Nodavirus protein A’s interdomain elbow controls RNA replication organelle formation and function
Helena Jaramillo-Mesa, Megan Bracken, Hong Zhan, Mark Horswill, Timothy Grant, Johan A den Boon, Paul Ahlquist
TL;DR
This study reveals how a small segment of a nodavirus protein controls the formation and function of RNA replication organelles.
Contribution
The paper identifies a 17-amino-acid 'elbow' segment in protein A that regulates replication organelle assembly and function.
Findings
The elbow segment coordinates interactions between subunits in the proto-crown structure.
Elbow subsegments separately activate RNA capping and polymerase domains.
The elbow and a flexible linker control conformational changes in protein A.
Abstract
Positive-strand [(+)RNA] viruses replicate their RNA genomes in poorly understood membrane-associated replication organelles (ROs). Cryo-electron microscopy of nodaviral ROs revealed that viral RNA replication protein A, with polymerase and RNA capping domains, forms a “crown” of two stacked 12-mer rings at the RO’s opening to the cytosol, providing powerful foundations for analyzing RO formation and function. The lower proto-crown is a ring of 12 polymerases with RNA capping domains clustered to form a central floor. The upper crown mirrors the polymerase ring but has RNA capping domains projecting radially outward. Here, we identify a critical protein A “elbow” segment of only 17 amino acids that coordinates most interactions between crown floor subunits. Our extensive mutational and genetic complementation analyses reveal that distinct elbow subsegments cooperatively support…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8- —UW-Madison Center For High Throughput Computing (CHTC)
- —UW-Madison
- —Advanced Computing Initiative
- —Wisconsin Alumni Research Foundation10.13039/100001395
- —Wisconsin Institutes for Discovery
- —NSF10.13039/100000001
- —Morgridge Institute for Research
- —Rowe Center for Research in Virology
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsBacteriophages and microbial interactions · Plant Virus Research Studies · Viral Infections and Immunology Research
Introduction
Positive-sense RNA [(+)RNA] viruses are a remarkably diverse, high impact class of viruses, encompassing pathogens of major medical, veterinary, and agricultural significance. These include pandemic coronaviruses, cancer-causing hepatitis C virus, flaviviruses such as dengue virus, emerging viruses such as chikungunya virus (CHIKV), and most plant-infecting viruses. For all (+)RNA viruses, RNA genome replication occurs in membrane-associated replication organelles (ROs) that concentrate viral and host factors, protect RNA intermediates from innate immune recognition, produce new (+)RNA genomes, and release them for translation, further replication, and virion packaging. Despite their central role in the viral replication cycle and potential as viral control targets, ROs remain poorly understood. Illuminating RO formation and operation, thus, is valuable for developing more effective antiviral strategies, utilizing (+)RNA viruses as tools for gene transfer, amplification, and expression, and enlightening viral evolution.
Flock house virus (FHV), the best-studied member of the Nodaviridae, is an informative model for many aspects of (+)RNA virus biology [1]. FHV’s 4.5 kb genome encodes protein A, the sole protein required for RNA replication (Fig. 1A), protein B2, an RNAi inhibitor, and capsid protein. Although FHV naturally infects insect cells, transfected FHV RNA supports genome replication and virion assembly in mammalian, nematode, plant, and fungal cells [2–4]. Its compact genome, defined replication machinery, and broad replicative competence have made FHV a highly productive system for elucidating and potentially harnessing molecular mechanisms of RNA replication. Like many (+)RNA viruses, FHV ROs are formed as invaginated membrane vesicles, also called spherules, with necked connections to the cytoplasm (Fig. 1B) [5–9]. For FHV, spherules form on outer mitochondrial membranes [10].
Architecture of the nodavirus RNA RO and organization of protein A. (A) Linear map of FHV multifunctional RNA replication protein A (998 amino acids), highlighting its domain organization. The N-terminal region includes two membrane-binding regions (purple), the MTase-GTase RNA capping domain (yellow), the central channel domain (red), and the elbow segment (green). The C-terminal region contains the RNA-dependent RNA polymerase domain (blue) followed by a poorly conserved C-terminal domain (CTD, gray). The N- and C-terminal regions are connected by a flexible linker (orange) located immediately downstream of the elbow region. Three active site residues and their respective inactivating mutations are indicated: H93A and D141A in the capping domain and D692E in the polymerase domain. The amino acid sequence of the elbow segment is shown. (B) Cryo-electron tomographic (cryo-ET) structure of the mature nodavirus RNA replication complex (RC) crown, with protein A domains colored as in panel (A). The left perspective view shows the complete mature crown. The right cross-sectional view highlights the crown’s gating of the neck of the RO vesicle, with the outer mitochondrial membrane depicted in gray. Distinct conformations of protein A in the 12-mer upper ring, 12-mer lower ring, and central density are visible, with the upper- and lower-ring subunits separated in the right to highlight their arrangement. The lower ring N-terminal domains form the floor and the polymerase-sized central density (blue) likely corresponds to the active Pol responsible for positive-strand synthesis; scale bar: 50 Å. (C) Left panel: single-particle cryo-EM structure of the 12-mer proto-crown assembly intermediate, shown as a surface diagram. The elbow region is highlighted in green and depicted as a stick diagram within a surface envelope. The right panel shows two adjacent subunits (subunit n and subunit n − 1) in detail. Domains are colored as in panel (A): polymerase (blue), RNA capping (yellow), central channel (red), and elbow (green). As shown, the elbow is a 17-amino-acid segment (aa 379–395) embedded in the interface between floor subunits, adjacent to the capping domain, directly connected to the central channel, and immediately preceding the flexible linker (orange). For visual simplicity, only the linker from subunit n − 1 is included.
Recent FHV studies provided the first 3D cryo-electron microscopy (cryo-EM) tomographic images of (+)RNA virus ROs [11]. These and other cryo-EM analyses [12, 13] revealed that FHV spherules are packed with filaments of the viral double-stranded RNA (dsRNA) replication intermediates and that the cytosolic side of the spherule neck is gated by a “crown” structure formed by two stacked 12-mer rings of the 998-aa, multidomain viral replication protein A. This crown also releases filaments into the cytosol, which are consistent with newly synthesized viral genomic and subgenomic (+)RNAs. Subsequent work showed that in ROs of alphaviruses, coronaviruses, and likely other (+)RNA viruses, viral RNA replication proteins also assemble ringed crowns sharing multiple principles with nodavirus crowns [14–21].
As shown in Fig. 1A, protein A’s C-terminal half includes an RNA-dependent RNA polymerase (Pol) domain while its N-terminal half includes two membrane association domains, a 5′ RNA capping (Cap) domain and a “central channel” domain described below. The central channel and RNA capping domain folds are strongly conserved between alphaviruses and nodaviruses [13], whose capping domains share an unusual methyltransferase-guanylyltransferase (MTase-GTase) mechanism in which GTP is 7-methylated, and resulting m^7^GMP is covalently bound and transferred to nascent viral positive-strand RNA 5′ ends [13, 22–24]. Between the N-proximal RNA capping/central channel domain and the Pol domain is a flexible linker contributing plasticity to protein A conformational transitions [13] (Fig. 1A and C).
The nodavirus crown is built from at least three conformations of viral protein A [13]. In the crown’s lower 12-mer ring, the N-proximal MTase-GTase RNA capping domains form a toroidal, membrane-proximal floor that constitutes the most stable structural core of the entire crown and RO complex (Fig. 1B). The associated membrane binding domains fold together below the floor’s outer edge, while the C-proximal Pol domains form an interlocked ring above the outer edge. In the absence of viral RNA templates, this lower ring forms as an early assembly intermediate, the “proto-crown,” whose protein structure was determined by single particle cryo-EM (Fig. 1C) [13]. In the crown’s upper, 12-mer protein A ring, the Pol domains assemble atop the lower Pol ring, forming a stacked “central turret.” Unlike the lower ring, in the upper ring the protein A N-terminal domains extend radially outward as “legs” whose outer edges again interact with the underlying membrane (Fig. 1B). Since the central turret’s 24 Pol domains are separated from the viral dsRNA template by the crown floor and RO membrane vesicle, positive-strand RNA synthesis—the only synthetic activity of mature ROs—requires an additional form of Pol at or below the crown floor [12]. This active Pol likely corresponds to a Pol-sized electron density in the center of the crown floor that, in exact parallel with the sole, active nsp4 Pol of CHIKV crowns [18], is docked to a ring of 12 “central channel” domains, which is further enclosed by a ring of 12 MTase-GTase RNA capping domains (Fig. 1B and C) [13, 15].
