View all hidden authors and organizations

Contributed by Paul Ahlquist, June 9, 2020 (submitted for review on April 2, 2020; reviewed by Ralf Bartenschlager and Richard J. Kuhn)

Positive-strand RNA [(+)RNA] viruses constitute the largest genetic category of viruses, including many high-impact pathogens, such as SARS-CoV-2 (COVID-19 pandemic coronavirus), MERS CoV, Zika virus, Chikungun Ya fever, dengue fever, and hepatitis C virus. (+) RNA viral genome replication always occurs in virus-induced, membrane-bound organelles, called RNA replication complexes, which is an attractive potential target for antiviral drugs with a wide range of activities. In order to better understand, control and beneficially use (+) RNA viruses, there is an urgent need to define the structure and function of RNA replication complexes at the molecular level. This study uses cryo-electron microscopy and complementary methods to provide a previously unavailable view of the native, near-atomic resolution, and well-characterized Noda virus RNA replication complex, and promotes the structure of the (+) RNA virus genome replication complex , Organization, stability and basic understanding of function.

For plus-strand RNA [(+)RNA] viruses, the main goal of antiviral therapy is genomic RNA replication, which occurs in the poorly understood membrane-bound viral RNA replication complex. The recent cryo-electron microscopy (cryo-EM) of the Noda virus RNA replication complex showed that the virus double-stranded RNA replication template is coiled in the 30 to 90 nm invagination of the outer mitochondrial membrane, and its neck-shaped pore leading to the cytoplasm is formed by Contains a 12-fold symmetric, 35 nm diameter "crown" complex of multifunctional viral RNA replication protein A. Here we report the optimization of cryo-EM tomography and image processing to increase the crown resolution from 33 to 8.5 Å. This decomposes the crown into 12 different vertical segments. Each segment has 3 main subdomains: a membrane-connected basal lobe and a parietal lobe, which together form a central turret with a diameter of approximately 19 nanometers, and a sub-domain from the base. The outstretched legs of the leaves are connected to the membrane with a diameter of approximately 35 nm. Although the diameter of replicated vesicles varies greatly, the resulting two loops of membrane interaction sites restrict the vesicle neck to a highly uniform shape. Labeling protein A with a His tag combined with 5 nanometers of nickel and nano-gold allows the cryo-EM tomography of the C-terminus of protein A to map to the parietal lobe, which is closely related to the predicted structure of the protein's C-proximal polymerase domain A. These And other results indicate that the crown contains 12 copies of protein A, which are arranged from bottom to top in the direction of N to C. In addition, apical polymerase localization has important mechanical implications for template RNA recruitment and (-) and (+) RNA synthesis.

Positive-stranded [(+)RNA] viruses contain single-stranded messenger RNA and replicate their genome in the cytoplasm of infected cells without any DNA intermediates. This (+) RNA viral genome replication always occurs in complexes associated with the inner membrane of the host cell, and these complexes have undergone an amazing virus-induced rearrangement. Many (+) RNA viruses, including human and animal alphaviruses, the larger alphavirus-like superfamily, many flaviviruses, Nodaviruses, and other viruses, replicate their genomes in "small balls" of 30 to 120 nanometers, These globules are host membranes whose sites vary with specific viruses (1, 2). In addition, some coronaviruses produce similar globules, which are connected to double-membrane vesicles in space and time, and the latter are more often associated with coronavirus RNA replication (3). This RNA replication complex integrates replication factors and templates, coordinates the successive steps of (-) and (+) RNA synthesis, and isolates double-stranded RNA (dsRNA) and uncapped RNA replication intermediates from innate immunosensing . The central effector of these processes are non-structural viral genome replication proteins, which target and reshape membranes, recruit additional viral and cellular proteins and viral RNA templates, and provide enzymatic activity for viral RNA synthesis and the usual 5'm7G capping Function.

Classical electron tomography greatly elucidates the dramatic membrane rearrangement associated with this type of (+)RNA virus replication complex (4⇓ ⇓ –7). However, due to the multiple limitations of traditional electron microscopy (EM) sample preparation and heavy metal staining visualization, such studies lack information about the organization of viral proteins and RNA at these sites. At the other extreme, proteins have been purified by X-ray crystallography and single-particle cryo-electron microscopy (cryo-EM) (8⇓ ⇓ ⇓ ⇓ ⇓ ⇓ ⇓ ⇓ –17 ). Although these structures provide important insights into the mechanism of (+) RNA viral genome replication, such studies often lack information about the organization of proteins involved in natural, membrane-bound RNA replication complexes.

To make up for this gap, we turned to high-resolution cryo-electron tomography (cryo-ET), which can preserve the natural structure of the (+)RNA virus replication complex, not only reveals the membrane structure, but also reveals the virus replication protein. Structure and organization, as well as the RNA in the replication complex. For this type of research, the most studied model at present may be Noda virus chicken house virus (FHV). FHV is a simple (+)RNA virus that expresses three RNAs (genomic RNA1 and RNA2 and subgenomic RNA3) and four proteins (protein A, capsid, B1 and B2) (Figure 1) (18). Protein A (998 aa) is the only viral protein required for globule formation and RNA replication (19). It has two known enzymatic domains: N-proximal RNA capping domain and C-proximal RNA dependent Sex RNA polymerase (RdRp). Protein A also has two membrane binding domains: a complete N-terminal transmembrane domain, which is consistent with the significant membrane interaction after deleting the transmembrane fragment, and the predicted membrane interaction region in the iceberg domain (Figure 1 ) (20⇓ ⇓ –23). Finally, the function of protein A in self-interaction and RNA replication requires capping, iceberg, and various multimerization regions within the RdRp domain (24).

Genomic organization and protein ORF of FHV. The bipartite FHV genome (+)RNA1 and (+)RNA2 express protein A and capsid protein, respectively. Protein A is the main FHV RNA replication protein and is involved in various functions such as membrane binding (transmembrane domain and possible membrane interaction domain in the iceberg region), replication complex formation, RNA 5'capping and RdRp enzyme activity . The various functional domains of protein A are highlighted in the ORF, and their amino acid coordinates are marked above the ORF. Note that the core RNA capping domain and the adjacent iceberg domain shown both seem to contribute to the RNA capping function (20). Protein A replicates viral RNA to form a negative sense product, which may be dsRNA. In the process of negative strand synthesis, the co-terminal (-)RNA3 at the 3'end of FHV RNA1 is synthesized, and then replicated to produce (+)RNA3. RNA3 encodes the B1 and B2 proteins. B1 is a priority nuclear protein with unknown function, and B2 is a mature RNAi inhibitor.

