Assessing Protein-Ligand Binding Modes Via Ensemble Molecular Dynamics

Molecular dynamics (MD) simulations of biomolecular phenomena require extensive sampling. Fortunately, advances in computational hardware and algorithmic efficiency facilitate the acquisition of ensemble datasets and there is a growing need for analytical approaches to assess datasets of this nature. In this study, we characterize conformations of butyrylcholinesterase (BChE) in complex with a number of inhibitors. The Folding@Home distributed computing network allowed for the acquisition of over 100 μs of all-atom MD sampling for each of thirteen BChE-inhibitor complexes and the enzyme sans inhibitor, yielding a total simulation time of nearly 1.5 ms. For each complex, post-equilibration structures were characterized by the magnitude and type of contact between inhibitor functional groups and BChE residues. These descriptors were then clustered via K-means and the resulting clusters were described via contact tables, which convey the results of massive datasets in a comprehensive, concise manner accessible to our collaborators, who have provided experimentally measured K_I values for the studied inhibitor. This approach, in tandem with ensemble-level sampling, provides a novel atomic-level picture perspective, improving our capacity to propose more efficacious inhibitors.