Influenza Viral Envelope Simulation Reveals Novel Druggable Pockets on Surface Glycoproteins

Abstract

Influenza virus infection continues to be a major healthcare issue, with 3-5 million cases of severe disease reported and 300,000-500,000 deaths worldwide each year. Variation in the sequences and structures of the two major surface glycoproteins, hemagglutinin (HA) and neuraminidase (NA), from strain to strain complicate prophylactic and therapeutic approaches. Integrative modeling techniques combining cryo-electron tomography with x-ray crystallography, homology modeling, and protein-protein docking have led to the construction of a whole-virion model of influenza. Using the petascale computing machine Blue Waters, we have performed molecular dynamics (MD) simulations of this virion model. Principal component analysis reveals that simulation of the complete virion allows exploration of greater glycoprotein conformational space as compared to single glycoprotein simulations. Further, Markov state models constructed from the trajectories of the glycoproteins reveal novel druggable pockets in both HA and NA that can be targeted for the development of novel treatments for influenza virus infection. The identification of potentially druggable pockets previously unseen in simulations of individual glycoproteins demonstrates the utility of modeling in the sub-cellular scale and beyond to inform drug discovery.