Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Potential conflict of interest: Nothing to report.
Supported by NIH grants U19 AI088791, R01 DA013324, and R01 DA024565.
The contribution of humoral immune responses to spontaneous control of hepatitis C virus (HCV) infection remains unclear. We assessed neutralizing antibody (nAb) responses during acute HCV infection to determine whether infection outcome is associated with the nAb response, specifically, its timing or breadth (neutralization of multiple genotype-matched variants). A representative genotype 1 HCV pseudoparticle (HCVpp) library, consisting of 19 genetically distinct genotype 1 HCVpp that comprise the natural variability of genotype 1 E1E2 sequences, was used to assess anti-genotype 1 nAb responses during acute infection in at-risk persons followed prospectively. Neutralization of individual library HCVpp by the last viremic plasma sample obtained before clearance was compared to either 1-year post-initial viremia or clearance time-matched specimens obtained from subjects developing persistent infection. In persistently infected persons nAb responses were delayed then progressively broadened, whereas in persons who controlled viremia broader responses were detected early and contracted after clearance of viremia. Surprisingly, the breadth of anti-genotype 1 nAb responses was not dependent on subjects' infection genotype. Also, individual library HCVpp neutralization sensitivity was not associated with any known E2 sequence determinants. Interestingly, two single nucleotide polymorphisms in the HLA-DQ locus were associated with nAb breadth. Conclusion: Control of HCV infection is associated with more rapid development of a broad nAb response, independent of the infection viral genotype, providing further evidence for the role of nAb in controlling HCV infection and the potential benefit of generating broad anti-HCV nAb responses by vaccination. (Hepatology 2014;59:2140–2151)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
A hepatitis C virus (HCV) vaccine that prevents infection remains elusive. Identification of immunological mechanisms of spontaneous resolution of HCV infection may inform rational vaccination strategies. While a broad and potent cellular immune response has been demonstrated to play an important role in spontaneous control of HCV infection, the contribution of humoral immune responses to spontaneous control of HCV infection remains unclear. Prior experimental evidence suggests that humoral immunity, specifically neutralizing antibodies (nAb), contributes to an effective immune response to HCV infection. Control of HCV infection was associated with an increased incidence and potency of nAb responses during the first 6 months of infection and the presence of nAb has been shown to drive HCV envelope sequence evolution during acute infection.[2, 3]
One of the obstacles to investigating the role of nAbs in HCV infection outcome is that the unique global and intrapersonal sequence variability of HCV complicates selection of a representative sequence to use in studies. Seven genotypes of HCV have been identified with extensive inter- and intragenotype envelope amino acid sequence diversity (up to 50% and 20%, respectively). Additionally, each HCV infection comprises a quasispecies of HCV sequences consisting of genetically distinct yet related sequences driven by immune selection and restricted by fitness constraints.[3, 5] nAb studies using a single isolate may underestimate the incidence or magnitude of nAb responses due to this high level of sequence diversity.
We have begun to address this issue by constructing a novel library of HCV pseudoparticles (HCVpp) expressing 19 genetically distinct, genotype 1, HCV envelope glycoproteins derived from primary human isolates. This library models the natural envelope sequence variability of the most prevalent and incident subtypes of genotype 1 and was used to analyze the breadth (capacity to neutralize multiple isolates of a single genotype) of nAb responses in individuals with HCV infection. We used this model to study nAb responses in a cohort of persons with acute HCV infection. We demonstrated that earlier appearance of broader nAb responses was associated with spontaneous control of HCV infection compared to individuals who progress to persistent infection, independent of infection genotype. Interestingly, we found that the neutralization sensitivity of library HCVpp was not associated with any known E2 sequence determinants and two single nucleotide polymorphisms (SNPs) in the HLA-DQ locus were associated with nAb breadth. Taken together, these data demonstrate that control of HCV infection is associated with more rapid development of a broad nAb response, providing further evidence for the role of nAb in control of HCV infection.
