Dominant influence of an HLA-B27 restricted CD8+ T cell response in mediating HCV clearance and evolution

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Virus-specific CD8+ T cell responses play an important role in the natural course of infection; however, the impact of certain CD8+ T cell responses in determining clinical outcome has not been fully defined. A well-defined cohort of women inoculated with HCV from a single source showed that HLA-B27 has a strong association with spontaneous clearance. The immunological basis for this association is unknown. However, the finding is especially significant because HLA-B27 has also been shown to have a protective role in HIV infection. We report the identification of an HLA-B27 restricted hepatitis C virus (HCV)-specific CD8+ T cell epitope that is recognized in the majority of recovered HLA-B27 positive women. In chronically HCV-infected individuals, analysis of the corresponding viral sequence showed a strong association between sequence variations within this epitope and expression of HLA-B27, indicating allele-specific selection pressure at the population level. Functional analysis in 3 chronically HCV-infected patients showed that the emerging variant viral epitopes represent escape mutations. In conclusion, our results suggest a dominant role of HLA-B27 in mediating spontaneous viral clearance as well as viral evolution in HCV infection and mechanistically link both associations to a dominant novel CD8+ T cell epitope. These results support the central role of virus-specific CD8+ T cells and the genetically determined restriction of the virus-specific T cell repertoire in HCV infection. (HEPATOLOGY 2006;43:563–572.)

The hepatitis C virus (HCV) is a small positive-stranded RNA virus within the Flaviviridae family that persists in 70% of infected individuals. Growing evidence suggests that the cellular host immune response against HCV plays an important role in the outcome of infection. Indeed, viral clearance is temporarily associated with sustained HCV-specific CD4+ and CD8+ T cell responses that accumulate in the infected liver.1–3 In addition, the important antiviral role of virus-specific CD8+ T cells has recently been directly demonstrated by CD8+ cell depletion studies in chimpanzees.4 The mechanisms that lead to the failure of the virus-specific T cell response and the persistence of HCV in most patients are still not well understood.5 Several different pathways, such as a primary failure to induce T cells, functional exhaustion, or the emergence of viral escape mutations, have been discussed.5–7 Indeed, a growing body of evidence suggests that T cell escape mutations develop early during acute HCV infection and that they remain fixed in the circulating quasispecies for several years.8–17 However, the development of escape mutations during HCV infection is not universal. For example, viral clearance can occur with minimal epitope variation before resolution,10 and studies in chronically HCV-infected chimpanzees and humans have shown that not all epitopes restricted by HLA class I alleles undergo variations to produce escape mutations.9, 10, 18 The factors that determine the occurrence of escape mutations in HCV infection are not well defined.

Because CD8+ T cells target epitopes restricted by several different HLA class I alleles, it has been suggested that the HLA background may be an important factor in determining the outcome of infection. Indeed, a recent study in this journal, performed in women who were infected with HCV from a single source, has revealed that the HLA class I alleles A03, B27, and Cw01 occurred more frequently in persons with viral clearance compared with chronically infected women. Importantly, the strongest association between any HLA allele and viral clearance has been observed for HLA-B27.19 Although this finding needs to be confirmed in other cohorts of HCV-infected subjects, it is important to note that HLA-B27 also has a protective role in HIV infection, indicated by its association with slow disease progression, for example, delayed CD4+ T cell decline and onset of symptoms, as well as with low viral loads.20–23 In accordance with these findings, a key immunodominant HLA-B27 restricted HIV-specific CD8+ T cell epitope has been described,24 and the emergence of viral escape mutations within this immunodominant HLA-B27 restricted epitope has been shown to be associated with HIV disease progression.25–27

In the case of HCV, however, HLA-B27 restricted virus-specific CD8+ T cell epitopes have not been previously identified. Thus, it has been impossible to determine the basis for the association between HLA-B27 and viral clearance and to gain better insights into the mechanisms of viral persistence in the setting of a protective HLA-allele. In this study, we identified a dominant HLA-B27 restricted CD8+ T cell epitope. Importantly, our results indicate that this HLA-B27 restricted CD8+ T cell response mechanistically links the associations between HLA-B27 and spontaneous viral clearance as well as HLA-B27 and viral evolution. Thus, these results support the central role of virus-specific CD8+ T cells and the genetically determined restriction of the virus-specific T cell repertoire in human HCV infection.

