoophilic species that also readily feeds on humans [ 230 ]. Tshinaev [ 315 ] reported from Latyshaev [ 317 ] (reference unavailable) that, in Uzbekistan 'individuals that who sleep out during the summer on the flat roofs of houses and on towers are not attacked by this mosquito'. Conversely, Ramsdale & Haas [ 289 ] stated that An. superpictus in Turkey has a 'marked preference for animals but feeds on man in their absence...', but still described An. superpictus as an 'unusually dangerous mosquito' for those people who spend nights out in the open, away from villages and towns. Again, no primary data were found describing biting location. However, An. superpictus appears to be opportunistic in its feeding habits and will enter houses to feed [ 250 ], but is generally regarded as exophagic [ 318 , 319 ].

Discussion The BRT model has been applied to contemporary data on the occurrence of 13 DVS in Africa, Europe and the Middle East using the most comprehensive database of DVS occurrence currently available. These maps and the underlying database will be made available in the public domain. We stress that the predictive maps produced will not be perfect representations of the true geographical distributions of these species but nevertheless, they represent a substantive step in improving our knowledge of the range of those DVS studied. One particular issue for any environmental-niche based mapping technique is predictions in areas that, although environmentally suitable, may not contain the vector for other biogeographical reasons. The model predicts, for example, the presence of An. arabiensis throughout Madagascar. Anopheles arabiensis is a species commonly associated with dry, savannah-type habitats and is considered absent, or at least, is rarely encountered in the humid climate of the eastern coast of Madagascar (Manguin, unpub. obs.). Conversely, An. nili has never been recorded in Madagascar (Manguin, unpub. obs.) but, as identified in the predicted distribution, there are areas where conditions are suitable for An. nili to become established if it were to be introduced. Biases in collection location, variation in sampling methodologies, limited data for some species or an absence of data over large areas of suspected occurrence all contribute to uncertainty in the final predictions. Yet despite these limitations, the maps represent the first attempt to model DVS distributions across Africa, Europe and the Middle East using a combination of extensive occurrence data combined with contemporary EO distributions. All this information should be triangulated when evaluating the utility of the maps which are best considered as the beginning of an on-going process of understanding, describing and better predicting the range of these DVS. This process may be further complicated by the ever evolving revision of taxonomic status of a number of the African DVS that may lead to further stratification of the occurrence data and revisions of the predictions. This is particularly important where newly identified forms are associated with varying bionomics relevant to their control; the molecular and chromosomal forms of An. gambiae are but one example. Bionomics The behavioural plasticity, large geographic ranges, and changing taxonomic categorisation, in particular with the African DVS, present challenges when summarising the bionomics of individual species. Moreover, conclusions drawn about behavioural characteristics based on biased sampling may mask the true variability in a population and behavioural adaptation to human influences, such as insecticide use or environmental disturbance, can also influence local variation in species bionomics. The bionomics data are again viewed as a significant compendium but with the caveat that expert, local knowledge should always complement the information provided. Future work This is the second in a series of three publications describing the distribution and relevant bionomics of the global DVS of malaria. The first publication [ 2 ] detailed the DVS of the Americas and the final publication will examine the DVS of the Asian Pacific region (Sinka et al : The dominant Anopheles vectors of human malaria in the Asia Pacific region: occurrence data, distribution maps and bionomic précis, unpublished). Together, these three publications are intended to provide a baseline set of data and maps and summarise the current knowledge of the bionomics of the 41 DVS identified as the primary vectors of P. falciparum and P. vivax malaria.

Conclusions The maps and data presented here, and those relating to the DVS of the Americas [ 2 ], and the Asian Pacific region, will be available on the MAP website [ 64 ] in accordance with the open access principles of the MAP (please contact authors for details). These data and maps are provided as a dataset to be improved and built upon. Undoubtedly, the process of species distribution mapping will improve, environmental and climatic spatial data will become available at higher resolutions, and more refined understanding of the ecology that limits a given DVS distribution attained. The single most important factor, however, will be more spatially comprehensive occurrence data and this exercise has been additionally valuable in identifying the paucity of information in large areas in Africa, Europe and the Middle East. An increasing willingness to share data between research groups and national malaria control programmes has been instrumental in this initiative and is critical to its sustained future.

Background This is the second in a series of three articles documenting the geographical distribution of 41 dominant vector species (DVS) of human malaria. The first paper addressed the DVS of the Americas and the third will consider those of the Asian Pacific Region. Here, the DVS of Africa, Europe and the Middle East are discussed. The continent of Africa experiences the bulk of the global malaria burden due in part to the presence of the An. gambiae complex. Anopheles gambiae is one of four DVS within the An. gambiae complex, the others being An. arabiensis and the coastal An. merus and An. melas . There are a further three, highly anthropophilic DVS in Africa, An. funestus , An. moucheti and An. nili . Conversely, across Europe and the Middle East, malaria transmission is low and frequently absent, despite the presence of six DVS. To help control malaria in Africa and the Middle East, or to identify the risk of its re-emergence in Europe, the contemporary distribution and bionomics of the relevant DVS are needed. Results A contemporary database of occurrence data, compiled from the formal literature and other relevant resources, resulted in the collation of information for seven DVS from 44 countries in Africa containing 4234 geo-referenced, independent sites. In Europe and the Middle East, six DVS were identified from 2784 geo-referenced sites across 49 countries. These occurrence data were combined with expert opinion ranges and a suite of environmental and climatic variables of relevance to anopheline ecology to produce predictive distribution maps using the Boosted Regression Tree (BRT) method. Conclusions The predicted geographic extent for the following DVS (or species/suspected species complex*) is provided for Africa: Anopheles ( Cellia ) arabiensis , An. ( Cel. ) funestus* , An. ( Cel. ) gambiae , An. ( Cel. ) melas , An. ( Cel. ) merus , An. ( Cel. ) moucheti and An. ( Cel. ) nili* , and in the European and Middle Eastern Region: An. ( Anopheles ) atroparvus , An. ( Ano. ) labranchiae , An. ( Ano. ) messeae , An. ( Ano. ) sacharovi , An. ( Cel. ) sergentii and An. ( Cel. ) superpictus* . These maps are presented alongside a bionomics summary for each species relevant to its control.

List of abbreviations AUC: Area Under the operating characteristic Curve; AVHRR: Advanced Very High Resolution Radiometer; BRT: Boosted Regression Trees; COI: (mitochondrial) Cytochrome Oxidase 1; DEM: Digital Elevation Model; DVS: Dominant Vector Species; EO: Expert Opinion; EVI: Enhanced Vegetation Index; GIS: Geographic Information System; IRS: Insecticide Residual Spraying; ITNs: Insecticide Treated Bednets; ITS2: Internal Transcribed Spacer 2; IVCC: Innovative Vector Control Consortium; LST: Land Surface Temperature; MAP: Malaria Atlas Project; MIR: Middle Infrared Radiation; MODIS: MODerate Resolution Imaging Spectroradiometer; NDVI: Normalized Difference Vegetation Index; PCR: Polymerase Chain Reaction; TAG: Technical Advisory Group; TFA: Temporal Fourier Analysis; WRBU: Walter Reed Biosystematics Unit. Competing interests The authors declare that they have no competing interests. Authors' contributions SIH conceived the study and managed its design and implementation. MES wrote the first draft of the manuscript and assembled the occurrence data with assistance from CWK and RMO, CWK also digitised and edited all the expert opinion maps. WHT designed and maintained the databases and implemented the map figures. APP implemented the BRT scripts for predictive mapping. PWG processed the environmental and climatic data grids, with assistance from TVB. All TAG members (MJB, SM, MC, CMM and JH) provided data and advice in updating the EO range maps. Experiments were devised by SIH and MES and implemented by MES. All authors participated in the interpretation of results and in the writing and editing of the manuscript. MJB, MC, SM, HCJG, CMM and REH advised on bionomics and nomenclature issues, and provided additional comments and input to the manuscript. Supplementary Material

Acknowledgements We wish to thank Rosalind Howes, Edward Haynes, Philip Mbithi, Owen Yang, Carolynn Tago, and Elisabeth Thiveyrat for primary data abstraction. We also thank the Technical Advisory Group for extended support over the duration of the project (in addition to co-authors Michael Bangs, Sylvie Manguin, Maureen Coetzee, Ralph Harbach, Janet Hemingway and Charles M. Mbogo, these include, Theeraphap Chareonviriyaphap and Yasmin Rubio-Palis). MES is funded by a project grant from the Wellcome Trust (#083534) to SIH. SIH is funded by a Senior Research Fellowship from the Wellcome Trust (#079091) which also supports CWK and PWG. APP and WHT are funded by a Wellcome Trust Principal Research Fellowship (#079080) to Professor Robert Snow. This work forms part of the output of the Malaria Atlas Project (MAP, http://www.map.ox.ac.uk ), principally funded by the Wellcome Trust, U.K.