Materials and Methods
Subjects were selected from the Baltimore Before and After Acute Study of Hepatitis (BBAASH) in which the incidence, immunology, and virology of HCV infection in young injection drug users is prospectively monitored, as previously described. The study was approved by the Institutional Review Board at the Johns Hopkins School of Medicine.
Determination of Infection Outcome
Subjects were assigned to Clearance and Persistence groups based on infection outcome. Clearance was defined as undetectable HCV RNA assessed by the COBAS Taqman reverse-transcription polymerase chain reaction (RT-PCR) quantitative assay for a period of at least 60 days in individuals with detectable anti-HCV antibodies in the setting of incident infection. Persistence was defined as viremia for more than 1 year with the same virus as determined by Core-E1 phylogenetic analysis. The midpoint between sample collection dates was used to estimate date of onset and resolution of viremia. Clearance and Persistence subject samples were selected using two sets of selection criteria: 1-year and time-matched.
Plasma samples at 365 days (± 60 days) postdetection of initial viremia were selected for analysis. For HCV infections that were controlled in less than 365 days, the last sample in which viremia was detectable (± 60 days) was selected for analysis.
Persistence and Clearance subjects were randomized and two Persistence subjects were matched with one Clearance subject based on rank order. For Clearance subjects, the last viremic specimen (viremia duration less than 365 days) or a specimen obtained at 365 days (± 60 days) post-initial viremia detection was selected for analysis of neutralization. Persistence subject visits were selected by matching viremia duration (± 60 days) to the Clearance subject.
Genotype 1 HCVpp Library
Source of E1E2 Genes
Plasma samples obtained from HCV-infected subjects in the BBAASH cohort, Irish Anti-D cohort, and Swan Project were used to construct a library of genotype 1 E1E2-expressing lentiviral pseudoparticles using a high-throughput production and screening approach.
HCV E1E2 Expression Plasmids
The E1E2 region was PCR-amplified from complementary DNA (cDNA) reverse transcribed from viral RNA purified from subject plasma and cloned into the expression vector pcDNA3.2/V5/Dest (Invitrogen) using Gateway technology in a one-tube BP/LR reaction. Details are described in the Supporting Methods.
High-Throughput HCVpp Production and Infectivity Screening
HCVpp were produced from purified cDNA clones by lipofectamine-mediated transfection of HCV E1E2 and pNL4-3.Luc.R-E- plasmids into HEK293T cells in a 96-well plate as described in the Supporting Methods.
Selection of HCVpp for Inclusion in Library
Infectious constructs were sequenced, aligned using ClustalX(11), and a neighbor-joining amino acid phylogenetic tree with pairwise deletion was constructed using Mega. Unique infectious sequences were identified based on phylogeny and the most ancestral sequence from each clade was selected for inclusion in the library. Protein sequence divergence was determined using the Protdist Phylip program included in the BioEdit software package. Novel E1E2 sequences have been submitted to GenBank, accession numbers KJ187972 through KJ187990.
Comparison of Library and Global E1E2 Sequence Diversity
Library HCVpp were produced by lipofectamine-mediated transfection of E1E2 expression plasmids in HEK-293T cells as described in the Supporting Methods. Heat-inactivated plasma samples were tested for neutralization of library HCVpp at a single dilution as previously described, detailed in the Supporting Methods. HCVpp were tested undiluted if infectivity, in the absence of antibodies, was less than 1.5 × 106 relative light units (RLUs) or diluted to less than 1.5 × 106 RLUs with media. Positive detection of neutralizing antibodies was noted when neutralization was ≥50%.
Analysis of Viral Determinants of Neutralization Sensitivity in Library HCVpp E2 Sequences
Predicted N-linked glycosylation sites for each library HCVpp were determined using NetNGlyc 1.0 (http://www.cbs.dtu.dk/services/NetNGlyc/). Monoclonal antibody epitopes were determined based on previous studies.[14-22] Epitopes that contain the same amino acid positions were counted as one (HC84.1 and HC84.2; HC84.22 and HC84.23; Fab e137 and Fab e20; AR3B and AR3D; and 95-2 and HCV1). An epitope was considered intact when the amino acid sequence matched the published sequence.