Abbreviations

HCV, hepatitis C virus; PCR, polymerase chain reaction; PBMC, peripheral blood mononuclear cells; IL-2, interleukin-2.

Patients and Methods

Subjects and Study Samples.

The following HLA-B27 positive patients were enrolled in this study (Table 1): (a) A 23-year-old man with a history of intravenous drug use presenting to the University Hospital of Freiburg with symptoms of acute hepatitis, alanine aminotransferase titers peaking at 1,891 U/L (normal, <50 U/L), documented fresh anti-HCV seroconversion and positive HCV RNA (genotype 1a). HCV RNA was cleared 18 weeks after the onset of symptoms. (b) 6 Irish women who were infected with HCV (genotype 1b) from a single source by the injection of anti-D immunoglobulin in 1977 but cleared the infection spontaneously [positive anti-HCV, negative HCV polymerase chain reaction (PCR)]. (c) 7 patients with chronic HCV infection (genotype 1b) who presented to the University Hospital of Freiburg and 1 chronically infected patient (C1) from the Irish cohort. As negative controls, 9 randomly selected, HLA-B27 negative, chronically HCV-infected patients who presented to the University Hospital of Freiburg were studied. Fifty milliliters of EDTA anticoagulated blood was obtained after informed consent and in agreement with federal guidelines and the local ethics committee. Blood from the Irish cohort was shipped overnight to the University Hospital of Freiburg. Peripheral blood mononuclear cells (PBMC) were isolated on lymphocyte separation medium-density gradients (PAA Laboratories, Pasching, Austria) and washed 3 times with phosphate-buffered saline (Gibco, Karlsruhe, Germany). Experiments were performed with fresh cells or frozen cells stored in liquid nitrogen. Patients C6 and C8 underwent liver biopsy for diagnostic reasons; after informed consent, and in agreement with federal guidelines and the local ethics committee, fragments of liver biopsy specimen were put into RPMI 1640 medium (Gibco) containing 10% fetal calf serum and processed as described in the following sections.

Table 1. Patients Characteristics
Clinical SettingPatient IDSexAgeHLA TypeGenotypeViral Load (IU/mL)ALT (U/L)
  • NOTE. Patients R1-6 and C1 are members of the Irish cohort; patients A1 and C2-8 presented to the University Hospital of Freiburg.

  • *

    Peak ALT value. 265 U/mL at time of experiments.

  • HCV PCR positive.

Acute-resolving infectionA1M23A26, A29, B27, B351a2,440,2381,891*
Resolved infectionR1F60A1, A24, B27, B371bNeg. 
 R2F57A2, A29, B27, B491bNeg. 
 R3F60A3, A26, B27, B491bNeg. 
 R4F54A2, A3, B7, B271bNeg. 
 R5F51A2, A11, B7, B271bNeg. 
 R6F47A1, A2, B8, B271bNeg. 
Chronic infectionC1F58A1, A2, B8, B271bND 
 C2M54A2, A11, B27, B351b90,32640
 C3F27A2, B18, B271b157,117115
 C4M25A2, A26, B7, B271b28,23987
 C5F60A1, A25, B27, B571b1,326,95228
 C6F53A26, A33, B14, B271b39,09962
 C7M41A2, A11, B27, B381b810,04064
 C8F70A1, A10, B7, B271b1,550,67093

Peptides.