Here, we show that a 17-amino acid “elbow” segment of protein A (residues 379–395) plays pivotal roles in crown assembly and regulating protein A functions. Strategically positioned at the interface between floor subunits, this elbow coordinates the majority of intersubunit interactions that assemble the crown floor (Fig. 2), the most stable foundation of both the proto-crown and full crown [13]. It directly follows the conserved central channel domain and, importantly for conformational transitions, directly precedes the flexible linker connecting the RNA capping and polymerase regions (Fig. 1A and C). By combining targeted mutagenesis, genetic complementation, EM-based assembly assays and other approaches, we show that, despite its small size, the elbow is not a passive structural element but a master regulator that encodes multiple separable functions in defined segments. Mutational analyses establish the elbow as a crucial linchpin driving crown floor assembly. Moreover, distinct elbow subsegments independently activate the RNA capping and Pol domains, likely in different protein A conformations, licensing these functions to act in particular spatial and temporal contexts of the RO. As part of these actions, the elbow regulates protein A conformational changes by controlling floor assembly/disassembly, providing a genetically economical alternative to the protease-mediated processing strategies many (+)RNA viruses and retroviruses use to regulate their replication. These findings reveal new principles in RNA virus replication, control and evolution.
The elbow is embedded within the inter-subunit interfaces of the proto-crown floor. (A) Top-down view of the 12-mer proto-crown N-terminal floor, with four complete adjacent protein A subunits labeled n − 2 to n + 1. Color coding follows Fig. 1A, with alternating lighter and darker shades used to distinguish individual subunits. The elbow segments (aa 379–395) are highlighted in green, illustrating their embedded position at the interfaces between subunits. (B) Close-up view of the floor interface between subunits n (darker) and n − 1 (lighter), highlighting the elbow of subunit n (darker green) and its contacts with the adjacent n − 1 subunit. The two subunits are displayed slightly separated to better illustrate the network of interactions. Green lines indicate residues forming inter-subunit contacts with the n-elbow region (residues 379–395). Color coding of the protein structures follows Fig. 1A. (C) Residue-level contact network showing interactions of the n-elbow (green circles, residues 379–395) with residues in the adjacent n − 1 subunit (red and yellow circles, residues 81–318). Circle coloring (green, red, and yellow) follows Fig. 1A. Connections are color-coded by interaction type: green, inter-subunit contacts; orange, salt bridges; and blue, hydrogen bonds. (D) 90° rotations of the individual subunits, illustrating the n-elbow’s position within each subunit interface.
Materials and methods
Plasmid construction
All elbow mutants were generated using the homologous recombination–based cloning kit GenBuilder Plus (GenScript) with gene fragments synthesized by Twist Biosciences. Inserts were cloned into plasmid pIE1hr5-FHVptnA(rr) [25], which expresses a nonreplicable messenger RNA (mRNA) bearing the protein A open reading frame but lacking essential viral replication signals, for use in Drosophila melanogaster S2 cells. Alanine block elbow mutants also were cloned into pOET-FHVptnA(rr)ALFA [25], encoding the same nonreplicable mRNA, but for expression in Spodoptera frugiperda Sf9 cells for recombinant baculovirus generation. This pOET vector includes a C-terminal ALFA tag to enable downstream affinity purification of protein A. All constructs were verified by Sanger sequencing of the FHV copy DNA (cDNA) inserts and ∼80–100 nt of flanking sequence, and many were subjected to full plasmid sequencing using the Oxford Nanopore GridION platform (Functional Biosciences, Madison, WI). Construction of plasmids encoding the full-length, replicable RNA1 template with an early frameshift mutation and a D692E polymerase-null mutation that prevent functional protein A expression, as well as plasmids encoding the single amino acid enzymatic mutants (H93A, D141A, D692E) and the double enzymatic mutant (H93A/D692E), is described in [25].
Cell culture
D. melanogaster S2 cells were maintained at 28°C in either Schneider’s Drosophila Medium (Gibco/ThermoFisher) supplemented with penicillin, streptomycin, amphotericin B and L-glutamine, and 10% inactivated bovine calf serum (BCS, Cytiva/Hyclone) or Express Five serum-free medium (Express Five SFM, Gibco/ThermoFisher) supplemented with penicillin, streptomycin, amphotericin B, and L-glutamine. S. frugiperda Sf9 cells were maintained in Sf-900 III serum-free medium (Sf-900 III SFM, Gibco/ThermoFisher) supplemented with penicillin, streptomycin, amphotericin B and L-glutamine, either in stationary culture or shaking at 120 rpm, at 28°C.
DNA transfection
D. melanogaster S2 cells were seeded in 6-well plates at a density of 1 × 10^6^ cells/ml in 2 ml of Express Five SFM per well and incubated overnight at 28°C. Prior to transfection, Express Five SFM was replaced with 1 ml of Schneider’s medium supplemented with penicillin, streptomycin, amphotericin B, L-glutamine, and 10% BCS. Transfection mixtures were prepared by complexing 1.5 µg of total plasmid DNA (1 µg of RNA1 template plasmid and 0.5 µg of protein A expression plasmid[s]), 6 µl of Trans IT-Insect Transfection Reagent (Mirus Bio), and 100 µl of Opti-MEM (Gibco/ThermoFisher). The mixture was incubated for 15–20 min at room temperature and then added to the cells. After 1 h at 28°C, 500 µl of supplemented Schneider’s medium was added to each well. Cells were incubated at 28°C for 65 h before being harvested for protein and RNA extraction.
RNA and protein extraction
Total RNA was extracted from half of the transfected cells using a Maxwell RSC48 automated extraction system and the Maxwell RSC SimplyRNA Cells Kit (Promega), following manufacturer’s instructions. Total protein was harvested from the remaining cells using 300 µl of prewarmed urea/sodium dodecyl sulfate (SDS)-based cracking lysis buffer [8 M urea, 5% SDS, 40 mM Tris pH 7.0, 0.1 mM EDTA, 25% glycerol, 572 mM BME, and 0.2% (w/v) of Orange G], as described previously [13].