Genomic DNA was extracted from peripheral blood mononuclear cells using the QIAamp DNA Blood Mini Kit (Qiagen Sciences, Germantown, MD). SNPs were genotyped using custom TaqMan SNP Genotyping Assays (Life Technologies, Carlsbad, CA).
Wilcoxon rank-sum test, Wilcoxon signed rank-sum test, Kruskal-Wallis one-way analysis of variance (ANOVA) on Ranks and Fisher's Exact test (where appropriate, based on descriptive statistics) were used to evaluate statistically significant differences between groups.
Construction of a Representative Genotype 1 HCVpp Library
A total of 156 infectious HCVpp expressing unique E1E2 proteins (113 subtype 1a, 43 subtype 1b) derived from 33 different subjects (21 subtype 1a, 12 subtype 1b) were identified using a high-throughput production and screening approach. Phylogenetic analysis identified 12 distinct subtype 1a and 7 subtype 1b clades (Fig. 1A). Amino acid distance analysis revealed median (IQR) distances of 0.14 (0.12-0.18) for ancestral subtype 1a HCVpp E1E2 sequences and 0.17 (0.16-0.19) for ancestral subtype 1b HCVpp E1E2 sequences included in the library.
In order to assess how accurately our HCVpp library represents genotype 1, we compared the E1E2 sequence diversity of the library to 374 subtype 1a and 260 subtype 1b reference sequences. The diversity profile of subtype 1a (Fig. 1B, n = 12) and subtype 1b (Fig. 1C, n = 7) HCVpp E1E2 sequences was strikingly similar to the diversity of reference sequences. Additionally, the divergence between subtype 1a and 1b reference sequences was equivalent to the divergence between subtype 1a and 1b library sequences (Fig. 1D). These data strongly suggest that our genotype 1 HCVpp library accurately represents the natural diversity of E1E2 sequences.
Appearance of Broad nAb Responses Is Delayed in Subjects Who Fail to Control HCV Infection
Initially, 31 subjects who developed persistent infection and 21 subjects who controlled infection were selected for analysis of nAb breadth in specimens collected during acute infection. Age, sex, race, follow-up interval, HCV infection genotype, and IFNL3-related rs12979860 allele frequencies were similar between these Persistence and Clearance groups (Table 1). The duration of infection in Clearance subjects was significantly shorter than Persistence subjects, since the infection duration in eight Clearance subjects was less than 230 days.
Table 1. Demographic and HCV Infection Data of Clearance Subjects Versus Persistence Subjects of One-Year and Time-Matched Visits
In order to examine whether the incidence and breadth of anti-genotype 1 nAb responses during acute infection is associated with infection outcome, we first compared neutralization of library HCVpp by Persistence subject plasma samples obtained 1 year following initial detection to Clearance subject plasma samples obtained 1 year following initial detection of viremia or the last viremic sample (infection duration <1 year). Neutralization of at least one HCVpp was observed in 76.2% of Clearance subjects and 80.6% of Persistence subjects (Fig. 2A; P = 0.74). The median breadth of nAb responses at 12 months of infection or last viremia (Fig. 2B) in the Clearance group was similar to the median breadth of nAb responses at 12 months of infection in the Persistence group (P = 0.757). However, in this analysis there was a 2-fold lower duration of infection in the Clearance group, because more than 50% of Clearance subjects experienced less than 6 months of viremia.
In order to control for the effect of infection duration, nAb breadth was also assessed in Clearance subjects at last viremia or 1 year of infection compared to time-matched Persistence subjects. Therefore, 21 Clearance subjects were matched with 42 Persistence subjects and samples from each of these Persistence subjects were time-matched based on infection duration (Table 1). As intended, the infection duration of Clearance and time-matched Persistence subjects was not different.