Overlapping peptides derived from HCV strain H77 (genotype 1a) and strain J4 (genotype 1b) spanning 18 amino acids and overlapping by 11 amino acids were obtained from the NIH AIDS Research and Reference Reagent Program, Germantown, MD. Additional peptides were synthesized with a free amine NH2 terminus and a free acid COOH terminus by standard Fmoc chemistry by Genaxxon Bioscience, Stafflangen, Germany. Peptides corresponding to the newly identified HLA-B27 epitopes (Table 2) including variant peptides were over 90% pure. The peptides were dissolved and diluted as previously described.3

Table 2. HLA-B27 Restricted Epitopes
   t1/2 of Predicted HLA Binding (h)
  1. NOTE. Genotype 1b (consensus): the sequence composed of the amino acid residues most frequently occurring in the Los Alamos HCV Database in genotype 1b sequences. Importantly, these sequences are also the most frequent genotype 1b sequences in the database.

  2. Genotype 1b (infecting virus): the infecting virus in the Irish cohort (GenBank/EMBL accession number AF313916).

  3. Genotype 1b (J4): genotype 1b overlapping peptides were derived from strain J4.

P7 780–788Genotype 1a (H77)G R W V P G A A Y5,000
 Genotype 1b (consensus and infecting virus)G R L V P G A A Y1,000
 Genotype 1b (J4)G R L A P G A A Y1,000
NS3 1492–1501Genotype 1aG R G K P G I Y R F1,000
 Genotype 1b (consensus)G R G R R G I Y R F1,000
 Genotype 1b (infecting virus)G R G R T G I Y R F1,000
 Genotype 1b (J4)G R G R S G I Y R F1,000
NS5B 2820–2828Genotype 1aA R H T P V N S W200
 Genotype 1b (consensus and infecting virus)A R H T P V N S W200
 Genotype 1b (J4)A R H T P I N S W200
NS5B 2841–2849Genotype 1a/1b (consensus, infecting virus, and J4)A R M I L M T H F1,000
NS5B 2936–2944Genotype 1aG R A A I C G K Y1,000
 Genotype 1b (consensus)G R A A T C G K Y1,000
 Genotype 1b (infecting virus)G R A A N C G K Y1,000
 Genotype 1b (J4)G R A A T C G R Y1,000

Tetramers.

The HLA-B27 tetramers were synthesized as previously described, using the peptides specified in Table 2 (purity >90%)(1).

Antibodies.

Anti-CD8 PE and APC, anti-interferon gamma (IFN-γ) FITC antibodies as well as isotype PE, APC and FITC (all BD PharMingen, San Jose, CA) were used according to the manufacturer's instructions.

ELISPOT Assay.

An IFN-γ ELISPOT kit (Diaclone, Besancon, France) was used according to the manufacturer's instructions and as previously described.18 Peptides were added at a final concentration of 10 μg/mL per peptide and were combined in pools of 10 peptides, with each peptide being contained in two different pools in a matrix setup, allowing the identification of single positive peptides. Tests with peptide pools were performed in duplicates; six wells without peptides served as negative controls; 4 wells each stimulated with phytohemagglutinin and phorbol 12-myristate 13-acetate (Sigma-Aldrich, Seelze, Germany) served as positive controls. Responses were considered positive if the number of spots per well was at least 3 times as high as the mean of negative controls.

Intracellular IFN-γ Staining.

Procedures were performed essentially as previously described.3 Briefly, PBMC, CD8+ T cells or expanded peptide-specific T cell lines (0.2 × 106 per well, 96-well plate) were stimulated with peptides (10 μg/mL) or peptide loaded autologous or allogeneic B-LCL in the presence of 50 U/mL human recombinant interleukin 2 (rIL-2; Hoffmann La Roche, Basel, Switzerland) and 1 μL/mL brefeldin A (BD PharMingen). For viral escape analysis experiments, increasing concentrations of consensus and variant peptides were used as indicated in Fig. 3. After 5 hours of incubation (37°C, 5% CO2), cells from each well were blocked with IgG1 antibodies and stained with antibodies against CD8. After permeabilization with Cytofix/cytoperm (BD PharMingen), cells were stained with antibodies against IFN-γ (BD PharMingen) and fixed in 100 μL CellFIX (BD PharMingen) per well before FACS analysis.

