Viral Adaptation to Host Immune Responses Occurs in Chronic ...

Viral Adaptation to Host Immune Responses Occurs in Chronic Hepatitis B Virus (HBV) Infection, and Adaptation Is Greatest in HBV e Antigen-Negative Disease Christopher P. Desmond,a,b Silvana Gaudieri,c,d Ian R. James,c Katja Pfafferott,c Abha Chopra,c George K. Lau,e,f Jennifer Audsley,b Caroline Day,g Sarah Chivers,g Adam Gordon,a,g Peter A. Revill,h Scott Bowden,h Anna Ayres,h Paul V. Desmond,i,j Alexander J. Thompson,h,i Stuart K. Roberts,a Stephen A. Locarnini,h Simon A. Mallal,c,k and Sharon R. Lewinb,l,m Department of Gastroenterology, The Alfred, Melbourne, Australiaa; Department of Medicine, Monash University, Melbourne, Australiab; Institute of Immunology and Infectious Diseases, Murdoch University, Perth, Western Australia, Australiac; Centre for Forensic Science and School of Anatomy and Human Biology, University of Western Australia, Crawley, Western Australia, Australiad; Research Centre of Infection and Immunology, The University of Hong Kong, Hong Konge; Department of Medicine, Queen Mary Hospital, The University of Hong Kong, Hong Kongf; Department of Gastroenterology, Box Hill Hospital, Box Hill, Australiag; Victorian Infectious Diseases Reference Laboratory, Melbourne, Australiah; Department of Gastroenterology, St. Vincent’s Hospital, Melbourne, Australiai; Department of Medicine, University of Melbourne, Melbourne, Australiaj; Department of Clinical Immunology and Biochemical Genetics, Royal Perth Hospital, Perth, Western Australia, Australiak; Infectious Diseases Unit, The Alfred, Melbourne, Australial; and Centre for Virology, Burnet Institute, Melbourne, Australiam

Hepatitis B virus (HBV)-specific T-cell responses are important in the natural history of HBV infection. The number of known HBV-specific T-cell epitopes is limited, and it is not clear whether viral evolution occurs in chronic HBV infection. We aimed to identify novel HBV T-cell epitopes by examining the relationship between HBV sequence variation and the human leukocyte antigen (HLA) type in a large prospective clinic-based cohort of Asian patients with chronic HBV infection recruited in Australia and China (n ⴝ 119). High-resolution 4-digit HLA class I and II typing and full-length HBV sequencing were undertaken for treatment-naïve individuals (52% with genotype B, 48% with genotype C, 63% HBV e antigen [HBeAg] positive). Statistically significant associations between HLA types and HBV sequence variation were identified (n ⴝ 49) at 41 sites in the HBV genome. Using prediction programs, we determined scores for binding between peptides containing these polymorphisms and associated HLA types. Among the regions that could be tested, HLA binding was predicted for 14/18 (78%). We identified several HLAassociated polymorphisms involving likely known anchor residues that resulted in altered predicted binding scores. Some HLAassociated polymorphisms fell within known T-cell epitopes with matching HLA restriction. Enhanced viral adaptation (defined as the presence of the relevant HLA and the escaped amino acid) was independently associated with HBeAg-negative disease (P ⴝ 0.003). Thus, HBV appears to be under immune pressure in chronic HBV infection, particularly in HBeAg-negative disease.

G

lobally, infection with hepatitis B virus (HBV) is common, and despite an effective vaccine, the number of persons with chronic HBV infection is increasing (9). Broad and effective HBVspecific T-cell responses are required to clear acute HBV infection (7, 37), but circulating HBV-specific T cells are rarely detected in chronic HBV infection (7, 8). It is therefore unclear whether the adaptive immune response plays a role in controlling viral replication in chronic HBV infection and whether changes or adaptations in HBV are selected due to immune pressure. In individuals with chronic HBV infection, as many as 1011 virus particles are produced per day (21, 30). HBV reverse transcriptase lacks proofreading activity, resulting in mutation rates estimated at 1.5 ⫻ 10⫺5 to 5 ⫻ 10⫺5 nucleotide substitutions per site per year in HBV e antigen (HBeAg)-positive individuals (32). Due to overlapping open reading frames (ORF), HBV genome evolution is constrained to maintain essential protein functions required for replication (28). There is some evidence that changes in HBV sequences can be selected due to host immune pressure (2, 5, 23) and that the HBV mutation rate is even higher in HBeAgnegative individuals (15, 33), possibly due to increased immune pressure prior to HBeAg loss (4, 10, 15) or due to higher levels of virus replication in the absence of HBeAg, given the known inhibitory effect of HBeAg on HBV replication at the level of nucleocapsid maturation or stability (13). This suggests that the host immune response plays an important role in HBV evolution. The human leukocyte antigen (HLA) molecules determine

0022-538X/12/$12.00

Journal of Virology

p. 1181–1192

how foreign peptides are presented to T cells. Population-based genetic approaches have identified statistically significant associations between specific HLA types and polymorphisms (adaptations) in human immunodeficiency virus (HIV) and hepatitis C virus (HCV) (12, 29, 31, 45). Many HLA-associated viral polymorphisms in HIV- and HCV-infected patients were within known T-cell epitopes, consistent with immune pressure (12, 29). These sites of adaptation were subsequently used to identify novel T-cell epitopes for both HIV and HCV (16, 20, 36). In this study, we examined the relationship between HBV polymorphisms and specific HLA types in an Asian population in order to find evidence of viral evolution of HBV at a population level and to determine whether viral adaptation was associated with clinical outcomes. Additionally, we aimed to use these data to identify putative novel HBV epitopes in an Asian population.

Received 5 June 2011 Accepted 12 October 2011 Published ahead of print 9 November 2011 Address correspondence to Sharon Ruth Lewin, [email protected]. Copyright © 2012, American Society for Microbiology. All Rights Reserved. doi:10.1128/JVI.05308-11

jvi.asm.org

1181

Desmond et al.

