Like other (+)RNA viruses, most (∼70%) of the SARS-CoV-2 genome encodes functions for RNA replication, underscoring the importance of this process for understanding and controlling these viruses. RCs support genome replication by organizing viral RNA replication proteins, viral RNA templates, specific host factors required for RNA replication, and successive reproductive steps. The RC-bounding membranes sequester RNA replication templates and intermediates from translation, virion assembly, RNA decay, and host defenses such as RNA interference and interferon-stimulated antiviral responses.

Although infection by coronaviruses induces several types of membrane rearrangements, multiple lines of evidence identified the dsRNA-containing DMVs as the viral RNA synthesis sites (2, 6). However, because DMVs lacked known openings, it was unclear how new (+)RNA genomes copied from dsRNA templates in the DMV interior could transit to the cytoplasm to be translated, packaged into virions, and potentially form new RCs.

Wolff et al. provide a compelling solution to this conundrum by using advanced cryo–electron tomography (cryo-ET) to identify a cylindrical protein pore complex traversing DMV double membranes in cells infected by SARS-CoV-2 or another coronavirus (see the figure). They also showed that this pore contains six copies of the large viral transmembrane protein nsp3 (nonstructural protein 3), which is essential for RNA replication and induces the formation of DMVs with viral nsp4. Consistent with the interaction of nsp3 with multiple viral replication proteins, the authors imaged frequent, apparently dynamic interaction of the pore’s DMV luminal and cytoplasmic sides with other macromolecules. Thus, the pore may interact with the viral RNA polymerase and other luminal RNA replication factors to guide newly synthesized RNAs to the cytoplasm, where the interaction of nsp3 with the viral nucleocapsid protein may facilitate RNA packaging into new virions.

The DMV pore is also an attractive solution for coronaviral RNA release because of similarities with a product RNA-release channel first identified in nodaviruses, a well-characterized model for (+)RNA virus replication. In addition to DMVs, another prominent class of RCs is necked spherular membrane invaginations (spherules) that are formed by numerous families of (+)RNA viruses including flaviviruses, such as Zika virus (7, 8), alphaviruses, such as chikungunya virus (9), and nodaviruses (10). Cryo-ET showed that the ∼50- to 80-nm-diameter nodavirus spherules (see the figure) contain viral dsRNA replication templates and that the cytoplasmic side of the spherule neck is surmounted by a ring, or crown, of 12 copies of nodavirus RNA replication protein A (11, 12). These crowns are frequently origins for cytoplasmic filaments that appear to represent new (+)RNA genomes being released. Crown-forming protein A contains all viral activities for RNA synthesis and shares distant sequence similarities to alphavirus RNA replication proteins (11–13).

Despite some differences in the membrane organization of DMV and spherule RCs, the DMV pores of coronaviruses show multiple parallels with nodavirus spherule crowns. These include their ringed multimeric structure, their apparent role as channels to release progeny RNA replication products, and potential involvement or interaction with active RNA synthesis. Consistent with these similarities, Wolff et al. also refer to the cytosolic portion of the coronaviral pore as a crown. However, it is important to note that this intracellular DMV RC crown is unrelated to the crown-like halo of virion envelope spike proteins that gave coronaviruses their name (14).

• Download high-res image
• Open in new tab
• Download Powerpoint

The recognition that RCs of coronaviruses and nodaviruses share fundamentally similar crown-like channels is notable given the considerable evolutionary separation of these viruses. Additional studies will further define the similarities and differences between these crowns, with functional and potentially evolutionary implications. Although the cytosolic portion of the coronavirus crown has sixfold symmetry, some membrane-interacting regions of the pore show rings of 12 similar electron-dense elements. Because the volume of the coronaviral pore complex shows that it must contain additional proteins beyond nsp3, the stoichiometry and symmetry of these regions remains uncertain. Even if only sixfold symmetric, do these 12-member rings bear any structural similarities to the 12-fold symmetric nodavirus crown?

Detailed interactions of these crowns with membrane lipids will also be of interest and might be more similar than seeming differences in DMV and spherule architectures suggest. The nodavirus spherule membrane folds at the neck into two sections that independently approach the crown, mirroring the connection of double membranes at nuclear pores. Similarly, hydrophobic surfaces on the coronaviral pore might induce lipids in the two DMV membranes to converge or interact, again approximating nuclear pores. Higher -resolution imaging of the nodavirus crown (12) bodes well for addressing such questions.

Additional questions include what other coronaviral and perhaps host proteins comprise the crown-like DMV channel. Viral nsp4 is one attractive candidate because it functions with nsp3 to form DMVs. Equally enticing is the identity of the dynamic RNA and protein interactions at both ends of the DMV channel and how these may promote RNA synthesis, transport, and virion assembly. Perhaps most important, recognizing conserved processes such as crown-mediated release of new genomic RNAs should provide a foundation for potentially broader-spectrum control, through pharmacologic or genetic means, of ubiquitous (+)RNA virus pathogens.

Go to Original – sciencemag.org