Epitope-focused immunogens targeting the hepatitis C virus glycoproteins induce broadly neutralizing antibodies

Abstract

Hepatitis C virus (HCV) infection causes ~290,000 annual human deaths despite the highly effective antiviral treatment available. Several viral immune evasion mechanisms have hampered the development of an effective vaccine against HCV, among them the remarkable conformational flexibility within neutralization epitopes in the HCV antigens. Here, we report the design of epitope-focused immunogens displaying two distinct HCV cross-neutralization epitopes. We show that these immunogens induce a pronounced, broadly neutralizing antibody response in laboratory and transgenic human antibody mice. Monoclonal human antibodies isolated from immunized human antibody mice specifically recognized the grafted epitopes and neutralized four diverse HCV strains. Our results highlight a promising strategy for developing HCV immunogens and provide an encouraging paradigm for targeting structurally flexible epitopes to improve the induction of neutralizing antibodies.

Fig. 1.. Pipeline for characterization of lead immunogen candidates.

(A) Epitopes were extracted from crystal structures of the respective epitope within E1 (residues 314 to 324) and E2 (residues 412 to 423, epitope I) in complex with their cognate Fab fragments (PDB 4N0Y and 4DGV). Multiple designs were generated for each of the epitopes by grafting their side chains onto suitable scaffolds followed by in silico structure–based design to optimize epitope display and antibody recognition. Side chains of epitopes were transplanted onto suitable scaffolds followed by further structure-based design to optimize epitope display and antibody recognition. (B and C) Fusion with the fluorophore mNeonGreen allowed rapid identification of soluble monomeric scaffolds that expose an accessible epitope by pull-down assay. Promising candidates were subjected to more extensive computational design to increase protein stability (i.e., thermostabilization). (D) Thermostabilized designs together with the parental scaffolds were subjected to biochemical and structural characterization, resulting in the identification of promising lead immunogens. (E) The latter were displayed on hepatitis B virus core antigen capsid-like particles to augment immunogenicity and used for immunization of mice. Neutralization potency of mouse antisera was evaluated against a representative panel of HCVcc. (F) Antigen-specific B cells were isolated from mice encoding a fully human IgG locus (human antibody mice) that were immunized with the immunogen NPs and monoclonal antibodies were isolated that broadly neutralize a more extensive reference panel comprising HCVcc of 13 distinct HCV strains (56) and specifically bind to the two displayed target epitopes.

Fig. 2.. Characterization of lead immunogen candidates for NP production.

(A) Grating-coupled interferometry (GCI)–derived kinetic binding parameters for Fab (ligand)–scaffold (analyte) interactions. Sensorgrams are shown with data in red and their respective fitting in black. The dissociation constants K_d for a 1-to-1 binding model or a mass transport binding model of the immunogen candidates–cognate Fab interactions are shown). Binding kinetics of E2_EpiI scaffolds E2_S2_1 and E2_S3_3 against Fabs AP33 and HCV1 using sE2 as control are shown. (B) Similarly, binding kinetics was also assessed for E1_S1 scaffold against Fab IGH526. (C and D) CD spectra reported as mean residual ellipticity (MRE) were measured for E1_S1 (C) and E2_S2_1 (D) at different temperatures. The data points extracted from negative maxima from each of the CD spectra are indicated as black dots and fitted with a curve in red; the apparent CD-T_m is derived from the slope. Typical positive maxima or negative minima of proteins with α-helical or β strand secondary structures are indicated with gray dotted lines. (E) Crystal structure of scaffold E1_S1 in complex with heavy chain (HC) and light chain (LC) of the IGH526 Fab fragment. (Right) A close-up view of the grafted epitope from the scaffold is overlaid with the reported epitope structure (PDB 4N0Y). Epitope residue side chains of the scaffold complex structure (purple) or the reported PDB model (pink) are highlighted as sticks, Cα atoms as spheres. The sequence of the grafted epitope in the scaffold highlights the mutated residues in purple (E1_S1).