Figure 3.

Viral sequence variations in the HLA-restricted epitope NS5B 2841-2849 abolish peptide-specific IFN-γ production by peripheral and intrahepatic CD8+ T cells. Upper row: PBMC from patients C1 (left), C6 (middle), and C8 (right) were stimulated for 2 weeks with peptide NS5B 2841-2849 and then tested for IFN-γ production after 5 hours' stimulation with the consensus peptide (filled diamonds) or the patient-specific variant (empty boxes) in serial dilution as indicated on the x-axis. Lower row: Intrahepatic CD8+ T cells of patients C6 (middle) and C8 (right) were tested for IFN-γ production after 5 hours' stimulation with the consensus peptide (filled diamonds) or the patient-specific variant (empty boxes) in serial dilution as indicated on the x-axis.

Tetramer Staining.

Tetramer staining of peptide-specific CD8+ T cells was performed by using 1 μg PE-labeled HLA-B27 tetramers containing the peptides specified in Table 2 exactly as previously described.18

CD8 Selection and Polyclonal Expansion of PBMC.

4 × 106 PBMC were resuspended in 2 mL phosphate-buffered saline and incubated with magnetic beads coupled to anti-CD8 antibodies (Dynabeads, Dynal, Oslo, Norway) for 20 minutes at 4°C. Bound CD8+ T cells were isolated by using a particle magnetic concentrator. The purity of CD8+ T cells was confirmed by FACS analysis and was greater than 95%. The CD8+ T cells were plated into one well of a 24-well plate (Corning, Corning, NY) in 1 mL complete medium (RPMI 1640 containing 10% fetal calf serum, 1% streptomycin/penicillin, and 1.5% Hepes buffer 1 mol/L) containing 100 U/mL IL-2 (Hoffmann-La Roche), 0.04 μg/mL anti-human CD3 monoclonal antibody (Immunotech, Marseilles, France), and 2 × 106 irradiated autologous PBMC as feeder cells. Twice per week, 1 mL medium was exchanged and 100 U/mL IL-2 added. After 2 weeks, the expanded CD8+ PBMC were tested for HCV-specific CD8+ T cell responses by ELISPOT and intracellular IFN-γ staining.

Stimulation of PBMC With Synthetic Peptides.

4 × 106 PBMC were resuspended in 1 mL complete medium (see above) and stimulated with peptide at a final concentration of 10 μg/mL and anti-CD28 (BD PharMingen) at a final concentration of 0.5 μg/mL. On days 3 and 10, 1 mL complete medium (see above) and recombinant IL-2 (Hoffmann-La Roche, Basel, Switzerland) at a final concentration of 20 U/mL was added to each well. On day 7, the cultures were restimulated with the corresponding peptide (10 μg/mL) and 106 irradiated autologous feeder cells. On day 14, the cells were tested for peptide-specific IFN-γ production after 5 hours of stimulation with autologous or allogeneic HLA-matched or mismatched B-LCL that have been incubated overnight with the corresponding synthetic peptides and washed extensively thereafter.

Isolation and Expansion of Intrahepatic CD8+ T Cells.

Isolation of liver-infiltrating CD8+ T cells was performed as previously described.28 The purity of CD8+ cells was confirmed by FACS analysis and was greater than 95%. The intrahepatic CD8+ T cells were then expanded exactly in the same way as described for CD8+ PBMC (see earlier discussion).

Target Cell Lines.

Autologous and allogeneic Epstein-Barr virus (EBV)-transformed B-LCL were established after informed consent from patients seen in our outpatient clinic as described.28 All target cells were maintained in complete medium (see previous paragraph).

DNA Amplification and Sequencing.

