The inherent sequence diversity of the hepatitis C virus (HCV) with the existence of multiple genotypes that differ up to 20% at the amino acid level represents one of the major obstacles for immune control. Accordingly, immune control of a heterologous virus challenge, particularly across genotypes, is difficult to achieve; however, the overall role of genotype-specific sequence differences has not yet been defined at the epitope level. The aim of this study was to determine the role of genotype-specific sequence differences for the CD8+ T cell response against HCV. We analyzed a cohort of anti-HCV–positive injection drug users infected with HCV genotype 1 (n = 17) or genotype 3 (n = 22) or undetectable HCV-RNA (n = 14) with overlapping peptides covering consensus sequences of NS3 from both genotypes. Importantly, the majority of HCV-specific CD8 T cells were specific for one genotype only indicating that sequence differences between genotypes are relevant at the epitope level. Interestingly, T cells active against both genotypes were significantly more frequent in HCV-RNA–negative subjects. Of note, we identified five subjects with undetectable viremia and coexistence of two T cell populations—one for each genotype—suggesting immune control of two different genotypes. Conclusion: We systematically analyzed the degree of cross-genotype reactivity of HCV-specific T cells and have shown that CD8 responses targeting different HCV genotypes can be primed in the same individual and that such responses potentially characterize a subgroup among injection drug users being protected from chronic HCV infection. (HEPATOLOGY 2009.)
Because the risk of infection with hepatitis C virus (HCV) by infusion of contaminated blood products has decreased substantially over the past two decades, injection drug use has become the most important transmission risk factor in Europe and North America.1, 2 The prevalence of HCV-specific antibodies in this patient group is 50%-90%.2 Importantly, not all patients exposed to HCV develop chronic infection with viral persistence, a subset of 20%-40% spontaneously controls viral replication and eliminates the virus.1 However, repetitive infections with HCV have been experimentally demonstrated in the chimpanzee model3, 4 and have been observed in high-risk groups for HCV exposure such as hemophiliacs or injection drug users,5, 6 suggesting a lack of sterilizing immunity even after spontaneous control of a previous infection. Interestingly, despite lack of sterilizing immunity, prospective studies suggest that previous clearance of HCV infection is associated at least with partial protection from viral persistence upon reinfection,7, 8 although a recent study did not support this finding.9
The exact determinants of a successful immune response against HCV are not fully understood. Multiple studies have suggested that the adaptive immune response plays an important role (reviewed by Bowen and Walker10). Functional and sustained virus-specific T cell responses against multiple epitopes are associated with spontaneous clearance of HCV. In turn, chronic HCV infection is often characterized by absence or only weak and dysfunctional T cell responses.10 Therefore, many vaccine candidates in preclinical studies aim to generate HCV-specific T cells.11, 12 However, the inherent sequence diversity of HCV precludes that all immunodominant targets of the T cell response in one isolate are conserved across all other isolates. At least six genotypes and multiple subtypes that differ up to 20% at the amino acid level exist.13 Studies from chimpanzees that previously cleared infection suggest that reinfection with a different genotype may be particularly difficult to control.4 Accordingly, immunization against a broad range of HCV genotypes is difficult to achieve,12 although the overall role of genotype-specific sequence differences was not yet defined at the epitope level.
In Europe, genotype 1a/1b (GT1a/1b) and genotype 3a (GT3a) are the most prevalent HCV genotypes, and GT3a is epidemiologically linked to intravenous drug use.2 The aim of this study was therefore to determine the impact of genotype-specific sequence differences of HCV on the CD8 T cell response and to define epitopes that are reactive across genotypes. We analyzed a cohort of injection drug users with overlapping peptides covering consensus sequences of HCV NS3 GT1b and GT3a, because this protein is an important target during acute and chronic HCV infection. The injection drug user cohort is characterized by a heterogeneous genotype distribution with GT1 and GT3 both being predominant.