Northern blotting
RNA was analyzed by northern blotting as described previously [13, 26]. Briefly, 500 ng of total RNA was mixed 1:2 (v/v) with RNA sample loading buffer (Sigma) and resolved on 1.5% denaturing agarose gels prepared with running buffer containing 20 mM (3-(N-morpholino) propanesulfonic acid (MOPS), 8 mM sodium acetate, and 1 mM EDTA, and supplemented with 2% formaldehyde. Following electrophoresis, RNA was transferred to nylon membranes by overnight passive upward transfer using 10× SSPE (1.5 M NaCl, 0.1 M NaH_₂_PO_4_, 0.01 M EDTA, pH 7.4). After transfer, blots were rinsed in 5× SSPE, and UV crosslinked twice using auto-settings on a UVP CL-1000 UV crosslinker. RNA blots were pre-hybridized for 1 h at 42°C in NorthernMAX solution (Ambion), then hybridized for at least 3 h at 42°C in fresh NorthernMAX containing 500 ng of biotinylated RNA probe. Probes for RNA1 and RNA3 detection were transcribed from a linearized plasmid corresponding to FHV RNA1 nucleotides 2719–3007 using a MAXIscript T7 transcription kit (Ambion) containing 0.5 mM ATP, CTP, and GTP, and a mixture of 0.3 mM UTP and 0.2 mM Bio-16 UTP (Ambion). After hybridization, membranes were washed for 20 min in a high-salt buffer (2× SSPE, 0.2% SDS), and then for 20 min in a low-salt buffer (0.2× SSPE and 0.2% SDS) at 42°C. Blots were blocked in Intercept (PBS) Blocking Buffer (LI-COR Biosciences) supplemented with 1% SDS for 1 h at room temperature, then incubated for 30 min at room temperature in a 1:10 000 dilution of IRDye680RD-streptavidin in Intercept (PBS) Blocking Buffer containing 1% SDS. Membranes were washed three times with PBS with 0.1% Tween, and once with PBS. Signals were detected using a LI-COR Odyssey digital imaging system. Band intensities were quantified using LI-COR Image Studio software (version 5.2). The high sensitivity, resolution, and dynamic range of this system yielded distinct, well-separated, well-discriminated signals for all bands of interest, despite widely varying intensities and lane proximity. To display images from the resulting digital files in the figures, image intensities were chosen to allow visualizing both RNA1 and RNA3 bands in all lanes with significant RNA replication. Since RNA3 signals were stronger than RNA1, the RNA3 bands in adjacent lanes of the displayed figures are sometimes saturated. This represents a limitation in visual image display, but not a limitation in the dynamic range of the underlying digital image data, which strongly differentiates adjacent bands in spatial resolution and signal intensity. This is illustrated in Supplementary Fig. S1, which displays the northern blots of Fig. 4A–C at lower intensities, showing the separation of adjacent RNA3 bands at the cost of reducing or eliminating visualization of corresponding RNA1 bands. To measure relative RNA3 band intensities, nonoverlapping regions of interest enclosing the bands were selected at such lower display intensities. The robustness of this approach is evident in the resulting histograms, as lanes with no detectable RNA3 signal consistently show values at or near zero even when flanked by strong bands, demonstrating no meaningful bleed-over from neighboring lanes (e.g. lanes 3–5 in Fig. 4B and C). All measurements of relative RNA3 levels were performed using at least three independent biological replicates.
Western blotting
Protein samples were analyzed by western blotting as described previously [26] with minor modifications. Samples were run on NuPAGE 4%–12% Bis–Tris gels (Invitrogen) in 1× MOPS buffer before transferring to Immobilon-FL PVDF membrane (Millipore) at 25V for 9 min. Membranes were blocked in Intercept (TBS) Blocking Buffer (LI-COR) for 1 h and incubated with rabbit anti-FHV Protein A and mouse anti-Tubulin (MilliporeSigma) antibodies at 1:2000 in Intercept (TBS) Blocking Buffer (LI-COR) supplemented with 0.2% Tween-20 for 1 h. Membranes were washed three times with Tris-buffered saline with 0.1% Tween 20 (TBST) and incubated with secondary antibodies conjugated to IRDye680RD or IRDye800CW at 1:10 000 in Intercept (TBS) Blocking Buffer with 0.2% Tween-20 + 0.1% SDS. Blots were washed three times in TBST before imaging on the LI-COR Odyssey Imaging System.
Baculovirus generation
Recombinant baculoviruses expressing wild-type (wt) protein A or protein A alanine block mutants, each containing a C-terminal ALFA tag, were generated using the pOET1-based transfer plasmids together with the FlashBac Baculovirus Expression System (Mirus Bio) as previously described [13]. The resulting culture medium containing recombinant virus was harvested and centrifuged to remove cellular debris. To amplify the recombinant virus to a working titer, 500 µl of the clarified supernatant was added to Sf9 cells at 1 × 10^6^ cells/well in six-well plates containing 2 ml/well Sf-900 III SFM. Cells were rocked at room temperature for 1 h and then incubated at 28°C for 5 days. After 5 days, the medium was harvested by centrifugation at 1000 × g for 10 min, and 2 ml of the clarified supernatant was used to inoculate Sf9 cells seeded at 1 × 10^6^ cells/ml in 40 ml of Sf-900 III SFM in a shaking flask, incubated at 28°C for 5 days. A third passage was generated similarly with 2 ml of clarified culture used to inoculate a final 40 ml of Sf9 culture at 1 × 10^6^ cells/ml and incubated with shaking at 28°C for 5 days. Passage three cultures were harvested by centrifugation and used as inocula for proto-crown isolation. Baculovirus titers were determined by Expression Systems in Davis, CA.
ALFA-tag-based purification of Protein A proto-crowns
For recombinant baculovirus infection, 5 × 10^8^ Sf9 cells were pelleted and resuspended in 20 ml of Sf-900 III SFM supplemented with penicillin, streptomycin, amphotericin B, and L-glutamine. Recombinant baculovirus expressing C-terminal ALFA-tagged wt protein A or alanine block mutants was added at a multiplicity of infection (MOI) of 10. Cells were incubated with shaking at 120 rpm for 1 h at 28°C, after which the culture volume was increased to 400 ml with additional Sf-900 III SFM and incubated with shaking at 120 rpm at 28°C for 3 days. After 3 days, 1.5 ml aliquots were set aside for protein expression analysis by western blotting. These analyses, together with additional checks performed during purification, confirmed that all samples exhibited comparable protein A-ALFA expression levels and yields. The rest of the culture was used for mitochondrial isolation using Qiagen’s Qproteome Kit with minor modifications. Briefly, infected Sf9 cells were pelleted for 5 min at 500 × g and washed sequentially with PBS and 0.9% NaCl to remove residual medium. Cells were then resuspended in Qproteome lysis buffer supplemented with 25× Roche protease inhibitor. Cells were incubated on ice for 10 min and centrifuged to remove cytosolic supernatant. The resulting pellets were resuspended in Qproteome Disruption buffer supplemented with 25× Roche protease inhibitor and mechanically disrupted by passing 10 times through a 25-gauge needle. Cellular debris was removed by centrifugation at 1000 × g for 10 min at 4°C, and mitochondria-enriched fractions were obtained following repeated centrifugation and washes according to the manufacturer’s instructions for high-purity mitochondrial isolation. Isolated mitochondrial pellets were resuspended to a final concentration of 45 mg/ml in a 1× mix of RSB Hypo Buffer, a 2:3 mixture of 2.5× MS Homogenization Buffer (12.5 mM Tris–HCl pH 7.5, 525 mM mannitol, 175 mM sucrose, and 2.5 mM EDTA) and 1× RSB Hypo Buffer (10 mM HEPES pH 7.5, 10 mM NaCl, 0.2 mM EDTA), supplemented with 2.5 mM disuccinimidyl glutarate (DSG, ThermoFisher), and gently rotated at 4°C for 30 min to cross-link protein complexes. Cross-linking was quenched with Tris pH 8.0 at a final concentration of 25 mM, and samples were rotated for an additional 15 min at room temperature. An equal volume of solubilization buffer (20 mM Tris pH 8.0, 300 mM MgCl_2_, 4% n-Dodecyl-B-D-maltoside (DDM), 4% glycerol, 1× cOmplete Protease Inhibitor, and 25 U benzonase) was added, and samples were gently rotated at room temperature for 30 min. Insoluble material was removed by centrifugation at 21 000 × g for 15 min at 20°C, and the clarified supernatant containing solubilized ALFA-tagged protein A and proto-crowns was collected. For affinity purification, 100 µl of 50% ALFA Selector PE resin slurry (NanoTag Biotechnologies) was equilibrated in 4 bed volumes of Wash Buffer (150 mM NaCl, 20 mM Tris pH 8.0, 1% glycerol, 0.05% DDM) and allowed to settle to remove excess buffer. Clarified lysates were added to the resin and gently rotated for 30 min at room temperature. The resin–lysate mixture was transferred to a 2 ml Pierce Centrifuge Column (ThermoFisher), allowing the resin to settle without centrifugation, and the flow-through was discarded. Resin was washed with 12 bed volumes of Wash Buffer to remove nonspecifically bound material. ALFA-tagged protein A was eluted in two bed volumes of Elution Buffer (Wash Buffer supplemented with 0.2 mM ALFA peptide) by gently mixing every 5 min over 25 min. Eluates were collected in low-binding tubes (USA Scientific) and concentrated to ∼30 µl using Vivaspin 500 50 kDa MWCO PES columns (Cytiva) pre-rinsed with Wash Buffer, using sequential centrifugation steps from 2400 × g up to 4400 × g at 4°C. Concentrated samples were either used immediately for downstream applications, such as negative-stain EM, or stored at −80°C.