FHV forms its globular RNA replication complex as a 30 to 90 nm invagination of the outer mitochondrial membrane (OMM) (5, 25, 26). Using this FHV model system, our group has recently demonstrated that mitochondria isolated from FHV-infected Drosophila cells maintain intact and highly active RNA replication complexes. When imaged by frozen ET, these complexes produce positive effects on RNA, Proteins, and membranes in replication complexes (25). Among other findings, these studies revealed a striking, 12-fold symmetrical "crown" complex containing protein A, located above the cytoplasmic side of the globular neck, and using preliminary sub-tomograms to average the crown structure. For imaging, the initial resolution is 33 Å. 25).

In this study, we combined multiple advances in sample preparation, cryo-ET image acquisition, image processing, and sub-tomogram averaging to increase the crown resolution by nearly four times to 8.45 Å. The results revealed the unique structure of the crown. Each of the 12 different crown segments has three sub-domains, two membrane contacts, and multiple crown construction interactions with the flanking segments. The C-proximal RdRp domain was mapped to the apex of the crown using the engineering tag in the chimeric protein A with RNA replication capability. The results show that the crown is mainly (if not completely) composed of 12 copies of viral protein A, elucidating the protein A-membrane and protein A-protein A interaction, stabilizing the globules, and revealing that each protein A is usually from the substrate N to the crown of the top C domain. This positioning has different mechanical meanings for RNA template recruitment, (-)RNA synthesis, and (+)RNA synthesis and capping. Overall, these findings provide a solid foundation and specific hypotheses for further experiments, and provide substantial mechanism significance for a broader understanding of (+) RNA viral genome replication.

Frozen ET samples were produced by infecting Drosophila S2 cells with purified FHV, followed by mitochondrial isolation and freezing. The relatively thick nature (>200 nm) of the mitochondrial sample on the frozen grid reduces the signal-to-noise ratio in the tomography (27, 28). In order to minimize the impact of this sample thickness and increase the high-resolution information in cryo-tomography, we have adopted a variety of emerging practices to improve cryo-ET data collection and processing, including motion correction through dose segmentation (29), using Dose symmetric tilt scheme (30), lower 3.5-μm defocus and 3D contrast transfer function correction (31).

Figure 2 A and B show the sample portion of the reconstructed tomography, showing many small balloon vesicles adjacent to the OMM, containing densely coiled fibrils and covered by a corona (Figure 2A, C, and E). In different positions of the same tomographic image, the crown structure can be seen from the top view (Figure 2 B, D and F). Even without sub-fault averaging, many key structural features, such as the central turret, the regularly spaced projections from the central turret, and the central density within the turret, are obvious.

The frozen ET procedure has been improved to visualize the FHV replication complex in the mitochondria. (A) Part of the reconstructed tomographic image showing the side view of the FHV replication complex. The various balls and OMM are highlighted with black arrows. (B) A cross-section of the reconstructed tomographic image, showing the top view of the OMM with the FHV replicated composite crown. (C and D) The illustrations of A and B show a magnified view of a single sphere and three top views of the crown, respectively. (E) The false color representation of C highlights the various densities and subcellular regions of the replication complex. Blue, coronal side view; white, OMM; red, densely coiled inner fibers; translucent orange, cytoplasm; and translucent green mitochondrial membrane space (IMS). (F) False color representation of D, highlighting the top view of OMM (translucent white) and crown (blue).

A total of 4,640 crowns from 59 tomographic images were used for sub-tomographic image averaging. First, 1,608 manually picked crowns and IMOD/PEET software (32, 33) were used to generate a 24 Å resolution sub-tomogram average crown template. This template is used to identify more crowns in the tomographic image through the automatic template matching function in emClarity (34) (Movie S1), which increases the total crown images by nearly three times to 4,640. In addition to increasing the output of crown images, this automatic selection overcomes potential user bias in particle selection, recognizes crowns in multiple directions, registers crown orientations without user input, and uses low defocus ( Low contrast, higher resolution) tomographic images.

Use emClarity software suite (34) to perform further sub-tomogram averaging. The crown has a 12-fold symmetry (c12) (25), which is evident in the top view of a single crown protruding from the central turntable at regular intervals (Figure 2D), especially when examining multiple z-series facades To better view the complete circle (movie S2). The initial sub-tomogram without imposing any symmetry produced an average of 10.37 Angstroms of the crown structure, showing c12 symmetry (SI appendix, Figure S2 AC). Therefore, in order to further improve the resolution, we applied c12 symmetry during the iterative sub-tomographic averaging process, resulting in a crown image with a resolution of 8.45 Å, determined by the Fourier crust correlation (SI appendix, Figure S2D). Our visual inspection and preliminary calculation of sub-categories failed to reveal structures lacking 12-fold symmetry. Therefore, an incomplete crown (if present) does not represent an important structural category.

Figure 3 and movie S3 show how the approximately four-fold increase in resolution compared to the previous image (25) shows how the crown is composed of three main sub-domains. The central turret with a diameter of about 19 nm consists of a basal domain or leaflet connected to the OMM (Figure 3A, inset, and Figure 3B) on its foundation, and supports the parietal leaf directly above. The parietal lobe is about 6 nanometers high and about 4.5 nanometers wide, while the basal lobe is about 7 nanometers high and 4.5 nanometers wide. In addition, the left side of the outer surface of the basal lobe anchors a leg domain that extends radially outward from the basal lobe about 7 nm, and then bends downward to form a second connection with the OMM with a diameter of about 35 nm (Figure 3 C And E). Previous images of the lower resolution of the crown showed that the lower part of this leg was a projection slightly inclined towards the central turret, although the connection to the basal lobe was not resolved (movie S3) (25). It is worth noting that the improved imaging used here makes this connection to the basal lobe visible even in the main tomographic image, before sub-tomographic image averaging (Figure 2D).

The subfault map of the crown structure reveals important new structural details on average. (A) Side view of the high density threshold of the crown (contour level = 4.2). The gray dashed line through A, B, and C is mapped to OMM as a reference point. (Inset) The area in A marks the three structurally different areas in the crown: parietal lobe, basal lobe, and legs. (B) Cross-sectional view of the low-density threshold crown (from C), as a transparent white shell, encased by the blue high-density crown (from A). The asterisk indicates the unusual density bridge between the dense head group regions of the lipid bilayer, and the dashed line below the crown depicts the remodeled OMM continuing to enter the main body of the globular RNA replication vesicle. (C) The low density threshold of the crown (profile level = 2). The side view shows the additional density contribution of the membrane. (D) The top view of the high-density threshold crown highlights the 12-fold symmetry of the crown complex. (E) Bottom view of high-density threshold crown.