RNA was extracted from sera with the QIAamp Viral RNA Mini Kit (Qiagen, Valencia, CA) and subjected to reverse transcription (ThermoScript Reverse Transcriptase, Invitrogen, Carlsbad, CA) and two rounds of amplification (Expand High Fidelity PCR System, Roche, Mannheim, Germany). The following combination of primers was used: RT—primer SV349: 5′-CCTATTGATCTCARCTGGAG (genomic location in accordance to the PrimAlign tool of the Los Alamos HCV database [http://hcv.lanl.gov]: 1999-2019); PCR I: primers SV349 and SV348: 5′-TCATGYTCCTCCAATGTGTC (genomic location 8694-8713); PCR II: primers SV346: 5′-TCCTCCAATGTGTCRGTCGC (genomic location 8700-8719) and SV347: 5′-AGGTAGGTCAAGTGGYTCAATG (genomic location 7798-7818). The amplified DNA was gel purified with the QIAquick Gel Extraction kit (Qiagen) and subjected to direct sequencing in both directions (Dye Terminator DNA Sequencing kit, Perkin-Elmer, Boston, MA).

Results

Identification of HLA-B27 Restricted HCV-Specific CD8+T Cell Epitopes.

In a first set of experiments, we set out to identify HLA-B27 restricted, HCV-specific CD8+T cell epitopes. For this purpose, PBMC of an acutely HCV-infected HLA-B27 positive patient who subsequently cleared the virus (subject A1, Table 1) were obtained 2 weeks after the onset of symptoms and tested for HCV-specific CD8+ T cell responses by a comprehensive ELISPOT assay with overlapping peptides (genotype 1a sequence H77), and positive peptide-specific responses were confirmed by intracellular IFN-γ staining procedures. A total of six epitopes were recognized in this patient, supporting the notion that multispecific CD8+ T cell responses are associated with spontaneous viral clearance. Additional HLA restriction analysis showed that five of those epitopes were restricted by HLA-B27. Subsequently, all five HLA-B27 restricted epitopes were fine-mapped to different regions of the virus (Table 2). Using the bimas score (http://bimas.dcrt.nih.gov/molbio/hla_bind), the HLA binding of all five HLA-B27 epitopes was calculated to have half-lives between 200 and 5,000 hours (Table 2), which is in the same range as previously described HLA-B27 restricted HIV epitopes.29 The assays performed for the identification and fine mapping of all epitopes are displayed as an example for epitope NS5B 2841 in Fig. 1.

Figure 1.

Fine-mapping of the HLA-B27 restricted epitope NS5B 2841-2849. (A) Functional titration of the minimal epitope NS5B 2841-2849 versus the overlapping peptide (olp) NS5B 2836-2853. PBMC from patient A1 were stimulated for 2 weeks with the overlapping peptide NS5B 2836-2853 and then tested for IFN-γ production after 5 hours' stimulation with the overlapping peptide or the predicted minimal epitope NS5B 2841-2849 in serial dilution as indicated on the x-axis. B HLA restriction of the NS5B 2841-2849 specific CD8+ T cell response. A peptide NS5B 2841-2849 specific CTL line was tested for peptide-specific IFN-γ production after 5 hours' stimulation with autologous or allogeneic HLA-matched or mismatched peptide loaded (black bars) or unloaded (white bars) EBV cell lines. The CD8+T cell responses are expressed as the percentage of CD8+ T cells that produce IFN-γ after 5 hours of stimulation with the different peptide-loaded or -unloaded EBV cell lines. PBMC, peripheral blood mononuclear cells; IFN-γ, interferon gamma; EBV, Epstein-Barr virus.

HLA-B27 Restricted NS5B 2841 Epitope-Specific CD8+ T Cell Responses Are Dominant in Resolved and Chronic HCV Infection.