MATERIALS AND METHODS Subjects. HBV-infected individuals of ethnic Chinese origin were recruited from the Queen Mary Hospital in Hong Kong, China (n ⫽ 141), and from St. Vincent’s Hospital (n ⫽ 51), The Alfred (n ⫽ 11), and Box Hill Hospital (n ⫽ 18) in Melbourne, Australia. Chronic HBV infection was defined by the presence of detectable HBV surface antigen (HBsAg) on two occasions more than 6 months apart, and all individuals were naïve to antiviral therapy. All individuals were HIV and HCV antibody negative. Clinical, biochemical, immunological, and virological details were obtained from prospectively maintained clinic databases and/or retrospective case record review. The study was undertaken with written informed consent from each individual, and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, and was approved by the local ethics committees. HBV DNA quantification. In Melbourne, HBV DNA was quantified with the HBV Digene Hybrid Capture II microplate assay (Digene Diagnostics, Beltsville, MD) or the Versant HBV DNA 3.0 assay (Siemens Healthcare Diagnostics, Tarrytown, NY), in accordance with the manufacturer’s instructions. The lower limit of detection (LLD) of the Digene Hybrid Capture II microplate assay was approximately 24,912 IU/ml, and that of the Versant HBV DNA 3.0 assay was 357 IU/ml. For samples from Hong Kong, HBV DNA was quantified using an in-house TaqMan assay amplifying the core region, with an LLD of 77 IU/ml. Therefore, an LLD of 357 IU/ml was used for all individuals. Alanine aminotransferase (ALT) levels were considered increased if they were ⱖ40 U/liter for the Box Hill Hospital and Alfred cohorts, ⱖ35 U/liter for the St. Vincent’s cohort, or ⱖ80 U/liter for the Hong Kong cohort. PCR and sequencing. HBV DNA was extracted from 200 ␮l of serum using the QIAamp DNA Mini kit (Qiagen, Dusseldorf, Germany) according to the manufacturer’s instructions. The purified DNA was eluted in a final volume of 50 ␮l. HBV DNA was amplified by PCR using the PicoMaxx high-fidelity PCR system (Stratagene, La Jolla, CA) and the method adapted from Günther et al. (14). The sequencing assay has an LLD of roughly 2,000 IU/ml, because a high-fidelity enzyme is required to amplify a relatively large piece of DNA. Samples with viral loads greater than 2,000 IU/ml generally required only one round of PCR. A second round of PCR was performed for samples with low viral loads (between 357 and 2,000 IU/ml) by amplifying two overlapping DNA fragments with primers P1 and 1798* (nucleotides [nt] 1799 to 1820) (3=-CCAACTGCATGGCCTG AGGATG-5=) and primers JM* (nt 1676 to 1696) (3=-TTGGGGTGGAG CCCTCAGGCT-5=) and P2, using 2 ␮l of the first-round product as the template with 25 cycles. Consensus sequences were determined for each genotype and HBV protein, and the amino acids (aa) for each protein and genotype were compared in order to determine percentages of difference per genotype and protein. Polymorphism was defined as the overall percentage of nonconsensus amino acids at a particular residue. Cloning. Cloning was undertaken for mixed populations (those for which the secondary peak was ⬎20% of the main peak) in 14 patients (1 with genotype B and 13 with genotype C). Cloning was performed using the Topo XL TA cloning kit (Invitrogen, Mt. Waverley, Australia), and on average, 5 clones were selected and 2 to 4 clones were sequenced for each patient. The sequences of the individual clones for each subject were then used in the analysis. HLA typing. High-resolution HLA class I (HLA-A, -B, and -C) and II (HLA-DRB1) typing was performed as described previously (48). Allele assignments (Table 1) were made using the Assign program (Conexio Genomics, Fremantle, Australia). This program utilizes the database of HLA allele sequences on the IMGT/HLA website (http://www.ebi.ac.uk /imgt/hla/). Generally, ambiguities were resolved following sequencing with allele-specific subtyping primers. However, for a few individuals, identity at exons 2 and 3 was consistent with the presence of rare alternative and null alleles, and in these rare cases, the allele assignment was based on the most commonly expressed allele and haplotype combination in the relevant population (http://www.allelefrequencies.net/).

1182

jvi.asm.org

TABLE 1 Frequencies of the most common HLA types in an HBVinfected Asian cohort and their associations with genotypes B and C No. (%) of patients with: HLA type

Genotype B

Genotype C

A* 1101 3303 0201 2402 0203 0207

28 (48) 14 (24) 10 (17) 12 (21) 6 (10) 12 (21)

19 (37) 10 (19) 12 (23) 14 (27) 14 (27) 8 (15)

B* 4001 1502 4601 1301 3802 5801

17 (28) 20 (33) 19 (31) 8 (13) 4 (7) 8 (13)

17 (32) 9 (17) 8 (15) 8 (15) 10 (19) 6 (11)

Cw* 0702 0801 0102 0304 0302

23 (38) 25 (41) 22 (36) 9 (15) 8 (13)

25 (47) 13 (25) 12 (23) 16 (30) 6 (11)

DRB1* 0901 0803 1602 0301 0701 1401

19 (32) 6 (10) 7 (12) 6 (10) 5 (8) 7 (12)

19 (35) 11 (21) 10 (19) 6 (11) 7 (13) 3 (6)

Phylogenetic analysis. Deduced HBV aa sequences from the different proteins were aligned using ClustalX. Phylogenetic analysis of the HBV polymerase (Pol) protein was carried out using the neighbor-joining method based on the p-distance model with pairwise deletion and the modified Nei-Gojobori method (49). The mean genetic distances (mean numbers of differences/sequence length) between and within genotype Band genotype C-infected individuals were determined using the same alignments. All analyses were performed using MEGA, version 3.1 (44). Statistical methods. (i) Association between specific HLA alleles and viral polymorphism. Statistical analysis of specific associations between HLA alleles and viral polymorphism, including assessment of any impact of founder effects, followed the methods described by Rauch et al. (36). Analyses were carried out in TIBCO Spotfire S⫹ (Tibco Software, Inc.). Because this study was the first to examine HBV viral adaptation using this approach, a nonconservative cutoff was used to identify viral adaptations, based on a Fisher P value of ⬍0.05 and a cluster-corrected MantelHaenszel P value of ⬍0.1. While false discovery rates and associated q-values as described by Rauch et al. (36) can be obtained for these data, the relatively small sample sizes precluded sensible estimation, and they are not reported here. Odds ratios (OR) for association were compared across genotypes via log-linear models. (ii) Correlation between host viral adaptation and clinical parameters. For all HBV proteins, an individual’s “adaptation score” was defined as the number of residues for which the individual had at least one of the relevant HLAs with a defined association and the escaped aa was present. Relationships between the adaptation score and the clinical parameters ALT level, HBeAg status, and HBV DNA level were assessed via analyses appropriate to the nature of the data; the continuous variable ALT level