Patients with a history of injection drug use who reported to be HCV-positive were recruited in the ward for inpatient detoxification treatment of drug addicts as well as in the clinic for opiate substitution treatment at the Department of Addictive Behavior and Addiction Medicine, Rhine State Hospital Essen, Hospital of the University of Duisburg-Essen, after written informed consent was obtained. By this approach, 53 anti-HCV–positive patients were identified, including 17 infected with GT1, 22 infected with GT3, and 14 HCV-RNA–negative subjects. A complete list of all subjects is provided in the Supporting Material. The study was approved by the local ethics committee according to the guidelines of Helsinki. All patients were HLA-typed using standard molecular techniques.
Analysis and Alignment of HCV Sequences.
Amplification of viral RNA and sequencing was performed as described.14 A total of eight HCV GT1a isolates, 44 GT1b isolates, and 42 GT3a isolates were included in the analysis. All sequences have been submitted to GenBank (accession no. FJ864731-FJ864816). To attain a similar total number of HCV GT1a sequences, an additional 25 NS3 sequences of German or Swiss origin from the public database (http://hcv.lanl.gov/) were included. Phylogenetic analysis of all sequences support the genotyping results and suggests similar homology levels between sequences in each genotype group (data not shown).
Overlapping HCV Peptides.
Two overlapping peptide sets (each 85 peptides, 15-18 amino acids in length) covering local consensus sequences from HCV GT1b and GT3a (Supporting Material) were generated with PeptGen (available at http://hcv.lanl.gov/) and synthesized (≥70% purity) together with an additional 12 previously defined optimal epitopes.
Strategy for Analysis of CD8 T Cell Responses and the Level of Cross-Genotype Reactivity.
Antigen-specific T cells were expanded in vitro in the presence of peptide pools for 10 days before restimulation with the same peptide pool and staining of intracellular interferon gamma (IFNg). The reactive peptide in positive pools was determined the next day by way of restimulation with individual peptides. All individual responses were confirmed in a second series of cultures using cryopreserved peripheral blood mononuclear cells (PBMCs) and fine-mapped if possible. To determine the degree of cross-genotype reactivity, PBMCs were cultivated in the presence of the GT1b–GT3a sequence of every reactive peptide. After 10 days, both cultures were resimulated with the GT1b and GT3a peptide before intracellular IFNg staining. Lack of cross-reactivity between genotypes was assumed when no specific T cells were expanded in the presence of one peptide in at least two attempts or the frequency of CD8+IFNg+ T cells upon restimulation in the presence of the highest concentration (10 μg/mL) of the nonreactive peptide was more than 10-fold below the frequency of the reactive peptide. Partial cross-genotype reactivity was assumed when specific T cells were expanded in the presence of both peptides and when the frequency of CD8+IFNg+ T cells in the presence of the highest concentration (10 μg/ml) of one peptide was less than 10-fold reduced but there was a more then 10-fold difference in the peptide concentration needed to stimulate 50% of the maximum response (SD50). Full cross-genotype reactivity was assumed when specific T cells were expanded in the presence of both peptides and when the frequency of CD8+IFNg+ T cells in the presence of the highest concentration (10 μg/mL) of one peptide was less than 10-fold reduced and there was a less then 10-fold difference in the SD50.
A detailed description of the strategy for detection of HCV-specific CD8 T cells and of the analysis of cross-genotype reactivity is provided in the Supporting Material.
Nonparametric Kruskal-Wallis tests and Fisher's exact tests were performed using GraphPad Prism 4.0 software (GraphPad Software, San Diego, CA). To avoid overestimation of the total breadth and magnitude, only CD8 responses against overlapping peptides were included in the analysis in Fig. 1.
Substantial Sequence Differences Between HCV GT1 and GT3.