Next, we determined the immunodominance of the identified HLA-B27 restricted epitopes by studying a cohort of women inoculated with anti-D immunoglobulin in 1977 that had been contaminated with HCV from a single source (genotype 1b). In this cohort, HLA-B27 was shown to hold the strongest association with successful viral clearance.19 Six HLA-B27+ women with successful viral clearance were tested for the presence of HLA-B27 restricted CD8+T cell responses. Because these subjects have been infected with genotype 1b and the infecting virus has been sequenced (GenBank/EMBL accession number AF313916), HLA-B27 restricted peptides corresponding to the inoculum were also synthesized and tested to rule out the possibility that potential discrepancies between synthetic peptides used for functional analysis and the sequences encoded by the virus account for a lack of the detection of T cell responses (Table 2). Five of the six women mounted a CD8+ T cell response against the HLA-B27 restricted NS5B 2841 epitope but not against any of the other epitopes (Fig. 2). This response was detectable by tetramer and peptide specific intracellular IFN-γ analysis. In addition, the NS5B 2841 specific CD8+ T cells proliferated significantly after two rounds of peptide specific stimulation (Fig. 2A). To further determine whether this HLA-B27 restricted CD8+ T cell response is the dominant targeted epitope, we performed a comprehensive analysis with overlapping peptides (genotype 1b sequence J4) and previously identified epitopes restricted by other HLA-types. These results showed that the NS5B 2841 specific CD8+ T cell response was indeed the strongest response in all five patients with a detectable CD8+ T cell response (Fig. 2C).

Figure 2.

NS5B 2841-2849 specific CD8+ T cell responses in acute-resolving/resolved HCV infection (A) and chronic HCV infection (B) and their immunodominance in comparison with other targeted epitopes (C). (A-B) NS5B 2841-2849 specific CD8+ T cell responses were tested in a patient with acute-resolving HCV infection (patient A1), patients with resolved HCV infection (patients R1-6), and patients with chronic HCV infection (C1-8). CD8+ cells were stained with the corresponding HLA-B27 tetramer (left column) and tested for peptide-specific IFN-γ production after 5 hours of peptide stimulation (middle column). PBMCs were also stained with the corresponding HLA-B27 tetramer after 2 weeks of stimulation with peptide NS5B 2841-2849 (right column). (C) Peripheral CD8+ T cells from patients R1-6, C1, C6, and C8 as well as intrahepatic CD8+ T cells from patients C6 and C8 were tested for peptide-specific IFN-γ production in a comprehensive analysis using overlapping peptides (olp) as well as previously described epitopes matched for the HLA alleles present in a given patient. Positive responses are shown as % IFNγ+ cells/CD8+ cells. No responses were detected in the peripheral CD8+ T cells of patient R6 and C8. IFN-γ, interferon gamma; PBMC, peripheral blood mononuclear cells.

Next, we studied the HLA-B27 restricted CD8+T cell response in chronically HCV-infected HLA-B27 positive patients (genotype 1b). Because most HLA-B27+ women within the well-defined Irish cohort cleared the virus, only one chronically infected person (C1) from this cohort could be enrolled in this study. To increase the study population, additional 7 HLA-B27+, chronically HCV-infected patients from the University of Freiburg were enrolled who showed similar clinical and virological characteristics (Table 1). For this analysis, we used peptides corresponding to the different genotype 1b sequences displayed in Table 2. These different peptides were used to minimize the likelihood of missing responses because of a possible mismatch between J4 and the sequences of the infecting viruses. Importantly, HLA-B27 restricted CD8+ T cell responses were observed in three of these patients, and they targeted only the NS5B 2841 epitope (Fig. 2B). Thus, this response was more frequently detectable in subjects with resolved infection than in chronically infected patients (5/6 versus 3/8; not significant; P = .14; Fisher's test). Additional analysis indicated that the HLA-B27 restricted epitope was the dominant response in all three patients (Fig. 2C).

Evidence for HLA-B27 Associated CD8+ T Cell Selective Pressure on the B27-NS5B 2841 Epitope.

