Clustering of SARS-CoV-2 membrane proteins in lipid bilayer membranes

The accumulation of viral structural proteins along the ER-Golgi intermediate compartment (ERGIC) membrane leads to SARS-CoV-2 self-assembly and budding, driven by the interactions between these proteins, RNA and the ERGIC membrane. The membrane protein (M) is believed to interact with other structural proteins and form clusters needed for the induction of membrane curvature that facilitates virion formation. However, the role played by direct and membrane-mediated interactions between M proteins and their interactions with other proteins in the clustering process remains unclear. Here, we utilize a combination of all-atom molecular dynamics (MD) simulations, continuum modeling and experiments to show that M-M interactions are sufficient to drive clustering in ERGIC-like lipid bilayers in the absence of other proteins or RNA. Using all-atom MD simulations we were able to estimate the membrane thinning induced by M proteins and the resulting membrane-mediated M-M interaction. Combining this with a continuum model that describes the evolution of M protein density in a planar lipid membrane, we identified the existence of a critical, direct M-M interaction energy needed for cluster assembly at a given density. By comparing the model predictions with analysis of atomic force microscopy images of M protein clusters in supported lipid bilayers, we were able to estimate the direct M-M interaction energy and found it to be significantly larger than the membrane mediated interaction energy. Our work therefore establishes that M protein interactions are sufficient to drive clustering and provides a quantitative understanding of the role played by direct and membrane-mediated interactions of M proteins in viral assembly and budding.