Anti‐envelope antibody responses in individuals at high risk of hepatitis C virus who resist infection

Summary Injection drug users uninfected by hepatitis C virus (HCV) despite likely repeated exposure through high‐risk behaviour are well documented. Factors preventing infection in these individuals are incompletely understood. Here, we looked for anti‐HCV‐envelope antibody responses in a cohort of repeatedly exposed but uninfected subjects. Forty‐two hepatitis C diagnostic antibody‐ and RNA‐negative injection drug users at high risk of exposure were studied and findings compared to healthy controls and cases with chronic HCV infection. Purified IgGs from sera were tested by ELISA for binding to genotype 1a and 3a envelope glycoproteins E1E2 with further testing for IgG and IgM reactivity against soluble E2. Virus‐neutralizing activity was assessed using an HCV pseudoparticle system. Uninfected subjects demonstrated significantly greater IgG and IgM reactivities to envelope glycoproteins than healthy controls with IgG from 6 individuals additionally showing significant neutralization. This study is the first to describe humoral immunological responses targeting the HCV envelope, important for viral neutralization, in exposed uninfected individuals. A subset of these cases also had evidence of viral neutralization via anti‐envelope antibodies. In addition to confirming viral exposure, the presence of specific anti‐envelope antibodies may be a factor that helps these individuals resist HCV infection.


| INTRODUCTION
Hepatitis C virus (HCV) is a major cause of liver morbidity and mortality worldwide. 1 Approximately 75% of infected individuals proceed to chronic infection. 2 In developed countries, the major route of transmission remains the use of injection drugs and sharing of related paraphernalia by injection drug users (IDUs). 3 Symptomatic acute infection is rare, and HCV has the potential to spread undetected within these populations. While effective antiviral medication is now available, 4 treatment remains costly and ineffectively implemented with the prevalence of HCV in IDUs rising in England. 5 It is clear that global eradication of HCV will not be possible unless robust preventative strategies are also developed. 6 Studying individuals who resist infection can help inform prevention strategies and vaccine design.
In natural infection, successful clearance of HCV infection has been associated with the activation of elements of both the adaptive and the innate immune system (e.g. HCV-specific B and T cell, NK cell, dendritic cell and interferon responses [7][8][9][10] ). There is growing evidence that a robust antibody response targeting HCV virion *Both authors contributed equally to the work and are joint first authors.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. envelope glycoproteins (E1 and E2) responsible for virus entry into host cells can contribute significantly to viral clearance in acute infection. [11][12][13][14][15][16] Rapid onset of anti-HCV envelope antibodies 17 capable of neutralizing diverse strains of HCV is associated with acute clearance. Furthermore, broadly neutralizing antibodies may contribute to resolution of infection even once HCV has become established. 11,18 In the search for a prophylactic vaccine, it is especially relevant to study individuals who appear to have natural resistance to HCV infection. Individuals who regularly inject drugs and share injecting equipment are at very high risk of HCV exposure with seroprevalence rates of up to 90% reported in long-term users.
Resistance to HCV infection is increasingly well documented in IDUs who remain uninfected despite a long history of unsafe injection practices. 19 Individuals at high risk of exposure with no evidence of past or current infection have been termed exposed but uninfected (EU).
EU cohorts are immunologically distinct from both healthy controls and those who spontaneously clear HCV infection (see Table 1). [20][21][22] Up to 60% of EU individuals display HCV-specific adaptive T-cell responses. Furthermore, they display enhanced NK cell, IL6/IL8 and TNF-α activity compared to HCV-infected IDUs. 23,24 They are also genetically distinct from spontaneous resolvers and those with chronic HCV 9 with the combination of KIR2DL3, HLA-C1 over represented in both the EU and spontaneous resolver groups while the prevalence of the IL28B polymorphism is similar in EU to chronically infected cohorts. 22,[24][25][26] While this EU cohort is defined by the lack of antibody response to HCV core and nonstructural proteins in a diagnostic assay, the role that anti-envelope neutralizing antibodies might play in their protection from infection has not yet been explored. Given the upregulation of neutralizing antibodies in those who acutely clear HCV infection on multiple exposures, it is possible that such antibodies may have a significant protective effect in the IDU EU population. 12 Therefore, we aimed to determine the presence of functional anti-HCV envelope antibody responses in a cohort of IDUs who remain uninfected despite their high risk of repeated exposure to HCV. A cohort of current IDU not known to have HCV infection or other   blood borne viruses was recruited between 2003 and 2014 from a variety of locations in Plymouth, UK, as previously described. 21,22 All individuals completed a confidential structured questionnaire to collect demographic data and a detailed injecting history. This included age at first injection, duration of injecting behaviour, frequency of injecting episodes, current injecting behaviour, frequency of sharing intravenous paraphernalia (needles, syringes, filters, spoons and water), frequency of sharing with a contact known to have HCV infection and risk of non-IDU HCV exposure. For this study, we included only those judged to be at substantial risk of HCV exposure based on a >1 year history of injecting drug use and regular sharing of needles and related paraphernalia. All individuals had serum and whole blood samples drawn for laboratory analysis and clinical details entered in respective databases. Where possible, follow-up clinical information and clinical samples from recruited individuals were obtained.