Journal of Virology

Viral Adaptation and Immune Response to HBV

FIG 1 Patient recruitment and exclusions. VL, viral load; ⫹ve, positive.

was analyzed (on the log scale) via linear regressions, categorical HBeAg status via Fisher tests, and HBV DNA level by using Cox regression to accommodate right censoring of the higher values. Peptide-HLA ligation. Putative novel HLA class I and II epitopes were screened using the BIMAS (34), SYFPEITHI (35), and Immune Epitope (47) databases of predictive algorithms of peptide-HLA interaction (when the relevant HLA type was available in the databases). Cutoffs for “good” epitopes were set at scores of 18 and 50 for SYFPEITHI and BIMAS, respectively, as described previously (36). These algorithms estimate the strength of ligation between a specific aa sequence and a defined HLA type. These algorithms have been validated previously by comparing the predictive binding of a particular aa, based on its side chain structure, with its actual binding (34). For each open reading frame, screening was undertaken by testing HLA-sequence binding using the aa sequences surrounding a site associated with a specific HLA allele (⫾ 13 aa for HLA class I and ⫾ 17 aa for HLA class II).

RESULTS

Study recruitment and participants. Patient recruitment and demographic characteristics are described in Fig. 1 and Table 2. HBV polymorphism rate. The polymorphism rate at each residue was calculated as the proportion of patients infected with

HBV who had a nonconsensus residue at that site. The HBV polymorphism rate differed at each residue across the genome, ranging from 0% to 49% for genotype B and 0% to 53% for genotype C (Fig. 2). We then compared the degree of aa variation between the consensus genotype B and C sequences for each HBV protein and found differences at 91/843 (10.8%) sites for Pol, 49/400 (12.2%) sites for pre-S, 17/154 (11%) sites for X protein, and 0/211 (0%) sites for core. The difference for core between genotypes B and C was lower than those for the other proteins (P ⬍ 0.00001). The intergenotype mean genetic distance (0.09) was greater than the intragenotype mean genetic distance (0.03). The synonymous and nonsynonymous genetic distances within each genotype across the proteins were similar, reflecting the maintenance of the basic structure and function of the proteins within both genotypes (Table 3). As with HCV (36), these genotype differences are likely to have an impact on potential sites of escape from HLA-restricted immune pressure across the genome. Therefore, all subsequent analyses were performed separately for genotypes B and C. Phylogenetic analysis. We then analyzed all HBV sequences in order to determine the relatedness of the sequences to each

TABLE 2 Demographic characteristics of individuals included in the final analysis Value Patient characteristic

Genotype B

Genotype C

Total

No. of patients No. (%) male No. (%) HBeAg positive Median (range) ALT concn (IU/liter) No. immunotolerant/no. immunoactive (%)a Median (range) HBV DNA concn (IU/ml)

62 34 (55) 35 (57) 47 (8-500) 24/11 (69) 1.20 ⫻ 107 (446-⬎1.7 ⫻ 107)

57 41 (72) 40 (70) 65 (24-591) 31/9 (78) 2.69 ⫻ 105 (689-⬎1.7 ⫻ 107)

119 75 (63) 75 (63) 59 (8-591) 55/20 (73) 2.85 ⫻ 106 (446-⬎1.7 ⫻ 107)

No. (%) with subgenotype: 1 2 3 4 5

1 (2) 44 (71) 4 (6) 13 (21)

12 (21) 14 (24)

31 (55)

a

Immunotolerant patients are HBsAg positive and HBeAg positive, with an ALT level less than twice the upper normal limit. Immunoactive patients are HBsAg positive and HBeAg positive, with an ALT level more than twice the upper normal limit. The sum of immunotolerant and immunoreactive patients as a percentage of all patients is given in parentheses.

January 2012 Volume 86 Number 2

jvi.asm.org 1183

Desmond et al.

FIG 2 HBV polymorphism rates for patients infected with genotype B or C. The x axis shows the aa position for each HBV protein. Vertical bars indicate the proportions of sequences with nonconsensus residues for genotype B (above the horizontal line at zero on the y axis) and genotype C (below the horizontal line at zero on the y axis) for Pol (A), pre-S (B), core (C), and X protein (D).

other. Phylogenetic analysis revealed clustering within the main genotype, with strong bootstrap support (Fig. 3). There was further clustering of sequences within each specific genotype, mainly representing subgenotypes. The methods used accounted for the clustering within the tree, and therefore, we concluded that any HLA-associated viral polymorphisms were not likely to be due to aa changes specific to a subcluster. The

lack of clustering (outside the genotype and subgenotypes) confirmed that sequences from subjects recruited from each site could be combined for HLA/polymorphism analysis but that genotype B and C sequences should be analyzed separately (as discussed above). HLA-associated viral polymorphisms representing putative viral adaptations. We identified 49 statistically significant asso-

TABLE 3 Mean genetic distances within and between HBV genotypes No. of substitutions/sitea for the following protein: Pol

X

Core

Pre-S

Comparison

Synonymous Nonsynonymous Synonymous Nonsynonymous Synonymous Nonsynonymous Synonymous Nonsynonymous

Within genotype Genotype B Genotype C

0.047 0.049

0.016 0.015

0.022 0.023

0.018 0.015

0.043 0.048

0.011 0.011

0.036 0.034

0.01 0.012

Between genotypes 0.174

0.058

0.075

0.061

0.109

0.012

0.135

0.067

a

Calculated as the mean number of synonymous differences per synonymous site or the mean number of nonsynonymous differences per nonsynonymous site, based on pairwise comparison of all sequences within and between genotypes. “Synonymous site” means that nucleotide changes would not result in an amino acid change, while “nonsynonymous site” means that nucleotide changes would result in an amino acid change. A modified Nei-Gojobori method (49) using the p-distance model was used to perform calculations.