CC BY

no

2022-01-12 15:21:45

Parasit Vectors. 2010 Dec 3; 3:117

oa_package/47/e1/PMC3016360.tar.gz

PMC3016361

21156061

Background Ticks are blood-sucking ectoparasites that may do harm to their host by severe blood loss, but the main danger of ticks remain their capability to transmit a wide variety of pathogens comprising viruses, bacteria and protozoa [ 1 ], causing diseases in humans and animals [ 2 ]. The miscellaneous vectorial capacity of ixodid ticks is due to their long-term co-evolution with the pathogens that they transmit, extended lifespan (up to years), the long-lasting blood feeding of all parasitic stages on different hosts and other aspects of tick biology [ 3 ]. Moreover, the transmission of pathogens is facilitated by the modulation of host haemostatic, inflammatory and immune responses, mediated by the inexhaustible pharmacology of the molecules present in tick saliva [ 4 ]. The castor bean tick, Ixodes ricinus is the most important arthropod disease vector in Europe, notorious for transmitting the tick-borne encephalitis virus and the Lyme disease spirochetes of the genus Borrelia [ 5 ], supporting the importance of genome sequencing of the closely related American deer tick, Ixodes scapularis [ 6 ]. During blood feeding, Borrelia sp . spirochetes multiply in the tick gut lumen and change the expression pattern of their outer surface proteins, allowing them to exit the gut and migrate to salivary glands, which they enter to ultimately infect the mammalian host [ 7 ]. These data suggest that transmission of Borrelia sp . and other tick borne pathogens can be potentially controlled via an efficient impairment of blood feeding and digestion. In our previous work, we have demonstrated in semi-engorged I. ricinus females that intestinal digestion is based on an evolutionarily conserved multienzyme complex of cysteine and aspartic peptidases [ 8 ], orthologous to those characterized in other parasites, including platyhelminthes [ 9 , 10 ] and nematodes [ 11 ]. We have also mapped the hemoglobinolytic cascade revealing the successive involvement of individual enzymes [ 12 ]. Hemoglobin digestion in the I. ricinus gut is initiated by generation of large fragments by aspartic peptidase of the cathepsin D-type (IrCD), supported by the cysteine class peptidase of the papain-type cathepsin L (IrCL) and legumain-type asparaginyl endopeptidase (IrAE). The cleavage of large hemoglobin fragments is further mediated by another papain-type cysteine endopeptidase cathepsin B (IrCB) and completed by the amino- and carboxydipeptidase activity of cathepsin C (IrCC) and IrCB, respectively [ 12 ]. Since the genetic screening [ 8 ] as well as the proteomic characterization of the I. ricinus digestive pathway [ 12 ] was performed for a single-time point towards the end of the slow feeding period (the sixth day post-attachment), several principle questions concerning the roles of the individual peptidases during tick feeding remain to be answered: (i) what is the dynamic course of overall hemoglobinolysis and the activity of component peptidases in relation to the feeding phases? (ii) is the hemoglobinolytic pathway (network) preserved during the entire blood feeding? (iii) is the digestion regulated at the transcriptional, translational or pro-enzyme activation level? (iv) is the site of action of hemoglobinolytic enzymes spatially restricted to the gut digestive cells throughout the whole feeding period? In order to shed more light into the process of blood uptake and digestion during tick feeding on the host, we performed complex dynamic analysis of the gene expression, activity and molar concentrations of the major digestive peptidases. In addition, indirect immunofluorescent localization of the major peptidase of the hemoglobinolytic pathway, IrCB, made it possible to follow the development of the digestive machinery in the gut cells from unfed to fully engorged females. This work provides a first, comprehensive insight into the dynamics of the digestive apparatus in an important tick model indicating the potential of blood feeding impairment as an efficient tick control point.

