Structure-based and AI-guided redesign of Sotrovimab for enhanced recognition of SARS-CoV-2 BA.2 and the E340G escape mutant

Introduction The rapid evolution of SARS-CoV-2 Omicron variants has substantially compromised the efficacy of several monoclonal antibody therapies, including Sotrovimab, through mutations that disrupt antibody–antigen recognition. Developing antibodies with improved recognition of emerging variants and resistance-associated mutations is therefore essential for maintaining therapeutic effectiveness. Methods Here, we employed an integrated computational framework combining structure-guided antibody engineering, artificial intelligence (AI)-assisted sequence design, molecular dynamics (MD) simulations, and binding free-energy analyses to redesign Sotrovimab against the Omicron BA.2 receptor-binding domain (RBD) and the clinically relevant E340G escape mutant. Residue-level energetic profiling was used to identify destabilizing interface residues, followed by mutational redesign and ProteinMPNN-guided sequence optimization. The resulting candidates were evaluated through 500 ns all-atom MD simulations, complemented by conformational analyses, MM/PBSA binding free-energy calculations, residue-level energy decomposition, and dynamic cross-correlation analyses. Results and Discussion The top-ranked redesigned Sotrovimab candidates retained the characteristic immunoglobulin fold while exhibiting improved structural stability, compactness, and reduced conformational fluctuations. Principal component analysis and free-energy landscape calculations further revealed more confined and energetically favorable conformational states, suggesting enhanced dynamic compatibility with the target RBD. MM/PBSA calculations indicated substantially more favorable binding free energies for both redesigned candidates against the BA.2 RBD and E340G mutant. Residue-level energy decomposition and dynamic cross-correlation analyses further demonstrated improved energetic complementarity and intermolecular communication across the antibody–antigen interface. These improvements were accompanied by enhanced hydrogen-bonding, hydrophobic, and van der Waals interactions, together with increased engagement of the conserved N343-linked glycan, potentially providing additional stabilization to the antibody -antigen complex. Among the redesigned candidates, Sotrovimab-design-2 consistently exhibited the most favorable structural and energetic characteristics. Conclusion Collectively, these findings demonstrate the potential of integrating AI-guided sequence optimization with structure-based antibody engineering to improve antibody recognition of SARS-CoV-2 variants and resistance-associated mutations. This framework provides a computational foundation for developing mutation-resilient therapeutic antibodies against evolving viral pathogens.