Negative-stain and electron microscopy image acquisition
Copper grids (CF300-Cu-50, Electron Microscopy Sciences) coated with carbon were glow-discharged for 30 s at 20 mA using a PELCO easiGlow system (Ted Pella, Inc.). Three microliters of concentrated proto-crown sample were applied to each grid and allowed to incubate for 1 min. Excess sample was blotted using Whatman grade 2 filter paper. Grids were then washed by touching the grid face to a fresh 20 µl drop of Wash Buffer (150 mM NaCl, 20 mM Tris pH 8.0, 1% glycerol, 0.05% DDM), blotting on Whatman grade 2 filter paper and repeating both steps. Staining was performed by briefly touching each grid to a fresh 20 µl drop of 2% uranyl acetate (Electron Microscopy Sciences), blotting on Whatman grade 2 filter paper, and repeating both steps. Finally, grids were incubated face-down on a fresh drop of 2% uranyl acetate for 1 min, then slowly blotted on Whatman grade 2 filter paper to remove excess stain. Grids were air-dried for 1 min and stored in grid boxes within a vacuum desiccator until imaged. Imaging was performed on a Talos F200C Transmission Electron Microscope equipped with a OneView camera (ThermoFisher) at ×57 000 magnification, corresponding to a pixel size of 2.646 Å, and with a defocus of ∼2 µm. Images were collected using SerialEM software (Mastronarde, 2005). For each condition, at least three independent experiments were imaged, with a minimum of 40 micrographs examined per condition, covering approximately 50 µm² of grid area.
2D classification and 3D reconstruction
Image processing was performed using CryoSPARC v4.7.1 [27] following standard single-particle analysis workflows. Micrographs with an astigmatism threshold greater than 150 Å were excluded from further analysis. Approximately 120 particles were manually selected and extracted with a box size of 168 pixels. These particles were then aligned to each other, and those sharing similar contour and orientation were grouped through 2D classification. The resulting 2D classes represent averaged projections of multiple particles, providing an improved signal-to-noise ratio (SNR). Classes dominated by random background noise were discarded prior to downstream analyses. Several well-defined 2D classes were chosen as templates for template-based particle picking across all micrographs using the same box size. After template picking, the results were manually inspected, followed by four iterative rounds of 2D classification to remove residual background classes and further enhance SNR, ultimately yielding a small number of high-quality 2D classes representing different projected views of the proto-crown. For protein A mutants lacking proto-crowns, blob-based particle picking was used in CryoSPARC with a maximum diameter of 200 Å and a minimum diameter of 100 Å. Extracted particles (box size 168 pixels) were subsequently subjected to four rounds of 2D classification, as described above. Following the final 2D classification step, particles corresponding to one side view and one top view of the proto-crown were selected to generate an ab initio 3D reconstruction in CryoSPARC. In this procedure, an initial low-resolution 3D model is produced de novo from the 2D particle images without the use of any external structural references. The algorithm assigns random initial orientations to the selected particles and iteratively refines orientations and translations to converge to a final 3D density that represents observed 2D projections. The resulting ab initio model served as the starting volume for subsequent homogeneous refinement, which further improved particle alignment accuracy and map resolution using all selected particles. During refinement, C12 symmetry was imposed to enhance the signal and generate a 3D density map of the proto-crown. Subsequent visualization and analysis were conducted using UCSF ChimeraX [28].
Results
Plasmid-based trans-RNA replication assays
To assay FHV RNA replication and the effects of selected protein A changes, we used a trans-replication assay in which S2 cells of Drosophila, a natural host for FHV infection, were co-transfected with two plasmids: one expressing protein A from a nonreplicable mRNA bearing the protein A open reading frame but lacking 3′ and 5′ viral signals essential for replication template activity, and another expressing a full length, replicable RNA1 template with an early frame-shift mutation preventing protein A expression (RNA1 fs) (Fig. 3A). This uncouples protein A expression from RNA replication, so that mutational defects in protein A RNA replication functions do not inhibit protein A expression. By largely stabilizing expression levels across different engineered protein A mutants, this greatly facilitates assessing the effects of protein A changes on RNA replication, and bypasses possible effects of coding changes on RNA template activity [25]. Another advantage is that protein A mutations are effectively irreversible since the protein is expressed from a stable DNA template, whose mRNA is not replicated and therefore is neither subject to the high mutation rates of viral RNA nor under genetic selection for reversion.
Plasmid-based trans-RNA replication assay. (A) Schematic of the trans-RNA replication assay in Drosophila S2 cells. Co-transfection of two plasmids separates protein A expression from RNA replication template functions: the left protein A plasmid expresses functional protein A from a nonreplicable mRNA lacking viral 5′ and 3′ replication signals, and the right RNA1 fs plasmid expresses a full-length RNA1 template containing an early frameshift (fs) to prevent translation of protein A. This approach largely stabilizes expression levels of different protein A mutants and allows more direct assessment of their effects on RNA replication, measured by genomic RNA1 and subgenomic RNA3 accumulation. Color coding follows Fig. 1A. (B) Northern blot analysis validating the trans-RNA replication assay, with RNA and protein collected ∼65 h post-transfection. The top panel shows a western blot detecting protein A, with tubulin as a loading control. The bottom panel shows the northern blot: Lanes 1 and 2 show no RNA1 or RNA3 signals when either the protein A plasmid or RNA1 fs plasmid is transfected alone. Lane 3 shows strong replication of both genomic RNA1 and RNA3 when wtwt protein A is co-expressed with the RNA1 fs template, confirming robust replication in trans. Lane 4 shows that co-expression of the RNA1 fs template with a protein A deletion mutant lacking the 17–amino acid elbow (Δ379–395) completely abolishes RNA replication, confirming that the elbow region is essential for FHV RNA replication.
Plasmid-based trans-replication assays have been extensively used in FHV research for decades and have been shown to reproduce all known aspects of FHV infection, including RNA1 replication and RNA3 production and their dependence on particular *cis-*acting sequences [29–31]; trans-activation of RNA2 replication by RNA3, with concomitant suppression of RNA3 levels [30, 32–34]; induction of the highly specific RO spherule invaginations and precise crown structures on outer mitochondrial membranes [26, 35]; interaction of a fraction of FHV RNA interference suppressor protein B2 with protein A [26]; and many other features.
Accordingly, expressing wt protein A with the frameshift RNA1 template yielded strong northern blot signals for replicated RNA1 fs, far above minor background from the individual plasmid transfections, and even stronger signals for subgenomic RNA3 (Fig. 3B, lanes 1–3). As a relevant control, deleting the 17-aa elbow (∆379–395) abolished RNA replication as expected, confirming that the elbow is essential (Fig. 3B, lane 4).
Elbow alanine substitutions reveal impacts of key single and additive aa changes
To begin assessing the elbow’s importance in RNA replication, we dissected the impact of individual and combined aa changes in the elbow region by conducting unbiased alanine scanning mutagenesis at multiple resolutions. As an initial, lower resolution scan, we spanned the 17-aa elbow region with four consecutive alanine replacement blocks: 379–383A_5_, 384–388A_5_, 389–393A_5_, and 394–395A_2_ (Fig. 4A). The first three blocks replace successive five aa segments with alanines, while the last was limited to two residues since aa 395 is the elbow C-terminus and last residue rigid enough to be built definitively into the structure before the flexible linker region. To further assess the effects of larger scale mutation of the elbow C-terminus, we augmented this analysis with an overlapping mutant, 391–395A_5_ substituting alanines for the last five elbow residues. Hereafter, these five mutants are referred to as alanine block mutants and, for brevity, each individual mutant protein A derivative will be referred to by its genotype, e.g. 379–383A_5_, 384–388A_5_, etc. Using the trans-replication system of Fig. 3, we co-transfected S2 cells in parallel with plasmids expressing each protein A alanine block elbow mutant plus the frameshift RNA1 template (RNA1 fs), and measured the products of RNA replication (genomic RNA1 and subgenomic RNA3) by northern blotting three days post-transfection. RNA3 levels were used as the measure of RNA replication since in this system RNA3 is free of any plasmid-derived transcript background. All five mutants severely impaired RNA replication, with even the highest RNA3 accumulation, induced by protein A mutant 394–395A_2_, barely detectable at ≤ 1% of the wt control (Fig. 4A). Thus, sequences throughout the elbow are crucial for RNA replication.
