Structure-Guided Engineering of High-Affinity Antibodies Against Zika Virus Using Deep Learning and Molecular Dynamics.

Zika virus (ZIKV) remains a global health threat, for which no licensed antiviral treatment has been available. In this study, we employed in silico approaches to optimize monoclonal antibodies targeting the Zika virus envelope protein (ZIKV E) in the Domain III (DIII) region, which is crucial for receptor binding and virus entry. A high-resolution crystal structure of ZIKV E in complex with the neutralizing antibody ZV-64 was used as a template for designing a library of antibody variants through targeted double-point mutations. The variants were systematically evaluated for stability, binding affinity, solubility, and protein-protein interaction potential using FoldX, DeepPurpose, SoluProt, and molecular docking. Among all the mutants, Variants-213 and -206 were identified as the top candidates, exhibiting the most favorable predicted binding affinity and solubility compared to the control antibody. The molecular dynamics simulations further revealed the structural stability of the two mutant variants, in which Variant-206 showed a predicted binding energy (-76.90 kcal/mol) along with higher conformational flexibilities. The findings demonstrate the use of computational antibody engineering to identify potentially high-affinity therapeutics against ZIKV, providing a foundation for future experimental validation and therapeutic development against ZIKV.