Next, we determined the epitope sequence of the predominant virus variants present in the sera of all HLA-B27 positive (n = 8) and randomly selected HLA-B27 negative (n = 9) persons with chronic HCV infection (all patients HCV genotype 1b). Seven of eight (88%) HLA-B27 positive patients exhibited sequence variations within the dominant HLA-B27 restricted NS5B 2841 epitope relative to the genotype 1b consensus sequence (Table 3). In contrast, only two of the nine (22%) HLA-B27 negative patients harbored variations, strongly suggesting an HLA-B27 mediated selective pressure against this viral region (P = .015; Fisher's test). Mutations in this region are also uncommon in the Los Alamos database; only 23 of 126 (18.3%) sequences of genotype 1a or 1b viruses show sequence variations in this epitope region, strongly implying that the mutations seen in HLA-B27+ individuals are attributable to selection by epitope-specific CD8+ T cells (P < .0001; Supplementary Fig. 1 [available at the HEPATOLOGY website [www.interscience.wiley.com/jpages/0270-9139/suppmat/index.html]). Phylogenetic analysis showed, additionally, that the mutations in HLA-B27+ subjects arose independently and are not a result of local/geographical clonal outgrowth (data not shown).

Table 3. Sequence Variations in the NS5 Region Corresponding to Epitope NS5B 2841–2849 (gray box) in HLA-B27 Positive and Negative Patients
inline image

Impact of Amino Acid Substitutions on T Cell Recognition.

Finally, we tested the possible influence of the observed amino acid substitutions within the NS5B-2841 epitope on the HLA binding affinity and T cell recognition. Of note, all observed variations (Table 3) did not lead to a decreased HLA binding affinity in the bimas score (data not shown) and did not affect anchor position 2 or 9. To further assess the impact of amino acid substitutions on T cell recognition, T cell lines derived from patients C1, C6, and C8, who had all detectable NS5B 28411 specific T cells, were tested for IFN-γ production in response to serial dilutions of the wild-type and variant peptide (Fig. 3, upper panel). Importantly, all variant peptides showed a significant decrease in IFN-γ production. The infecting virus of patient C1 (Irish cohort) has been analyzed and shown to contain the wild-type sequence in this epitope. In two of the three patients (C6 and C8), intrahepatic polyclonally expanded CD8+ T cells could also be tested (Fig. 3, lower panel). Again, variant peptides showed a significant decrease in IFN-γ production, demonstrating that viral escape is also detectable without prior peptide specific expansion.

Discussion

A careful analysis of the clinical outcome and HLA-background in a cohort of women who were infected with HCV from a single source19 has recently indicated that the HLA class I allele B27 occurred significantly more frequently in persons with viral clearance compared with chronically infected women. In this study, we describe an HLA-B27 restricted epitope within the NS5B protein that is recognized in most women from this cohort who spontaneously cleared the virus. Indeed, five of the six women tested responded to this epitope, and it was the dominant response in all women. Such a clear dominance of a single epitope-specific CD8+T cell response has not been shown for any other HLA-allele.30 Thus, these results suggest that a single HLA-B27 restricted CD8+ T cell response may explain the observed association between spontaneous viral clearance and HLA-B27, although this finding needs to be confirmed in other cohorts of HCV-infected patients. It is also important to note that the persistence of a dominant HCV-specific CD8+ T cell response in resolved patients does not necessarily reflect the situation during acute HCV infection. However, most data available so far in the field of HCV immunology suggest that the induction of a T cell response occurs early in infection, that the development of a T cell response is arrested thereafter, and that the repertoire of T cell responses persists indefinitely after viral clearance, as has been shown in the human and chimpanzee model as late as 7 years after recovery.1, 2, 4, 31, 32 In addition, our results support the observation that T cells persist decades after viral clearance.33

The important biological role of the NS5B-specific CD8+ T cell response is further supported by our results in chronically HCV-infected, HLA-B27+ subjects. Almost all chronically infected patients expressing the HLA-B27 allele demonstrated sequence polymorphisms within the NS5B epitope. The chronically HCV-infected HLA-B27+ patient without variation from the consensus sequence (C7) did not have a detectable CD8+ T cell response against the NS5B epitope, suggesting that viral escape may not have been required in this patient. The frequent detection of sequence variations within HLA-B27+ patients was in clear contrast to HLA-B27 negative patients, who expressed primarily prototype sequences, suggesting that selective pressure leading to escape was associated with the HLA-B27 allele. Indeed, functional analysis performed in three persons indicated that the variant viral sequences represent viral escape mutations (Fig. 3). Clearly, longitudinal and prospective studies in HLA-B27+ acutely HCV-infected patients who progress to viral persistence are needed to actually prove viral escape.