Global alanine-scanning mutagenesis of the protein A elbow identifies amino acid contributions to RNA replication. (A–C) Alanine substitutions were introduced across the 17-amino acid elbow (aa 379–395) as blocks (5 alanines, A), pairs (2 alanines, B), or single residues (1 alanine, C). Top panels: western blot detecting protein A, with tubulin as a loading control. Middle panels: northern blot analysis of RNA1 and RNA3 replication following co-transfection of mutant protein A plasmids with the RNA1 fs template. Bottom panels: bar graphs summarizing RNA3 replication relative to wt control across three or more experimental replicates. (D) Summary diagram mapping replication values from block, pair, and single alanine substitutions onto the elbow sequence. Color-coded gradients indicate functional impact: white represents mutations that abolish RNA3 replication (0%), and blue represents full replication comparable to wt (100%), allowing visualization of residues and segments critical for RNA replication. (E) Structure mapping of elbow (aa 379–395) characteristics. The surface diagrams use the indicated color gradients to show RNA3 replication levels (% of wt) induced by single alanine substitutions [gradient as in panel (D)], electrostatic potential (negative: red, positive: blue), and hydrophobicity (hydrophilic: blue, hydrophobic: yellow) to highlight functional features. Gray shading of the arrows indicates amino acids facing the back, while white shading indicates amino acids facing the front.
To further resolve important residues or regions within the elbow, we conducted alanine scanning mutational analysis at 2-residue and 1-residue resolutions. As shown in Fig. 4B–D, most double or single alanine substitutions retained replication-dependent RNA3 accumulation to >50% of wt protein A levels. However, alanine pair mutants 379–380A_2_ and 391–392A_2_ resulted in loss of detectable RNA3 accumulation (Fig. 4B, lanes 4 and 10), and 389–390A_2_ suppressed average RNA3 accumulation to 4% of wt (Fig. 4B, lane 9). Further focusing these results, the single alanine scan revealed that K379A and Y390A suppressed average RNA3 accumulation to 0% and 6% of wt, respectively (Fig. 4C, lanes 4 and 15), while mutants K391A and P395A inhibited average RNA3 yield to 29% and 19% of wt protein A, respectively (Fig. 4C, lanes 16 and 20). Although alanine substitution represents a larger physicochemical change for some residues than others, sensitivity to alanine substitution did not correlate simply with the nature of the side chain replaced. In contrast to the effects of K379A, K391A, and P395A, e.g. K380, P384, P387, and P392 tolerated alanine substitution with minimal effects, confirming that, in addition to the nature of the starting side chain, the observed phenotypes must reflect site-specific sequence context, protein folding and interaction constraints.
Taken together, these three distinct resolution scans indicate that, while some defects observed in the 5-alanine block scan might be attributable to single amino acids (K379 in 379–383A_5_, Y390 and K391 in 389–393A_5_, and P395 in 394–395A_2_), there are also significant additive effects. For example, while aa 384–388 tolerated 1- or 2-alanine substitutions with relatively little activity loss, replacing all five aa with alanine abolished RNA replication. The relation of these additive effects to multiple, separable elbow functions is explored further below.
To seek further insights into the contributions of key aa to elbow function in RNA replication, we used surface models of the elbow structure to map and compare alanine substitution RNA replication activity, electrostatic potential and hydrophobicity (Fig. 4E). Most notable was the association of positive charge with sensitivity to alanine substitution, since K379A and K391A, at the only two positively charged elbow aa, inhibited RNA replication to 0% and 29% of wt, respectively. No correlation between hydrophobicity and alanine substitution sensitivity was observed.
Distinct elbow segments activate polymerase, RNA capping, and other protein A activities
To illuminate how elbow mutations interfere with RNA replication, we tested for possible effects of elbow mutations on the crown’s RNA capping or Pol functions using genetic complementation assays. This approach is based on our recent finding that protein A RNA capping null [Cap(-)] and polymerase null [Pol(-)] mutants that are individually replication-defective due to active site mutations can complement each other’s RNA replication defects in trans, restoring RNA replication [25]. Thus, neither RNA capping nor Pol activities are required in all copies of protein A, nor are both required together in any single copy.
To test elbow mutations for effects on RNA capping or Pol activities, we co-transfected S2 cells in turn with plasmids expressing one of the five replication-impaired alanine block elbow mutants of Fig. 4A (i.e. alanine substitutions 379–383A_5_, 384–388A_5_, 389–393A_5_, 394–395A_2_, or 391–395A_5_), the frameshift RNA template, and a protein A enzymatic mutant lacking capping or polymerase activity due to point mutations H93A or D692E (Fig. 5A) or both (Fig. 5B). H93A is an RNA capping-negative mutation that alters a histidine, conserved among nodavirus and alphavirus RNA capping proteins, that covalently binds the m^7^GMP product of the viral MTase for transfer to the nascent RNA 5′ end [24, 36, 37]. D692E is a polymerase-blocking mutation within the highly conserved “GDD” RNA-dependent RNA polymerase motif [38–40]. These active site mutants and their defects are highly specific, independent of each other, show no evidence of significant impacts on protein A structure, and support proto-crown assembly [25]. As shown in control lanes of Fig. 5A (lanes 4 and 5) and 5B (lane 4), protein A bearing either or both of these enzymatic mutations lacked any detectable RNA replication activity.
Genetic complementation between protein A elbow alanine block mutants and RNA capping or Pol null mutants. (A and B) Trans-replication assays testing complementation of the five protein A alanine block elbow mutants co-transfected with either the Cap(-) (H93A) (A), Pol(-) (D692E) (A), or Cap(-) (H93A)/Pol(-) (D692E) enzymatic mutant (B). Top panels: western blots detecting protein A, with tubulin as a loading control. Middle panels: northern blots showing RNA1 and RNA3 replication. Bottom panels: bar graphs summarizing RNA3 replication relative to wt control (lane 3). Expected replication-null activity is observed for single Cap(-) or Pol(-) mutants (A, lanes 4 and 5) and for the Cap(-)/Pol(-) double mutant (B, lane 4) when transfected alone. (C) The five protein A alanine block elbow mutants were co-transfected with either the protein A H93A or D141A substitution to compare complementation activities between distinct Cap(-) variants (H93A disrupts m7GMP binding; D141A disrupts SAM binding). The bottom bar graph summarizes RNA3 replication as a percentage of the wt control (lane 3). (D) Structure mapping of elbow mutant complementation phenotypes. The surface diagrams segmented by alanine block groups use the indicated color gradients to show RNA3 replication levels (% of wt) resulting from complementation (A and B) between each indicated alanine block mutant and one of the following protein A enzymatic null mutants: Cap(-) (H93A) (white to yellow), Pol(-) (D692E) (white to cyan), or Cap(-) (H93A)/Pol(-) (D692E) double mutant (white to dark blue).
Intriguingly, all five alanine block mutants exhibited distinct complementation behaviors with H93A and D692E (Fig. 5). At one extreme, co-expressing the 384–388A_5_ mutant complemented the RNA replication-inactive D692E Pol(-) mutant to 65% (±30%) of wt protein A-directed RNA replication levels; the H93A Cap(-) mutant to 33% (±15%) of wt (Fig. 5A, lanes 9–11); and the H93A + D692E Cap(-) Pol(-) double mutant to 50% (±8%) of wt (Fig. 5B, lanes 7 and 8). To calibrate these results, it should be noted that co-expressing the highly specific, active site H93A Cap(-) and D692E Pol(-) mutants together only complements RNA replication to 27% (±5%) of wt protein A [25]. Thus, when co-expressed with enzymatically inactive single or double active site mutants, the 384–388A_5_ elbow mutant can supply in trans considerable Pol and RNA capping activity that must be derived from its own Pol and capping domains. Accordingly, the inability of 384–388A_5_ to independently support RNA replication (Fig. 4A, lane 5) must primarily be due to loss of some elbow function or functions distinguishable from RNA capping and Pol enzymatic functions, such as crown assembly. This function is then restored through trans complementation by co-expressing the H93A Cap(-), D692E Pol(-), or H93A + D692E Cap(-) Pol(-) mutants, presumably by their provision of a wt elbow sequence.