In addition to the vertical connection of the basal and parietal lobes, the horizontal interaction connects each pair of adjacent apical domains and each pair of adjacent basal domains to form a crown ring (Figure 3A, inset and Figure 3D). Although obvious, if the electron density threshold is increased to selectively show the highest density regions (movie S4), these circular interactions are not as pronounced as the vertical base-to-apical domain connection and disappear faster. The discussion considers the potential relationship between these lateral interactions and the multimerization interactions mapped within protein A (24). The electron density map is sectioned step by step through continuous top and side view planes to show that internal density changes indicate protein secondary structure and further illustrate the different properties of the top, base, and leg domains (Figure 4 and Movie S5).

The two-dimensional slice of the coronal electron density map shows the higher resolution substructure. (A and B) High-density threshold side view of the crown. The dashed line indicates that the slice shown in B has been rotated by 90°. (C and D) High-density threshold top view of the crown. The dashed line represents the slice shown in D after 90° rotation. (Scale bars in B and D, 10 nm.)

Although the diameter of potentially replicating vesicles varies in several categories from ~30 to 90 nm (Figure 2A) (25), the sub-tomogram averaging the 4,640 crowns used also closely strengthens the correlation with the crown. The structure of the adjacent vesicle neck (Figure 3) B and C). Therefore, unlike the different curvature of the vesicle body, the membrane shape and curvature of the neck are highly uniform. This close alignment of the membrane neck structure and the crown means that the crown is not simply passively located at the top of the vesicle neck. On the contrary, like the neck of a pressurized balloon, strong coronal interaction seems to play a major role in forming and stabilizing the neck, thereby preventing high-energy replicating vesicles from exploding due to electrostatic repulsion of tightly packed viruses. dsRNA.

Especially with the improvement of the overall resolution, the electron density threshold of imaging (Figure 3C) can be adjusted to distinguish between the higher electron density protein crown (Figure 3A) and the adjacent lower electron density lipid bilayer. Figure 3B shows a cross section in which a higher density protein crown (opaque blue) is wrapped in a lower density threshold image (transparent white), where the main extra volume comes from the flanking flanks of the membrane neck of the replicated vesicles. OMM. It is worth noting that the polar head groups on each side of the lipid bilayer have a higher electron density than the fatty acyl chain in the middle, causing the membrane to appear as two parallel layers about 3 nm apart (35). Interestingly, in the area between the two membrane contacts of the crown, the two head group layers appear to be closer to each other or bridged by unknown density (Figure 3B, asterisk). The possible reasons for this effect were considered in the discussion.

Inside the central turret, near the level of the OMM, the top appears to be partially or mostly enclosed by the electron density grid or the bottom (Figure 3 B, D, and E). Before averaging the sub-tomograms, a similar local density at this level can also be clearly seen in the side view of a single crown (Figure 2C and E). In the average structure, the center of the floor is occupied by a more slender cylindrical density (Figure 3B), which may be related to the outer filaments, which usually enter the extra mitochondrial space from the crown of tens of nanometers and may represent Newborn offspring RNA (25). Interestingly, the top views of individual crowns (Figures 2D and F) often show asymmetric densities in the central turret, which can be averaged by subfractal maps and may have a significant impact on the "floor" density in Figure 3B. Due to their significant heterogeneity, efforts to reconstruct these center densities asymmetrically have not yet produced informative local sublevel map averages. It is currently uncertain whether, in addition to any protein components, a part of the floor density may be contributed by locally restricted fragments that occupy replicating vesicles and are believed to represent the internal filaments of the viral genome dsRNA (25). 2 A and C).

Protein A is the only viral protein required to form the Noda virus RNA replication complex, and it is also the main component of the crown complex (19, 22, 24, 25). In addition, protein A is a highly self-multimerizing OMM-related protein that supports RNA synthesis and RNA capping enzyme activity in the virus life cycle (19, 22⇓ –24) (Figure 1).

In order to better understand the organization and function of protein A in the crown, we tried to further map the position, direction and stoichiometry of protein A with the help of appropriate tags. Although in many positions of protein A, even short epitope tag insertion will eliminate RNA replication, we found that the C-terminus received a 16-aa GFP11 peptide tag while retaining 80% of the cis-replication activity of WT RNA (Figure 5A) And B). As part of the split-GFP system (36), this tag can supplement fluorescence with its binding partner GFP1-10. In addition, the C-terminal tag is advantageous because the C-terminal is adjacent to the protein A RdRp domain, which is a domain of high interest (Figure 1). The biologically active protein A-GFP11 fusion protein is expressed by a DNA plasmid that transcribes FHV RNA1 from the baculovirus immediate early (IE1) promoter in transfected Drosophila S2 cells, and initiates the replication of genomic RNA1 and its product subgenomic RNA3 (Figure 5B). Maintain the expression of RNA1-derived subgenomic RNA3 and its encoded B2 RNAi inhibitory protein to avoid otherwise effective RNAi inhibition of replication (37). However, in the related plasmid pRNA1-GFP11-ΔB1, the translation of protein B1, the other product of RNA3, was knocked out by the M897L change at the start methionine codon of B1 (Figure 5A). This is done because the B1 ORF and the C-terminal 102 aa of protein A are in frame, so B1 expression will produce a second GFP11-labeled protein, which complicates the positioning of GFP11-labeled protein A. Previous results indicate that knocking out out B1 expression does not inhibit FHV infection and RNA replication in Drosophila DL-1 cells (38).