1184

jvi.asm.org

Journal of Virology


FIG 3 Phylogenetic analysis of polymerase amino acid sequences for patients recruited from Hong Kong (HK) and Victoria, Australia (VIC).

ciations between the HLA type and viral polymorphisms. These included 37 associations for genotype B (18 for Pol, 6 for the envelope protein, 8 for core, and 5 for X protein), corresponding to 30 residues, and 12 associations for genotype C (3 for Pol, 2 for the envelope protein, 3 for core, and 4 for the X protein), corresponding to 11 residues (Table 4). Four of the associations had an OR of ⬍1, indicating that the consensus sequence had the adapted (or “escaped”) aa (12, 29), and in one case, two different residues occurred with equal frequency in the consensus sequence. There was clustering of associations (involving different HLA types and genotypes) at certain amino acids. For example, the polymorphism at Pol aa 470 was associated with HLA-A*0201 (genotype B) and HLA-A*2402 (genotype C) (Table 4). Two or more HLA types were associated with polymorphisms at Pol aa 104 (genotype B) and 602 (genotype B), core aa 89 (genotype B) and 180 (genotype B), and X protein aa 131 (genotype C), 47 (genotype B), and 143 (genotype B). HBV genotype and HLA-associated viral polymorphisms. There was no overlap in specific 2-digit or 4-digit HLAassociated viral polymorphisms, and overall, there was no evidence of a difference in HLA distribution between genotypes B and C (P ⬎ 0.1). However, some HLA types were overrepresented in genotype B (including HLA-A*0206 [P ⫽ 0.038], HLA-B*1502 [P ⫽ 0.034], and HLA-DRB1*1202 [P ⫽ 0.04]), and others were underrepresented in genotype B (HLA-A*0203 [P ⫽ 0.02]). We then examined the frequency of variation for each of the 49 HLA (4-digit)-associated viral polymorphisms (corresponding to 41 sites) for the alternate genotype. Ten of the 49 (20%) HLA associations could not be compared between the genotypes, since fewer than 5 patients infected with the alternate genotype carried the specific HLA type. Of the remaining 39 residues that could be compared, 12 (31%) were completely conserved in the alternate genotype, even though the specific HLA type was present in patients with that genotype. Furthermore, 21 of the 41 (51%) sites differed in the consensus sequence between the genotypes. The differences in the polymorphism and escape pattern between genotypes are highlighted in Fig. 4.


HLA-associated viral polymorphisms within known HBV T-cell epitopes. We then looked to determine if the HLAassociated viral polymorphisms we identified were located in known HBV epitopes. We examined all published epitopes from all HBV genotypes (summarized and reviewed in reference 8). When we examined prior reports that also used 4-digit HLA typing, we identified two HLA-associated viral polymorphisms within previously published HBV epitopes with the same HLA restriction. The HLA-A*0201-associated polymorphism at aa 195 within the genotype B envelope protein occurred within the known HLA-A*0201-restricted epitopes pre-S1 aa 188 to 196 (S aa 14 to 22) (7, 42) and pre-S1 aa 194 to 202 (S aa 20 to 28) (22) (Fig. 4 and 5; Table 5) (position 195; OR, 15; confidence interval [CI], 1.3 to 264; P ⫽ 0.01). The pre-S1 aa 194-to-202 (S aa 20-to-28) epitope has been demonstrated to elicit specific cytotoxic T-lymphocyte responses in HLA-A2 Chinese patients with a pre-S1 aa 194-to-202 (S aa 20-to-28) sequence identical to that of our patients (22). Pre-S1 aa 195 involved the primary anchor position within the HLA-A*0201 T-cell epitope, where mutation could potentially abrogate HLA binding. The additional HLAA*0201-associated polymorphism at aa 384 within the genotype C envelope protein occurred within the known HLA-A*0201restricted epitopes pre-S1 aa 381 to 390 (S aa 207 to 216) (38) and pre-S1 aa 382 to 390 (S aa 208 to 216) (18) (Fig. 5; Table 5) (position 384; OR, 5; CI, 0.8 to 35; P ⫽ 0.04). The pre-S1 aa 382-to-390 (S aa 208-to-216) epitopes (18) had been identified in genotype C-infected individuals. When we examined the list of HLA associations based on 2-digit typing (data not shown), we found a further three possible HLA allele-specific viral polymorphisms that flanked known CD8 T-cell epitopes, including HLA A*11-associated polymorphisms at core aa 116 (OR, 7.3; CI, 1.1 to 479; P ⫽ 0.01) and aa 126 (OR, 5.3; CI, 0.8 to 356; P ⫽ 0.096) and an HLA A*02-associated polymorphism at X protein aa 143 (OR, 0.05; CI, 0 to 0.6; P ⫽ 0.006). The association between HLA A*02 and X protein aa 143 favored conservation of the wild-type sequence. These associations did not reach statistical significance in the 4-digit analysis, most likely because the HLA-A*11 and HLA-A*02 alleles have more than one main subtype, which were analyzed separately in our 4-digit analysis. These data therefore support the validity of our populationbased approach to identifying novel T-cell epitopes. HLA-associated viral polymorphisms within putative novel HBV T-cell epitopes. In order to determine whether HLAassociated viral polymorphisms not located within known epitopes were of potential biological significance, we used the Web-based epitope prediction programs (SYFPEITHI, BIMAS, and the Immune Epitope database) to determine predicted binding of the viral sequence and associated HLA type. Predicted putative epitopes were identified within all HBV proteins (Table 5). In addition, we identified putative aa escape polymorphisms that were associated with lower predicted HLA binding in several of these epitopes, many located at likely anchor residues (Table 5). Although not yet tested in functional assays using T cells from patients with chronic HBV, these data provide strong supportive evidence that many of the HLA-associated polymorphisms newly identified in this study are likely to be within HBV T-cell epitopes. Host viral adaptation and clinical parameters. We then asked whether the degree of adaptation of HBV was associated with any clinical parameters at a population level. We defined adaptation as

jvi.asm.org 1185

Desmond et al.

TABLE 4 Associations between amino acid residues and HLA types for individuals infected with genotype B or genotype Ca Proteinb Genotype B Pol

Core

Pre-S

X

Genotype C Pol

Core

Pre-S

X

Residue

HLA

Consensusc

OR

CI

P

17 35 73 93 104 104 104 246 266 284 470 480d 568 602 602 678 692 743 78 89 89 106 116 142 180 180 68 195e 214 250 335 387 47 47 87 143 143

A*1102 DRB1*1202 B*4001 DRB1*1201 B*4601 A*0207 DRB1*0901 A*0207 A*1101 A*0207 A*0201 DRB1*0901 DRB1*1602 A*0206 C*0801 A*1101 DRB1*0701 DRB1*0901 A*0206 A*1101 DRB1*1602 B*4001 B*4001 B*4601 A*1102 A*1101 A*0206 A*0201 A*1101 B*1301 A*1101 DRB1*1602 A*0201 C*1502 DRB1*0701 A*1101 DRB1*1501

E R K K N N N R H A N N A S S S P K S L L E S E R R T L N C Y M T T R C C

11 17 0.1 10 6.4 5.0 4.8 6.6 14 6.6 11 4.4 15 8.7 3.9 4.0 7.7 6.0 8.6 0.1 10 7.4 0.1 7.8 8.5 5.0 7.7 15 5.0 9.0 16 15 8.7 11 0.1 17 8.8

