Nodavirus RNA replication crown architecture reveals proto-crown precursor and viral protein A conformational switching

Abstract

Positive-strand RNA viruses replicate their genomes in virus-induced membrane vesicles, and the resulting RNA replication complexes are a major target for virus control. Nodavirus studies first revealed viral RNA replication proteins forming a 12-fold symmetric "crown" at the vesicle opening to the cytosol, an arrangement recently confirmed to extend to distantly related alphaviruses. Using cryoelectron microscopy (cryo-EM), we show that mature nodavirus crowns comprise two stacked 12-mer rings of multidomain viral RNA replication protein A. Each ring contains an ~19 nm circle of C-proximal polymerase domains, differentiated by strikingly diverged positions of N-proximal RNA capping/membrane binding domains. The lower ring is a "proto-crown" precursor that assembles prior to RNA template recruitment, RNA synthesis, and replication vesicle formation. In this proto-crown, the N-proximal segments interact to form a toroidal central floor, whose 3.1 Å resolution structure reveals many mechanistic details of the RNA capping/membrane binding domains. In the upper ring, cryo-EM fitting indicates that the N-proximal domains extend radially outside the polymerases, forming separated, membrane-binding "legs." The polymerase and N-proximal domains are connected by a long linker accommodating the conformational switch between the two rings and possibly also polymerase movements associated with RNA synthesis and nonsymmetric electron density in the lower center of mature crowns. The results reveal remarkable viral protein multifunctionality, conformational flexibility, and evolutionary plasticity and insights into (+)RNA virus replication and control.

Keywords: RNA replication complex; RNA replication crown; cryo-EM; nodavirus; positive-strand RNA virus.

Fig. 1.

Nodavirus genome replication overview. (A) General mechanism of RNA genome replication by (+) strand RNA viruses. Incoming (+) RNA is used as a template for (−) strand synthesis, yielding a dsRNA replicative intermediate that templates synthesis of many copies of (+) strand genomic RNA for packaging in progeny virions. (B) The bipartite FHV RNA genome encodes RNA replication protein A on RNA1 (3.1 kb) and the capsid protein precursor on RNA2 (1.4 kb). Subgenomic RNA3, transcribed from RNA1, encodes RNAi suppressor B2 and an overlapping ORF B1, comprising the C-terminal 102 aa of protein A. Aligning FHV protein A and the alphavirus RNA replication polyprotein highlights conserved RNA capping and polymerase domains and FHV’s lack of helicase and protease domains. The N-terminal half of protein A, denoted segment A_N, and alphavirus nsP1 show particular parallels. (C) Cryo-ET imaging of a nodavirus RC spherule in infected Drosophila S2 cells with the mitochondrial outer membrane in dark blue, invaginated spherule membrane in white, interior spherule fibrils corresponding to viral dsRNA in red, and the protein A-containing crown complex in light blue. (D) Reconstruction of the 12-fold symmetric nodavirus RC crown by subtomogram averaging (2).

Fig. 2.

Baculovirus-launched nodavirus RNA1 replication. (A) Northern and western blots of Nodavirus RNAs and proteins in Sf9 cells at 48 h post-infection with baculoviruses AcR1 or AcR1∆3′ at an m.o.i. of 20. AcR1 launches wt RNA1 that self-replicates through cis-recognition by its product replicase protein A and produces subgenomic RNA3 and its translation product protein B2. AcR1∆3′, lacking the RNA1 3′ noncoding region, fails to launch self-replicating RNA1, but its expressed protein A supports replication of a trans-supplied RNA1 derivative (RNA1fs) unable to produce protein A due to an N-proximal frameshift. The FHV RNA probe also detects the primary mRNA transcript expressed from the baculovirus DNA genome and a regularly-observed virus-specific RNA species roughly twice the size of sgRNA3 (asterisk). (B) Cryo-ET imaging shows that AcR1-launched wt FHV RNA1 induces active RNA RCs bearing spherule membrane vesicles and mature crowns, while (C) AcR1∆3′-expressed protein A, lacking a functional template RNA, forms crown-like structures on the OMM but no replication vesicles. Arrowheads: black = OMM, white = inner mitochondrial membrane, red = RC vesicle, yellow = crown-like structures. Higher magnification top views of the OMM show AcR1-induced crowns with leg-like extensions (yellow arrowheads), and “leg-less” and more tightly packed (white arrowheads) AcR1∆3′ crowns.

Fig. 3.

Subtomogram averages of (A) the baculovirus AcR1-induced, RNA replication-active mature crown (green), and (D) the AcR1∆3′-induced proto-crown (pink). Upper images show side views superimposed on low-threshold density contours of the OMM, with top views below. The superposition in (B) and cross-sectional views in (C) show how the proto-crown corresponds directly to the floor and basal lobe sections of the mature crown.

Fig. 4.