The C-terminus of protein A is exposed to the mitochondria in cells that replicate FHV. (A) The RNA1-GFP11-ΔB1 construct contains three different changes: 1) M897L amino acid change to knock out B1 expression, 2) *999R amino acid change to extend Protein A ORF, and 3) C-terminal extension of Protein A ORF Express the 16-aa GFP11 sequence (amino acids 1,008 to 1,023). RNA1-His-ΔB1 destroys the GFP11 sequence and adds a His tag at amino acid positions 1010 to 1015 in protein A. (B) Northern blot analysis of RNA isolated from cells transfected with the designated FHV RNA1 expression plasmid to compare the replication ability of the encoded protein A fusion in cis RNA. fs = RNA1fs, GFP11 = RNA1-GFP11-ΔB1, His = RNA1-His-ΔB1, WT = RNA1-WT. Ribosomal RNA (rRNA) shown as loading control. RNA1fs is a full-length RNA1 derivative in which the engineered frameshift blocks the expression of protein A (39). (C) Wide-field fluorescence microscope of S2 cells. (Top) cells transfected with the expression plasmids RNA1-WT and GFP1-10 and (bottom) cells transfected with the RNA1-GFP11-ΔB1 and GFP1-10 plasmids. Immunostained protein A is shown in red, and GFP fluorescence is shown in cyan. The merged image includes the three channels on the left. (D) In ​​vitro complementation of FHV-WT or FHV-GFP11-ΔB1 mitochondria infected with S2 cells by FHV-WT or FHV-GFP11-ΔB1. Fluorescence of MBP-GFP1-10 and GFP purified from Escherichia coli was measured over time. (E) Wide-area immunofluorescence analysis of S2 cells infected with FHV-WT or FHV-His-ΔB1. S2 cells coexist with antibodies against protein A and His tag.

In order to assess whether the C-terminus of protein A exposes the FHV RNA replication site on the mitochondria, we co-transfected Drosophila S2 cells with a plasmid expressing GFP1-10 and pRNA1-WT expressing WT RNA1 (Figure 5C, top) , Or pRNA1-GFP11-ΔB1 (Figure 5C, bottom). Fluorescence microscopy showed that protein A-GFP11, instead of WT protein A, successfully supplemented GFP1-10, and the resulting GFP fluorescence was co-localized (Pearson correlation coefficient [PCC] = 0.88 ± 0.1, 88 cells were analyzed), As expected for protein A immunofluorescence (Figure 5C). In order to test whether GFP1-10 can complement OMM-related protein A on the pre-formed replication complex, we used pRNA1-GFP11-ΔB1 to generate infectious FHV virus particles, which express the protein A-GFP11 fusion during infection , And can be in the body (movie S6). We isolated mitochondria from S2 cells lacking GFP1-10 and infected with WT FHV or FHV-GFP11-ΔB1 virus particles. These samples of mitochondria were added or not added with maltose binding protein (MBP)-GFP1-10 purified from E. coli, and GFP fluorescence was measured over time. As shown in Figure 5D, mitochondria from RNA1-GFP11-ΔB1 FHV-infected cells are complementary to GFP1-10, but not a negative control lacking GFP11 or GFP1-10. GFP fluorescence continues to increase over time. In summary, these results indicate that the mitochondria in FHV-infected cells carry a large amount of protein A, the C-terminus of which is exposed to the cytoplasm and accessible to the GFP1-10 protein, rather than being enclosed behind a barrier of membranes or other protein domains.

In order to determine the position of the exposed C-terminus of protein A in the crown, we combined Ni-NTA-nano-gold labeling and EM. This Ni-NTA-nano-gold label has successfully mapped the position of hexahistidine (His)-labeled protein domains in different protein complexes. Part of the reason is that the high electron density of gold makes the labeling complex easy to identify. Used for sub-fault map averaging (40⇓ ⇓ – 43). Ni-NTA-nano-gold also provides relatively accurate His tag positioning, because no additional protein is involved, and the distance from the gold particle to the tag is only 1.8 nm.

Therefore, we generated the pRNA1-His-ΔB1 expression plasmid by replacing 4 aa in the GFP11 tag of protein A-GFP11 with histidine residues to generate the His tag (Figure 5A). The resulting RNA1-His-ΔB1 retained 58% of the WT RNA replication activity in the transfected S2 cells (Figure 5B). Using pRNA1-His-∆B1, we produced infectious FHV-His-∆B1 virus particles, expressed protein A His tag fusion when S2 cells were infected, and confirmed the presence of His tag by immunofluorescence. As shown in Figure 5E, S2 cells infected with FHV-His-ΔB1 and antibodies against protein A and His tag produced colocalization signals (PCC = 0.94 ± 0.05, 100 cells analyzed), while S2 cells infected with WT FHV stained Has antibodies against protein A, but not against the His tag.

In order to label the His tag on protein A with gold nanoparticles on the crown, we infected S2 cells with FHV-His-ΔB1 or WT FHV virus particles, and isolated mitochondria 16 hours after infection. The mitochondria were incubated with 5-nm Ni-NTA-nano-gold particles at 4°C for 1 hour, and then quickly frozen (Figure 6A). Cryo-ET imaging shows highly specific, high-density nano-gold labeling of mitochondria from cells expressing protein A-His instead of WT protein A (Figure 6B). In addition, for the mitochondria of cells infected with FHV-His-∆B1 virus particles, the nano-gold label especially appeared on the OMM segment decorated with small balls, but not on the OMM segment without small balls (movie S7). This confirms that protein A can only be detected at the RNA replication site on OMM, which is consistent with our previous immunogold labeling results using antibodies against protein A (25). Please note that in this experiment, intact, unsliced ​​mitochondria were used. Since OMM and the globule membrane prevent Ni-NTA-nano-gold from entering the globule (discussion), it will not detect any possible presence in the replica globule的protein A.

The site-specific gold nanoparticles map the protein A polymerase domain to the distal lobes of the crown. (A) The experimental protocol for infection, isolation and natural labeling of His-tagged protein A with 5-nm Ni-NTA-nano-gold for cryo-ET imaging. (B) Reconstructed tomogram of mitochondria separated from FHV-WT (left) or FHV-His-ΔB1 (right) after nanogold labeling. The white arrows indicate the 10 nm reference gold used for alignment during the tomographic reconstruction of the two samples. The yellow arrow indicates 5-nm Ni-NTA-nano-gold. (Inset on the right) shows two separate spheres marked by nano-gold. (C) Representative nanogold labeled sub-tomograms from the higher resolution frozen ET dataset. (D) The average of sub-tomograms of nano-gold labeled and unlabeled crowns. Unlabeled = 141 unlabeled crowns used for averaging. Unmarked + Nano-Gold Marking = 141 unmarked crowns + 184 gold-marked crowns are included in the average. Overlay = The grid view of the blue unmarked crown and the yellow unmarked + nano-gold marked crown are superimposed. The black arrow highlights the main additional density contributed by the nanogold appearing on the crown.

As shown in Figure 6B, the right inset and the representative example in Figure 6C, a single globule with a nanogold label uniformly displays the mark above the neck of the RNA replication vesicle (i.e., in the crown, usually in the tomographic area). It is directly visible in the scan) at about 15 nm from the surface of the OMM. Combining the 1.8 nm connector distance between the nano-gold and His tag and the 14 nm height of the central turret from the membrane, the nano-gold is roughly positioned at the apex of the crown.