1.04–167 1.4-infinity 0.00–0.60 1.17–702 1.28–49 0.88–31 0.94–30 0.96–56 1.03-infinity 0.96–56 1.73–706 1.21–20 2.24–219 1.36–116 1.03–18 0.91–29 0.80–112 1.38–34 0.89–92 0.0–0.86 1.02–129 1.15–100 0.00–1.08 0.89–539 0.88–127 0.99–61 1.22–103 1.31–264 0.99–61 0.97–90 1.18-infinity 2.24–219 0.90–129 0.67–145 0.00–0.92 1.34-infinity 0.91–95

0.023 0.009 0.005 0.017 0.010 0.036 0.048 0.028 0.023 0.028 0.003 0.021 0.002 0.010 0.033 0.049 0.040 0.012 0.033 0.023 0.024 0.016 0.044 0.034 0.033 0.035 0.014 0.013 0.035 0.027 0.016 0.002 0.032 0.049 0.020 0.022 0.031

470 584 841 15 113 129 227 300 384e,f 36 42 131 131

A*2402 B*3802 C*0102 A*2402 DRB1*0405 DRB1*1202 DRB1*0803 A*0203 A*0201 B*1502 A*3303 C*0801 DRB1*0901

Y N R P L L L I N T S I I

5.4 8 5.9 7.3 7.9 6.9 9.9 7.5 5 6.3 7.9 6.7 8.2

0.76–42 1.23–530 0.80–51 0.93–76 0.76–98 0.97–101 1.52–91 1.08–76 0.83–35 1.07–44 0.79–120 0.85–70 1.24–117

0.050 0.021 0.044 0.030 0.045 0.027 0.007 0.020 0.042 0.020 0.042 0.037 0.013

a Residue, location of the polymorphism; consensus, the amino acid at that site in the consensus sequence. The odds ratio (OR), confidence interval (CI), and P values represent associations between the 4-digit HLA type and the presence of a consensus versus nonconsensus amino acid. b Pol, polymerase; pre-S, envelope; X, X protein. c A, alanine; R, arginine; N, asparagine; C, cysteine; E, glutamic acid; H, histidine; I, isoleucine; L, leucine; K, lysine; M, methionine; P, proline; S, serine; T, threonine; Y, tyrosine. d Amino acids N and D occur with equal frequency. e HLA-associated viral polymorphism within a known HBV T-cell epitope. The HLA allele matched the known HLA restriction of the epitope. f The association just fails the Mantel-Haenszel condition and is included for reference.

1186

jvi.asm.org

Journal of Virology


FIG 4 Genotype variation within envelope epitopes. Shown are different variation patterns for genotypes B and C within the published HLA-Aⴱ0201-restricted envelope epitopes pre-S1 aa 188 to 196 (S aa 14 to 22) (7, 44) and pre-S1 aa 194 to 202 (S aa 20 to 28) (33) containing an HLA-Aⴱ0201-associated viral polymorphism site (position 384; OR, 5; CI, 0.8 to 35.3; P ⫽ 0.04 [Table 4]). The proportion of nonconsensus residues within and flanking the epitope is indicated for individuals carrying the HLA-Aⴱ0201 allele (filled bar) and for those not carrying this HLA allele (open bars). This likely reflects divergent cellular immune pressures acting on the virus. Boldface residues represent the significant HLA-associated amino acid polymorphism sites. The published epitope is shown above each graph. SYFPEITHI scores for consensus amino acids are given in parentheses.

the number of residues for which the individual had at least one of the relevant HLAs with a defined association and the putative escaped aa was present. Among patients infected with genotype B, we found that adaptation was significantly lower in HBeAgpositive than in HBeAg-negative patients (P ⫽ 0.0155) (Fig. 6). This difference remained significant after adjustment for gender,

recruitment site, ALT level, and HBV DNA level (P ⬍ 0.001). Among patients infected with genotype C, we also observed lower adaptation in HBeAg-positive than in HBeAg-negative patients, but this difference did not reach significance in an unadjusted analysis (P ⫽ 0.15) (Fig. 6). However, after adjustment for gender, recruitment site, and HBV DNA level, lower adaptation was significantly associ-

FIG 5 Associations between HBsAg amino acid polymorphisms and HLA alleles. Significant (P ⬍ 0.05) associations of HLA alleles with viral polymorphisms within HBsAg may mark relevant immunological sites within the virus. Abbreviations: pre-S1, envelope gene encoding the large surface glycoprotein; pre-S2, envelope gene encoding the middle surface glycoprotein; HBsAg, hepatitis B surface antigen; tp, terminal protein; rt, reverse transcriptase; rh, RNase H. (A) Published HLA class I-restricted epitopes (9). (B) Published HLA class II-restricted epitopes (9). NA, not available. (C) Specific HLA associations with viral mutations within the HBsAg (upper row, genotype B; lower row, genotype C) whereby variation from the consensus amino acid is overrepresented in the HLA-positive group. HLA alleles are shown at the positions of association, with odds ratios given below. Associations within HLA-matched published epitopes are in boldface. (D) Consensus sequences of HBsAg amino acids for genotypes B (upper row) and C (lower row). Sequences are numbered from the start of HBsAg (upper line) and from the start of the surface gene (lower line). We found highly significant associations (P ⬍ 0.05) between HLA Aⴱ0201 and Env residues 195 and 384. These matched known published HLA Aⴱ0201 epitopes.


jvi.asm.org 1187

Desmond et al.