In this time-controlled analysis, there was a trend for a greater percentage of Clearance subjects neutralizing at least one HCVpp (76.2% versus 52.4%, P = 0.101; Fig. 3A). In addition, the breadth of nAb responses was greater in Clearance subjects than in time-matched Persistence subjects (P = 0.007; Fig. 3B). Taken together, these results suggest that the appearance of broad nAb responses is delayed in subjects who fail to control HCV infection. Endpoint plasma titers of nAb were similar between subjects with Clearance and Persistence in a subset of subjects representing the range of neutralization breadth (Supporting Fig. 3).
Overall, 60.3% of plasma samples neutralized at least one HCVpp in the genotype 1 HCVpp library while the subtype 1a and subtype 1b HCVpp with the highest neutralization sensitivity was neutralized by only 41.3% and 39.7% of subjects, respectively, (subtype 1a, P = 0.05, subtype 1b, P = 0.03). These data indicate that the use of an HCVpp library for screening for nAb responses during acute infection is a more sensitive approach than screening with a single HCVpp.
Infection Genotype Does Not Affect Neutralization Breadth
To assess the influence of infection genotype on the breadth of nAb responses, neutralization data from the 1-year and time-matched analysis was stratified by infection genotype, regardless of infection outcome, and breadth of nAb responses was compared between infection genotypes. The breadth of nAb responses was similar in both 1-year (Fig. 2C) and time-matched subjects (Fig. 3C) infected with genotype 1, 2, or 3 viruses (1-year, P = 0.452; time-matched, P = 0.778). These data indicate that the breadth of anti-genotype 1 nAb responses is not influenced by infection genotype, and provide further evidence that HCV genotypes do not represent serotypes.
Stability of nAb Responses During Persistent Infection and Following Viremia Control
The breadth of nAb responses in Persistence subjects over time was compared in subjects who were included in both the 1-year analysis and time-matched analysis in order to assess whether nAb responses broadened over time in Persistence subjects during acute infection. The breadth of nAb responses in the 1-year analysis samples was significantly greater (P < 0.001) than the breadth of nAb responses in the time-matched analysis samples obtained 98 to 306 days earlier from the same subjects (Fig. 4A). Conversely, we compared the breadth of nAb responses in Clearance subjects at last viremia with samples obtained from these same subjects 1 year following initial detection of viremia. The breadth of nAb responses in the last viremic samples was significantly greater (P = 0.03) than the breadth of nAb responses in the 1-year samples obtained 97 to 279 days later from the same subjects (Fig. 4B). These data suggest that the breadth of nAb responses declines following control of viremia and increases during persistent infection.
Variation at Viral Sequence Determinants Is Not Associated With Neutralization Sensitivity
To explore mechanisms underlying variation in sensitivity to neutralization by plasma among HCVpp, we studied viral sequence determinants with respect to neutralization sensitivity. Since the degree of glycosylation of E2 may affect neutralization sensitivity, we examined whether the number of predicted N-linked glycosylation sites in the library HCVpp was associated with neutralization sensitivity. The number of N-linked glycosylation sites predicted was 11 in 83% of subtype 1a, and 10 in 86% of subtype 1b, HCVpp envelope sequences (Fig. 5A). There were two subtype 1a HCVpp envelope sequences in which only 10 N-linked glycosylation sites were predicted; similarly, there was one genotype 1b HCVpp envelope sequence with only 9 predicted N-linked glycosylation sites. These data suggest that natural variation in the number of N-linked glycosylation sites does not appear to have a major impact on neutralization sensitivity.
Specific amino acid positions in E2 have been shown to be important in HCV-CD81 interactions and disruption of this interaction is a mechanism of neutralization. In the current study, variant CD81 binding residues were observed in 67% of subtype 1a HCVpp E1E2 sequences (Fig. 5B). While the CD81 binding residues appear to be more variable in the HCVpp with the lowest neutralization sensitivity, variability at CD81 binding residues were also observed in the HCVpp with the highest neutralization sensitivity. Conversely, variability at CD81 binding residues were observed in 86% of subtype 1b sequences across the spectrum of neutralization sensitivities (Fig. 5C). These results suggest that the neutralization sensitivity is not strongly associated with natural variation in canonical CD81 binding residues.
