Binding Free Energy-Based Footprint Pharmacophore Model to Enhance Virtual Screening and Drug Discovery: A Case on Glycosidases as Anti-influenza Drug Targets

摘要

Frequent emergence of influenza virus strains resistant to current neuraminidase inhibitors is a global threat and demands for the discovery of new potent inhibitors. Virtual screening techniques have proved to be an effective approach in drug discovery. In this study, we present an approach to further enhance the potency of the typical pharmacophore-based virtual screening method by incorporating a MM/GBSA per-residue energy contribution footprint from molecular dynamics simulation, as opposed to the typical use of docking scores as a frontline screening strategy. The MM/GBSA per-residue energy footprint with highest contribution to the binding free energy was mapped on the reference drug and used to screen for compounds sharing structural similarity with the reference drug. The proposed approach was generated and used to screen the ZINC database for potent inhibitors against influenza neuraminidase. Seven of the novel compounds identified by the proposed approach, with ZINC18142090 being the top-ranked compound, showed higher binding affinities compared to that of known neuraminidase inhibitors zanamivir, oseltamivir and laninamivir. These novel compounds also formed interactions with the conserved active site residues Arg152, Arg292, Asn294, Arg371, Ile222, Arg224, Glu227, Glu276 and Glu277, thus implying a conserved selectivity and binding mode adopted by the obtained compounds. A strategic computational approach presented in this study could serve as a beneficial tool to enhance native virtual screening as well as novel drug discovery.