Hepatitis C virus (HCV) infection has been significantly impacted by the development of direct-acting antivirals (DAAs). However, challenges remain, including a lack of awareness of infection, high treatment costs, and potential for drug resistance. Therefore, a protective vaccine is essential for global HCV control. Previous research demonstrates that the early production of broadly neutralizing antibodies (bNAbs) is associated with spontaneous HCV clearance. Several potent bNAbs target a conserved epitope on HCV glycoprotein E2, using an unusual HCDR3 containing an intra-loop disulfide bond. Understanding the structural basis of these bNAbs is crucial for rational vaccine design, as this would facilitate the induction of desired bNAbs while avoiding non-neutralizing antibody responses. This study aims to computationally identify additional antibodies sharing these key features to enhance our understanding of bNAb-antigen interactions and guide vaccine development.

Several studies have shown a strong correlation between the early production of broadly neutralizing antibodies (bNAbs) and spontaneous clearance of HCV infection. These bNAbs often target conserved epitopes on the surface of the HCV glycoprotein E2, particularly the antigenic region 3 (AR3), which includes the front layer and the CD81 binding loop. However, the overlapping epitopes of bNAbs with other, less broadly neutralizing antibodies complicate vaccine design. Epitope-focused vaccine strategies aim to overcome this by inducing specific bNAbs while avoiding competition from non-neutralizing antibodies. High-resolution structural data of antibody-antigen complexes is valuable, but often insufficient to fully elucidate essential features. Therefore, identification and characterization of additional bNAbs are crucial to improve our understanding of the key structural and biophysical features for effective neutralization.

This study employed a combination of sequence-based and structure-based approaches to identify new antibodies with HCDR3 disulfide bond motifs. Initially, conventional sequence-based searches were conducted on antibody libraries from individuals who spontaneously cleared HCV infection. These searches focused on antibodies encoded by the VH1-69 gene segment, known to be associated with broadly neutralizing activity. However, conventional searches yielded limited results. Therefore, a homology modeling-based Position-Specific Structure Scoring Matrix (P3SM) approach was employed. This method uses known antibody-antigen structures as templates to identify structurally homologous antibodies in sequence databases, even with low sequence similarity. The P3SM approach uses Rosetta software for homology modeling and scores antibody sequences based on their predicted ability to adopt a specific conformation and form interactions with the antigen. Sequences with a C-X-G-G-X-C motif in their HCDR3, characteristic of the target antibody family, were prioritized. After homology modeling and manual evaluation of the models, eight HEPC3-like and two AR3C-like HCDR3 sequences were selected for experimental validation. Recombinant antibodies were generated, both chimeric (only the HCDR3 loop from a computationally predicted sequence was placed on the appropriate antibody framework) and native (full-length antibody sequences) versions. These antibodies were then tested for binding to a panel of HCV E2 ectodomains using ELISA. The neutralizing activity of the antibodies was assessed using HCV pseudoparticles (HCVpp). Furthermore, the binding site of selected antibodies was determined by using a knockout variant that contained alanine mutations that disable binding of HEPC3-like antibodies. Crystal structures of Fabs of HEPC3.1 and HEPC3.4 were determined to understand the structural features of the discovered antibodies. Molecular dynamics (MD) simulations were performed to investigate the conformational dynamics of the HCDR3 loops in the identified antibodies and compare them with those of previously characterized antibodies.

The study successfully identified new HCV E2-binding antibodies from individuals who spontaneously cleared HCV infection. Conventional sequence-based searches identified two HEPC3-like antibodies with minor HCDR3 mutations. The P3SM approach identified two additional, independently evolved HEPC3-like antibodies with low sequence homology. Seven out of twenty-two tested monoclonal antibodies (mAbs) bound to HCV E2 ectodomains. Specifically, both sibling Abs and five Abs derived from three HCDR3 sequences prioritized by P3SM showed binding to HCV E2 ectodomains in ELISA. The native antibodies displayed stronger binding and broader reactivity compared to their chimeric counterparts. All discovered antibodies neutralized multiple HCV strains in HCVpp neutralization assays, though with varying breadth and potency. The analysis of the crystal structures of HEPC3.1 and HEPC3.4 Fabs revealed a bent HCDR3 conformation. The study also found that proline and glycine residues in the HCDR3 loop may promote bent HCDR3 conformations, influencing the antibody's interaction with the antigen. Molecular dynamics (MD) simulations indicated different levels of flexibility in the HCDR3 loops of the studied antibodies, with HEPC3.1 and HEPC3.4 showing a preference for the bent conformation. The study also revealed that HCDR2 interactions may contribute to the stabilization of the bent conformation in some cases. The HCDR3 motif was reevaluated, revealing that the amino acid composition is less restricted than previously thought.

This study successfully employed computational approaches to identify new HCV-neutralizing antibodies from the antibody repertoires of individuals who naturally cleared HCV infection. The identification of antibodies with low sequence similarity to known broadly neutralizing antibodies highlights the power of structure-based methods. The findings demonstrate that both bent and straight HCDR3 loop conformations are prevalent in this antibody family, suggesting that both conformations could be effectively utilized in vaccine design. The study also suggests that proline and glycine residues at the base of the HCDR3 loop could be used to promote a bent conformation. The observed differences in binding and neutralization among the antibodies may be correlated with their flexibility, suggesting that the dynamic properties of the HCDR3 loop are crucial for effective antigen recognition and neutralization.

This study demonstrates the successful application of a structure-based computational approach in discovering new broadly neutralizing antibodies against HCV. The identification of antibodies with diverse HCDR3 sequences while still sharing a common structural motif provides valuable insight into the structural determinants for effective HCV neutralization. The findings support the development of epitope-focused vaccine strategies that can induce antibodies with both bent and straight HCDR3 loop conformations, increasing the likelihood of effective neutralization. Further research should focus on understanding the role of other CDR regions in shaping the overall antibody-antigen interaction.

The study was limited by the availability of antibody libraries and the limited number of antibodies selected for experimental validation. The use of a single light chain in the recombinant antibodies could potentially affect the binding and neutralization properties. The observed conformational preferences from molecular dynamics simulations may not fully represent the range of conformations adopted by the antibodies in vivo. Future studies with larger antibody libraries, diverse antibody-light chain pairings, and more extensive experimental characterization would further enhance our understanding of the structural determinants for HCV neutralization.