Similarly, the adjacent, most N-terminal alanine block mutant, 379–383A_5_, also complemented both Cap(-) and Pol(-) enzymatic mutants, though to more modest levels. Specifically, 379–383A_5_ complemented Pol(-) D692E to 21% (±15%) of wt protein A-directed RNA replication and Cap(-) mutant H93A to 5% (±5%) of wt (Fig. 5A, lanes 6–8). Thus, 379–383A_5_ retains significant Pol function and easily measurable RNA capping function, and its failure to detectably replicate RNA1 or synthesize subgenomic RNA3 (Fig. 4A) also must be due to loss of another function X, which again might be crown assembly. Relative to 384–388A_5_, the reduced complementation between mutant 379–383A_5_ and Cap(-) H93A or Pol(-) D692E might reflect that the 379–383A_5_ mutation partially inhibits capping or Pol functions in cis, or dominantly interferes with function X, impeding full restoration of function X by the wt elbow sequence provided by H93A or D692E.
In contrast to the two N-proximal elbow mutants, C-proximal mutants 389–393A_5_ and 394–395A_2_ showed highly specific but opposite defects in RNA capping and Pol function, respectively. Co-expressing 389–393A_5_ complemented the D692E Pol(-) mutant to 29% (±19%) of wt protein A RNA replication but failed to discernibly complement the H93A Cap(-) mutant or H93A + D692E Cap(-) Pol(-) double mutant (Fig. 5A, lanes 12–14 and Fig. 5B, lanes 9 and 10). The 389–393A_5_ mutant therefore supplies Pol function in trans, but cannot supply RNA capping activity in trans.
Conversely, the 394–395A_2_ mutant complemented the H93A Cap(-) mutant to 48% (±11%) of wt protein A RNA replication, but could not measurably complement the D692E Pol(-) mutant or H93A + D692E Cap(-) Pol(-) double mutant, implying loss of Pol function (Fig. 5A, lanes 15–17 and Fig. 5B, lanes 11–12). Not surprisingly, the 391–395A_5_ mutant spanning both preceding elbow mutants combined the defects of both, showing occasional, barely detectable (average < 0.5%) complementation of Cap(-) mutant H93A and no visible complementation of Pol(-) mutant D692E or the H93A + D692E double mutant (Fig. 5A, lanes 18–20 and Fig. 5B, lanes 13–14).
To further test the specific effects of these alanine block mutations, complementation tests were repeated with an alternate RNA capping null mutation, D141A, which changes an aa involved in binding methyl donor S-adenosylmethionine (SAM) [24]. As shown in Fig. 5C, complementation tests with Cap(-) mutant D141A closely paralleled the distinctive effects with Cap(-) mutant H93A for all five alanine block mutants. The strong match of these phenotypic patterns from well-separated mutations in distinct enzymatic steps confirms that, while many protein A mutations have pleiotropic effects, the RNA synthesis defects in H93A and D141A are restricted to their shared involvement in RNA capping. This validates that the failure of the 389–393A_5_ elbow mutant to detectably complement H93A or D141A, while nevertheless significantly complementing Pol(-) D692E, is due to a 389–393A_5_-induced loss of RNA capping. Overall, these complementation results show that the elbow is involved in multiple, separable functions, including independently activating RNA capping and Pol functions, and additional functions possibly related to assembling or stabilizing the proto-crown floor (Fig. 5D and Supplementary Fig. S2).
Elbow mutants 394–395A2 and 389–393A5 are functionally equivalent to RNA polymerase and capping active site mutants, respectively
In protein A’s proto-crown conformation, the elbow does not interact with the Pol domain and is separated from the D692E Pol active site mutation by ∼4 nm or more (Fig. 6A). Therefore, while mutant 394–395A_2_’s complementation of Cap(-) H93A but not Pol(-) D692E (Fig. 5A, lanes 15–17) suggested that 394–395A_2_ specifically interfered with Pol enzymatic function(s) blocked by D692E, it was not clear how this elbow mutation could selectively block Pol function.
Functional equivalence of specific elbow mutants to enzymatic active-site mutants. (A) Over the indicated range of plasmid ratios, S2 cells were co-transfected with plasmids expressing the Cap(-) mutant H93A and either the alanine elbow mutant 394–395A2 (lanes 3–10) or the double mutant 394–395A2-Pol(-), which also includes Pol(-) active-site mutation D692E (lanes 11–20). The diagrams on top illustrate each protein A mutant used in the complementation and the position of the 394–395A2 mutation relative to the Cap(-) H93A and Pol(-) D692E sites. (B) Over the indicated range of plasmid ratios, S2 cells were co-transfected with plasmids expressing the Pol(-) mutant D692E and either the alanine elbow mutant 389–393A5 (lanes 3–10) or the double mutant 389–393A5-Cap(-), which also includes Cap(-) active-site mutation H93A (lanes 11–20). The diagrams on top illustrate each protein A mutant used in the complementation and the position of the 389–393A5 mutation relative to the Cap(-) H93A and Pol(-) D692E sites. Top data panels: western blots detecting protein A, with tubulin as a loading control. Middle panels: northern blots showing RNA1 and RNA3 replication. Bottom panels: bar graphs summarizing RNA3 replication relative to the wt control (lane 2). Color coding of structural diagrams follows Fig. 1A, with burgundy highlighting the specific mutated sites.
Accordingly, to further test if elbow mutation 394–395A_2_ is functionally equivalent to D692E—which is understood to block all Pol functions in (-) and (+) genomic RNA and subgenomic mRNA synthesis—we constructed a protein A double mutant bearing 394–395A_2_ plus D692E (394–395A_2_-Pol(-)), and compared its ability to complement Cap(-) H93A to that of single mutant 394–395A_2_ (Fig. 6A). If 394–395A_2_ is functionally equivalent to D692E in blocking essentially all Pol RNA synthesis functions, adding D692E should have little or no effect on complementing H93A. However, if instead 394–395A_2_ e.g. inhibits a step distinct from Pol functions but more closely linked to them than to RNA capping, combining 394–395A_2_ and D692E should have additive loss-of-function effect(s) inhibiting H93A complementation relative to 394–395A_2_ alone.
In parallel, we constructed a protein A double mutant bearing 389–393A_5_ plus H93A (389–393A_5_-Cap(-)) (Fig. 6B). Elbow mutation 389–393A_5_ complements Pol(-) D692E but not Cap(-) H93A (Fig. 5A, lanes 12–14). Thus, comparing the ability of 389–393A_5_-Cap(-) versus 389–393A_5_ to complement RNA replication by D692E tests if elbow mutation 389–393A_5_ and Cap(-) mutation H93A are functionally equivalent. Relative to 394–395A_2_ and D692E, such equivalence seems more likely for 389–393A_5_ and H93A since (i) RNA capping activity appears to reside in the crown floor; (ii) in the crown floor structure, 389–393A_5_ and the H93A capping active site are close and directly connected by a web of adjacent residues (Fig. 6B); and (iii) interactions at and between similar lateral subunit interfaces are implicated in activating RNA capping in the 12-mer ring of the related CHIKV nsP1 RNA capping protein [41–43].
To better compare single and double mutant phenotypes, these complementation assays were conducted across a range of ratios between the two co-transfected mutant expression plasmids (Fig. 6). Consistent with the points noted above, complementation between Pol(-) D692E and 389–393A_5_ or 389–393A_5_-Cap(-) were virtually identical, showing only minor differences at the extreme ranges of the ratios tested, which might relate to small differences in accumulation of the single and double mutant proteins (Fig. 6B northern and western blots). Hence, elbow residues 389–393 are crucial to the activity of the adjacent RNA capping subunit, so that mutating these residues in 389–393A_5_ is functionally equivalent to the complete block of RNA capping activity by active site mutation H93A.