Broadly neutralizing human anti-HCV mAbs have been isolated by multiple laboratories and the epitopes targeted by these mAbs have been mapped using alanine scanning.[15, 17-22, 24] Since mutation of amino acids in these epitopes may result in escape from nAbs, we investigated whether the presence of intact epitopes in our HCVpp sequences was associated with neutralization sensitivity. The amino acid sequences of each HCVpp at the described epitope positions are presented in Supporting Fig. 1. For all subtype 1a HCVpp, the number of epitopes in which the amino acid sequence was different from the published epitope sequence was not significantly correlated with neutralization sensitivity (R2 = 0.06, P = 0.5; Fig. 5D,E). These results suggest that the observed natural (rather than alanine-mutagenesis) variation in known mAb epitopes is not associated with decreased neutralization sensitivity.
HLA Class II SNPs Are Associated With nAb Breadth
A previous genome-wide association study identified a number of SNPs that are associated with HCV infection outcome. Seven of these SNPs are located on chromosome 6, upstream of the HLA-DQB1 and HLA-DQA2 genes. Because one of these genes encodes a protein in the HLA-DQ heterodimer complex that is important in antigen presentation to CD4+ T helper cells, we investigated whether these SNPs were associated with differential breadth of nAb responses. All subjects included in the time-matched analysis were genotyped for each of these seven SNPs and nAb breadth was compared between SNP genotypes, regardless of outcome. The breadth of nAb responses was greater in subjects having the rs2395522 A allele (Fig. 6A) or the rs9275224 G allele (Fig. 6B). These two SNPs are genetically linked (D′ = 1.0, R2 = 1.0). No differences in nAb breadth were detected between genotypes of the remaining five SNPs (Supporting Fig. 2). These results suggest that HLA class II variation influences breadth of nAb responses during HCV infection.
Using a novel, representative library of genotype 1 HCV envelopes and a unique prospective cohort of persons infected with HCV, we performed the first complete assessment of HCV nAb breadth and control of HCV viremia in humans. In persons who cleared viremia, nAb responses were detected earlier and in greater breadth than in persons whose viremia persisted. In contrast, nAb responses contracted after viremia control while incrementally broadened in persons with persistent infection. These results provide further evidence for a role of nAb in control of HCV infection.
Previous studies have demonstrated that HCV infected individuals are able to generate cross-reactive nAb that target genotypes other than the infecting virus.[8, 26] Building on these studies, we show that the breadth of anti-genotype 1 nAb responses during acute infection was independent of the infection genotype. Therefore, nAb responses in subjects infected with genotype 1, 2, or 3 viruses, and possibly other genotypes, can be assessed using a genotype 1 HCVpp library. These results have implications for rational anti-HCV vaccine design, because the genotype of the E1E2 sequence selected for inclusion in the vaccine may not be important for generating antibodies with relevant neutralization activity.
Frequent sampling of study subjects allowed us to demonstrate that the similar breadth of nAb in the 1-year analysis was because persons with persistent infection continued to mount new nAb responses. The differences between 1-year and time-matched results can be explained by improved matching in the latter analysis, and reciprocal changes in nAb responses in the two groups. nAb responses rapidly waned after control of viremia, while the breadth of nAb increased during the first year in those with Persistence. In contrast, as previously reported using a subset of the same subjects, new T-cell responses are rarely detected more than 6 months after initial viremia. These kinetic changes agree with prior analyses of common-source outbreaks of HCV,[1, 28] validating our HCVpp panel approach and highlighting the importance of accurately identifying infection start and control dates when studying anti-HCV nAb responses.
Interestingly, the percentage of subjects who neutralized at least one heterologous HCVpp was similar in the Persistence and Clearance groups, while the breadth of nAb responses was associated with infection outcome. This is in agreement with previous reports demonstrating the neutralization of a single heterologous HCV isolate during acute infection was not associated with outcome.[29, 30] Therefore, the breadth but not simple presence of a heterologous nAb response may be an important immunological correlate of HCV infection control.
