Computational investigations of the binding mechanism of current influenza virus neuraminidase inhibitors: Correlation with experiment

Abstract

In the context of the recent pandemic threat by the worldwide spread of H5N1 avian influenza, novel insights into the mechanisms of ligand binding and interaction between various inhibitors (zanamivir-ZMV, oseltamivir-OTV, DANA, peramivir-PMV) and neuraminidases (NAs) are of vital importance for the structure-based design of new antiviral drugs. Computational methods are herein shown to be a viable partner to experiment in the investigation of the binding modes of known NA inhibitors. Using the crystal structures of inhibitors bound to either group-2 or group-1 NAs, the AScore/ShapeDockscoring was shown to identify the binding modes in agreement with the experiment for all inhibitors docked in their own NA/inhibitor crystal structures. To investigate the effect of small changes in protein structure on predicted binding modes, in a set of 132 docking experiments (11 inhibitors docked in 12 group-2 NA structures) AScore/ShapeDock identified the correct binding modes of 116 complexes. In a total of 88 docking experiments (8 inhibitors docked in 11 group-1 NA structures) AScore/ShapeDock predicted 80 binding modes correctly. A small vHTS experiment, reflected by mixing the known activity molecules to a set of randomly selected molecules, confirmed the ability of the AScore/ShapeDock approach to extract biologically active molecules from inactive ones. By both outperforming other docking methods used previously in the literature and being quite reproducible, the AScore/ShapeDock protocol is suggested to be convenient for designing novel H5N1-NA inhibitors. To shed more light on the high resistance of H5N1 strains to various NA inhibitors, the three-dimensional models of H5N1-NA andN9-NA were generated by homology modeling. Traditional residues within the active site throughout the family of NA protein structures were found to be highly conserved in H5N1-NA, and a subtle variation between lipophilic and hydrophilic environments in H5N1-NA with respect to N9-NA was observed. Besides, molecular bases of the mechanism of H5N1-NA resistance to oseltamivir were elucidated in a systematic fashion. Using the crystal structure of the complex of H5N1-NA with OTV (PDB ID: 2hu0) as the starting point, the following question was addressed and correlated with the experimental data: How mutations at His274 by both smaller side chain (Gly, Ser, Asn, Gln) and larger side chain (Phe, Tyr) residues influence the sensitivity of N1 to oseltamivir? The smaller side chain residue mutations of His274 resulted in slightly enhanced or unchanged NA sensitivity to OTV, while His274Phe and His274Tyr reduced the susceptibility of OTV to N1. In contrast to the binding free energies, the net charges of Glu276 and Arg224, making charge-charge interactions with Glu276, were established to be more sensitive to detecting subtle conformational differences induced at the key residue Glu276 by the His274X mutations. Thus, deeper insights into the possibility of developing viable drug-resistant mutants were possible.