Methods Ticks, feeding and tissue dissection Ixodes ricinus ticks were collected by flagging in localities around Ceske Budejovice, Czech Republic. One hundred I. ricinus females were allowed to feed naturally on four laboratory guinea pigs (25 females each) in the presence of the same number of males. At a specific time of feeding, namely 2, 4, 5 and 6 days post-attachment (p.a.), groups of 15 to 20 females were forcibly removed from each guinea pig in a proportional manner using forceps. The last group was allowed to accomplish feeding and drop off the host (7 to 8 days p.a.). Gut tissue from individual time points, including 25 of unfed females (day 0) were dissected under a binocular microscope taking care not to disrupt the epithelium. Gut tissue was subsequently washed from the excess host blood in phosphate buffered saline (PBS). Cleaned gut tissues were subsequently longitudinally divided into two halves and pooled for either RNA isolation or tissue extraction. A smaller number (3 - 4) of dissected tick gut tissues was processed independently for microscopy observations (see below). Laboratory animals were treated in accordance with the Animal Protection Law of the Czech Republic no. 246/1992 Sb, ethics approval number 137/2008. Isolation of total RNA and analysis of peptidases expression by quantitative real-time polymerase chain reaction For isolation of total RNA, the NucleoSpin ® RNA II kit (Macherey-Nagel) was used. Pooled gut tissue was placed into RA1 buffer, homogenized with a plastic pestle and further processed according to the manufacturer's instructions. Integrity of the resulted total RNA was analysed using agarose gel electrophoresis according to [ 31 ] and stored at -80°C until use. Single stranded cDNA was reversibly transcribed from 0.5 μg of total RNA using the Transcriptor High Fidelity cDNA Synthesis Kit (Roche). Resulting cDNA was diluted 20 times and 2 μl were used as a template for semi-quantitative real-time polymerase chain reaction (qRT-PCR) performed in a final volume of 25 μl using Dual labeled UPL probes (Roche) and specific primers designed online http://www.universalprobelibrary.com (Table 2 ). The reaction was carried out using the Rotor-Gene RG3000 PCR cycler (Corbett Research) and the amplification conditions were 95°C - 10 min followed by 40 cycles of 95°C - 15 sec and 60°C - 60 sec. All reactions were carried out in triplicates. Data were analyzed and quantified with the Rotor-Gene6 analysis software. Relative values were standardized to the gene for elongation factor 1α (ELF1A) [ 32 ] and normalized to the sample with the highest level of expression. Gut tissue extraction The second half of pooled gut tissues were dispersed in 300 μl of ice-cold 0.1 M sodium acetate buffer pH 5.0 in a 1.5 ml Eppendorf tube and homogenized by three times repetition of freezing in liquid nitrogen, thawing and grinding with a plastic pestle. An aliquot of the homogenate was boiled with reducing SDS-PAGE buffer for Western blot analysis [ 33 ]. The remaining homogenate was supplemented with CHAPS to a final concentration of 1%, incubated on ice for 30 minutes under agitation and then cleared by centrifugation (20,000× g, 4°C, 10 min). The supernatant aliquots (referred to as gut tissue extracts, GE) were stored at -80°C until use for activity measurements and active-site titration. Determination of the overall hemoglobinolysis and individual peptidase activities Overall hemoglobinolytic activity in gut tissue extracts was determined using the hemoglobinolytic fluorescamine assay as described previously [ 12 ]. The hydrolytic activity of IrCB, IrCC, IrAE, IrCL and IrCD was measured with the Infinite 200 M (Tecan) microplate fluorimeter using fluorogenic substrates and the shielding inhibitors presented in the Table 1 . The activity measurements were performed at 25°C in 0.1 M sodium citrate-phosphate (SCP) including 2.5 mM DTT (for cysteine peptidases) and 25 mM NaCl (for IrCC) at the respective pH optimum (Table 1 ). Gut extracts were appropriately diluted in the assay buffers to achieve roughly 1,500 RFU/min. Enzyme activity was measured for at least 10 minutes and calculated from the linear portion of the kinetic curve. All measurements were performed in triplicates. The measured activity was normalized per one tick gut using the formula: RFU/min/gut = (measured RFU/min × dilution factor)/(number of tick guts × 0.5). Determination of the peptidase molar concentrations by active-site titration The absolute molarity of individual peptidases in the gut tissue extract was determined by the stoichiometric titration method according to Barrett and Kirschke [ 34 ] and Knight and Barret [ 35 ]. The following inhibitors were applied as active-site titrants: CA-074 for IrCB, Ala-Hph-VS-Ph for cathepsin C, Aza-N-11a for IrAE and pepstatin for IrCD (Table 1 ), as described previously [ 12 ]. An aliquot of the GE was incubated with various amounts of the titrant for 30 min at 35°C in above mentioned SCP buffers at the respective pH optima (Table 1 ). Residual peptidase activities were measured using the assay systems as described in the previous paragraph, and plotted into the titration curves. Expression of recombinant IrCB and preparation of antibody for Western blot analysis and indirect immunofluorescence To recombinantly express IrCB proenzyme (GenBank ABO26563 ), the E. coli bacterial expression system and ChampionTM pET Directional TOPO ® Expression kit (Invitrogen) was used. The N-terminal (His) 6 -tagged fusion of IrCB was prepared using the pET100/D-TOPO ® expression vector and the following primers: forward, 5'-CACCCGTGAGATTCATCC-3'; reverse, 5'-AGCTATTCGTCCTTGGGG-3'. The resulting expression constructs were transformed into TOP10 cells (Invitrogen) and sequenced using the T7 forward and T7 reverse sequencing primers. The correct constructs were transformed into BL21 StarTM (DE3) E. coli . Expression was induced with 1 mM IPTG in LB medium. Isolated inclusion bodies were dissolved in a buffer containing 6 M guanidium hydrochloride and the fusion (His) 6 -tagged IrCB was purified using Ni 2+ -chelating chromatography in the presence of 8 M urea. The purified protein was renatured by dialysis against 0.4 M L-arginine, 0.15 M NaCl, 1 mM mercapthoethanol and gradually decreasing the concentration of urea (from 8 to 0 M) with final dialysis against 25 mM Tris-HCl pH 7.5. Rabbits were immunized with purified soluble recombinant IrCB (recIrCB) using a standard immunization protocol [ 36 ]. The immunoglobulin fraction was enriched by serum precipitation with caprylic acid [ 37 ] and the resulting antibody (about 0.5 mg protein/ml) referred to as Ra×IrCB Ig were used for Western blot analysis following reducing SDS-PAGE [ 33 ]. For the purpose of indirect immunofluorescence, the IrCB-specific antibodies were further purified by affinity chromatography using a column of 1 mg recIrCB coupled to 1.5 ml of CNBr-activated SepharoseTM 4B (GE-Healthcare) according to the protocol provided by the manufacturer. Sixty milliliters of Ra×IrCB Ig diluted in PBS to 0.15 mg/ml were loaded onto the column, washed with 50 ml of PBS and eluted with 0.2 M L-glycine, 0.15 M NaCl, pH 2.2. The fractions within the eluted peak were immediately neutralized with 1 M Tris-base and pooled. The aliquots of the resulting purified antibodies (150 μg protein/ml) referred to as Ra×IrCB ACP (for affinity chromatography purified) were stored at -20°C until use. Indirect immunofluorescent localization of IrCB in gut tissues by microscopy Dissected tissue was fixed overnight in a solution of 4% formaldehyde and 0.1% glutaraldehyde, in 0.1 M sodium phosphate buffer pH 7.3 at 4°C. Samples were subsequently agitated three times for 10 min in washing buffer (4% glucose in 0.1 M sodium phosphate buffer, pH 7.3). Microwave incubations in a water bath were carried out three times for 30 s each in a microwave oven set at the lowest power (80W). Samples were subsequently dehydrated under the microwave radiation using ascending ethanol dilutions, then infiltrated in LR White resin (London Resin Company, Ltd.) in ratios of resin : 95% ethanol 2:1, 1:1, 1:2 (1.5 h each, 4°C) and finally kept in pure LR White resin overnight at 4°C. Samples were then transferred to gelatin capsules (Polysciences, Inc.) and filled with resin, which was allowed to polymerize for 24 h at 50°C. Semi-thin sections (0.5 μm) were prepared, transferred onto glass slides and blocked with 1% BSA and 10% low fat dry milk in PBS-Tween (0.3% Tween-20) for 10 min. Incubation with the primary Ra×IrCBACP antibodies (30 μg/ml) in PBS was performed in a humid chamber for 3.5 h at room temperature. For negative control experiments, the primary antibody incubation was omitted. Sections were washed with PBS-Tween (four times 5 min) and then incubated in a Alexa Fluor ® 488 dye-conjugated goat anti-rabbit antibody (Invitrogen/Molecular Probes) diluted to 1:500 in PBS-Tween for 1 h at room temperature. After washing with PBS-Tween, the slides were counterstained with DAPI (4',6'-diamidino-2-phenylindole; 2.5 μg/ml; Sigma). Finally, sections were mounted in 2.5% DABCO (1,4-diazabicyclo[2.2.2]octane; Sigma) dissolved in glycerol and examined using the Olympus FW1000 confocal microscope and consequently processed with the Fluoview (FV10-ASW, Version 1.7) software.

