Design of a native-like secreted form of the hepatitis C virus E1E2 heterodimer

Abstract

Hepatitis C virus (HCV) is a major worldwide health burden, and a preventive vaccine is needed for global control or eradication of this virus. A substantial hurdle to an effective HCV vaccine is the high variability of the virus, leading to immune escape. The E1E2 glycoprotein complex contains conserved epitopes and elicits neutralizing antibody responses, making it a primary target for HCV vaccine development. However, the E1E2 transmembrane domains that are critical for native assembly make it challenging to produce this complex in a homogenous soluble form that is reflective of its state on the viral envelope. To enable rational design of an E1E2 vaccine, as well as structural characterization efforts, we have designed a soluble, secreted form of E1E2 (sE1E2). As with soluble glycoprotein designs for other viruses, it incorporates a scaffold to enforce assembly in the absence of the transmembrane domains, along with a furin cleavage site to permit native-like heterodimerization. This sE1E2 was found to assemble into a form closer to its expected size than full-length E1E2. Preservation of native structural elements was confirmed by high-affinity binding to a panel of conformationally specific monoclonal antibodies, including two neutralizing antibodies specific to native E1E2 and to its primary receptor, CD81. Finally, sE1E2 was found to elicit robust neutralizing antibodies in vivo. This designed sE1E2 can both provide insights into the determinants of native E1E2 assembly and serve as a platform for production of E1E2 for future structural and vaccine studies, enabling rational optimization of an E1E2-based antigen.

Keywords: E1E2; envelope glycoprotein; hepatitis C virus; scaffold; vaccine.

Fig. 1.

Design of sE1E2 constructs. (A) Schematic of mbE1E2, covalent linker sE1E2 constructs, and cleavable polyprotein constructs. Regions shown include tPA signal sequence (green box), E1 ectodomain (yellow boxes), E2 ectodomain (red boxes), wild-type TMDs (gray boxes), Gly-Ser linker (orange boxes), and various scaffolds replacing TMDs. E1E2 residue ranges for each region are noted according to H77 numbering. C-terminal His tags and furin cleavage sites are shown in boxes and labeled. The expected molecular weight of each construct is indicated, and molecular weights of expected oligomers for sE1E2.FD and sE1E2.CC are in parentheses. For molecular weight estimations, each N-glycan is approximated to be 2 kDa at each NxS/NxT sequon, a value within the molecular weight range of typical N-linked glycans (108). (B) X-ray structure of human c-Fos/c-Jun heterodimer (PDB ID code 1FOS); only the coiled-coil region that was used for the sE1E2.LZ scaffold is shown. c-Fos and c-Jun chains were colored to match the diagram of sE1E2.LZ. (C) X-ray structure of foldon domain (PDB ID code 4NCU). All chains are colored light blue to match the diagram for sE1E2.FD. (D) Model of CC1+CC2 heterohexameric peptide assembly. CC1 and CC2 chains are colored to match the diagram for sE1E2.CC. All structures were visualized in PyMOL (Schrodinger, LLC).

Fig. 2.

E1 and E2 Western blots of sE1E2 supernatant. HCV1 antibody at 5 μg/mL was used for the E2 Western blot. H-111 antibody at 10 μg/mL was used for the E1 Western blot. All sE1E2 supernatant samples were loaded under reducing conditions. Supernatants were concentrated 10 times prior to E1 Western blot. Molecular weights, in kilodaltons, of the Western blot markers closest to observed bands are indicated on the left. Expected band positions of E1, E2, and E1E2 are indicated with black triangles on right and labeled.

Fig. 3.