TABLE 5 Associations between amino acid residues and HLA types within potential novel epitopes Proteina and residues

Genotype

HLA

Sequenceb

Syc

Pol 261–269

B

A*1101

466–474

B

A*0201

671–679

B

A*1101

682–696

B

DRB1*0701

GSGPTHNCA GSGHTHICA GSGHTYNCA GSGPTNNCA GSGHTCNCA RIINNQHRT RIINDQHRT RIINHQHRT PTYKAFLSK PTYKAFLHK PTYKAFLRN PTYKAFLNK LNLYPVARQRPGLCQ LNLYPVVRQRPGLCQ LNLYPVARQRSGLCQ

14 17 13 14 13 15 15 15 21 25 15 25 16 10 16

Core 105–113

B

B*4001

174–182

B

A*1101

LEDPASREL LDDPASREL TTVVRPRGR TTVVRRRGR TTVVRQRGR TTVIRQRGR

23 13 20 19 19 18

7, 43

B

A*0201

7

B

A*0201

210–218

B

A*1101

335–343

B

A*1101

VLQAGFFLL VLQAGFFSL FLLTKILTI FSLTKILTI WTSLNFLGG WTSLSFLGG YLWEWASVR FLWEWASVR SLWEWASVR NILNPFLPLL NIVKPFIPLL NILSPFLPLL NILKPFLPLL ILNPFLPLL IVKPFIPLL ILSPFLPLL ILKPFLPLL

23 12 29 19 15 15 16 16 19 25 23 25 25 31 23 30 29

VVPTDHGAHL AVPPDHGAHL LVPADHGAHL AVPSDHGAHL ARRMETTVNAHRNLP ARRMETTVNAHGNLP ARRMETTVNAHWNLP GGCRHKLVC GGCRHKLVR RHKLVCSPAPCNFFT RHKLVRSPAPCNFFT

17 20 18 18 22 22 28 8 16 18 18

Pre-S 188–196d 194–202d

Reference(s)

381–390d

38

C

A*0201

382–390d

7, 18, 43

C

A*0201

44–53

B

A*0201

76–90

B

DRB1*0701

135–143

B

A*1101

138–152

B

DRB1*1501

X

a

Pol, polymerase; pre-S, envelope. Boldface indicates the HLA-amino acid association. c Sy, SYFPEITHI score based on algorithms of predicted peptide-HLA interaction. d HLA-associated viral polymorphism within a known HBV T-cell epitope. b

1188

jvi.asm.org

Journal of Virology


FIG 6 Associations between the adaptation score and clinical parameters for patients infected with genotypes B (left) and C (right). The parameters assessed included HBeAg status (A), ALT level (B), and HBV DNA load (C). (A and B) Data are shown as box-and-whisker plots, where the middle line represents the median; the edges of each box represent the 25th and 75th percentiles; and the whiskers represent the range. (C) Data are plotted as the proportion of cases (y axis) exceeding the corresponding log viral load on the x axis (Kaplan-Meier plots).

ated with HBeAg-positive status, as we found for genotype B (P ⬍ 0.02). Combined analysis of genotypes B and C indicated no difference between the genotypes for an association between HBeAg and adaptation (P ⬎ 0.2). In the combined analysis, HBeAg-positive status and adaptation were again strongly associated (P ⫽ 0.003).


There was no evidence that adaptation was associated with lower HBV DNA levels in genotype B (P ⫽ 0.45) or genotype C (P ⫽ 0.5) (Fig. 6). There was no significant relationship between adaptation and ALT levels for either genotype B or genotype C (P, 0.72 and 0.17, respectively) (Fig. 6).

jvi.asm.org 1189

Desmond et al.

DISCUSSION

We performed full-length HBV sequencing and high-resolution HLA typing for a large cohort of Asian individuals chronically infected with HBV genotype B or genotype C and identified 49 statistically significant associations between viral polymorphisms and HLA type, including associations within previously defined HLA-matched T-cell epitopes. Novel epitopes involving all HBV proteins and potential escape mutations were also identified. Of these putative epitopes, we were able to predict HLA binding for 14/18 (78%) associations, and we identified polymorphisms associated with reduced binding. Polymorphisms occurred in likely anchor residues, consistent with other escape mutations (6, 39). Finally, we found a significant association between viral adaptation within a host and the absence of HBeAg. Taken together, these data demonstrate that HBV is under significant immune pressure at a population level and that this is most prominent in HBeAg-negative individuals. This is the first large population-based study aimed at identifying novel HBV epitopes in a cohort of Asian individuals, who have HLA types vastly different from those of Caucasians. The majority of known HBV HLA class I-restricted epitopes have been identified in individuals with HLA-A2. However, 75% of individuals with chronic HBV infection reside in Asia (19, 24), where HLA-A24 is the most common HLA class I allele (43). Furthermore, HLA-A11 is present in 52% of Chinese individuals and 14% of Caucasians, and HLA-B40 is present in 32% of Chinese individuals and 15% of Caucasians (25) (results consistent with those obtained in this study [Table 1]). Additionally, there are substantial differences in allele subtypes. More than 95% of Caucasians with HLA-A2 have HLAA0201, whereas subtypes HLA-A*0203, HLA-A*0206, and HLA-A*0207 are present in 23%, 10%, and 45% of HLA-A2 individuals of Chinese origin, respectively (17). Given the degree of constraint on HBV viral evolution, largely as a result of overlapping ORF, we found a surprisingly high degree of viral sequence variability, and this differed for genotype B and genotype C. The mean intragenotype genetic distance for genotypes B and C was 0.03, similar to the intragenotype value of 0.05 previously reported for nonstructural protein 3 of HCV genotype 1 sequences (12, 36). Nonstructural protein 3 is the most conserved of the nonstructural proteins in HCV. The intergenotype genetic distance for HCV genotypes 1 and 3 was much higher than that observed for HBV (0.20 for HCV versus 0.09 for HBV), reflecting greater diversity overall for HCV than for HBV. Furthermore, in support of the hypothesis that viral escape is driven by host immune pressure and is restricted by viral (genotype) characteristics, we found substantial differences in the consensus sequences for HBV genotypes B and C that may abrogate HLA binding to critical anchor residues within viral epitopes (Fig. 4). We identified HLA-associated viral polymorphisms that were located within previously described epitopes, providing support for this population-based methodology. In a recent study of New Zealand Tongans with chronic hepatitis B (CHB) with known specific HLA types (mainly HLA-B*4001 and HLA-B*5602 in patients with genotype C and D infections), 13 sites in the HBV genome under significant positive selection pressure were identified, and 5 of these sites were associated with a specific HLA type (1). One of these significant associa-

