Rapid Identification of H3N2 Influenza A Virus HA-Targeting Antibody Epitopes Driven by Computational Antibody Similarity Measures

Abstract Antibodies are important therapeutic agents for the prevention and treatment of influenza, and hemagglutinin (HA) serves as the primary target for the development of therapeutic antibodies. However, the identification of epitopes remains an unmet need in the development of such therapeutic antibodies. This study investigates the characteristics of antibody–HA binding interfaces (i.e., epitopes and paratopes) through a computational analysis of all antibody–HA complexes with nonredundant experimentally determined structures from SAbDab, the largest reported antibody structure data set to date. After experimentally identifying and selecting H3N2 HA-specific binding antibodies, a reference antibody library was established. For anti-H3N2 HA antibodies with unknown structures, epitope grouping was conducted based on competitive binding assays that assess interference with the binding of antibodies from the reference library. Based on the complementary relationship between epitope and paratope structures, we selected the structure-based similarity measure most relevant to epitope convergence. Furthermore, we prioritized the experimental validation of binding interference between candidate antibodies and their most structurally similar reference antibodies. We present an integrated computational and experimental framework. This approach includes the construction of an H3N2-specific reference antibody set, computational modeling of antibodies, measurement of antibody similarity, and is complemented by antibody competition binding assays. The dry–wet iterative protocol presented here yielded accurate assessment of the neutralizing potency and breadth of the previously uncharacterized antibody P2A10. Together, the presented pipeline offers a powerful approach for identifying epitopes on anti-H3N2 HA antibodies and for epitope-targeted antibody discovery, while demonstrating the effectiveness of structural analytics in reducing experimental labor.