Even more informatively, complementation between Cap(-) H93A and 394–395A_2_ or 394–395A_2_-Pol(-) were similarly almost indistinguishable, again with only small differences at the extreme edges of the co-expression ratios tested (Fig. 6A). Thus, elbow aa 394–395 are so essential to Pol activity that their mutation to alanine reproduces the functional block of active site mutation D692E. The possibility that aa 394–395 exert these activating effects through interactions with Pol in a protein A conformation other than the proto-crown is considered in the Discussion.
Mutations across much of the elbow disrupt proto-crown assembly
Genetic complementation (Fig. 5) showed that elbow mutations 379–383A_5_ and 384–388A_5_ each retained significant RNA capping and Pol activity in cis, indicating that their failure to support detectable RNA replication (Fig. 4A) must be due to loss of an additional function. One prominent candidate for such additional protein A function is crown assembly, as the elbow coordinates most inter-subunit interactions within the crown floor, which in turn provides most of the stability of the first known crown assembly intermediate, the proto-crown [13]. Accordingly, we tested whether 379–383A_5_, 384–388A_5_, or other alanine block mutants inhibited proto-crown assembly, using our previously established proto-crown isolation procedure [13] and our recently developed negative-stain electron microscopy proto-crown assay [25].
Baculovirus-expressed wt protein A produced abundant proto-crown particles with the characteristic ring morphology, clearly visible in raw EM images (Fig. 7A) and further defined by 2D class averaging (Fig. 7B). As previously reported [13], while native crowns and proto-crowns imaged on mitochondrial outer membranes exhibit exclusively C12 symmetry in cryo-ET analyses, detergent-purified proto-crowns solubilized from membranes include both 12-mer and 11-mer rings (Fig. 7B), with no structural differences detected between C12 and C11 monomers. The C11 assembly likely reflects a detergent-extraction artifact rather than a native state [13]. From multiple independent experiments, 9980 proto-crowns were identified, corresponding to a density of 12.9 proto-crowns per µm² of grid area. As shown by the Fig. 7C superposition, the resulting reconstructed exterior envelope closely matched the high-resolution proto-crown structure from single-particle cryo-EM [13], as expected.
Electron microscopy analysis of proto-crown assembly phenotypes of elbow alanine block mutants. (A) Transmission electron microscopy (TEM) of wt protein A and alanine block elbow mutants after baculovirus expression, ALFA-tagged protein A purification and negative staining with uranyl acetate. Proto-crown density (proto-crowns/µm² of EM grid area) is shown for each alanine block mutant. Wt protein A forms abundant, characteristic ring-shaped proto-crown particles. Mutants 379–383A5 and 384–388A5 show a complete loss of proto-crown assembly, yielding only amorphous aggregates, whereas 389–393A5 and 391–395A5 show a 10-fold or greater reduced yield of proto-crowns. 394–395A2 is the only alanine block mutant that produces a near-wt yield of assembled proto-crowns; scale bar: 100 nm. (B) Representative 2D class averages from the TEM data for wt protein A and the 394–395A2 mutant. Both assemble characteristic ring structures corresponding to the native 12-mer and a detergent-induced 11-mer conformation. Mutants 379–383A5, 384–388A5, and 389–393A5 did not produce coherent class averages; scale bar: 10 nm. (C) 3D reconstructions of wt and 394–395A2 proto-crowns generated from the 2D class averages. At the resolutions visualized, the two reconstructions are structurally equivalent and closely match the previously determined atomic-resolution single-particle cryo-EM structure (shown superimposed within the density maps), confirming that alanine substitution of C-terminal elbow residues 394–395 does not interfere with proto-crown assembly and does not detectably alter the overall structure.
For all samples, at least 40 micrographs from multiple independent experiments were examined, corresponding to a minimum of ∼50 µm² of grid area per condition. Despite this extensive sampling, which for wt protein A would have revealed at least 650 proto-crowns, no proto-crown-like structures were detected for mutants 379–383A_5_ and 384–388A_5_. Overlapping mutants 389–393A_5_ and 391–395A_5_ yielded rare particles identified as proto-crowns by automated particle picking, with estimated densities not exceeding 0.9 and 1.4 proto-crowns/µm² for 389–393A_5_ and 391–395A_5_, respectively (Fig. 7 and Supplementary Fig. S3). For these two mutants, some of these rare particles appeared consistent with true proto-crowns (Supplementary Fig. S3A), whereas others were more ambiguous (Supplementary Fig. S3B), so that these densities are interpreted as upper limits rather than definitive measurements. Strikingly, the C-terminal elbow mutant 394–395A_2_ was the only alanine block mutant to generate a near wt yield of 7.2 proto-crowns/µm² (Fig. 7A and B). Accordingly, in addition to wt protein A, only mutant 394–395A_2_ produced robust class averages suitable for 3D reconstruction. The resulting image, averaged from 4534 proto-crowns, again closely aligned with the wt atomic-resolution proto-crown structure [13] (Fig. 7C). Mutants 379–383A_5_, 384–388A_5_, and 389–393A_5_ instead yielded heterogeneous, irregular particles and amorphous aggregates without consistent 2D classes, which are being investigated further.
Taken together, these findings confirm the crucial role of the elbow region in crown assembly, showing that wt elbow aa 379–393 have essential roles in proto-crown formation, whereas C-terminal elbow aa 394–395, essential for Pol activation (Fig. 5A), are replaceable by alanines for proto-crown assembly. Combined with the complementation data, these results highlight the modularity and multifunctionality of the elbow, with distinct subregions mediating separate protein A functions such as proto-crown assembly and capping activity versus polymerase activation. The Discussion considers how proto-crown assembly-defective elbow mutants participate in forming mixed crowns when RNA replication is restored by complementation with other co-expressed protein A mutants.
Intra-elbow complementation confirms distinct elbow subdomains for assembly, capping, and polymerase activation
Our assembly (Fig. 7) and complementation data (Fig. 5) sort the alanine block mutants into distinct functional groups. For example, contiguous mutants 379–383A_5_, 384–388A_5_, and 389–393A_5_ share blocks to proto-crown assembly, while downstream mutant 394–395A_2_ retains proto-crown assembly. To further test such functional relationships between elbow subregions, we performed pairwise complementation assays co-expressing each alanine block mutant with every other (Fig. 8). As expected, assembly-deficient mutants 379–383A_5_, 384–388A_5_, and 389–393A_5_ failed to restore RNA replication when co-expressed with each other (Fig 8, lanes 9–11), implying that multiple segments in aa 379–393 are needed to form stable proto-crowns.
Intra-elbow complementation confirms distinct elbow functional subdomains for proto-crown assembly, RNA capping activation, and polymerase activation. Pairwise complementation assays were conducted by co-transfecting S2 cells with pairs of plasmids expressing different elbow alanine block mutants to assess rescue of RNA replication defects. The table at the top shows the mutant combinations tested in the trans-replication assay (RNA1 fs template co-expressed with two protein A variants). Top panel: western blots detecting protein A, with tubulin as a loading control. Middle panel: northern blots detecting RNA1 and RNA3 replication. Bottom panel: bar graphs summarizing RNA3 replication relative to wt control (lane 3). Assembly-deficient mutants (379–383A5, 384–388A5, and 389–393A5) fail to complement one another (lanes 9–11), but are complemented by assembly-competent mutant 394–395A2 (lanes 12–14). Mutant 391–395A5 shows partial complementation with 384–388A5 and 379–383A5, but not with 389–393A5 or 394–395A2 (lanes 15–18).
As noted earlier, beyond their proto-crown assembly phenotypes, the Fig. 5 complementation results imply that mutant 384–388A_5_ is effectively [Cap(+)/Pol(+]); 379–383A_5_ and 389–393A_5_ are partially and completely Cap(-), respectively; 394–395A_2_ is effectively Pol(-); and 391–395A_5_—which spans the regions of Cap(-) 389–393A_5_ and Pol(-) 394–395A_2_—is effectively [Pol(-)/Cap(-)]. In strong validation of these inferred functional enzyme activity classifications, the results of the remaining intra-elbow complementation tests paralleled the Fig. 5 results derived from co-expressing the same alanine block mutations with definitive Cap(-) mutants H93A and D141A, definitive Pol(-) mutant D692E, and [Cap(-)/Pol(-)] mutant H93A + D692E.
