Application of per-residue energy decomposition to identify the set of amino acids critical for in silico prediction of COX-2 inhibitory activity

The enormous magnitude of scientific research carried out in the field of NSAIDs and cyclooxygenases (COXs) is known. They are crucial in pain management. COX-2 inhibitors have evolved over the years; from traditional NSAIDs to isoform-specific. The present study is aimed to identify a cluster of amino acids in the catalytic site whose energy contribution can better explain COX-2 inhibitory activity accurately than the binding energy of the whole protein. Initially, MD simulations (25 ns) and MM-PBSA calculations were performed for 8 diarylheterocyclic inhibitors. Per-residue energy decomposition studies were carried out to elucidate the energy contribution of each amino acid, and their correlation with COX-2 inhibitory activity was enumerated. A cluster of catalytic amino acids whose free energy sum has a high correlation with biological data was identified. The cluster of Gln178, Ser339, Tyr341, Arg499, Phe504, Val509 and Ala513 showed the correlation of -0.60. Further, the study was extended to a total of 26 COX-2 inhibitors belonging to different classes to validate the applicability of the cluster of amino acids identified. Results clearly suggest that the cluster of amino acids identified provide accurate screening method, and can be applied to predict COX-2 inhibitory activity of small molecules.