SEC of sE1E2.LZ and mbE1E2. Chromatographic traces for (A) sE1E2.LZ and (D) mbE1E2 shown in blue lines plotted with molecular weight standards shown in gray lines after elution from a Superdex 200 SEC column (Cytiva). Molecular weight estimates for the center of each peak are labeled based on comparisons with elution of high-molecular-weight standards (Cytiva), with molecular masses of 670, 440, 158, 73, and 44 kDa. The range for elution fractions F1 to F10 used for analysis is shown as a red line. Western blots of sE1E2.LZ for E2 (B), sE1E2.LZ for E1 (C), mbE1E2 for E2 (E), and mbE1E2 for E1 (F) under nonreducing conditions. HCV1 antibody was used to probe for E2, while H-111 antibody was used to probe for E1. Molecular weights, in kilodaltons, of the Western blot markers closest to observed bands are indicated on the left of each panel. All fractions had 250 ng loaded for improved visualization of size. For E1 Western blots, all fractions were concentrated 10 times prior to loading. Putative E1 monomer, dimer, and trimer populations shown in F are highlighted with red initials.

Fig. 4.

Analytical characterization of sE1E2.LZ and mbE1E2 size and heterogeneity. AUC profiles of (A) purified sE1E2.LZ with or without detergent β-OG and (B) purified mbE1E2. Shown are the distribution of Lamm equation solutions c(s) for the two proteins (blue or black lines). Calculated sedimentation coefficients for the peaks are labeled. Observed species for sE1E2.LZ approximately correspond to a heterodimer at 4.9 S, a dimer of heterodimers at 7.7 S, and higher-order aggregates at 10.3 S. Observed species for mbE1E2 approximately correspond to free E2 at 4.0 S, a dimer of heterodimers at 6.6 S, a trimer of heterodimers at 9.1 S, and a tetramer of heterodimers and higher-order aggregates at >10 S. (C) sE1E2.LZ and (D) mbE1E2 characterization with SEC-MALS. The chromatographs of each protein are shown as blue lines. For reference, chromatographs of molecular weight standards are shown as gray lines in C and D, corresponding to molecular masses of 670, 158, 44, 17, and 1.35 kDa. The MALS scattering sizes between the peak half-maxima are shown as red points, with the estimated molecular weight at the center of each peak labeled and size distribution of each range in parentheses. Based on calculated molecular weights of each heterodimer and SEC-MALS molecular size ranges, these peaks predominantly contain oligomers of (C) 1 to 2 sE1E2.LZ heterodimers and (D) 5 to 27 mbE1E2 heterodimers.

Fig. 5.

Initial antigenicity screening of sE1E2 designs in ELISA. mbE1E2, sE1E2.LZ, sE1E2GS3, and sE2 were coated on ELISA plates at a concentration of 2 μg/mL and tested for binding to a panel of E2 and E1E2 bnAbs, representing E2 antigenic domains E (HCV1), B (AR3A), and D (HC84.26.WH.5DL), as well as E1E2 domains AR4 (AR4A) and AR5 (AR5A). Binding was measured at 450 nm with an antibody concentration of 0.185 μg/mL. Negative controls shown are an unrelated antibody (CA45) or PBS.

Fig. 6.

Measurement of binding to the CD81 receptor by SPR. CD81 binding kinetic curves to (A) mbE1E2, (B) sE1E2.LZ, and (C) sE2 are shown. Kinetic (k_on, k_off) and steady-state (K_d; calculated as k_off/k_on) binding parameters were calculated based on 1:1 model and are shown in each panel.

Fig. 7.

Immunogenicity assessment of sE2, mbE1E2, and sE1E2.LZ. Six mice per group were immunized with sE2, mbE1E2, or sE1E2.LZ, and sera were tested for binding to (A) mbE1E2 and (B) H77C-pseudotyped HCVpp in ELISA. One mouse in the sE2-immunized group died prior to final bleed, and thus responses for five mice are shown for that group. Endpoint titers were calculated using Graphpad Prism, and geometric mean titers are shown for each group as black lines. (C) Neutralization of H77C HCVpp by immunized murine sera. ID₅₀ values were calculated in Graphpad Prism for individual mice, and average ID₅₀ titers for each immunized group are shown as black lines. The minimal serum dilution used for ID₅₀ measurement (1:64) is shown as a horizontal dashed line, for reference. P values between group endpoint titer or ID₅₀ values were calculated using Kruskal–Wallis analysis of variance with Dunn’s multiple comparison test (ns, not significant: P > 0.05; **P ≤ 0.01).