Results Morphological changes in the Ixodes ricinus gut during feeding correlate with overall hemoglobinolytic capacity Feeding of adult I. ricinus females on guinea pigs takes 7 to 8 days under in-house laboratory conditions. The feeding phases as known for the ixodid ticks [ 3 , 13 ] are schematically depicted and aligned with the corresponding observed gut morphological changes and overall hemoglobinolytic activities (Figure 1 ). During attachment, narrow gut lumen of unfed ticks is surrounded by the midgut epithelium formed by undifferentiated reserve cells (also referred to as replacement cells [ 13 ] or stem cells [ 14 ]) and digestive cells preserved from the preceding nymphal stage. On the second day post attachment (p.a.), the midgut lumen slowly expands with the uptake of the first portion of host blood and the reserve cells start to grow and differentiate into the initial digestive cells, also named prodigest cells [ 15 ]. The hemoglobinolytic activity in gut tissue extracts (GEs) of unfed as well as 2-day-fed ticks is barely detectable. On the fourth day of feeding, the digestive cells further enlarge and display signs of hemoglobin uptake and digestion. The first residual bodies containing waste aggregated heme products, formed in organelles called hemosomes [ 16 ], as well as large endosomes and lipid vacuoles, start to appear at this time point (Figure 1 ). During this phase of slow feeding, it is already possible to observe the detachment of the first digestive cells from the midgut epithelium and the initial increase in overall hemoglobinolytic activity. The most dramatic changes in the midgut morphology is evident towards the end of the slow feeding period, about the six days p.a.. At this stage, the digestive cells are fully extended, filled with residual bodies, large endosomes and lipid inclusions. The detached digestive cells are liberated into the lumen without any sign of lysis and most likely removed by defecation. The hemoglobinolytic activity in the gut tissue extract in semi-engorged ticks (six days p.a.) reaches about 65% of the maximum. The rapid-engorgement phase occurs during the last 24-48 hours prior to detachment. It has been referred to as the 'big-sip' [ 3 ] because mated females imbibe about two-thirds of the entire blood-meal within this phase. The appearance of the midgut in fully fed females differs from that observed during the slow feeding period. The midgut epithelium is overlaid with flattened sessile digestive cells that remain associated with the epithelium; reserve or initial digestive cells are not detected anymore. Noticeably, the overall hemoglobinolytic activity reaches its maximum in gut extracts of fully engorged ticks that just dropped off the host. The amount of component digestive peptidases markedly increases towards the end of slow feeding Two approaches were employed to quantify the component digestive peptidases in GEs: (i) the molar content of peptidases was determined by active-site titration with selective irreversible inhibitors; (ii) the enzymatic activity of peptidases was measured in a kinetic assay with specific fluorogenic substrates. The active-site titration was performed for all enzymes except for IrCL, due to the lack of an appropriate irreversible inhibitor as a titrant (Table 1 ). The concentrations of digestive peptidases proportionally increase in time and reach their maximum levels in fully fed ticks (Figure 2 ). IrCB is the dominant enzyme throughout the whole feeding period and it is the only enzyme which can be detected in the GE of unfed ticks. The second most abundant enzyme is the dipeptidyl peptidase IrCC. The amounts of IrCB and IrCC exceed the concentrations of the other peptidases in the network by more than one order of magnitude. The activity measurements of individual enzymes in the GEs were performed at their pH optima with specific fluorogenic substrates, preventing the confounding hydrolysis activities of other peptidases with the aid of shielding inhibitors (Table 1 ). In accordance with the active-site titration results, IrCB was the only enzyme giving measurable activities in the gut extracts from unfed ticks. As shown in the Figure 3 the activities of IrCD, IrAE, IrCB and IrCC, respectively, begin to increase after the second day of feeding, followed by an exponential increase between days 4 and 6 and finally reaching their maxima in fully fed ticks. The only enzymatic activity that proved to be an exception to this trend was that of IrCL, which peaks by the fifth day of feeding and then steeply declines in fully fed ticks (Figure 3 ). Digestive peptidases are regulated at the transcriptional level Determination of molar concentrations of individual digestive peptidases as well as their activity profiles during blood feeding prompted further studies to assess whether de-novo protein synthesis and/or activation of existing pro-enzymes is the mode of action underlying their increase. The determination of digestive peptidase mRNA levels by semi-quantitative real-time PCR (qRT-PCR) analysis demonstrated that the dynamic profiles of mRNA expression clearly precede the activities of their respective peptidases (Figure 3 ; bar graphs) implying that the major components of the digestive system are transcriptionally regulated. The mRNA levels of IrCB and IrCC reach their maxima already five days p.a., IrCD and IrAE mRNA levels peak six days p.a. and IrCL mRNA displays two expression maxima the fourth and the sixth day p.a.. For all enzymes, the mRNA expression drops dramatically in fully fed ticks, indicating that the further continual digestive phase in engorged ticks following their detachment from the host relies mainly on the enzymes synthesized during the slow feeding period. The digestive apparatus is localized intracellularly during the whole feeding period To demonstrate whether the hemoglobinolytic machinery remains intracellular during the whole blood feeding, immunolocalization of IrCB, the most abundant peptidase of the network, was performed. The separation of gut tissue extracts by SDS-PAGE followed by immunoblot analysis using the antibody prepared against the recombinant IrCB labeled a faint band of about 38 kDa and an intense double band of 33 and 31 kDa (Figure 4A ). The former band likely corresponds to the IrCB pro-enzyme having a theoretical mass (without glycosylation) of 35,725 Da [ 8 ], whereas the double band may correspond to the intermediate and fully activated form of IrCB, as has been previously described for schistosmal cathepsin B [ 17 ] or to a possible presence of another closely related isoform of IrCB that cross-reacts with the antibody [ 12 ]. Activated IrCB protein starts to appear at the second day of feeding and becomes quite intense from the fourth day on to the fully fed stage (Figure 4A ). This result is consistent with the quantitative results obtained by active-site titration and activity measurements. In order to minimize the background fluorescence of the gut tissues and to obtain a clear immunostaining signal, affinity purified anti-IrCB antibodies from the rabbit serum were used. Immunolocalization of IrCB in tick guts corresponding to specified feeding periods clearly demonstrates that IrCB gradually accumulates and remains restricted within the digestive cells during the entire duration of the tick feeding on the host (Figure 4B ). Traces of IrCB were observed already in the gut sections of unfed ticks, possibly as a remnant of the enzyme from the preceding nymphal stage. The increase of IrCB in the digestive cells becomes clearly evident by the fourth day of feeding. Towards the end of the slow feeding period (day six p.a.), most of the digestive cells are fully loaded with the enzyme and detached from the epithelium. No evidence of disintegration of the digestive cells and release of the free enzyme into the lumen was observed. This finding was further corroborated by the control measurement of activity of digestive enzymes in the gut lumen, which did not exceed 10% of the activities determined in GE (data not shown). In fully fed ticks, the last generation of digestive cells containing the IrCB remains associated within the midgut epithelium. Although the intensity of the IrCB signal within the sessile digestive cells in fully fed ticks seems to be lower compared to the day six p.a. (Figure 4B , panels FF vs. 6d), the remarkable enlargement of the gut tissue epithelium apparently accounts for the apparent maximal hemoglobinolytic capacity of the whole gut tissue extract in fully fed ticks.

