Computational Approaches to Predict Hepatitis B Virus Capsid Protein Mutations That Confer Resistance to Capsid...

Capsid assembly modulators (CAMs) are a novel class of antiviral agents in clinical development for the treatment of chronic hepatitis B. CAMs inhibit hepatitis B virus (HBV) replication by binding to a hydrophobic pocket, i.e., HAP pocket, between HBV capsid protein (Cp) dimer-dimer interfaces to misdirect its assembly into empty capsids or aberrant structures and designated as CAM-E and CAM-A, respectively. Because the emergence of CAM-resistant variants results in the failure of antiviral therapy, it is important to rationally design CAMs with a high barrier of resistance for development. To establish computational approaches for the prediction of Cp mutations that confer resistance to CAMs, we investigated the interaction of representative CAM-A and CAM-E compounds, BAY 41-4109 and JNJ-56136379, with wild-type and 35 naturally occurring mutations of Cp residues at the HAP pocket using molecular docking, prime molecular mechanics with generalized Born and surface area solvation (MM/GBSA) and molecular dynamics (MD) simulation methods. Out of nine publicly available HBV capsid or CpY132A hexamer structures in the protein database, molecular docking correctly predicted the resistance and sensitivity of more than 50% Cp mutations to JNJ-56136379 with structures 5D7Y and 5T2P-FA. MM/GBSA correctly predicted the resistance and sensitivity of more than 50% Cp mutations to BAY41-4109 with the structures 5E0I-BC and 5WRE-FA, and to JNJ-56136379 with the 5E0I-FA structure. Our work indicates that only the capsid or CpY132A hexamer structure bound with a CAM with similar chemical scaffold can be used for more accurately predicting the resistance and sensitivity of Cp mutations to a CAM molecule under investigation by molecular docking and/or MM/GBSA methods.