To confirm this, we collected a higher-resolution cryo-ET dataset (representative example in Figure 6C) and used IMOD/PEET on 141 unlabeled crowns individually or with 184 from the same tomographic scan. A combination of nano-gold-labeled crowns is subjected to iterative sub-tomography averaging to control any changes in the crown structure due to sample preparation. When performing sub-tomography of the labeled crown, we found that the high electron density nano-gold particles dominate the alignment process of the crown, causing the program to orient the labeling complex superimposed by the nano-gold particles, but the protein crown often rotates and does not align. In order to avoid this distortion, we use a defined mask to align the labeled complexes only by referring to the protein density of the corona, eliminating the influence of nano-gold on the alignment. The resulting arrangement is then used to average the sub-tomograms of the entire complex, including protein crowns and nano-gold, so that the density of nano-gold rotates around the entire crown ring to average. Comparing the average value of the unlabeled crown (Figure 6D, left) with the combination of unlabeled and nano-gold labeling (Figure 6D, center), the average additional density of nano-gold particles appears at the top, around and center of the crown Above the top lobe of the turret (Figure 6D, center and right). These results indicate that the C-terminus of protein A is located at or very close to the top of the parietal lobe.

The C-terminus of protein A is located at the top of the crown parietal lobe, adjacent to the RdRp domain (Figure 1). Despite the differences in amino acid sequence, the polymerase domain in (+)RNA viruses is structurally highly conserved (44). To date, there are no atomic or near-atom resolution structures available for nodaviral protein A or any of its functional domains. Therefore, we used the iTasser software suite (45) to predict the structure of the polymerase domain of FHV protein A (amino acids 453 to 896). On the left side of Figure 7A, the predicted structure of FHV polymerase (Pol-Pred) is shown, where the template entry, NTP entry, and product exit sites are shown, and the product exit hole faces the observer. iTasser predicts that the FHV protein A polymerase domain has a high degree of structural similarity with RNA polymerase from the Flaviviridae family, and has the highest similarity with the Pestivirus in this family. Also as shown in the matching direction in Figure 7A, the first four structurally similar polymerase domains come from classical swine fever virus (CSFV), bovine viral diarrhea virus (BVDV), Japanese encephalitis virus (JEV) and Zika virus ( ZIKV)) has 97.8% of CSFV structure aligned residues to 90.8% of ZIKV.

The protein A polymerase secondary structure is predicted and fitted to the EM density map. (A) iTasser predicted the secondary structure of the protein A polymerase domain (Pol-Pred, amino acids 453 to 896), shown on the left, and the product export side on the front. The first four iTasser structural analogs shown (from high to low, left to right) are in the same orientation as the polymerase domain predicted by FHV. (B) Fit the Pol-Pred manual C end up and down to the EM density map. The Pol-Pred structure shows rainbow colors from N to C end, from blue to red. The N-to-C color key is also displayed on the right. (C) The first two fits (based on correlation scores) obtained by fitting Pol-Pred to the segmented apical lobes using the rigid body docking tool in Segger.

As shown in Figure 7A, Pol-Pred and Viral RdRps are generally wider in the finger area than in the distal area of ​​the thumb. The EM density map of the apical lobe has a wider base that tapers toward the apex (Figure 3A, inset). Therefore, the best fit between Pol-Pred and the EM density envelope of the apical lobe is obviously the finger area at the bottom and the thumb area at the top (Figure 7B). Interestingly, this places the C-terminus at the top of the apical lobe (see the NC ribbon color gradient in Figure 7B), which matches the nanogold result (Figure 6D). In addition, we use the Segger rigid body docking method (46, 47) to find the best fit of the Pol-Pred with the highest correlation fit to the apical lobe by allowing 100 rotations around the three main axes. Consistent with our manual docking results, the first two fits provided by Segger also dock Pol-Pred with the finger field at the bottom and the thumb field at the top of the parietal lobe, although the difference between the two fits is that they rotate around 180° The y-axis (ie, the vertical axis in the plane of the figure) (Figure 7C). Since the two fitted butt scores are not significantly different, the current resolution of the crown structure does not seem to allow a clear prediction of the direction of the Pol-Pred around the y-axis (SI Appendix, Figure S3). However, in general, the nano-gold mapping and these modeling results are consistent with the general N to C direction of protein A from the base to the parietal lobe. The polymerase domain is located in the parietal lobe, and its N-terminus is located at the interface. Between parietal and basal lobe. The discussion further considered the mechanical significance of the position of this polymerase in the replication of the coronavirus genome.

This study combined improved sample preparation, frozen ET data collection, and image processing to increase the resolution of the Noda virus RNA replication complex by nearly four times. The resulting 8.45-Å structure shows that the crown is composed of 12 parts, each with top, base and leg domains and 2 membrane binding sites (Figure 3A, inset and Figure BD). Below we will discuss the relationship between these features and the multifunctional viral replication protein A, the mapping of the RdRp domain of protein A to the parietal lobe, and the mechanical influence on the continuous RNA replication step. It may be possible for many other (+)RNA viruses with similar characteristics. Has a wider impact. RNA replication vesicles.

The multiple aspects we found mean that the crown (Figure 3A, inset) is mainly or entirely composed of 12 copies of protein A. The nanogold tag maps the C-terminus of protein A to the top of the coronal apical domain (Figure 3A, inset). 6), its 12-fold repetition indicates that the crown contains 12 copies of protein A. Given that the average density of globular proteins is 1.35 g/cm3 (48, 49) and a 20% variation in the bulk density between proteins (50), protein A (998 aa) is expected to occupy a volume of approximately 125 to 150 nm3. This is a reasonable match with the estimated 150 nm3 volume of the single crown segment (top + base + leg domain) shown in Figure 3A, especially because Figure 7 and Movie S4 show that the electron density threshold of Figure 3A is highly conservative and may be overestimated the actual amount.

The tissue of each coronal segment with basal and parietal lobe and two membrane contact points (Figure 3A and B) also interacts with the two enzyme domains of protein A (methyltransferase-guanylate transferase [MTase-GTase] Parallel to RdRp) (Figure 1) and present understanding of protein A-membrane interactions. Protein A at its N-terminal 36 aa (23) has a mitochondrial targeting transmembrane domain and additional downstream membrane interaction capabilities, because after deleting the N-terminal transmembrane domain alone or deleting the flanking sequence of amino acid 245, 40 to 64 %Protein A is still separated from the membrane (23). One possible site for this downstream membrane binding is the iceberg region of protein A (amino acids ~216 to 391) (Figure 1), which is similar to the conserved membrane-bound iceberg region of the transalphavirus superfamily (20).