Discussion During previous studies we have described the major constituents of the digestive machinery of the hard tick I. ricinus and deciphered how they are organized into a hemoglobinolytic pathway [ 8 , 12 ]. Herein, we have focused on the dynamic profiling of the components of the tick digestive apparatus during the course of tick feeding. This information is particularly relevant since the feeding phase is decisive for pathogen transmission. We have provided a comprehensive insight into the gene transcription, activity and molar concentration of five digestive enzymes during blood feeding, as well as the localization of the IrCB, as the most abundant peptidase of the pathway. Placing these molecular data into the context of previous ultrastructural and histochemical studies on tick gut development performed in 1970's and 80's [ 3 , 13 , 18 ] improves significantly our understanding of the entire process of intracellular blood digestion. Morphological changes of the midgut epithelium observed during blood feeding of I. ricinus by light microscopy (Figure 1 ) basically correspond to the numerous previous reports published on gut morphology in other ixodid tick species [ 14 , 15 , 19 , 20 ]. In these studies, a variety of cells within the gut epithelium have been described and named according to their structure and/or putative functions, resulting in a rather disperse nomenclature in different tick species. Based on our observations (Figure 1 ), we tend to support the hypothesis of some authors [ 13 , 15 ] that various cell types represent digestive cells at different stages of development, reflecting their growth, differentiation and gradually changing function in endocytosis, lysosomal digestion, secretion and handling of waste products via hemosome formation [ 16 ]. This study also supports the recognition of distinct digestive phases previously based on the structural changes of the tick gut [ 18 ]. The initial continual digestive phase takes place during the slow feeding period starting one to two days p.a. to the host and is characterized by the gradual maturation and detachment of digestive cells from the gut epithelium [ 13 ]. During the rapid engorgement phase, when the mated female ingests about two thirds of the total blood meal, no more reserve cells or initial digestive cells are observed and the mature digestive cells remain associated with the epithelium. We have previously characterized the hemoglobinolytic pathway in semi-engorged I. ricinus females that were fed for six days, comprising of a cascade of enzymatic activities of aspartic peptidase IrCD, cysteine peptidase IrCL, asparaginyl endopeptidase IrAE, the major endo/exopeptidase IrCB and dipeptidyl peptidase IrCC [ 12 ]. The determination of the molar concentrations as well as activity measurements of the component peptidases presented in this work support that the hemoglobinolytic pathway described for the sixth day p.a. is preserved throughout the entire blood feeding period. The quantitative determination of the overall hemoglobinolytic activity indicates that digestion is almost negligible for the first two days after tick attachment on the host and dramatically increases between days four and six, progressing towards the end of the slow feeding period (Figure 1 ). The maximum level of hemoglobinolytic capacity (activity normalized to the entire tick gut) was recorded in the guts of fully fed ticks that just dropped off the host. The trend of an increase in total hemoglobinolysis matches with the concomitant activity profiles of the majority digestive enzymes, namely IrCB, IrCC, IrAE and IrCD. The significant up-regulation of their respective mRNA levels could be monitored already at the second day of feeding, peaking towards the end of the slow feeding period between days four and six and then steeply declining during the rapid engorgement phase. This result is consistent with an earlier histological observation demonstrating that RNA is mainly accumulated in the stem and prodigest cells (herein referred to as reserve and initial digestive cells, respectively), that disappear during the rapid engorgement stage [ 15 ]. The contrast of the significant down-regulation of mRNA expression to the peaking hemoglobinolysis and enzyme activities (except for IrCL) during the rapid engorgement suggests that the following second continuous digestive phase, i.e. blood digestion in fully fed ticks detached from the host, relies mainly on the enzymes synthesized and accumulated during the slow feeding period. Our results contradict the prevailing opinion that the rapid engorgement stage is associated with reduced hemoglobinolysis, which has been accordingly designated the reduced-digestion phase in the literature [ 18 , 20 , 21 ]. To our knowledge, the general recognition of the term 'reduced-digestion phase' originates from a 1973 report on the total proteolytic activity in gut homogenates of the hard tick Hyalomma excavatum [ 22 ]. The authors reported a low activity in both the initial stage of feeding as well as in fully engorged females. However, it is important to take into consideration that they related the proteolytic activity to the total protein concentration in the gut extracts without any compensation for the proteins of the host origin. To our experience, it is impossible to free mechanically the dissected gut tissues of significant host blood contamination, since substantial portion of the host proteins is already present inside the digestive vacuoles. Moreover, a gentle washing has to be applied to prevent an undesired detachment of digestive cells from the epithelium. Therefore, the host proteins inevitably remain the most abundant components in the gut homogenates from the later feeding stages. Despite its simplicity, normalization of hemoglobinolytic and enzyme activities to one tick gut is more reliable than relation to the total protein or host hemoglobin concentrations (based on absorbance at 415 nm) and reflects better the real total proteolytic potential of the gut. In conclusion, this approach leads to a novel interpretation of the blood digestion capacity during the 'big-sip' which disproves the earlier opinions. The only enzyme activity that drops significantly during the fast engorgement phase is that of IrCL. This finding was also supported by the Western blot analysis showing a substantial reduction of IrCL amount in fully engorged ticks (unpublished results). The reason for the decrease of IrCL activity in fully fed ticks remains still obscure. One possible explanation to this effect is a simultaneous increase of cathepsin L-specific inhibitors, such as cystatins found to be expressed in the midgut of the soft tick Ornithodoros moubata [ 23 ] or the hard tick Haemaphysalis longicornis [ 24 ]. Our results strongly imply that de novo protein synthesis of the digestive enzymes rather than activation of existing pro-enzymes underlies the overall increase of hemoglobinolytic activity in the gut of I. ricinus . The main evidence for this interpretation was provided by Western blot analysis of IrCB from gut homogenates dissected during the course of blood feeding (Figure 4A ). At all time points, IrCB was mainly detected as an activated mature enzyme and the proportion of the IrCB pro-enzyme and mature active enzyme did not seem to be altered significantly in time (Figure 4A ). The affinity purified antibodies made it possible to perform a clear immunolocalization of IrCB in I. ricinus midgut sections prepared at different time points of feeding (Figure 4B ). Similar results were obtained for IrCL and IrCD (data not shown), indicating that the entire digestive apparatus remains localized intracellularly in the digestive cells during the feeding. This is in accordance with our previous single time point localization of IrAE in the midgut of semi-engorged I. ricinus [ 25 ] and digestive enzymes localized in the digestive cells of other tick species [ 24 , 26 , 27 ]. On the other hand, it remains to be rigorously investigated whether or not intracellular digestion is preserved also in the guts of fully engorged ticks during the second digestive phase, following tick detachment and preceding oviposition. Based on the results given herein we can conclude that all the major constituents of the tick digestive pathway are regulated at the transcriptional level. Certainly, the mechanisms triggering and regulating the expression of digestive enzymes remain to be explored. The negligible gene expression and enzyme activity at the early stage of tick feeding make the digestive system extremely vulnerable target for efficient impairment, for example via antibodies presents in vaccinated animals. Therefore, digestive enzymes as well as yet unknown regulating factors are promising candidates as concealed antigens for the development of efficient 'anti-tick' vaccines capable to control ticks and associated transmission of tick-borne pathogens [ 28 , 29 ].

Conclusions Ticks are obligate blood-feeders capable of ingesting a blood-meal exceeding more than one-hundred times their own body weight. Efficient interference of the blood feeding and digestion processes may be a key to tick control and transmission of tick-borne pathogens. Despite its importance, our knowledge of the molecular mechanisms involved in blood digestion has been rather limited and dispersed across various ixodid tick species. In order to obtain a comprehensive view of blood digestion in a single tick species, Ixodes ricinus , we combined genetic and biochemical analyses to unravel the complex suite of gut-associated cysteine and aspartic peptidases [ 8 ]. We have demonstrated that the key components of the tick digestive apparatus, namely the cathepsins D, B, L, C and the asparaginyl endopeptidase, are remarkably similar to those found in phylogenetically distant parasites such as nematodes [ 11 ], platyhelminthes [ 9 ] and even protozoa such as malaria Plasmodium [ 30 ]. Moreover, we have deconvoluted how I. ricinus gut-specific peptidases function to mediate the gradual hydrolysis of hemoglobin [ 12 ]. The present study accomplishes our focus on this multi-enzyme network, as it provides a complex dynamic view of gene expression and enzyme concentration and activities during the course of tick feeding on the host. We have demonstrated that expression and protein synthesis of the component peptidases increase most dramatically towards the end of the slow-feeding period. Overall hemoglobinolysis as well as activities of most peptidases (except cathepsin L) reach their maxima during the rapid engorgement phase, which contradicts the established opinion that blood digestion is reduced during this phase. On the other hand, hemoglobinolysis as well as the activities of individual peptidases are negligible during the first two days after attachment to the host. This feature makes the digestive peptidases, and yet unknown molecules involved in their regulation, promising targets for tick control possibly leading to the rational development of efficient 'anti-tick' vaccines. The applied aspect of this work will be the subject of future reports.

Background Ticks are vectors of a wide variety of pathogens causing severe diseases in humans and domestic animals. Intestinal digestion of the host blood is an essential process of tick physiology and also a limiting factor for pathogen transmission since the tick gut represents the primary site for pathogen infection and proliferation. Using the model tick Ixodes ricinus , the European Lyme disease vector, we have previously demonstrated by genetic and biochemical analyses that host blood is degraded in the tick gut by a network of acidic peptidases of the aspartic and cysteine classes. Results This study reveals the digestive machinery of the I. ricinus during the course of blood-feeding on the host. The dynamic profiling of concentrations, activities and mRNA expressions of the major digestive enzymes demonstrates that the de novo synthesis of peptidases triggers the dramatic increase of the hemoglobinolytic activity along the feeding period. Overall hemoglobinolysis, as well as the activity of digestive peptidases are negligible at the early stage of feeding, but increase dramatically towards the end of the slow feeding period, reaching maxima in fully fed ticks. This finding contradicts the established opinion that blood digestion is reduced at the end of engorgement. Furthermore, we show that the digestive proteolysis is localized intracellularly throughout the whole duration of feeding. Conclusions Results suggest that the egressing proteolytic system in the early stage of feeding and digestion is a potential target for efficient impairment, most likely by blocking its components via antibodies present in the host blood. Therefore, digestive enzymes are promising candidates for development of novel 'anti-tick' vaccines capable of tick control and even transmission of tick-borne pathogens.

