Abstract
A key step in the emergence of human pandemic influenza strains has been a switch in binding preference of the viral glycoprotein hemagglutinin (HA) from avian to human sialic acid (SA) receptors. The conformation of the bound SA varies substantially with HA sequence, and crystallographic evidence suggests that the bound SA is flexible, making it difficult to predict which mutations are responsible for changing HA-binding preference. We performed molecular dynamics (MD) simulations of SA analogues binding to various HAs and observed a dynamic equilibrium among structurally diverse receptor conformations, including conformations that have not been experimentally observed. Using one such novel conformation, we predicted─and experimentally confirmed─a set of mutations that substantially increased an HA's affinity for a human SA analogue. This prediction could not have been inferred from the existing crystal structures, suggesting that MD-generated HA-SA conformational ensembles could help researchers predict human-adaptive mutations, aiding surveillance of emerging pandemic threats.