Previous studies have demonstrated that use of an autologous HCVpp results in greater detection of nAb responses than the use of a heterologous HCVpp.[1, 3, 31] However, the use of autologous HCVpp in studies examining nAb responses in the context of heterogeneous infection scenarios, such as intravenous drug use, is more challenging because an autologous HCVpp must be produced for each subject. We hypothesized that it would be feasible to study nAb responses during acute, non-single-source HCV infection in large subject populations by using an HCVpp library that encompasses the genetic variability of E1E2 observed in nature. This study demonstrates that screening for nAb responses using a HCVpp library is more sensitive than using a single HCVpp, because detection of neutralization was observed in ∼50% more subjects when screened using the panel than with the subtype 1a or 1b HCVpp with the highest neutralization sensitivity.
Using an HCVpp library that represents the natural diversity of genotype 1 E1E2 sequences reveals E1E2 sequence determinants that affect neutralization sensitivity. Variation in neutralization sensitivity was not explained by variation in predicted glycosylation state of library HCVpp, potentially reflecting a vital role for glycans in entry and maintenance of protein structure. Disruption of the interaction of HCV with any of the known entry factors is a possible neutralization mechanism for anti-HCV antibodies. CD81, a tetraspanin transmembrane protein, has been shown to facilitate HCV entry through an interaction between E2[33, 34] and the large extracellular loop of CD81, so variability of amino acids at these positions may alter neutralization sensitivity. While CD81 binding residues varied in some of the subtype 1a HCVpp E1E2 sequences with low neutralization sensitivity, these amino acids were essentially invariant in subtype 1b HCVpp across neutralization sensitivities.
Using hybridoma and antibody phage display technologies, human anti-HCV mAb that target conformational epitopes and neutralize HCV isolates of multiple genotypes have been isolated and characterized.[14-22] We found that natural variability in our genotype 1 HCVpp library at amino acid sites in these conformational epitopes was not correlated with sensitivity to neutralization by plasma. Possible explanations for this finding include incomplete mapping of these epitopes, differential targeting of epitopes during persistent and cleared infection, and infrequent targeting of these epitopes during acute infection. Future mAb-subject plasma competition analyses may help address these discrepancies.
Interestingly, the breadth of nAb responses was significantly associated with the presence of specific alleles of two SNPs located upstream of the HLA-DQB1 and HLA-DQA2 genes, outside of the protein coding regions. While the rs2395522 SNP has not been associated with any other disease outcome, the rs9275224 SNP has been associated with the development of systemic sclerosis and rheumatoid arthritis. These SNPs are located outside of the protein coding regions, so it is possible that they may be affecting regulation of gene transcription, or are linked to a mechanism in some other way. Since HLA class II proteins are necessary for CD4+ helper T-cell activation, it is possible that altered expression of these genes may contribute to increased nAb breadth. Further study is needed to clarify the mechanism by which these SNPs affect nAb responses.
Collectively, these results provide clear evidence for nAb responses in control of acute HCV infection, and identify early and broad nAb responses as an immunological correlate of protection against persistent HCV infection. This study also highlights the importance of controlling for duration of infection as well as the utility of using representative HCVpp libraries in assessing nAb responses during acute HCV infection. That control of viremia was associated with an earlier appearance of broader nAb responses and that nAb breadth was independent of infection genotype highlights the relevance of targeting envelope proteins in vaccination strategies to combat HCV-associated mortality and morbidity.
The authors thank Barbara Rehermann and Elizabeth Kenny-Walsh for providing samples for inclusion in the HCVpp library and the BBAASH study subjects and coordinators.
Role of Authors
W.O., Study concept and design; acquisition of data; analysis and interpretation of data; drafting of the article; A.S., B.W., R.L., B.F., Acquisition of data; J.B., Study concept and design; D.T., A.C., Critical revision of the article for important intellectual content, obtained funding; S.R., Obtained funding; study concept and design; critical revision of the article for important intellectual content.