Abbreviations GE: gut tissue extract; IrCB: Ixodes ricinus cathepsin B; IrCL: I. ricinus cathepsin L; IrCC: I. ricinus cathepsin C; IrCD: I. ricinus cathepsin D; IrAE: I. ricinus asparaginyl endopeptidase (legumain); p.a.: post-attachment; qRT-PCR: semi-quantitative real-time polymerase chain reaction; RFU: relative fluorescence units. Competing interests The authors declare that they have no competing interests. Authors' contributions ZF, PK, MM and DS designed the study and PK and MM were responsible for its coordination. ZF and DS expressed, purified and refolded the recombinant IrCB and prepared the antibodies. ZF performed and analyzed mRNA expression profiles by qRT-PCR. HF performed tissue dissections, microscope observations, and immunolocalization of IrCB. JK and PK measured and evaluated the enzyme activities and performed the Western blot analysis. MH determined the enzyme concentration by active-site titration and analyzed the data. PK and ZF wrote the paper with the critical input of MM, DS and MH. All authors have read and approved the final version of this manuscript.

Acknowledgements This work was supported by grants from the Grant Agency ASCR Nos. IAA 600960910 to P.K. and M.M. and KJB 600960911 to D.S., Grant Agency CR No. 206/09/H026 to H.F. and by Research Center LC06009 (MSMT CR). Research at the Institute of Parasitology BC ASCR, Institute of Organic Chemistry and Biochemistry ASCR and Faculty of Sciences USB is covered by Research plans from MSMT CR, Nos. Z602205728, Z40550506 and MSM6007665801, respectively. We thank Jan Kopecky for tick photographs and Alena Kopackova for graphical editing. We thank J. C. Powers (Georgia Institute of Technology, Atlanta) for providing selected peptidase inhibitors.

CC BY

no

2022-01-12 15:21:45

Parasit Vectors. 2010 Dec 14; 3:119

oa_package/72/20/PMC3016361.tar.gz

PMC3016362

21143943

Background Advancing age increases the risk of dementia [ 1 ], of which Alzheimer's disease (AD) is the most common cause. Available projections indicate that the number of people with dementia worldwide will increase from 35.6 million in 2010 to 115.4 million people by 2050 [ 2 ], with dementia expected to become the major social and economic health challenge of the 21 st century. Alzheimer's disease is characterized by a progressive deterioration in memory and other higher cortical functions that ultimately leads to a loss of independent living skills [ 3 ]. Despite the availability of palliative medication with cholinesterase inhibitors and memantine [ 4 ], there is currently no cure for AD. There is growing evidence that some environmental factors such as physical and cognitive activity [ 5 , 6 ], decrease the risk of AD [ 7 ] and improve the behaviour of patients [ 8 ], but it is unclear if such lifestyle interventions can also delay the progression of cognitive decline once the diagnosis of AD has been established [ 9 ]. Prospective cohort studies indicate that PA is associated with reduced incidence of dementia [ 10 ]. Middleton et al. [ 11 ] reported that women who were physically active across the life course had a lower prevalence of cognitive impairment in later life. The association between PA and cognitive function was evident even when exercise is limited to later life [ 12 ]. However, as Leone et al. [ 13 ] noted, these findings are limited by methodological issues such as survivorship bias and confounding, the latter arising because the exposure to PA is not random in observational studies. A recent randomised trial (RCT) showed that six months of PA decreased the rate of cognitive decline in older people with subjective memory complaints or Mild Cognitive Impairment (MCI) [ 14 ], a group that is at increased risk of developing AD [ 15 ]. Subsequently, Baker and colleagues [ 16 ] found that six months of high-intensity aerobic activity (75-85% of heart rate reserve) improved executive control processes in sedentary women with MCI. Possible mechanisms mediating the cognitive enhancing effect of PA include alterations of cerebral vascular functioning and brain perfusion, environment enrichment and stimulation of synaptogenesis [ 14 ]. A Cochrane review of PA trials for individuals with MCI is currently under way [ 17 ]. Few RCTs of PA in AD have been published to date. Most evidence in this area comes from studies conducted with nursing home residents or participants with moderate to severe AD, and the primary outcomes of interest tend to be functional status and/or mood [ 8 , 18 - 20 ]. The intervention provided in the largest trial, with 134 participants from five nursing homes, was a group-based multi-component exercise program [ 18 ]. After 12 months, participants in the exercise group experienced a slower decline in their performance of activities of daily living than those in the control group, but there was no change in behavioural or psychological symptoms (BPSD). In a pilot RCT, Steinberg et al. [ 19 ] examined the effect of a home-based, carer-supervised program of walking, strength training and balance exercises on cognition, secondary to measures of functional performance, in 27 people with AD. Although the small sample size likely precluded finding any significant differences between groups in cognitive functioning, the investigators found that the exercise program was safe, could be supervised by caregivers and had good compliance. The objectives of the present study are to conduct a methodologically rigorous RCT investigating the potential benefits of PA on cognition, well-being, function and BPSD in community-dwelling participants diagnosed with mild to moderate AD. We will also examine whether the intervention eases stress and burden of care, an area that is often neglected in such RCTs [ 21 ]. This paper describes the design of the Fitness for the Ageing Brain Study II (FABS II).

