A structural roadmap for the formation of the coronavirus nsp3/nsp4 double membrane vesicle pore and its implications for polyprotein processing and replication/transcription

Coronavirus replication is understood to occur within double membrane vesicles (DMVs) that arise during viral infection. Prior work has determined that these DMVs have characteristic pores formed from the non-structural viral proteins nsp3 and nsp4, which facilitate export of newly synthesized viral RNA. Yet how the replication machinery, which is comprised of the non-structural proteins nsp7 to nsp16, is recruited to the DMV remains a mystery. Working from AlphaFold and previously determined structures, we constructed a series of models that link formation of the DMV pore to nsp5 protease processing of the polyprotein and trapping of the cleaved products within the DMV itself. We argue that the initial pore is formed from 12 subunits of nsp3 and six subunits each of the intermediate uncleaved polyproteins pp1a' (nsp4-nsp10) and pp1ab' (nsp4-nsp16). Formation of this initial structure activates the protease function of alternating nsp5 subunits within a close-packed ring, facilitating the initial <i>trans-</i>cleavage of the nsp4-nsp5 linkage. Maturation of the pore follows, as does formation of canonical nsp5 dimers, which can process the remainder of the polyproteins. When protease activation occurs subsequent to closure of the DMV, the cleavage products, whose stoichiometry is consistent with the previously proposed nsp15-centered hexameric replication complex, will be trapped inside.IMPORTANCECoronaviruses, like many other positive-sense single-stranded RNA viruses, rely on intracellular membrane remodeling to create a protected environment for efficient replication to occur. The resulting double membrane vesicles (DMVs) have characteristic pores formed from the non-structural viral proteins nsp3 and nsp4, while enzymes responsible for RNA replication, including the polymerase nsp12, are contained inside. Recent structural work has elucidated the nature of the pores, but how the polymerase and other viral proteins are recruited to the DMV represents a major gap in our current knowledge. Here we present a novel, step-by-step structural model of how the pores initially form from the uncleaved polyprotein, how protease activity is initiated, and how the viral replication machinery is trapped within the DMV.