1190

jvi.asm.org

tions was between HLA-B*4001 and position 106 (equivalent to position 77 in reference 1), the same association found in our cohort, but by use of a different approach, identifying adaptation rather than selection. After adjusting for gender, recruitment site, and HBV DNA level, we found significantly lower adaptation in HBeAgpositive individuals for both genotypes B and C. These findings are consistent with those of another study comparing HBV sequences in individuals over a 20- to 35-year period, where HBeAg-positive asymptomatic carriers with very high levels of HBV DNA had highly conserved nucleotide sequences (33). In contrast, mutations were seen in HBeAg-negative carriers (mean, 20 mutations) and were distributed over all regions of the viral genome, occurring more frequently in putative CD8 T-cell epitopes (33). One potential interpretation of these findings and ours is that in the absence of HBeAg and its “suppressive” effect on the immune response, the HBV sequence evolves more rapidly (26, 27, 46). HBeAg leads to reduction of toll-like receptor-2 expression on monocytes (40) and inefficient release of the inflammatory cytokine tumor necrosis factor alpha. This promotes an imbalance of T helper 1 (Th-1)/Th-2 responses, the production of anti-inflammatory cytokines, such as interleukin 4 (IL-4) and IL-10, and the suppression of HBeAg/HBcAg-specific CD8⫹ T-cell responses (11). This “suppressive” cytokine profile is reversed following seroconversion from HBeAg to HBV e antibody (HBeAb), leading to increases in IL-12 and gamma interferon levels (a Th-1 cytokine profile), which may enhance CD8⫹ T-cell function (41). Increased activity of the innate and adaptive immune system during the HBeAg-negative phase may therefore allow for increased immune pressure, leading to an accumulation of mutations due to immune escape (4, 10, 15). However, this interpretation can be directly addressed only by examining epitopespecific responses and potential specific escape mutations over time prior to and following HBeAg seroconversion, which we were unable to do in this study. We were surprised not to identify an association between adaptation and other clinical parameters, such as the HBV DNA load. Given that HBV has a small and compact genome with multiple overlapping reading frames, any changes in one protein will lead to changes in another protein (that shares the overlapping reading frame). These secondary changes could potentially alter fitness and therefore may not necessarily result in a change in HBV DNA load. Our study had several limitations. First, despite the large sample size, our study had limited power to detect HLAassociated viral polymorphisms for rare HLA alleles. Second, although we checked for phylogenetic relatedness, we cannot exclude the possibility that in some instances an association between an HLA allele and viral polymorphism may be secondary to founder effects (3). Third, we identified putative epitopes and escape mutations only by using epitope prediction programs. Functional analysis to determine the presence of epitope-specific CD8 T cells in HLA-matched individuals with chronic or acute HBV infections will be needed to confirm the immunogenicity of these predicted epitopes. Changes in HLA binding in the presence and absence of the identified polymorphism should also be examined by functional testing. These experiments are under way. Fourth, to demonstrate true viral adaptation or immune evasion at the time of HBeAg se-

Journal of Virology


roconversion, it would be best to assess this in a longitudinal cohort, which was not possible with our current cohort. Finally, in this study, we assessed HBV DNA only from plasma, although it is possible that quasispecies variation or adaptation may be different in HBV DNA from infected hepatocytes, where antigen presentation likely occurs. Analysis and comparison of plasma- and liver-derived HBV sequences would be important in future studies. We have carried out a comprehensive analysis of viral polymorphisms and their associations with the HLA type across the full HBV proteome, with adjustment for potential phylogenetic relatedness between viral sequences and linkage disequilibrium between coinherited HLA alleles. We showed differences in patterns of viral adaptation to HLA-restricted immune pressure between HBV genotypes B and C. Greater viral adaptation was observed in HBeAg-negative individuals, consistent with the concept that the transition from HBeAg-positive to HBeAg-negative disease is associated with significant immune pressure on the virus. Despite the low frequency of circulating HBV-specific T cells, there is evidence that HBV is under substantial immune pressure. Understanding key epitopes associated with immune pressure may potentially lead to new therapeutic strategies for the management of chronic HBV infection. ACKNOWLEDGMENTS We acknowledge support from the Victorian Infectious Diseases Reference Laboratory, Australia; the National Health and Medical Research Council (NHMRC), Australia; and SeqHepB. S.R.L. is an NHMRC Practitioner Fellow. Funding for this work was provided in part by a Pillar Award from Roche Pty Ltd., Australia. The authors have no disclosures.

REFERENCES 1. Abbott WG, et al. 2010. Associations between HLA class I alleles and escape mutations in the hepatitis B virus core gene in New Zealandresident Tongans. J. Virol. 84:621– 629. 2. Bertoletti A, et al. 1994. Natural variants of cytotoxic epitopes are T-cell receptor antagonists for antiviral cytotoxic T cells. Nature 369:407– 410. 3. Bhattacharya T, et al. 2007. Founder effects in the assessment of HIV polymorphisms and HLA allele associations. Science 315:1583–1586. 4. Bozkaya H, Ayola B, Lok AS. 1996. High rate of mutations in the hepatitis B core gene during the immune clearance phase of chronic hepatitis B virus infection. Hepatology 24:32–37. 5. Carman WF, et al. 1997. Hepatitis B virus core protein mutations are concentrated in B cell epitopes in progressive disease and in T helper cell epitopes during clinical remission. J. Infect. Dis. 175:1093–1100. 6. Chen BF, et al. 2006. High prevalence and mapping of pre-S deletion in hepatitis B virus carriers with progressive liver diseases. Gastroenterology 130:1153–1168. 7. Chisari FV, Ferrari C. 1995. Hepatitis B virus immunopathogenesis. Annu. Rev. Immunol. 13:29 – 60. 8. Desmond CP, Bartholomeusz A, Gaudieri S, Revill PA, Lewin SR. 2008. A systematic review of T-cell epitopes in hepatitis B virus: identification, genotypic variation and relevance to antiviral therapeutics. Antivir. Ther. 13:161–175. 9. Dienstag J. 2008. Hepatitis B virus infection. N. Engl. J. Med. 359: 1486 –1500. 10. Fares MA, Holmes EC. 2002. A revised evolutionary history of hepatitis B virus (HBV). J. Mol. Evol. 54:807– 814. 11. Ferrari C, et al. 1990. Cellular immune response to hepatitis B virusencoded antigens in acute and chronic hepatitis B virus infection. J. Immunol. 145:3442–3449. 12. Gaudieri S, et al. 2006. Evidence of viral adaptation to HLA class I-restricted immune pressure in chronic hepatitis C virus infection. J. Virol. 80:11094 –11104.