| TheEUandcontrolcohorts
The exposed group was screened for the evidence of current or previous HCV infection using diagnostic tests for antibodies to core and nonstructural proteins (third-generation ELISA by Abbot IMx) and HCV RNA by qualitative PCR (Amplicor, Roche Diagnostics). Those individuals negative on both tests were termed exposed uninfected (EU) and were included in the study. Two individuals who were found to have developed HCV infection on subsequent testing were excluded. In addition, we separately recruited positive and negative control cohorts of individuals with chronic HCV infection (CHCV) and non-IDU healthy controls (HC) with no liver disease or history of HCV infection, respectively. The full details of these cohorts have been described elsewhere. 27 Study protocols conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by regional ethical committees. Informed written consent was obtained from each subject prior to entry in the study.
T A B L E 1 Immunological characteristics of exposed uninfected individuals KIR2DL3 genotype Increased frequency of KIR2DL3:HLA-C1 homozygosity Chronically infected individuals [26] No difference in frequency of KIR2DL3:HLA-C1 homozygosity Pre-infection samples from IDUs with subsequent seroconversion [25] IL12B CC allele (rs3213113) Increased frequency compared to healthy controls Healthy controls [22] Serum cytokine levels Elevated IL6, IL8 and TNF-α Healthy controls, spontaneous resolvers and chronically infected individuals [24] HCV-specific T cells Evidence of IFN-γ production and T-cell proliferation Healthy controls and chronically infected individuals [21,41] Natural Killer Cells Enhanced IL-2-mediated cytotoxicity. Increased NKp30 expression Pre-infection samples from IDUs with subsequent seroconversion [23] | 875

| GenerationofHCVpseudoparticle(HCVpp)and E1E2Lysate
As genotypes 1 (gt 1) and 3 (gt 3) account for the vast majority of HCV infections in the UK, envelope glycoprotein sequences from a standard HCV gt 1a and a UK derived gt 3 strain were used. HCV pseudoparticles bearing envelope proteins from gt 1a (strain H77c, accession number AF011751.1) and gt 3a (sequence UKN3a1.28 F4/2-35; closely related to accession number AY734984.1) were generated in HEK-293T cells as described previously. 28 Further details are available in Supplementary Methods. Following harvesting of pseudoparticles, cells were lysed in 1 mL of lysis buffer and centrifuged and the supernatant used for E1E2 ELISA assays as described below.

| GNAcaptureE1E2ELISA
IgG was purified from EU, CHCV and HC samples using Protein G IgG-specific spin columns (Thermo Scientific, UK; see Supplementary Methods) including 3 wash steps prior to elution ensuring removal of nonspecific serum factors that may interfere with the assay. Purified IgG was tested in an ELISA assay to detect antibodies to the E1 and E2 glycoproteins contained in the HEK-293T lysate supernatant. These ELISAs were performed as described previously with further detail in Supplementary Methods. 29 Absorbance readings were normalized to a multiple of the mean of the HC readings. Significant binding by ELISA was determined as absorbance values ≥2 times HC mean.
Reactivity to both gt 1a and gt 3a E1E2 was tested.

| Solublegt1aE2bindingELISAs:IgGandIgM
As IgM is difficult to purify, diluted serum was used to detect binding of IgG and IgM to purified HCV soluble gt 1a E2 (sE2). EU and control serum samples were diluted 1:50 in PBSTM and tested for binding to sE2 by ELISA. Absorbance readings were normalized to healthy control mean and analysed as above with further detail in the Supplementary Methods.