In addition, the lateral interaction that connects the apex to the apex and the basal to the basal lobe to form the crown ring (Figure 3A, inset) matches the previous mapping of multiple independent protein A multimerization domains in the MTase-GTase core, Iceberg, and RdRp Region (24). In these regions, alanine substitution mutations that inhibit protein A multimerization also inhibit RNA replication. The results of the study show that multiple interactions pull two or more copies of protein A together, with at least some necessary functions, but the nature and stoichiometry of the complex are still unclear. The results shown here indicate that 12 copies of protein A exist in the crown like wooden sticks in a barrel, and multimerization interactions connect the fragments like a hoop on the barrel.

The general direction of protein A in the crown is defined by the nanogold mapping (Figure 6) where we map the C-terminus to the top of the crown and the transmembrane nature of the N-terminus, so it must be located in one of the two membranes below the basal lobe and the leg region Point of contact (Figure 3B). Since the core region of MTase-GTase (amino acids ~91 to 215) (Figure 1) is directly connected to the N-terminal transmembrane segment, it must be adjacent to the membrane. These considerations mean that the N-terminal half of protein A, from the N-terminal transmembrane region, the MTase-GTase core to the ice region that may interact with the membrane, constitute the entire crown base region, including membrane interaction sites and basal lobes. Although the enzymatically active RNA capping regions, including the MTase-GTase core and some flanking iceberg sequences, may be candidates for occupying the basal lobe, assigning specific positions to these sequence elements in the basal coronal structure requires higher imaging resolution Or map the results further.

Therefore, the transmembrane anchoring of the MTase-GTase core and basement membrane and the C-terminus of the crown apex indicate that the C-proximal RdRp must occupy the remaining parietal lobe. This RdRps has a highly conservative fold. Figure 7 shows that the predicted flavivirus-like fold of FHV RdRp is very suitable for parietal lobe. Obviously, it prefers the narrower C-terminus at the top, which matches the nanogold pattern, and the wider N-bottom end. Again, it is consistent with the association between the AN end of the whole protein and the basement membrane. The substantial binding surface between the bottom of each apical domain and the top of the underlying basal domain (movie S4) indicates that there is a strong connection between these domains, but does not indicate whether the binding is intramolecular or intermolecular . Therefore, a single protein A may constitute a pair of vertically stacked basal and apical domains, or the basal lobe may combine with more spirals of the left or right parietal lobe (Figure 3A). Solving this intramolecular connection between the parietal and basal lobes may ultimately require the analysis and tracing of the α-carbon skeleton of protein A.

Each coronal segment interacts with the underlying membrane at two different points flanking the highest curvature of the lipid bilayer (Figures 3B and 8B). Through these contacts, the corona tightly confines the neck of the replicated vesicle to a constant diameter and curvature, although the diameter of the associated vesicle varies greatly (Figure 2A) (25), it is strengthened in the subtomogram average. The 12 repetitions of these membrane contacts in the crown may be important to provide sufficient force to maintain a strong local neck curvature against the highly deformed vesicle membrane and to replicate the back pressure of the densely coiled, electrostatically repelled dsRNA within the vesicle (Figure 2C And E) (25). In the membrane area between two protein A membrane contacts, the density of the two leaflets bridging the bilayer (Figure 3B, asterisk) may be related to the protein A transmembrane domain (23) or abnormal strain on the membrane. High curvature area.

The dodecamers of protein A are arranged to form a corona complex. (A) An oblique side view of the crown, with alternating parts of the blue and white 12-fold symmetrical crown. The floor density is displayed in transparent white with a black outline. (B) A single predicted protein A fragment is shown in the side view in two colors, the mapped polymerase domain is yellow, and the rest of the protein AN end density is blue. The rest of the crown is displayed in transparent white with a black outline, and the highlighted crown can be viewed without obstacles. The asterisk marks two independent membrane interaction sites for each coronal segment. The linear plot of protein A from the N-terminus to the C-terminus is shown on the right, with the same color scheme as the density plot. The asterisk in the protein A line graph indicates the well-characterized N-terminal transmembrane domain TM and possible membrane interaction domains in the iceberg region.

The positioning of RdRp at the apex of the crown has mechanical significance for the steps of the entire RNA replication cycle. This positioning seems to be particularly consistent with the early steps of RNA replication. The viral (+)RNA template is recruited to OMM before (-)RNA synthesis, and the protein A RdRp domain needs to be bound to its membrane binding site (39). The exposed top RdRp domain is great for capturing incoming template (+) RNA, especially because it is repeated 12 times. After (+)RNA template capture, (-)RNA synthesis is required to form Noda virus replication complex vesicles (19), which may be by directing dsRNA products into the membrane to expand the vesicles (25). Consistent with this, RNA polymerase is a powerful molecular motor that can generate force ≥30 pN (51). The dodecamer nature of the crown and the double loop at the membrane interaction site will strongly anchor RdRp because it exerts this force to expand the vesicle.

In contrast, it is difficult to explain the role of top RdRp in (+) RNA synthesis, if any. First, the dsRNA template used for (+)RNA synthesis is isolated in the replication complex vesicle, at least 9 nm from the RdRp of the crown tip. (+) RNA synthesized by the top RdRp requires dsRNA template to circulate out of the vesicle and return, complicating RNA synthesis and exposing dsRNA to IFN stimulation and recognition of the RNAi innate immune pathway, thus defeating the main benefit of dsRNA-protecting vesicles . Second, the m7G capped membrane junction core MTase-GTase domain for progeny RNA is located between the top RdRp and the vesicle-isolated dsRNA template. Since the new (+) RNA must be synthesized, capped, and released in this order, the product (+) RNA must also perform a complex circular path inside the corona before exiting to the cytoplasm.