Methods/Design Study Design FABS II is a single blind RCT (Figure 1 ). The CONSORT statement has been used as a framework for development of the methodology for this project. Participants 230 community-dwelling older adults diagnosed with AD will be recruited across three sites in Australia (Melbourne, Perth and Brisbane). Participants will be included in the study if they satisfy the following criteria: (i) diagnosed with probable or possible AD according to NINCDS-ADRDA criteria [ 3 ], (ii) score of 10 or greater on the Standardised Mini Mental State Examination (SMMSE) [ 22 ], (iii) community dwelling, (iv) no clinically significant depression, (v) understands written and spoken English, (vi) contact with a friend or family member (carer) for at least 10 hours per week, who is also willing to participate in the trial, and (vii) no other major neurological history or medical condition that contraindicates PA. Participants will be excluded if they have limited mobility (e.g. unable to walk alone or require a walking aid for balance), show evidence of clinically significant aphasia or pervasive depression, current history of alcohol dependence, have an unstable or life-threatening medical condition, or are participating in another RCT. Diagnosis of AD will be confirmed through medical records obtained from the participant's treating general practitioner and specialist. These records will be reviewed by members of the research team who are either experienced Geriatricians or Old Age Psychiatrists to confirm the participant's suitability for inclusion. This study is funded by the National Health and Medical Research Council of Australia. Ethics approval has been obtained from the Human Research Ethics Committee(s) of each participating site and the project complies with the Declaration of Helsinki. Recruitment and Screening Potential participants and their carers will be identified and recruited via local physicians, Memory Clinics and specialists. Additionally, flyers advertising the study and media activities will invite suitable older adults living in the community to consider participation. The research team will contact the participant and next of kin by telephone to check the suitability of the participant according to the study inclusion and exclusion criteria using a strict screening protocol. The 15-item Geriatric Depression Scale (GDS - 15) [ 23 ] is included as part of the phone screening to establish the presence of clinically relevant symptoms of depression, excluding potential participants with a score of 6 and higher. The name and contact details for the participant's GP and specialist will be obtained, and the participant will be asked to sign and return a release of medical information form to allow the study team to access their medical records. The participant's GP will also be asked to consent to their patient's involvement in the study. This is another way of determining the suitability of FABS II for the potential participant and ensuring his/her safety during the course of the trial. Assessments Participants will be assessed at baseline and after six months and twelve months in the study (see Table 1 for an overview). There will be two components to the assessments comprising cognitive measures and physical function measures/physical activity questionnaires. Unless stated otherwise, all measures will be administered at each time point. The research assistant (RA) administering the cognitive assessment (Cognitive RA) will be trained by either a Geriatrician or an Old Age Psychiatrist at each site. To ensure that the Cognitive RA remains blind to study group allocation, the physical assessments will be conducted by a separate PA RA. The PA RA will collect demographic and health information via participant interview. The PA RA at each site will be trained by the same chief investigator with a background in exercise physiology to ensure standardization of procedures across all three sites. Measures administered by the Cognitive RA The total score on the Alzheimer's disease Assessment Scale - Cognitive section (ADAS-cog) [ 24 ] will be the primary outcome measure of the study. The ADAS-cog consists of an 11-item battery of short neuropsychological tests and is widely used to monitor the progression of cognitive deficits in clinical trials with patients with AD. Here we will use a version incorporating a delayed verbal recall task. Higher scores indicate increased severity of cognitive impairment (maximum score is 85). The following nine assessments comprise the secondary outcome measures. The Cambridge Contextual Reading Test (CCRT) [ 25 ] includes the words of the National Adult Reading Test (NART), embedding them within appropriate semantic and syntactic contexts to provide a reliable measure of pre-morbid intelligence. The CCRT is only administered at baseline. The SMMSE [ 22 ] is a modified version of the traditional MMSE [ 26 ], which benefits from greater objectivity through more specific scoring examples as well as alternatives for repeated testing of the registration and delayed recall items. A score of less than 10 on the SMMSE will be used to exclude participants, as these individuals would most likely be too cognitively impaired to follow the intervention instructions. The Clinical Dementia Rating Scale (CDR) [ 27 ] is a global rating scale for dementia incorporating information from the primary carer across the domains of memory, orientation, judgment and problem-solving, community affairs, home and hobbies, and personal care. The subscores on each domain are ranked according to severity and an algorithm is used to compute the total score. The Quality of Life in AD (QoL-AD) [ 28 ] is a 13-item questionnaire which requires the participant and carer to independently rate the participant's quality of life across a number of domains on a scale ranging from poor (1 point) to excellent (4 points). The Neuropsychiatric Inventory (NPI) [ 29 ] is a well-established assessment of 12 different BPSD common in AD including delusions, hallucinations, agitation, depression, anxiety, euphoria, apathy, disinhibition, irritability, aberrant motor behaviour, night-time behaviour, appetite and eating disorders. The carer rates each symptom according to the frequency and severity as well as the degree of distress the symptom causes them. The Instrumental Activities of Daily Living scale (IADL) [ 30 ] is widely used to assess independent living skills and is useful in identifying improvement or deterioration over time. The carer rates the AD participant across eight functional abilities, including use of the telephone, shopping, food preparation, housekeeping, laundry, mode of transport, taking medications and handling finances. The Index of Independence in Activities of Daily Living scale (ADL) [ 31 ] is a six-item questionnaire where the carer is required to rate personal and domestic ADLs such as bathing, dressing, toileting, transfers, continence, and feeding. The number of activities in which the participant is deemed 'dependent' is counted. The Short Form-36v2 [ 32 ] is a 36-item questionnaire that assesses the health and well-being of the carer across eight dimensions; physical functioning, social functioning, role limitations due to physical problems, role limitations due to emotional problems, mental health, energy/vitality, pain, and general health perception. Health change over the past year is also assessed. For each dimension of the SF36v2, item scores are coded, summed and transformed on a scale from 0 (worst possible health state measured by questionnaire) to 100 (best possible health state). The Zarit Burden Interview (ZBI) [ 33 ] is a 22-item self-rating scale that examines the degree of burden experienced by the carer associated with functional and behavioural impairments of the person with AD. The items are phrased subjectively and focus on the affective response of the carer. Measures administered by the PA RA Resting blood pressure will be assessed after a five-minute resting period, with five measurements taken at two-minute intervals with a digital blood pressure monitor (model UA-767PC, A&D Co., Ltd., Saitama, Japan). Height, body weight, body composition and girths will be measured in light clothing and bare feet. Height will be measured using a fixed stadiometer. Body weight and composition (fat mass, fat free mass) will be assessed using a bio-impedance analyzer, following the procedures provided by the manufacturer (model TBF-300, Tanita Corporation of America, Inc., Illinois, USA). Waist and hip girths will be measured three times (median score used) using a retractable steel tape measure (Lufkin W606PM Cooper industries SC, USA) [ 34 ]. A step test using a 7.5 cm square wooden block will be used to assess dynamic balance [ 35 ]. The participant places one foot on and off the block as many times as possible in 15 seconds without using hand support. Performance for each leg stepping is assessed separately, and the lowest score will be recorded. The Timed Up and Go (TUG) test will be used to assess mobility [ 36 ]. Participants will be seated in a chair with arms resting on the arm rests and asked to get up and walk three meters as fast and safely as they can, turn around and return to a seated position. Functional lower limb strength will be measured using the sit-to-stand test (five chair stands) [ 37 ], with participants instructed to stand up and sit down as quickly as possible with or without use of their arms. Maximum voluntary hand grip strength (both hands) will be assessed using a Smedleys dynamometer [ 34 ]. The participant will stand against a wall. Holding the dynamometer vertically above the head, the participant will be instructed to squeeze as hard as possible while bringing the arm back down to their side. A 2-minute walk around a 20-metre measured course will be utilized to assess cardiovascular endurance [ 38 ]. Participants will be instructed to walk as fast as possible around the course as many times as they can in two minutes. Participants will be allowed to stop walking during the testing period if they needed to, and any rests will be included in the total 2-minute test period. The maximum distance walked, heart rate and rate of perceived exertion [ 39 ] will be recorded. Physical activity will be assessed using the Community Healthy Activities Model Program for Seniors (CHAMPS) physical activity questionnaire [ 40 ]. Assessment will be made of the frequency, minutes, and calorie expenditure per week of all, low, moderate, high and very high intensity activities. Participants will be provided with a pedometer (Digi-Walker SW-200, Yamax Inc., Tokyo, Japan) and asked to wear it for five weekdays and the weekend, following the baseline visit, to objectively measure their weekly PA. The pedometer will be put on immediately upon wakening and taken off just before retiring at night. It will also be removed whilst bathing or showering and the time off recorded on the pedometer diary. Participants and carers will be shown how to wear and use the pedometer and how to complete the diary. Participants will be instructed to maintain their usual activity pattern during the monitoring period. The stage of PA behavior will be assessed using the Stage of Change Instrument (SCI), a five-item questionnaire asking participants to indicate their current level of activity [ 41 ]. The frequency and minutes per week of regular moderate intensity PA will also be measured using this questionnaire. Exercise self-efficacy will be assessed with the five-item Self-Efficacy Questionnaire (SEQ) asking participants how confident they were to exercise in certain adverse conditions with responses on a 5-point scale ranging from "not at all sure" to "very sure" [ 42 ]. Satisfaction with Life will be measured using a five-item questionnaire with responses on a seven-point scale ranging from "strongly disagree" to "strongly agree" [ 43 ]. DNA Sample Collection APOE genotyping will be performed at baseline testing to adjust for APOE genotype in the final analysis and to investigate whether there is an interaction between APOE genotype and the benefits of PA influencing the primary and secondary outcomes. Participants will be asked to provide a 2 ml saliva sample using a DNA self-collection kit (Oragene DNA OG-250 Disc Format, DNA Genotek Inc, Ontario, Canada). They will be stored at -80°C at the Department of Clinical Pathology and Biochemistry at the Royal Perth Hospital. Samples from Brisbane and Melbourne will be mailed to Perth. All samples will be batched and processed once the final baseline assessment has been completed. Randomization and Blinding Random allocation to intervention or control group will be performed using Stata and the user-written program ralloc [ 44 ]. Allocation numbers will be drawn by an investigator not directly involved in the recruitment or assessment of participants. Each of the three study sites will be provided with sealed opaque envelopes numbered 1-100 and used in that order. After a participant completes his/her baseline assessment, the PA RA will open the next consecutive envelope. Inside the envelope there will either be the letter A (physical activity group) or B (control group). Due to logistic difficulties, participants will not be blind to the intervention and sham PA will not be used. However, the "clinical staff" involved in the collection of endpoints will not be aware of group allocation (single blind). Blinding will be supported by the performance of cognitive and physical assessments at different locations and explicit instructions to participants and clinical research staff not to discuss issues related to PA during the assessments. The PA RA will inform participants of their group allocation by telephone and will instruct them not to share that information with research staff involved in their future cognitive assessments. Intervention The intervention period will be 24 weeks. The intervention will comprise three components: the PA program, the behavioral intervention package, and telephone monitoring. Participants randomized to the intervention will return with their carer ("coach") for a PA workshop within two to four weeks of their baseline visit. During this 60-minute session, the PA RA will give participants their program manual and explain the participants' PA program and the behavioral intervention package. Physical activity program Participants will be advised to perform at least 150 min/week of moderate PA as per the new PA recommendations for older people [ 45 ]. Where walking is a suitable and acceptable option to the participant and coach, this will be a primary PA recommendation. Examples will be given on how this can be achieved. The program will be individualized based on the CHAMPS and SCI data, and the person's interests. Activities prescribed will take into account health problems or other limitations, and participants will be instructed to start slowly and progress gradually taking eight weeks to reach the target amount and intensity. Participants will be encouraged to achieve the 150 min/week by completing 3 × 50 minutes sessions (most PA classes range between 45-60 minutes). In previous studies with middle-aged and older women using this format, we have demonstrated good retention, adherence and improvements in cardiovascular fitness and blood pressure [ 46 ]. Participants who already perform 150 min/week of PA will be encouraged to increase their PA by adding one 50-minute session. To reduce the burden on coaches, some participants might choose to complete activities in community centers. The coach will be invited to perform physical activities alongside participants, if appropriate. In order to maintain standardization of the physical activity programs one CI will monitor the programs prescribed by the RA at each site. Further, to maintain quality control another CI with a PA background will review programs from a random selection on a regular basis from all sites. The PA program will include instructions on how to read the program, complete the activities, record their sessions, and exercise safely. Participants will be given a simple diary and the coach will be asked to verify the activity record the PA in the diary. The diaries, once completed, will be retained by the participant and returned at their six-month visit. After the first 24 weeks, participants will be asked to continue with their PA program for a further 24 weeks however, there will be no further monitoring or contact between the researchers and participants except for three newsletters. At the end of 12 months, intervention participants will be asked to complete a brief questionnaire regarding their program adherence for the previous six months. Adherence during the six month active intervention will be calculated from the number of sessions completed and recorded on the exercise diaries. This will be expressed as a percentage of the number of sessions completed relative to the number of sessions prescribed. Behavioral Intervention Participants and coaches will receive the same educational material about AD as the control group. The intervention program will be based on the Stages of Change model modified for PA [ 41 ], which has been shown to be effective in increasing and maintaining PA in middle-aged and older women [ 46 ]. This model proposes that individuals move through five stages when they adopt a new behavior. From participants' SCI data, a behavioral program will be developed to assist them to move to the next stage of PA. Strategies to increase adherence to the program will be discussed during the workshop and worksheets will be included in a manual. Topics will include tips on how to exercise, benefits and costs of exercising, rewards of exercise, goal setting, time management, etc. During the 24-week trial, the PA RA will mail newsletters at regular intervals to participants, which contain additional motivational information. This will also be reinforced during the five telephone calls made at regular intervals. Participants and coaches will be contacted by telephone for a standardized and structured 15- minute interview to monitor and give feedback on their progress and encourage their continuing adherence. Participants will also be given a report after their follow-up assessments. Giving feedback about progress and increasing participants' perceived benefits of being more physically active has been shown to increase program adherence [ 47 ]. Content and Program Evaluation At the end of the intervention period, the intervention group participants and the coach will be asked to complete a brief interview questionnaire on the content and processes of the program including how easy the program was to understand and follow, and any barriers that were encountered. Control Group Control group (usual care) participants and carers will receive educational material about AD and recommendations for a healthy lifestyle (other than PA). Participants will be contacted by telephone at the same frequency as the intervention group to ensure that the control and intervention group have similar treatment except for the actual intervention. Conversation for this group will be limited to their general health and will not include discussion about physical activity. These activities, together with the benefit of follow-up assessments will strengthen the adherence or ongoing participation of the usual care group. At the conclusion of the study, these participants will be offered the opportunity to attend a session on PA. Statistical Methods Participants who drop out during the trial will be invited to return for the follow-up assessments. We will use imputation by chain equations (ICE) to estimate possible missing outcomes in an intention-to-treat analysis (primary analysis). We will use multilevel regression models to take into account repeated measures and intra-individual variability (mixed models). These models will be adjusted for confounding should the randomization produce unbalanced groups. Sample Size and Power Calculation In a pilot study conducted by this research team, the ADAS-Cog score of participants in the control group (usual care, n = 10) increased 4.47 ± 6.36 points in six months (indicating cognitive decline) whereas the patients in the intervention group (physical activity, n = 12) increased 2.21 ± 4.88 points. Based on this pilot data, recruiting 115 participants in each of the two groups (total = 230) at baseline will result in a power of 80% to detect a difference of 2.2 points between the groups (alpha = 0.05, 2-tailed). This sample size also allows for an estimated 15% drop out rate at follow-up.

