Improved Specificity Prediction of Small Molecule Myosin Inhibitors through Ensemble-Based Molecular Docking

Abstract

Computational calculations of small molecule-protein interaction using molecular docking algorithms are powerful tools for the study of ligand recognition and drug discovery. However, accounting for protein flexibility remains still a challenge. Here we show the combination of physics-based molecular dynamics simulations with time-efficient docking that yield a highly improved correlation of the in silico predictions with experimentally determined inhibitory potencies of halogenated carbazoles towards different isoforms of the motor protein myosin. Analysis of protein dynamics revealed marked structural changes of the allosteric target pockets in different isoforms. Ranking of the predicted binding affinities of a series of halogenated carbazoles towards different myosin isoforms using molecular docking against clustered ensembles of protein conformations correlated well with IC50 values, measured using a spectroscopy-based ATPase assay, and gave high potencies of carbazole derivatives in the low micromolar range. Our results demonstrate the superior, but still resource-efficient performance of our combined molecular dynamics - molecular docking method in computing the specificity behavior of small molecules between different myosin isoforms, and highlight the importance of including protein flexibility in the in silico predictions.