These challenges indicate that the RdRp activity of (+)RNA synthesis may be provided by another non-corona form of protein A, which has better access to the dsRNA template. When the embedded and sliced ​​infected cells were immunogold-labeled with antibodies against protein A, evidence of protein A non-corona pools in the Noda virus replication complex appeared, revealing a large number of markers inside the RNA replication vesicles (5). This protein A in the replication vesicle, whether in the inside of the vesicle neck or further away, may have a function similar to that of the reovirus core synthetic (+) RNA containing dsRNA, in which the internal RdRp replication is isolated in the core In the dsRNA template, the nascent (+)RNA is released through the coronal pentamer ring of the viral protein as the exit and RNA capping channel (52, 53). It may be consistent with this that the top view of the crown in the reconstructed tomography usually shows the asymmetric density near the inner bottom of the crown (Figure 2 B, D, and F). These densities may be related to the long fibrils that often emerge from the crown, as expected, the release of new offspring (+) RNA into the cytoplasm (25), the replacement pool of protein A, or both. The results presented here provide a solid foundation for further structural and functional studies to solve these and other basic problems in (+)RNA virus genome replication.

All FHV RNA1 WT expression plasmids are derived from the pIE1hr5/PA backbone, as previously described (19), except for one modification. By introducing a self-cutting hammerhead ribozyme sequence (54) upstream of RNA1 5'UTR, the RNA1-WT plasmid was further optimized to support higher levels of RNA1 replication. In order to generate plasmid RNA1-GFP11-ΔB1, first the Drosophila codon-optimized GFP11 sequence (encoding aa RDHMVLHEYVNAAGIT*) (36) was inserted into the expression plasmid at nucleotide position 3061 (FHV RNA1 coordinates) by overlapping PCR. The T3034G mutation (FHV RNA1 coordinates) was also introduced to change the protein A stop codon to arginine. Second, because the introduction of the GFP11 sequence disrupted the FHV RNA1 3'UTR, the first 23 nucleotides of the 3'UTR were replicated downstream of the GFP11 stop codon to restore the complete 3'UTR sequence to achieve optimal RNA1 replication. Finally, in order to knock out protein B1 expression, the A2728C mutation (FHV RNA1 coordinates) was introduced by overlapping PCR to knock out B1 expression, which also introduced the M897L change in the protein A ORF. To generate RNA1-His-ΔB1, the expression plasmid RNA-GFP11-ΔB1 was modified to introduce M1011H, V1012H, L1013H, and E1015H residue changes in the protein A ORF, thereby generating a His tag. The RNA1-fs plasmid has been previously described (39). The FHV RNA2 expression plasmid (pMT-RNA2) was designed by exchanging the IE1 promoter with the metallothionein (MT) promoter in the previously described plasmid pIE1hr5-RNA2 (19).

The GFP1-10 Drosophila expression plasmid (pIE1-GFP1-10) was constructed by codon-optimizing the GFP1-10 nucleotide sequence (36) and synthesizing custom DNA gBlock (Integrated DNA Technologies). The resulting gBlock sequence was amplified by PCR and cloned into the pIE1hr5/PA backbone downstream of the IE1 promoter using HindIII restriction sites.

In order to generate the E. coli expression vector of GFP1-10, pIE1-GFP1-10 was digested with EcoRV and Eco53kI, and the resulting GFP1-10 fragment was inserted into the XmnI digested pMAL-c5x (New England Biolabs) vector. The plasmid pMAL-GFP1-10 expresses GFP1-10 fused with MBP at its N-terminus under the control of the isopropyl-β-d-thiogalactopyranoside (IPTG) inducible tac promoter.

Drosophila S2 cells were obtained from ATCC ([Drosophila line 2, D. Mel (2), SL2], ATCC CRL-1963) and were supplemented with penicillin, streptomycin, amphotericin B and l-glutamine (E0F ). Drosophila DL-1 cells were provided by Paul Friesen (University of Wisconsin-Madison, Wisconsin) and maintained at 28°C in Schneider Drosophila Medium (Gibco) supplemented with 10% FBS, penicillin, and streptomycin And amphotericin B (S10F).

Wild-type FHV infection has a multiplicity of infection (MOI) of 10, while FHV-His-ΔB1 infection has a multiplicity of infection of 5 on Drosophila S2 cells. For all infections, 108 cells were pelleted and resuspended in 1 mL of S10F medium. Then add the purified virus at the specified MOI, and infect the cells on a rotary shaker at 150 rpm for 1 hour. Subsequently, 14 mL of S10F was added to the cells and transferred to a T75 flask, and incubated at 28°C for 16 to 18 hours. After incubation, the cells were harvested and high-purity mitochondria were isolated using the Qpr​​​​​​​​​​​ We found that previous mitochondrial isolation technology resulted in an unexpected ~10 kDa protease digestion product of protein A. We systematically screened protease inhibitors to find an effective protease inhibitor mixture that can protect protein A from digestion during each step of high-purity mitochondrial isolation. SI appendix, Figure S1). Based on our findings, we replaced the protease inhibitor provided by the supplier with a Halt-free EDTA-free protease inhibitor (ThermoFisher). For WT FHV infection, the final mitochondrial particles are resuspended in the mitochondrial storage buffer provided by the supplier with an A280 absorbance of 10. For FHV-His-ΔB1 infection, the final mitochondrial particles were gently washed 3 times and resuspended in a customized nano-gold binding buffer (50 mM Hepes pH 7.2, 136 mM KCl, 2% sucrose and 10 mM imidazole), A280 The absorbance is 10.

Transfected or infected Drosophila S2 cells are used for immunofluorescence imaging. The transfection and infection procedures are as described in the previous section. As previously mentioned, the cells were harvested and processed for immunofluorescence imaging (55). For protein A immunostaining, use a 1:500 dilution of rabbit antiserum against protein A [antibody R1194 (26)]. For His tag immunostaining, mouse monoclonal clone H8 antibody (ab18184, Abcam) diluted 1:100 was used. Perform wide-field epi-fluorescence microscopy on a Nikon Ti microscope, and use NIS-Elements software (Nikon) to acquire images. Additional image processing is done using the Fiji (ImageJ) software package (56). PCC is calculated using the EzColocalization plug-in of ImageJ (57), where -1 = complete anti-colocalization, 0 = non-colocalization, and 1 = complete colocalization.

For rapid freezing and screening, use PELCO easiGlow (Ted Pella) to conduct a 60-second glow discharge on a 200 mesh Quantifoil freezing grid (2/2) (Cat. No.: Q2100CR2, Electron Microscopy Sciences). Three microliters of mitochondrial preparation mixed in volume 1:1 with 10 nm BSA Gold Tracer (electron microscopy science) was applied to the freezing grid. Vitrobot (FEI, ThermoFisher) was used to vitrify the frozen grid in liquid ethane in a blotting chamber at 100% humidity and 22°C. The freezing parameters are set as follows: 1) Tweezers offset 0, 2) Waiting time 2 seconds, 3) Blotting time 2 seconds, 4) Draining time 10 seconds. In order to screen the ice quality and sample density, a TF30 transmission electron microscope (FEI, ThermoFisher) equipped with a 300 kV field emission gun, a post-column energy filter (Gatan) and a Gatan K2 peak was used to inspect the vitrified frozen grid direct electron detector.