Fig. 3.. Characterization of capsid-like immunogen NPs.

(A to C) GCI-derived binding kinetics for Fab-NP interactions. Sensorgrams are shown with data in red and their respective fits in black. Kinetic parameters for a 1-to-1 binding model or a mass transport binding model did not yield any productive values because of the absence of observable dissociation due to the avidity effect caused by the multivalent nature of the NPs. (D) Negative stain electron microscopy of NPs presenting either scaffold lacking GB1 (delGB1). Scale bar, 200 nm.

Fig. 4.. NP mix induces bnAbs in C57BL6 mice.

A single-dose neutralization assay was used to investigate the induction of nAbs against a panel of four different viruses that differ in neutralization sensitivity (Gt1b_J4, Gt2a_Jc1, Gt3a_S52, and Gt5a_SA13). (A to D). Female C57BL6 mice were immunized with the indicated immunogen(s) in two batches [N1 = 5 (dark blue), N2 = 12 (light blue)], and the neutralization efficiency of the individual mouse sera is reported as % infectivity (defined as the % infectivity normalized to mice immunized with control NPs). Colored horizontal bars indicate the arithmetic mean per immunization protocol. Warm colors indicate immunization protocols with a significant mean change in % infectivity (two-sided one-sample t test against a value of 100% infectivity, false discovery rate–corrected for multiple comparisons). Neutralization efficiency was analyzed with linear mixed-effects regression indicating that inter-mouse variability accounted for 54% of total variance (intra-class correlation coefficient, ICC = 0.54). Neutralization of NPmix was compared to sE2 and control by general linear tests with Dunnett-correction for multiple comparisons. To satisfy linear model assumption, raw % infectivity values were Box-Cox transformed (λ = 1.4) for statistical analysis. n.s., P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Fig. 5.. Neutralization of HCVcc genotypes with sera from Human antibody mice and isolated mAbs from B cell sequencing immunized with NPmix or sE2 and ELISA.

(A) Human antibody mice were immunized with sE2 GT2a_3 or NPmix. Sera from individual mice were evaluated in a single-dose neutralization assay against the panel of viruses described in Fig. 4. (A) Scatterplots show individual % infectivity (normalized to mice immunized with sE2) along with the arithmetic mean per immunization protocol as in Fig. 4. Neutralization efficiency was analyzed with linear mixed-effects regression indicating that inter-mouse variability accounted for 40% of total variance (ICC = 0. 4). Follow-up linear models per genotype compared mean % infectivity of mice immunized with NP mix versus sE2. *P ≤ 0.05 (B) A single-dose neutralization assay at 50 μg/ml against a reference panel of 13 HCVcc strains (56) was performed with human mAbs isolated from mice immunized with NPmix (N01 to N15) and two well-characterized reference bnAbs (dark gray). Individual % infectivity (normalized to infection in the presence of phosphate-buffered saline) is shown, and each dot represents an independent HCVcc strain mean value (scatterplot) and the median is highlighted. mAbs with the broadest neutralization capacity are colored (N08, orange; N15, cyan). Human bnAbs were further analyzed in ELISA against (C) wt (1W4K/2K6Z) or epitope scaffolds E1_S1 (1W4K_08) and E2_S2_1 (2K6Z_01). Parental nAbs IGH526 and HCV1 were used as positive controls and an unrelated human IgG1 antibody as negative control. (D) Broad recognition of sE2 via bnAbs N08 and N15 was tested by ELISA using sE2 H77c (GT1a), J8 (GT2b), ED43 (GT4a), and UKN2b_2.8 (GT2b). As sE2 UKN2b_2.8 was not recognized by bnAb N08, the well-characterized bnAb HCV1 was used as positive control in this assay. ELISA results are presented as the mean of three biological replicates with three technical replicates each.