The vast majority of immunology studies of HCV infection in recent years focused on analyses of T cell responses directed against GT1-derived antigens. Although HCV GT3a infection is common in Europe, only four full GT3a genomes are available in the public database.15 We therefore decided to sequence local GT3a isolates to construct a more robust consensus sequence. Complete NS3 of a total of 42 HCV GT3a isolates has been sequenced, and the derived consensus sequence was compared with a local HCV GT1b consensus sequence (based on 44 isolates). In this comparison, 108 of all 631 (17.1%) residues of NS3 differed between the GT1b and GT3a consensus sequence (Supporting Fig. 1). However, more important for the impact on T cell immunology are the differences that occur within 9mers as the usual epitope length for specific CD8 T cells. We therefore counted the number of amino acid differences for each of the 623 9mers in NS3. Interestingly, only 23% of all 9mers were identical between both consensus sequences. Of note, 77% of these 9 amino acid windows harbored at least one amino acid difference and 48% harbored at least two amino acids difference (data not shown), indicating that differences in targeted epitopes between HCV GT1b and GT3a may be common.
Comparison of HCV-Specific CD8 Responses Against GT1 and GT3.
The observed sequence difference between HCV GT1b and GT3a raised the possibility that the immune responses directed against these genotypes differ. To address this, we used overlapping peptide sets based on these NS3 consensus sequences and analyzed HCV-specific CD8 T cell responses in a cohort of injection drug users. A total of 53 subjects were analyzed, including 14 subjects with undetectable viremia, 17 subjects with HCV GT1 infection, and 22 subjects with HCV GT3 infection (Table 1). Of note, although seven of 14 (50%) of the HCV-RNA–negative subjects were not typeable with an HCV serotyping assay, serotype 1 was detected in four subjects, serotype 3 in two subjects, and a mixed GT1/GT3 serotype in one additional subject, indicating heterogeneous genotype exposure in this group (Table 1). With the overlapping peptide sets, a total of 27 individual epitopes was identified (one additional epitope was only detected with an optimal peptide) corresponding to a total of 57 CD8 responses (Table 2). HCV-specific T cells were detected in seven of 17 (41.2%) subjects infected with GT1, in nine of 22 (40.9%) subjects infected with GT3, and 12 of 14 (85.7%) subjects with undetectable viremia. The total strength and number of HCV-specific T cell responses identified with overlapping peptides was higher in the HCV-RNA–negative group when compared with the GT1- or GT3-infected groups (Fig. 1A,B) (P < 0.01). There was no difference between the strength and magnitude of the T cell response directed against GT1 or GT3 peptides in HCV-RNA–negative subjects. Interestingly, in the subjects with GT1 or GT3 infection, there was a trend toward preferential targeting of the heterologous genotype, although this trend was statistically not significant.
Degree of Cross-Genotype Reactivity of CD8 T Cell Epitopes.
To further address the degree of cross-genotype reactivity, individual epitopes were fine-mapped if possible, and additional cultures were set up in which frozen PBMCs were stimulated with the GT1 and GT3 sequence of each individual reactive peptide. After 10 days, each of these cultures was restimulated with the mapped GT1 and GT3 epitope peptide before intracellular IFNg staining. This approach enabled a classification of the epitopes into those with no, partial, or full cross-genotype reactivity (see Supporting Methods for definition).
Only three of 28 epitopes detected in this study had an identical consensus sequence. An example is shown in Fig. 2. We identified three subjects that target amino acids 1373-1380 (subjects 42, 62 and 67). This epitope was recently reported as an HLA-B51–restricted epitope.16 In line with this, all three subjects here are HLA-B51–positive. The alignment of GT1 and GT3 sequences indicates that this region is highly conserved in both genotypes, which potentially points to an attractive target for immune therapies. Of note, two of three subjects targeting this epitope are HCV-RNA–negative and the single subject with chronic infection harbors a sequence variant (VPFYGKAI) (Supporting Table 1). This variant is only partially cross-reactive with the prototype-specific response, suggesting that it represents an escape variant.