| Pseudoparticleneutralizationassays
In the subset of EU individuals displaying significantly elevated E1E2 binding compared to HC on ELISA, pseudoparticle (pp) neutralization assays were conducted as previously described. 30 Briefly, purified subject IgG was added to 40 μL of HCVpp-containing medium prepared as above. Purified IgG was added at a concentration of 400 μg mL −1 for screening of the EU cohort, for some individuals where serum was scarce, this was reduced to 200 μg mL −1 . The mouse monoclonal anti-E2 antibody AP33 31 and HC IgG were included as positive and negative controls, respectively. After incubation for 1 hour, the IgG-HCVpp mixture was added to a 96-well plate preseeded with Huh-7 cells. Following incubation for 3 hour, the inoculum was replaced with fresh medium, and after a further 72-hour incubation, luciferase activity in infected cells was detected as a marker of HCVpp infectivity using the Bright-Glo Luciferase kit (Promega, UK). Virus neutralization was defined as 50% reduction in pseudoparticle infectivity as measured in relative light units (RLU) using a Chameleon II plate reader.
Ability to neutralize both gt 1a and gt 3 was tested. For those samples with apparent neutralizing activity, further neutralization assays were conducted using serial dilutions of subject IgG.

| E2competitionELISA
For EUs with neutralizing activity and sufficient IgG available, competition ELISAs were performed using a panel of well-characterized monoclonal antibodies targeting known CD81-binding epitopes (Table S1). CHCV samples with known neutralizing activity were used as positive controls.
Purified IgG was incubated on E2-coated Immulon II plates at a concentration of 200 μg mL −1 in PBSTM. Subsequently, biotinylated antibodies to known conformational epitopes were added to these wells in addition to control wells with no EU IgG. Streptavidin-HRP was added and binding measured as previously. The reduction in relative binding of each biotinylated antibody (calculated as percentage reduction in absorbance) on addition of EU IgG compared to control was determined. PBSTM and neutralizing IgG from the CHCV cohort were used as negative and positive controls, respectively.

| Clinicalcohorts
Forty-two EU subjects, 8 gt 1 chronically infected patients (CHCV) and 8 healthy controls (HC) were studied. By definition, all EU cases consistently tested negative for HCV RNA and anti-HCV antibodies using standard diagnostic tests as described. In 5 EU individuals, serial samples were tested. Demographics of the cohorts are detailed in Table 2.

| ReactivityofIgGtoHCVenvelope
The median absorbance readings for the HCV gt 1a E1E2 capture ELISA were significantly higher in the EU cohort than in the HC group (P=.038) (Fig. 1a). While there was a trend towards overall higher absorbance readings against gt 3a E1E2 in the EU group than in controls, this did not reach significance (P=.067, Fig. 1b). However, three samples showed levels of binding to gt 3a lysate comparable to chronically infected controls.

| HCVexposureriskandE1E2reactivity
EU IgG absorbance levels to gt 1a E1E2 and gt 3a E1E2 were analysed for association with reported injection behaviour with a significant correlation between IgG reactivity against gt 1a E1E2 and greater lifetime injecting episodes (Spearman R=.38, P=.02, Fig. 1c), although this association was not seen with gt 3a reactivity (Fig. 1d).

| IgGandIgMsE2ELISA
To eliminate reactivity against HEK antigens and medium components, diluted serum samples from all individuals were tested for IgG reactivity to gt 1a sE2. For two of the subjects, serum from 2 separate time points was tested. There was significantly higher absorbance of IgG from EU to sE2 than HC (P<.01, Fig. 2a). Serum was also tested for IgM binding to sE2. The EU cohort showed significantly higher quantities of IgM binding to sE2 than the HC group (P=.001 Wilcoxon rank sum, Fig. 2b) with several individuals displaying levels comparable to the CHCV group (data not shown).

| Positivityacrossmultipleassays
We set a cut-off of absorbance ≥2 times the HC mean value as indicat- displaying absorbance readings ≥2 times the HC mean (Fig. 2a, Table S2) with 2 more showing above average binding to sE2 but below this cutoff value. Seven further individuals with significant responses against sE2 showed elevated reactivity against gt 1a lysate, but the absorbance attained did not reach the cut-off value. Only four samples showed significant binding to sE2 alone. It should be noted that for the vast majority of assays the value of 2 times the HC mean was higher (therefore a more stringent cut-off) than HC mean + 2 times standard deviations.