The data set used to generate the 8.5 Å crown subtomogram average was obtained at the Pacific Northwest Cryo-EM Center (Portland, Oregon) facility. For this data set, the vitrified freezing grid was placed in the box of an autoloader, which was loaded into a Titan Krios (Thermo Fisher Scientific) operating at liquid nitrogen temperature, equipped with a field launch gun operating at 300 kV , Post-column energy filter (Gatan) runs with zero loss, and Bioquantum K3 direct electronic detector with 5,760 × 4,092 pixel sensor (Gatan). After collecting the low-magnification 100x full-grid montage, collect the medium-magnification montage of the area with thinner vitrified ice and abundant samples at 2,000 times. Then, select the area of ​​interest from the middle montage for the tilt series collection. The tilt series are collected from -54° to +54° or -60° to +60°, using a grouped dose symmetric tilt plan starting from 0° (30), with a defocus increment of -3.5 µm and a 3° dose rate set to 15 e-/unbinned pixels per second. The cumulative dose of the entire tilt series is 180 e-/Å2, the cumulative dose of each tilt is 4.86 e-/Å2 and 4.39 e-/Å2, the total exposure/tilt of -54 is 0.67 and 0.6 s ° to +54° and -60° to +60° tilt range, respectively. Each tilt is divided into 20 frames with a pixel size of 1.078 Å per pixel in the super-resolution mode on the Bioquantum K3.

A cryo-tomogram of a His-labeled FHV-infected mitochondrial sample collected by Titan Krios (ThermoFisher) at the Janelia Farm Research Park (Howard Hughes Medical Institute) equipped with a post-column energy filter (GIF) and K2 peak top direct electron detector . Use the dose symmetric tilt scheme (30) to collect tilt series from -60° to +60° in 3° increments with -3.5 µm defocusing, starting from 0°, and the dose rate set to 8 e-/unbinned pixels per second. The cumulative dose of the entire tilt series is 180 e-/Å2, and the cumulative dose of each tilt is 4.39 e-/Å2. The exposure time is set to 3.7 seconds. In the super-resolution mode of the K2 peak, each tilt is divided into 10 frames, and the pixel size is 1.35 Å per pixel.

A single frame containing tilt is transferred to the High Throughput Computing Center (CHTC) platform of the University of Wisconsin-Madison at the University of Wisconsin-Madison, and these frames are processed using MotionCor2 v1.2.0 (29) to correct for drift. Each tilted CTF estimation was performed by using emClarity 1.4.3 (34), and by using the automatic template search function in emClarity to select particles, using a 24 Å initial average model from 27 tomographic scans of 1,608 crowns generated by IMOD/PEET ( 32). Subsequent 3D CTF correction, tomographic image reconstruction, and sub-tomographic image averaging were performed using emClarity on the SLURM-based server of the Pacific Northwest National Laboratory.

For His-tag mapping using Ni-NTA-nanogold, a single set of frames of the oblique series are uploaded to the CHTC platform, and the frames are processed using the Unblur v1.02 algorithm in the software suite CisTEM v1 to correct the drift. 0.0 (58). The CTF correction of each tilt series is performed by using the CTFhaseflip function, and the Etomo/IMOD software v4.9.12 (59) is used to generate tomographic images on the local workstation (Dell). Use IMOD to manually select particles from each tomographic image, and use the StalkInit function in PEET/IMOD to define the direction of the crown (32, 33). Use PEET/IMOD on the local workstation to average the sub-tomograms.

Use Chimera (University of California, San Francisco) (60) and IMOD to perform all the image rendering on the sub-tomogram average density map. Use the "Segment Map" and "Fit to Segment" Chimera plug-ins in the software suite Segger v1.9.5 (46, 47) to perform density map segmentation and rigid body docking.

Other methods are provided in the SI appendix Supplementary Methods.

The electron density map of Noda virus replication protein A crown complex has been deposited in EMDataBank, https://www.ebi.ac.uk/pdbe/emdb/, and the registration number is EMD-22129 (61).

We thank Yu Zhiheng, Rick Huang, Xiaowei Zhao, and Doreen Matthies of the HHMI Janelia Cryo-EM facility for their assistance in microscope operation and data collection; Alex Kvit for continuous support of Tecnai TF30 cryo-tomography sample screening and data collection; let- Jean-Yves Sgro provided assistance to the University of California, San Francisco, Chimera; and Claudia Lopez, Lauren Hales-Beck, Craig Yoshioka and Harry Scott of the Pacific Northwest Cryoelectron Microscopy Center assisted in the arrangement, sample preparation and data collect. Part of this research was supported by NIH Grant U24GM129547, and was conducted at the Pacific Northwest Cryo-Electron Microscopy Center of Oregon Health and Science University, and was interviewed through EMSL (grid.436923.9), which was sponsored by the office Biological and environmental research in scientific user facilities of the Office of Energy. This research was conducted with the computing resources and assistance of the High Throughput Computing Center (CHTC) of the Department of Computer Science, University of Wisconsin-Madison. CHTC is supported by the University of Wisconsin-Madison, Advanced Computing Project, Wisconsin Alumni Research Foundation, Wisconsin Discovery Institute, and NSF, and is an active member of the Open Science Grid, which is supported by NSF and the Office of Science of the U.S. Department of Energy. PA is a researcher at the HHMI and Morgridge Institute, and thanks these institutes, NIH, and the John W. and Jeanne M. Rowe Virology Research Center for their support.

↵1N.U. and HZ have made the same contribution to this work.

Author contributions: NU, HZ and PA design research; NU, HZ and JP conducted research; MH and JAdB contributed new reagents/analysis tools; NU, HZ, JP, MH, JAdB, PA analysis data; NU, HZ Wrote this paper with PA.

Reviewers: RB, Heidelberg University; and RJK, Purdue University Institute of Inflammation, Immunology and Infectious Diseases.

The author declares no competing interests.

Data storage: The electron density map of Noda virus replication protein A crown complex has been deposited in EMDataBank, https://www.ebi.ac.uk/pdbe/emdb/ (accession number EMD-22129).

This article contains online support information https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2006165117/-/DCSupplemental.

This open access article is distributed under the Creative Commons Attribution-Non-Commercial-No Derivative License 4.0 (CC BY-NC-ND).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.