However, 25 of 28 epitopes harbored at least one amino acid difference between the consensus sequences of GT1b and GT3a. Figure 3 shows examples of the different degrees of cross-genotype reactivity of these epitopes. An example of a CD8 response without cross-genotype reactivity from subject C11 infected with GT3 is shown in Fig. 3A. This subject targets two CD8 epitopes in NS3, including one located in positions 1193-1201. The consensus sequence differs in five positions between genotype 1b and genotype 3a. Interestingly, although the subject is infected with GT3a, only T cells reactive with the GT1b sequence of the epitope were detected after in vitro expansion, indicating that the autologous GT3a sequence is not targeted and not cross-reactive. Of note, sequences from local GT1b and GT3a isolates support that the differences between GT1b and GT3a are specific for each genotype (data not shown).
An example of a cross-genotype reactive CD8 T cell response characterized by targeting of a region that differs between GT1 and GT3 is shown in Fig. 3B. Subject R37 has a CD8 response directed against the HLA-A68–restricted epitope A68-1175. The GT1 and GT3 sequence in the HLA-A68–restricted epitope differ in position 2 and 3 of the epitope. However, T cells were successfully expanded in the presence of the GT1 and GT3 sequence, and both cultures reacted with both peptide variants; this finding suggests full cross-genotype reactivity, which was confirmed in peptide titrations (data not shown). Other fully cross-genotype reactive CD8 responses were directed against the HLA-B8–restricted epitopes B8-1395 and B8-1611 (Table 2). An example of a CD8 response with partial cross-recognition is shown in Fig. 3C. T cells from subject R42 directed against the epitope A2-1273 were successfully expanded in the presence of the GT1b and GT3a sequence and showed cross-reactivity in the presence of high peptide concentrations. However, additional peptide titrations revealed preferential targeting of the GT1b sequence (Fig. 3D).
In summary, the majority of epitopes (19 of 28; 67.9%) detected in this study did not show any cross-genotype reactivity between GT1 and GT3. This includes 13 epitopes (46.4%) that were only detected in GT1 and six epitopes (21.4%) that were only detected in GT3 (Fig. 4). Six of the total of 28 epitopes (21.4%) showed cross-genotype reactivity. This includes three epitopes in which the targeted sequence is identical in GT1 and GT3 and three additional epitopes that targeted both genotype-specific variants. Two additional epitopes showed partial cross-reactivity between both genotypes.
Coexistence of Distinct T Cell Populations in the Same Individual Targeting both Genotypes.
One epitope was unique because two reactive T cell populations—each of which is predominantly targeting only one genotype—were detected in the same subject. Three subjects showed a response against amino acids 1627-1635, including two subjects with undetectable HCV-RNA and one with HCV GT1 infection and low viral load (1,785 IU/mL). The targeted epitope represents an HLA-B13–restricted epitope (Fig. 5A,B). The consensus sequence of both genotypes differs in two positions. Interestingly, two CD8 T cell populations specific for this epitope coexisted. One T cell population was expanded in the presence of the GT1 sequence of the epitope (RLGAVQNEV) and did not target the GT3 sequence (RLGPVQNEI); the other T cell population was expanded in the presence of the GT3 sequence and did not target the GT1 sequence. Importantly, the A to P substitution in position 4 of the GT3a sequence of the epitope is completely absent from all 44 GT1b sequences and is present in all 42 GT3a sequences, suggesting that both epitope sequences are specific for each genotype (data not shown). The same pattern was observed in a third subject, indicating that they all have been exposed to both genotypes and that two distinct T cell populations have been primed. Of note, the subject with detectable HCV GT1a viremia harbored a variant (RLGAVQNEA) that is similarly targeted when compared with the prototype (data not shown).
An additional four subjects (one with GT1 infection and three HCV-RNA–negative subjects) were identified with at least two T cell populations targeting different regions in both genotypes, but each lacking cross-genotype reactivity. Two examples are shown in Fig. 5C-F. Subject R13 targets three epitopes in NS3, including the HLA-A3–restricted epitope A3-1227 and a novel epitope in GT3 spanning amino acids 1265-1274. In the presence of the GT1 sequence of the A3-1227 epitope, T cells were expanded that did not cross-react with the GT3 sequence of the epitope (Fig. 5C). Of note, in this case there is only one amino acid different between both genotypes in position 1 of the epitope. In contrast, T cells targeting the 1265-1274 epitope were only expanded in the presence of the GT3 peptide sequence (Fig. 5D). Interestingly, this region entirely overlaps with an HLA-A11–restricted epitope (A11-1265) described previously in GT1. However, the subject is HLA-A11–negative and five of ten residues differ in the nonreactive GT1 sequence of this epitope, further supporting that this is truly a GT3-specific epitope. Similarly, subject R29 has four CD8 responses, including T cells specific for the GT1b sequence of the epitope A2-1073 and T cells specific for the GT3a sequence of the epitope B27-1492 (Fig. 5E,F).