Taken together, our results suggest that a single HLA-B27 restricted CD8+ T cell response may explain both the association between HLA-B27 and spontaneous viral clearance as well as HLA-B27 and sequence polymorphisms within the NS5B protein. Of note, such a dual effect of a single epitope-specific CD8+ T cell response has not been previously reported in the field of HCV immunology. These results also support the notion that viral escape mutations at least within a certain immunodominant epitope may contribute to viral persistence and evolution as has been suggested by other investigators. For example, Timm et al.13 recently observed a strong association between sequence variation within an immunodominant HLA-B8 restricted NS3 epitope and expression of HLA-B8, also supporting reproducible allele-specific selection pressures at the population level. In another study, Ray et al.15 also used the unique approach of comparing the sequences of viruses from 22 chronically HCV infected women with the sequence of the single common virus that initiated the infection more than 20 years earlier. In that study, the expression of HLA-B7, HLA-B25, and HLA-B37 alleles were found to be linked to the presence of mutations in epitopes presented by these alleles. Together, the combined results clearly indicate that certain HLA-restricted T cell selection pressures drive HCV evolution.

The important question remains why most HCV-infected HLA-B27 positive patients clear the virus and why some go on to viral persistence that seems to be associated with the emergence of viral escape mutations. It is possible that viral factors or certain characteristics of the adaptive immune response may contribute to the development of viral escape mutations. For example, it has been hypothesized that a major factor underlying the apparently less variable nature of certain regions of the genome of mutable RNA viruses may be viral fitness cost; for example, mutations within certain regions will lead to loss of protein structures or enzymatic functions.34 It is interesting to note that the NS5B region is conserved in the general population of HLA-B27 negative patients (Table 3). It is also possible that immunological factors, such as T cell receptor diversity35 or a lack of a sufficient CD4+T cell help11 contribute to escape from CD8+ T cell responses. The latter has been elegantly shown in the chimpanzee model in which an incomplete control of HCV replication by memory CD8+ T cells occurred in the absence of adequate CD4+ T cell help and was associated with the emergence of CD8+ T cell viral escape mutations.11

The mechanisms behind the protective role of HLA-B27 in different viral infections are less understood. Possible explanations include a molecular mimicry between foreign and self-peptides presented by HLA-B27, the high binding capacity of HLA-B27, the unusual tendency of HLA-B27 to misfold, the involvement of non-canonical forms of HLA-B27 such as heavy chain homodimers, the failure of B27 ligands to engage KIR3DL1, leading to an increased natural killer activation or linkage disequilibrium with other immune response genes.36, 37.

In summary, our results indicate that a single dominant HLA-B27 restricted CD8+T cell response may explain the associations between HLA-B27 and spontaneous viral clearance as well as HLA-B27 and sequence polymorphisms within the NS5B protein. The latter has also been shown independently by Timm et al. (personal communication). By correlating the amino acid sequence polymorphisms with the expression of certain HLA class I alleles in chronically HCV-infected patients, they found the strongest association between expression of HLA-B27 and polymorphisms within the NS5B 2841 epitope at the population level, further supporting the important biological role of the T cell response against this epitope region. Taken together, these results suggest that the outcome of infection and viral evolution are significantly influenced by HCV-specific CD8+T cells that are restricted by a protective class I allele such as HLA-B27.

Acknowledgements

The authors thank Drs. Frank Chisari and Stefan Stevanovic for helpful discussions and Nadine Kersting for excellent technical assistance. The overlapping peptides were provided by the NIH AIDS Reagent Programm, MD, and the vaccinia viruses were kindly provided by Dr. Charles Rice, Rockefeller University, New York, NY. The authors thank the patients for donating blood.

Ancillary