| Competitionwithantibodiesto knownepitopes
IgG from EU individuals with evidence of neutralizing activity was tested in a competition assay with a small panel of well-characterized conformation-sensitive anti-E2 antibodies. Three of these (HC1, HC11 and CBH7) recognize amino acid residues/regions that are critical for the interaction of E2 to CD81, the host entry factor essential for virus entry into target cells [13][14][15][16] (Table S1). Although chronically infected samples with neutralizing activity tended to compete with antibodies to these regions, we did not see any evidence of significant competition in the EU samples (Fig. 4). While there was evidence of binding to pure E2 in these samples at >2× control samples (Fig. 2a), it was considerably weaker than in individuals with chronic infection which bound to viral glycoproteins at >20 times the strength of HC samples (data not shown).

| Durationofantibodydetectionovertime
For 5 EU individuals, serial samples from time points separated by a period of at least 1 year were studied. Their E1E2 reactivity values relative to HC samples were consistent over the time periods studied, either remaining at HC levels or being consistently elevated (Fig. S1).
In the one individual with significant E1E2 and neutralizing responses studied serially, these responses remained detectable over time but with a diminution in strength at later time points.

| DISCUSSION
Long-term injection drug users who recurrently, persistently and frequently share injection equipment but who remain HCV antibody negative with undetectable HCV RNA represent an optimal cohort for advancing knowledge and understanding of mechanisms of natural protection from HCV infection. These individuals appear resistant to infection. Unlike HIV where resistance may be conferred by mutations in host entry proteins, no such variants have been identified in HCV exposed uninfected cases. 32-34 Therefore, it is likely that these subjects resist infection through immune-mediated mechanisms.
This study is the first to report evidence of humoral immunological responses targeting the HCV envelope, potentially a key area for viral neutralization, in exposed uninfected individuals. In those who resolve acute infection, detectable neutralizing antibody responses are seen to reduce over a period of months to years F I G U R E 4 EU IgG fails to compete with conformational antibodies to known epitopes on gt 1a E2. IgGs from individuals with neutralizing ability in the HCV pp system were selected for testing by competition ELISA to determine competitive binding with monoclonal antibodies to known conformational epitopes on E2 (Table S1). Samples from chronically infected individuals (CHCV) with known neutralizing activity were also included as positive controls. Percentage reduction in absorbance of the monoclonal antibodies was calculated and plotted. Significant competition would be expected at a level of 50% inhibition | 879 following viral clearance although they may be restored in subsequent episodes of infection to aid more rapid viral control. 12 The presence of potentially protective humoral responses in our cohort is seen most strikingly in cases with the greatest likelihood and frequency of HCV  40 it is also conceivable that the EU antibody responses are directed at CD81 binding regions in E2 but do not bind with sufficient avidity to avoid being 'competed off' by the conformational antibodies derived from chronically infected individuals. It is also possible that these individuals, in common with most acutely infected cases, develop antibodies which predominantly target the Highly Variable Region-1 region of the E2 protein or regions of E1 essential for entry. An alternative explanation may be that due to high levels of diversity in E2 amino acid sequences, these individuals may raise antibodies to a local E2 sequence which only provides weak cross-reactivity with the H77 sequence at CD81 binding epitopes, and future work should aim to explore the breadth of E1E2 reactivity within these individuals.
In conclusion, this is the first report of HCV envelope-specific humoral immune responses in a cohort of exposed uninfected injection drug users who remain HCV PCR and EIA negative despite repeated risk of exposure. In addition to adaptive humoral immune responses to envelope proteins, some individuals produce neutralizing anti-envelope antibodies which may contribute to host immunity.
These neutralizing antibodies do not appear to compete with commonly targeted CD81 binding sites and may therefore recognize novel epitopes. Our data complement previous reports of HCV-specific T-cell responses 9,26 and upregulated innate immune responses in exposed uninfected cases. 25 Together, these studies provide robust evidence that such individuals have been exposed to HCV, but are resistant to developing established infection. Whether any one of these responses is central to resisting infection, or whether a combination of upregulated innate immunity together with adaptive T-and B-cell responses is needed is not yet known. Further study of the antibody responses present in EU individuals will enhance our understanding of the role of antibodies in HCV infection and inform future preventative vaccine design.