Taken together, five subjects with undetectable viremia were identified in which two T cell populations—one for each genotype—coexisted, suggesting immune control of two different genotypes.
CD8 Responses Active Against both Genotypes Are Predominantly Detected in HCV-RNA–Negative Subjects.
The risk behavior and heterogeneous genotype distribution of injection drug users suggest that exposure to different genotypes may be common. An important question is therefore if CD8 responses that are effective against both genotypes provide some degree of protection from chronic HCV infection. Interestingly, we found CD8 responses active against both genotypes predominantly in HCV-RNA–negative subjects. Nine of 14 (64.3%) subjects with undetectable viremia harbored T cells that are active against GT1b and GT3a (Fig. 6). This includes T cells that are directed against fully cross-reactive epitopes (four subjects) and T cells that are genotype-specific, but two T cell populations specific for each genotype coexisted (five subjects). In contrast, T cells active against both genotypes were only detected in six of 39 (15.3%) subjects with detectable HCV-RNA.
The antiviral immune response against HCV infection does not achieve sustained control of viral replication in the majority of subjects with acute infection.1 However, clinical observations suggest that successful clearance of one HCV infection increases the likelihood of spontaneous control of a second infection,7, 8 raising hopes that a vaccine may be beneficial by substantially decreasing the rate of viral persistence. The sequence diversity of HCV is believed to represent one of the major obstacles to immune control. We analyzed the impact of genotype-specific sequence differences on the CD8 T cell immune response in a cohort of injection drug users infected with GT1 and GT3. Of note, the majority of CD8 responses are genotype-specific. However, we identified several subjects with undetectable viremia and evidence for immune control of two different genotypes.
Much effort has been spent in recent years to improve vectors for antigen delivery or to define more efficient vaccine strategies. Interestingly, in HCV vaccine design, few studies have addressed which protein or peptide sequences are optimal for immunization.16 This aspect seems particularly important in the presence of multiple sequence variants in circulating isolates worldwide. The importance of sequence differences within the host at the level of quasispecies has already been highlighted. The quasispecies serve as a pool from which variants that harbor beneficial mutations are selected. Accordingly, rapid selection of immune escape and drug resistance mutations is observed.17–20 Notably, the sequence differences between HCV genotypes are not the result of selection within individuals. Most differences are specific for the genotype and reproducibly observed across all other isolates of the same type. Our study shows that these sequence differences are relevant in T cell immunology, because the majority of T cell responses are directed against only one genotype and are not cross-reactive with the other genotype.
Cross-reactive CD8 responses are likely beneficial to achieve a protective immune response against different HCV genotypes. CD8 responses directed to epitopes that are conserved across genotypes and responses that equally recognize different genotype-specific variants of the epitope fall into this category; in this study, we identified CD8 responses from both. For example, the HLA-B51–restricted epitope IPFYGKAI1373-1380 is highly conserved across all genotypes but also within different isolates of the same subtype, indicating that sequence variation is not well tolerated in this region. CD8 responses that are able to recognize different genotype-specific sequence variants were detected against three epitopes. It has been suggested that the existence of high-avidity CD8 T cells that are able to cross-react with different sequence variants is associated with control of viremia.21 Although the total number of such responses is too small to draw solid conclusions in this study, it is striking that CD8 T cells cross-reactive with both genotype-specific variants were only observed in HCV-RNA–negative subjects.
