Elucidating Ephrin-Induced Intersecting Signaling Pathways in the Nipah Virus G Protein using Machine Learning

The fusion of Nipah viruses with host cells is facilitated by two of their membrane proteins, the attachment protein (G) and the fusion protein (F). G binds to specific ephrin receptors on the host membrane. Ephrin binding changes the configurational density of G that activates F, which, in turn, mediates fusion. To understand how ephrin binding causes G to activate F, we use molecular dynamics in conjunction with machine learning1 and filter out the set of residues in the G head domain whose configurational densities are shifted equivalently by different ephrins, B2, B3, and a double mutant of B2. These three ephrins all trigger viral fusion, but with different potencies. We find that these three ephrins induce statistically equivalent shifts in the configurational densities of about one-quarter of the residues in the G head domain. This surprisingly expansive communal change in G includes most of the residues that have been shown experimentally to be important to F activation. This suggests that this set of residues contain the signaling pathways that connect the G-ephrin interface to the G stalk domain that activates F. The distribution of these residues in the G head domain is consistent with two models of signal transduction: one in which the ephrin-binding signal transduces to the F-activating G stalk domain via changes in the head-stalk interface, and the other in which the signal transduces via changes in the G head domain dimer interface. In general, this study shows how machine learning can be utilized along with molecular simulations to filter out quantitatively conserved patterns in changes in protein structure and dynamics.1) Leighty RE.; Varma S. Quantifying Changes in Intrinsic Molecular Motion Using Support Vector Machines. J Chem. Theor Comput. 2013, 9, 868-875.