Reverse epitope mapping of the E2 glycoprotein in antibody associated hepatitis C virus

Abstract

The humoral immune system responds to chronic hepatitis C virus (HCV) infection by producing neutralising antibodies (nAb). In this study we generated three HCV pseudoparticles in which E1E2 glycoprotein sequence was targeted by the host humoral immune system. We used patient derived virus free Fabs (VF-Fabs) obtained from HCV genotype 1a (n = 3), genotype 1b (n = 7) and genotype 3a (n = 1) for neutralisation of HCVpp produced in this study both individually and in combination. Based on the available anti-HCV monoclonal nAb mapping information we selected amino acid region 384-619 for conformational epitope mapping. Amongst our notable findings, we observed significant reduction in HCVpp infectivity (p<0.05) when challenged with a combination of inter genotype and subtype VF-Fabs. We also identified five binding motifs targeted by patient derived VF-Fab upon peptide mapping, of which two shared the residues with previously reported epitopes. One epitope lies within an immunodominant HVR1 and two were novel. In summary, we used a reverse epitope mapping strategy to identify preferred epitopes by the host humoral immune system. Additionally, we have combined different VF-Fabs to further reduce the HCVpp infectivity. Our data indicates that combining the antigen specificity of antibodies may be a useful strategy to reduce (in-vitro) infectivity.

Fig 1. Varying degree of infectivity of E1E2 glycoproteins derived from AAV HCVpp.

HCVpp generated from E1E2 associated with AAV showed varying degree of infectivity. HCVpp generated from phCMV-ΔC/E1/E2 H77 were used as reference HCVpp. No envelope control reproducibly gave RLU values less than 100, therefore a cut-off of 1000 RLU was used to determine the infectivity of a clone (***p<0.0001, one way ANOVA, Dunnett’s posthoc test) The X axis depicts infectious HCVpp clones (shaded boxes, n = 3/6) in this study. Y axis denotes infectivity in relative light units (RLU). GT: genotype

Fig 2. Mutation in the E1 glycoprotein affects the infectivity of HCVpp.

Six mutant clones were generated for highly infectious HCVpp1b-1-2 as follows, 1b-1-2_I388V, 1b-1-2_V395A, _T292A1b-1-2_{I388V, T292A}1b-1-2_V395A, 1b-1-2_DM and _T292A1b-1-2_DM (DM corresponds to double mutant I₃₈₈V and V₃₉₅A). Wild type clones HCVpp1b-1-2 and HCVpp1b-1-3 were used as controls. As a reference infectivity value of HCVpp1b-1-2 was set to 1. Mutation in the E1 glycoprotein reduces the infectivity of HCVpp1b-1-2. Pairwise mutation at position 292 and 388 (_T292A1b-1-2_I388V) and _T292A1b-1-2_DM corresponds to wild type HCVpp1b-1-3 (*p<0.0001, one way ANOVA, Dunnett’s posthoc test). The X axis depicts infectious mutant HCVpp clones in this study. Y axis denotes % infectivity.

Fig 3. Patient derived VF-Fab display varying degree of neutralising ability to different E1E2 HCV pseudoparticles.

VF-Fab were purified from antibody-virus complex. VF-Fab1a-1-1, VF-Fab1a-1-2, VF-Fab 1a-1-3 were purified from a patient infected with HCV genotype 1a. VF-Fab1b-1-2, VF-Fab1b-1-3 and VF-Fab1b-10-1 were purified from a patients infected with genotype 1b whose common source of infection was HCV infected anti D immunoglobulin [25]. VF-Fab1b-5-1 were purified from a blood transfusion patient infected with genotype 1b. VF-Fab3a-1-1 was purified from a patient infected with HCV genotype 3a (Table 1). HCVpp incorporating E1E2 derived from genotype 1a (HCVpp1a-1-3, HCVppH77), 1b (HCVpp1b-1-2 and HCVpp1b-1-3) were pre-incubated with different concentrations (0.006 to 0.4 mg/ml) of purified VF-Fabs prior to infection of Huh7 cells. No envelope control was used to normalise the data. The neutralising activity of the VF-Fab is expressed as percentage of inhibition of the infectious titres. VF-Fab1b-5-1 bottom values were constraint to zero for that data set only. Each experiment was repeated three times. IC₅₀ for each VF-Fab is detailed in Table 3. Error bars indicate standard deviation.

Fig 4. VF-Fab from sera without detectable AAV shows neutralisation activity.

VF-Fab1b-4-1, VF-Fab1b-6-1, VF-Fab1b-8-1 were purified from three unrelated patients infected with HCV genotype 1b that were negative for presence of AAV. HCVpp incorporating E1E2 derived from genotype 1a (HCVpp1b-1-3, HCVppH77) and 1b (HCVpp1b-1-2, HCVpp1b-1-3) were pre-incubated with different concentrations (0.006 to 0.4 mg/ml) of purified VF-Fabs prior to infection of Huh7 cells. A no envelope control was used to normalise the data. The neutralising activity of the VF-Fab is expressed as percentage of inhibition of the infectious titres. Each experiment was repeated three times. IC₅₀ for each VF-Fab is detailed in Table 3. Error bars indicate standard deviation.

Fig 5. Conformational epitope mapping of E2 glycoprotein region starting from amino acid 384–619.

A. The multiple sequence alignment (MSA) of the E2 glycoprotein. HCVpp1b-1-3 is a reference sequence; HCVpp1b-1-1, HCVpp1a-1-3 and HCVppH77 were the infectious pseudoparticles. AF011751 is an H77 strain. aa marked with *: targeted by patient derived VF-Fabs. The colour code of the epitopes represent the position of the epitope in 3D structure of E2 glycoprotein (Table 4); aa in dotted box: linear epitope targeted by AP33 mouse MAb [35]; aa in red box: residues on E2 interact with AR3C HuMAb [13]; aa marked with ‡: residues important for CD81 binding [13]. B. The 3D model of HCV-E1E2 glycoprotein shown in white cartoon with flexible non-modelled E2 N-termini (384–412) labelled with spheres. Sequence ₄₂₈NCNDSLNTGFLAALFYTHRF₄₄₇ is highlighted in yellow, ₅₃₉LLNNTRPPRGNWF₅₅₀ in red and ₅₉₉SGPWLTPRCM₆₀₈ in blue (Pepscan Presto; Lelystad, Netherlands)

Fig 6. Neutralisation of pseudotyped virus with combination of VF-Fabs.

HCVpp1a-1-3, HCVpp1b-1-2, HCVpp1b-1-3 and control HCVppH77 were tested for neutralisation using the average Log IC₅₀ (mg/ml) for the VF-Fab in combinations (Table 5). The data is obtained from three independent experiments. Data for all the HCVpp were grouped together for statistical analysis. Significance was tested using one-way ANOVA with Dunnett t test (p<0.05–0.001, Table 5). Each VF-Fab combination was compared against individual VF-Fab at the concentration used in combination experiments.