13. Guidotti LG, et al. 1996. The hepatitis B virus (HBV) precore protein inhibits HBV replication in transgenic mice. J. Virol. 70:7056 –7061. 14. Günther S, et al. 1995. A novel method for efficient amplification of whole hepatitis B virus genomes permits rapid functional analysis and reveals deletion mutants in immunosuppressed patients. J. Virol. 69: 5437–5444. 15. Hannoun C, Horal P, Lindh M. 2000. Long-term mutation rates in the hepatitis B virus genome. J. Gen. Virol. 81:75– 83. 16. Kawashima Y, et al. 2009. Adaptation of HIV-1 to human leukocyte antigen class I. Nature 458:641– 645. 17. Krausa P, Browning MJ. 1996. HLA-A2 polymorphism and immune functions. Eur. J. Immunogenet. 23:261–274. 18. Lee HG, et al. 1997. Peptide-specific CTL induction in HBV-seropositive PBMC by stimulation with peptides in vitro: novel epitopes identified from chronic carriers. Virus Res. 50:185–194. 19. Lee WM. 1997. Hepatitis B virus infection. N. Engl. J. Med. 337:1733–1745. 20. Leslie A, et al. 2005. Transmission and accumulation of CTL escape variants drive negative associations between HIV polymorphisms and HLA. J. Exp. Med. 201:891–902. 21. Lewin SR, et al. 2001. Analysis of hepatitis B viral load decline under potent therapy: complex decay profiles observed. Hepatology 34: 1012–1020. 22. Liu HG, et al. 2008. A mutant HBs antigen (HBsAg)183–191 epitope elicits specific cytotoxic T lymphocytes in acute hepatitis B patients. Clin. Exp. Immunol. 151:441– 447. 23. Maini MK, et al. 2000. T cell receptor usage of virus-specific CD8 cells and recognition of viral mutations during acute and persistent hepatitis B virus infection. Eur. J. Immunol. 30:3067–3078. 24. Merican I, et al. 2000. Chronic hepatitis B virus infection in Asian countries. J. Gastroenterol. Hepatol. 15:1356 –1361. 25. Middleton D, Menchaca L, Rood H, Komerofsky R. 2003. New allele frequency database: http://www.allelefrequencies.net. Tissue Antigens 61: 403– 407. 26. Milich D, Liang TJ. 2003. Exploring the biological basis of hepatitis B e antigen in hepatitis B virus infection. Hepatology 38:1075–1086. 27. Milich DR, Chen MK, Hughes JL, Jones JE. 1998. The secreted hepatitis B precore antigen can modulate the immune response to the nucleocapsid: a mechanism for persistence. J. Immunol. 160:2013–2021. 28. Mizokami M, et al. 1997. Constrained evolution with respect to gene overlap of hepatitis B virus. J. Mol. Evol. 44(Suppl. 1):S83–S90. 29. Moore CB, et al. 2002. Evidence of HIV-1 adaptation to HLA-restricted immune responses at a population level. Science 296:1439 –1443. 30. Nowak MA, et al. 1996. Viral dynamics in hepatitis B virus infection. Proc. Natl. Acad. Sci. U. S. A. 93:4398 – 4402. 31. Okazaki T, et al. 2006. Possible therapeutic vaccine strategy against human immunodeficiency virus escape from reverse transcriptase inhibitors studied in HLA-A2 transgenic mice. J. Virol. 80:10645–10651. 32. Orito E, et al. 1989. Host-independent evolution and a genetic classification of the hepadnavirus family based on nucleotide sequences. Proc. Natl. Acad. Sci. U. S. A. 86:7059 –7062. 33. Osiowy C, Giles E, Tanaka Y, Mizokami M, Minuk GY. 2006. Molecular evolution of hepatitis B virus over 25 years. J. Virol. 80:10307–10314. 34. Parker KC, Bednarek MA, Coligan JE. 1994. Scheme for ranking potential HLA-A2 binding peptides based on independent binding of individual peptide side-chains. J. Immunol. 152:163–175. 35. Rammensee H, Bachman J, Emmerich N, Bachor O, Stevanovic S. 1999. SYFPEITHI: database for MHC ligands and peptide motifs. Immunogenetics 50:213–219. 36. Rauch A, et al. 2009. Divergent adaptation of hepatitis C virus genotypes 1 and 3 to human leukocyte antigen-restricted immune pressure. Hepatology 50:1017–1029. 37. Rehermann B, et al. 1995. The cytotoxic T lymphocyte response to multiple hepatitis B virus polymerase epitopes during and after acute viral hepatitis. J. Exp. Med. 181:1047–1058. 38. Rehermann B, Pasquinelli C, Mosier SM, Chisari FV. 1995. Hepatitis B virus (HBV) sequence variation of cytotoxic T lymphocyte epitopes is not common in patients with chronic HBV infection. J. Clin. Invest. 96:1527–1534. 39. Riedl P, et al. 2006. Distinct, cross-reactive epitope specificities of CD8 T cell responses are induced by natural hepatitis B surface antigen variants of different hepatitis B virus genotypes. J. Immunol. 176:4003– 4011. 40. Riordan SM, Skinner N, Kurtovic J, Locarnini S, Visvanathan K. 2006. Reduced expression of toll-like receptor 2 on peripheral monocytes in patients with chronic hepatitis B. Clin. Vaccine Immunol. 13:972–974.

jvi.asm.org 1191

Desmond et al.

41. Rossol S, et al. 1997. Interleukin-12 induction of Th1 cytokines is important for viral clearance in chronic hepatitis B. J. Clin. Invest. 99:3025–3033. 42. Sette A, et al. 1994. The relationship between class I binding affinity and immunogenicity of potential cytotoxic T cell epitopes. J. Immunol. 153: 5586 –5592. 43. Sobao Y, et al. 2001. Identification of hepatitis B virus-specific CTL epitopes presented by HLA-A*2402, the most common HLA class I allele in East Asia. J. Hepatol. 34:922–929. 44. Tamura K, Dudley J, Nei M, Kumar S. 2007. MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24:1596 –1599. 45. Timm J, et al. 2007. Human leukocyte antigen-associated sequence poly-

1192

jvi.asm.org

46. 47. 48. 49.

morphisms in hepatitis C virus reveal reproducible immune responses and constraints on viral evolution. Hepatology 46:339 –349. Visvanathan K, et al. 2007. Regulation of Toll-like receptor-2 expression in chronic hepatitis B by the precore protein. Hepatology 45:102–110. Wang P, et al. 2008. A systematic assessment of MHC class II peptide binding predictions and evaluation of a consensus approach. PLoS Comput. Biol. 4:e1000048. Witt CS, et al. 2002. Common HLA-B8-DR3 haplotype in Northern India is different from that found in Europe. Tissue Antigens 60:474 – 480. Zhang J, Rosenberg HF, Nei M. 1998. Positive Darwinian selection after gene duplication in primate ribonuclease genes. Proc. Natl. Acad. Sci. U. S. A. 95:3708 –3713.

Journal of Virology