Discussion Few RCTs have tested the effect of physical activity in community dwelling people with AD, and none have assessed the beneficial effects of physical activity in terms of family caregiver burden and wellbeing [ 21 ]. By building on the success of the FABS I trial involving people with MCI [ 14 ], the FABS II study attempts to address this gap in the literature. The study will also evaluate the effectiveness of the strategies used to promote short-term and longer-term adherence to physical activity in this population. The findings have the potential to inform practitioners about successful strategies and provide the impetus for translation into community programs. Further, the trial not only focuses on the participant's global cognitive and clinical symptoms, functional level and quality of life but also on the indirect positive effects that may be experienced by the caregiver. From the 12-month follow-up assessment, we also hope to provide evidence regarding the continuation of such benefits. Given the projected rising global incidence of AD over the next several decades, the FABS II trial is timely. If the current study protocol should prove successful, this intervention will represent an affordable and safe method, in combination with standard pharmacological treatments, to alleviate the symptoms of AD. We anticipate that the results of this study will contribute to the overall understanding and management of cognitive and behavioural symptoms of AD as well as potentially improving the quality of life of patients and their carers.

Background Observational studies have documented a potential protective effect of physical exercise in older adults who are at risk for developing Alzheimer's disease. The Fitness for the Ageing Brain II (FABS II) study is a multicentre randomized controlled clinical trial (RCT) aiming to determine whether physical activity reduces the rate of cognitive decline among individuals with Alzheimer's disease. This paper describes the background, objectives of the study, and an overview of the protocol including design, organization and data collection methods. Methods/Design The study will recruit 230 community-dwelling participants diagnosed with Alzheimer's disease. Participants will be randomly allocated to two treatment groups: usual care group or 24-week home-based program consisting of 150 minutes per week of tailored moderate physical activity. The primary outcome measure of the study is cognitive decline as measured by the change from baseline in the total score on the Alzheimer's disease Assessment Scale-Cognitive section. Secondary outcomes of interest include behavioral and psychological symptoms, quality of life, functional level, carer burden and physical function (strength, balance, endurance, physical activity). Primary endpoints will be measured at six and twelve months following the baseline assessment. Discussion This RCT will contribute evidence regarding the potential benefits of a systematic program of physical activity as an affordable and safe intervention for people with Alzheimer's disease. Further, if successful, physical activity in combination with usual care has the potential to alleviate the symptoms of Alzheimer's disease and improve its management and the quality of life of patients and their carers. Trial Registration Australia New Zealand Clinical Trials Registry ACTRN12609000755235

Competing interests The authors declare that they have no competing interests. Authors' contributions All authors are members of the FABS II research team and participated in the implementation of the study. The physical activity and behavioral intervention program was developed by KC. EC is the project coordinator and drafted the manuscript. NL, KC, OA, LF, DA, GB, KH, CB, DL and KA conceived of the study, participated in its design and coordination and critically reviewed the manuscript. MI helped to draft the manuscript and contributed to its critical review. ER contributed to critical review of the manuscript. All authors read and approved the final manuscript.

Acknowledgements We are most grateful to all volunteers taking part in the study and to our research staff. This project is supported by a project grant from the National Health and Medical Research Council (572563) to NL, KC, OA, LF, DA, GB and KH and a Chair Start-Up Support grant to NL from the University of Melbourne.

CC BY

no

2022-01-12 15:21:45

Trials. 2010 Dec 10; 11:120

oa_package/83/1d/PMC3016362.tar.gz