Multiple Molecule λ-Dynamics: Probing Drug Resistance with Concurrent Protein and Ligand Perturbations

Abstract

Though commonly used in drug discovery, alchemical free energy calculations have not been extensively used to explore issues of drug resistance caused by missense mutations to a drug target. Unlike traditional methods, λ-dynamics (λD) can evaluate multiple modifications within a single simulation; however, perturbations on more than one molecule, e.g., in a ligand and receptor, have not been performed previously. In an approach referred to as Multiple Molecule λ-Dynamics (MMλD), simultaneous ligand and protein perturbations are performed in a single simulation to sample a small series of ligands bound to native and T315I mutant Abl kinases, a protein target in chronic myelogenous leukemia associated with drug resistance. MMλD agreement with conventional λD calculations and experiments is high, with mean unsigned errors of 0.21 and 0.94 kcal/mol, respectively. Protein sequence specific ligand conformational sampling is also identified. Collectively, this work demonstrates that MMλD is a valuable tool for drug resistance drug discovery.