One epitope (HLA-B13–restricted) was particularly interesting because we identified three subjects with two coexisting T cell populations targeting the same region. One was directed against the GT1 variant, and a second was directed against the GT3 variant. This suggests exposure to both antigens and priming of two distinct T cell populations; however, it is unclear whether simultaneous exposure to both genotypes or two separate infections with different genotypes occurred. Although mixed infections with two HCV genotypes have been reported,22 they are overall rarely detected, even in high-risk groups,22, 23 arguing against simultaneous exposure to two HCV genotypes as a common event. It is unclear why coexistence of two T cell populations was only observed for the B13-epitope in our study. Interestingly, HLA-B13 is associated with delayed disease progression in human immunodeficiency virus, and this was mechanistically linked to T cell responses targeting HLA-B13–restricted epitopes in human immunodeficiency virus gag.24 The impact of HLA-B13 on disease outcome is unclear, as the low frequency of this allele in many populations may have precluded a solid analysis.
In a recent survey in local drug consumption facilities, 63% of drug users reported as being HCV-positive, and 22% reported needle sharing within the previous month.25 Although the risk profile was not specifically addressed in the present study, the previously reported risk behavior suggests that multiple exposures to HCV may occur. The comparison of genotype-specific T cell responses in the present study allows conclusions about exposure to different HCV genotypes. Immunological evidence for exposure to different genotypes was previously reported for CD426–28 and CD8 T cells.28, 29 These studies based their conclusions on the discrepancy between T cell responses and the genotype-specific sequence of the autologous virus in the targeted region. We similarly detected a mismatch between the genotype of the autologous virus and the CD8 response in 10 of 39 subjects with chronic infection (data not shown). In two cases, this was supported by serotyping results through the detection of serotype 1 in GT3-infected subjects. However, a discrepancy between serotype and genotype was detected in only four of 39 subjects with chronic HCV infection. One additional subject had a mixed GT1/GT3 serotype, and five subjects were not typeable (Table 1). It was suggested that the response against the autologous virus may be preferentially impaired, and responses against a previously resolved heterologous genotype persist.28 In line with this, there was a trend toward a stronger CD8 response against a heterologous genotype in the present study, although this trend was not statistically significant. Importantly, the strength of these putative memory responses in subjects with chronic HCV infection did not reach the same magnitude as has been observed in HCV-RNA–negative subjects. Of note, we detected only three epitopes in which the autologous sequence fully matched with the mapped response. In the majority of cases where the CD8 response corresponded to the genotype of the isolated virus, a divergent epitope sequence was encoded by the autologous virus. In the majority of these variants (5 of 8 [62.5%]), T cell recognition was decreased compared with the prototype, which is consistent with previous studies.30, 31
In the present study, we also identified five subjects with undetectable viremia and coexistence of distinct CD8 responses against each genotype, suggesting immune control of two different genotypes. Importantly, T cells active against both genotypes were more frequent in subjects with undetectable viremia (64.3% in HCV-RNA–negative subjects versus 15.3% in HCV-RNA–positive subjects). Although we underestimate the true number of CD8 responses as any epitope outside NS3 would have been missed, this raises the possibility that broad responses directed to different genotypes characterize a subgroup of individuals among injection drug users that are seemingly protected from chronic HCV infection. Longitudinal studies are needed to confirm this.
Genotype-specific sequence differences are important in the antiviral immune response against HCV. In the present study, we systematically analyzed the degree of cross-genotype reactivity of HCV-specific T cells at the epitope level. Although the majority of these cells show only limited cross-reactivity, we were able to identify several subjects with T cells active against both genotypes. Interestingly, T cells active against both genotypes were preferentially detected in HCV-RNA–negative subjects. This demonstrates that CD8 responses targeting different HCV genotypes can be primed in the same individual and that these responses are potentially linked to protection from chronic infection. In the face of a heterogeneous genotype distribution in many areas of the world, these findings have important implications for vaccine design.
We thank Anna Eckart for her excellent technical assistance and all patients for donating blood. We thank Christoph Neumann-Haefelin and Arthur Kim for kindly reviewing the manuscript.