Integrating RFantibody and molecular dynamics to discover high-affinity Nanobody inhibitors targeting dengue virus NS3 protease

Abstract

Dengue fever remains a major global health challenge due to the lack of effective antiviral drugs and the limited efficacy of available vaccines. The dengue virus NS3 protease plays a vital role in viral replication and is highly conserved across serotypes, making it an attractive drug target. In this study, we first used RFantibody, an Artificial intelligence (AI)-driven framework for single-domain antibody design, to generate 100 nanobody candidates targeting the catalytic triad (His51, Asp75, Ser135) of the NS3 protease. The binding affinities of these complexes were predicted using PRODIGY. Based on the results, the top-six ranking nanobody–protease complexes were selected for further evaluation, five of which successfully underwent 100-ns molecular dynamics simulations using GROMACS. Analyses of stability, compactness, and hydrogen bonding showed that most designed nanobody complexes maintained stable conformations and formed more hydrogen bonds than the reference NS3–aprotinin complex. Binding free energy calculations using the MM/GBSA method confirmed that several designed nanobodies—particularly complexes 16, 40, and 78—exhibited much stronger binding energies (approximately −50 kJ/mol) than the reference complex. Per-residue energy decomposition and alanine scanning identified key residues in the complementarity-determining regions (CDRs), especially in CDR3, that contributed significantly to binding through hydrophobic and hydrogen bond interactions. Overall, our results demonstrate that combining AI-driven nanobody design with molecular simulations can effectively identify high-affinity inhibitors targeting the dengue virus NS3 protease, providing a promising strategy for developing novel antiviral therapeutics.