Single particle cryo-EM proto-crown structure. Top, side, and bottom views of the C12 symmetric floor section in crowns from baculovirus AcR1-infected cells are shown merged with 12 copies of the Protein A Pol domain locally refined from the C12 proto-crown. (A) EM density maps with individually colored floor and basal lobe segments. (B) Ribbon diagrams of the protein A structure. The lower floor segments include domains for membrane-association (purple), capping (yellow), and central ring (red) and a functionally unassigned region (dark green). The upper Pol domain includes thumb (gold), palm (light green), and fingers (blue) domains. (C) Surface electrostatics calculated using APBS plugin in PyMol.

Fig. 5.

Structure of the nodavirus protein A proto-crown and selected comparisons to alphavirus nsP1. These and other zoomed-in views in figures below were taken from the highest resolution protein A structure obtained, which was the baculovirus-expressed C11 ring, but as noted in the Results no significant variations were detected between C11 and C12 subunit structures. (A) Fit of the atomic build of the FHV proto-crown in central cross-sections of the cryo-ET densities of the mature replication crown and the proto-crown (Left side) and atomic build of the CHIKV nsP1 crown (Right side) (13). Overlaying light gray segmentations define the OMM for the FHV crowns; positioning of the CHIKV crown on the plasma membrane (PM) is inferred from the detergent belt in the published structure. Color coding is as defined in the legend to Fig. 4B. (B) Twelve copies of the N-terminal half of FHV protein A (segment A_N, aa 1 to 452) form a flat floor in the mature crown and the proto-crown, while the equivalent CHIKV nsP1 forms a cone-shaped crown. (C) Superposition of the capping domains of FHV protein A (yellow) and CHIKV nsP1 (cyan). The close superposition of FHV and CHIKV capping and central channel domains and their distinct orientations in their respective crowns are further illustrated in Movie S3. (D) Positions of SAH and GTP in the capping domain (yellow) of the FHV proto-crown floor were estimated based on superposition of the FHV protein A structural elements on the published CHIKV nsP1-ligand interacting structures and fungal pathogen E. cuniculi. The SAM/SAH binding site is composed of R100, R104, N122, and V162, while D161, Y165, P186, W215, and R271 bind GTP. Note the 16 Å distance between the GTP α phosphate and GMP acceptor site H93 (dark purple). (E) Similarity of central ring domains in FHV protein A and CHIKV nsP1. Side- and top-views of the FHV protein A central ring domain (red) including an 18 aa linker loop (green), super-imposed on the CHIKV nsP1 RAMBO domain (cyan). Three central helical bundles (α1, α2, and α3) are sandwiched between two conserved upstream beta strands (β1 and β2) and the downstream polymerase linker in FHV protein A, whereas in CHIKV nsP1, the central helical bundles (α1, α2, and α3) are connected with two upstream beta strands (β1 and β2) followed by a membrane-binding loop (MB Loop, purple).

Fig. 6.

N-end of the polymerase linker is interposed between adjacent floor segments. (A) Bottom and (B) top views of the crown floor density, showing the first 18 aa (dark shaded region of each colored segment) of the 74 aa linker (protein A aa 379 to 452) between each floor segment and its corresponding Pol domain. See main text for further details of the linker path under and between adjacent floor segments, rising to closely approach the Pol domain. Following these 18 aa, the remainder of the linker is too flexible to visualize by cryo-EM.

Fig. 7.

Structural comparison of Pol domains of FHV protein A, Dengue virus NS5 (PDB 4C11), and Sindbis virus nsP4 (PDB 7VW5). (A) Annotated ribbon diagram of the FHV protein A polymerase, in a radial view from the crown’s central axis, showing the finger (dark blue), palm (light green), and thumb regions (gold), the catalytic site (GDD, circled in red) and the RNA template entry area. (B) Superposition of FHV protein A Pol and Dengue virus NS5 (blue). (C) Superposition of FHV protein A Pol and Sindbis virus nsP4 (cyan).

Fig. 8.

Domain assignment in the upper mature crown. (A) Superposition of the Pol ribbon diagram with both lobes of the central turret of the mature crown cryo-ET density, consistent with independent identification of these apical (2) and basal (Figs. 3–5) lobes as stacked Pol domains. (B) As shown in Figs. 3–5 and matching the A_N ribbon diagram superposition shown, the 12 floor segments (green) are the N-proximal A_N portions of protein A (aa 1 to 452, Fig. 1B) that are contiguous with each of the 12 basal Pol domains that constitute protein A’s C-proximal portion. Similarly, as detailed further in the main text, the legs (orange) are singled out as the A_N segments linked to the 12 basal Pol domains by numerous criteria, including their 12-fold repetition, position, volume, appropriately oriented membrane binding, the correspondingly good superposition of the A_N ribbon diagram illustrated, and the absence of any other unassigned density.

Fig. 9.

Asymmetric density in the crown floor central channel. (A) Top and (B) side views of the cryo-ET density of the RNA replication-active mature crown from FHV-infected cells refined by subtomogram averaging without imposed symmetry. In both graded density (Left) and density-thresholded views (Right), red arrowheads point to an asymmetric density within the 7 nm diameter crown floor central channel, distinct from the floor segment densities of Figs. 3–6. This density (colored teal in the right side views) occupies a volume closely similar to that of the apical and basal polymerase lobes.