Thus, as shown in Fig. 8 lanes 12–14, inferred Pol(-) mutant 394–395A_2_ complemented upstream [Cap(+)/Pol(+)] mutant 384–388A_5_ to high levels (∼80% of wt RNA replication) and Cap(-) elbow mutants 379–383A_5_ and 389–393A_5_ to intermediate levels, yielding a pattern of RNA replication signals mimicking complementation levels of the same three mutants by definitive Pol(-) mutant D692E (Fig. 5A, lanes 8, 11, and 14). Similarly, inferred [Cap(-)/Pol(-)] mutant 391–395A_5_ exhibited partial and selective complementation of all four remaining alanine block mutants (Fig. 8, lanes 15–18) that closely mimicked their relative complementation by [Cap(-)/Pol(-)] mutant H93A + D692E (Fig. 5B, lanes 5–12). Specifically, 391–395A_5_ and H93A + D692E both complemented 384–388A_5_ to ∼50% of wt RNA replication, more weakly rescued 379–383A_5_ to 10%–20% of wt levels, but failed to restore replication of 389–393A_5_ or to raise 394–395A_2_ replication above its ≤1% of wt residual replication activity.
These extensive parallels reinforce the prior conclusions from independent data above that the 17 aa elbow contains multiple regions with discrete, critical functions, and in particular that elbow aa 379–393 contribute cooperatively to proto-crown assembly, that aa 379–393 also contribute to RNA capping with 389–393 being the most critical, and that aa 394–395 are crucial for activating RNA polymerase activity by a mechanism separable from proto-crown assembly.
Discussion
Defining the structure, assembly and operation of the ringed “crown” complexes of (+)RNA virus genome replication proteins should greatly advance understanding and control of these viruses, and constructive applications of their unique gene transfer, amplification and expression pathways [11, 14, 15, 17, 44]. In this study, we explored a striking structural element, designated the elbow, of the archetypal RNA replication crown from the nodavirus FHV [11, 13]. This elbow segment serves as the major connection between adjacent N-proximal domains of multimerized nodaviral RNA replication protein A’s in the crown floor, the highly stable foundation of both the proto-crown, a single-ring, 12-mer protein A assembly intermediate, and the final mature crown, a double ring 24-mer of protein A (Fig. 1).
Although the elbow is only 17 aa long, mutations within it showed an unexpectedly diverse range of individual replication phenotypes and complementation behaviors, revealing its critical interactions with multiple sub-processes in genome RNA replication. Among these were anticipated effects on proto-crown assembly and surprisingly selective effects on 5′ RNA capping and polymerase activities. Together these results show that the elbow is a linchpin of proto-crown assembly and that elbow interactions in and beyond crown floor assembly regulate protein A conformation and multiple crown functions in genome replication. Below we discuss the nature and interrelation of these effects, their possible mechanistic bases, and emerging connections with other viruses.
Crucial elbow roles in proto-crown assembly
Results from multiple approaches in this study combine to show that the elbow is a cornerstone for proto-crown assembly. As initial evidence for such a role, elbow alanine scanning mutants 384–388A_5_ and 379–383A_5_ individually failed to support any detectable RNA replication (Fig. 4) but retained strong (384–388A_5_) or significant (379–383A_5_) abilities to complement in trans active site null mutants in protein A’s 5′ RNA capping or RNA Pol domains (Fig. 5 and Supplementary Fig. S2). These elbow mutants thus retained all sites and intramolecular interactions needed for the enzymatic functions of RNA synthesis and 5′ RNA capping, so that their loss of RNA replication must reflect another defect.
An obvious alternative was a defect in crown assembly, since the elbow participates in the majority of interactions between the adjacent copies of protein A’s N-proximal membrane interaction/RNA capping domain that form the proto-crown floor (Fig. 2). Negative staining assays confirmed that elbow mutants 379–383A_5_, 384–388A_5_, and 389–393A_5_ blocked formation and recovery of stable proto-crowns (Fig. 7). Further underscoring this shared defect in stable proto-crown assembly, all three mutations failed to complement each other to support RNA replication, but complemented proto-crown assembly-competent elbow mutant 394–395A_2_ (Figs 7 and 8). Moreover, for these three assembly-defective mutants, the strongly additive nature of replication defects revealed by single, double and quintuple alanine scanning mutants (Fig. 4; see also Results) is consistent with the elbow’s multiplicity of inter-subunit interactions (Fig. 2B and C).
Complementation of proto-crown assembly-defective mutants 379–383A_5_, 384–388A_5_, and 389–393A_5_ by various co-expressed protein A enzymatic mutants (Fig. 5) shows that the functional mutant pairings must supply RNA capping and polymerase activities and also support proto-crown assembly. That these mutant elbows participate in forming proto-crowns when mixed with some wt elbows, but not when expressed alone, implies that their proto-crown assembly defects do not completely block interaction between two adjacent proto-crown subunits. Among other possibilities, these mutant elbows may have partial, additive defects sufficient to block assembly when present in all subunits but tolerable in a lower fraction, such as reduced interaction affinity, altered ring curvature or other perturbations at the interface between subunits. Tolerance for occasional imperfections in subunit–subunit interactions is increased because the proto-crown only has to remain intact until further stabilized by addition of the upper crown ring (Fig. 1B) to form the full, mature crown. A further illustration of such tolerance and its possible specificity for particular allelic combinations is provided by 384–388A_5_ and 391–395A_5_, which are respectively defective and reduced in proto-crown assembly, but complement each other to ∼50% of wt RNA replication (Fig. 8, lane 16).
The elbow mediates protein A conformational switching with genome economy
Protein A’s conformations in the upper and lower crown rings differ principally in the position of the N-proximal membrane interaction/RNA capping domains, which in the lower ring interact to form the toroidal floor but in the upper ring separate and translate radially outward to form the crown legs (Fig. 1B). Thus, protein A’s transition between the lower proto-crown and the upper ring conformations must involve breaking the elbow’s intermolecular floor-forming interactions and re-positioning the elbow in new but as yet incompletely defined interactions in the upper crown, making the elbow a key switch between these states. The elbow is also immediately followed by the flexible linker preceding the Pol domain (Fig. 1C), which likely also contributes to positional shifts between Pol and the remainder of protein A, and may contribute to regulating Pol activity (see below). Hence, protein A conformational transitions may involve coordinated movements of the juxtaposed elbow and flexible linker.
Whether primordial or lately evolved, the conformational switching features of the elbow and flexible linker appear to be adaptations providing the unusually compact replication machinery of nodavirus protein A with structural flexibility in the absence of virally-encoded protease(s) and site-specific cleavages, which drive conformational changes in the genome replication proteins of alphaviruses, flaviviruses, coronaviruses, retroviruses, and others. While differing in genetic coding economy, both mechanisms enable conformational transitions that sequentially engage successive phases of genome replication, such as RO nucleation and template recruitment by the nodavirus proto-crown, negative- and in turn positive-strand RNA synthesis by the mature crown, etc. [15, 45–48].
Specific elbow mutations selectively block 5′ RNA capping or polymerase activity
While elbow mutant 384–388A_5_ strongly complemented both Cap(-) and Pol(-) active site mutants, some elbow mutants completely lost the ability to complement RNA capping or Pol function. Thus, in addition to the elbow’s roles in crown assembly and conformational switching, certain elbow mutations surprisingly abolished the activity of the flanking RNA capping or distal Pol domain as effectively and selectively as direct mutation of the most crucial active site residues. Mutant 389–393A_5_, e.g. complemented active site Pol null mutant D692E to 29% of wt RNA replication, but did not measurably complement active site capping null mutants H93A or D141A (Fig. 5A and C). Strikingly, these effects were completely reversed in immediately adjacent elbow mutation 394–395A_2_, which complemented capping null mutant H93A to 48% of wt protein A RNA replication but failed to complement Pol null mutant D692E (Fig. 5A and D). The functional equivalence of 394–395A_2_ to a full Pol activity block was confirmed by the 394–395A_2_–D692E double mutant retaining similarly strong complementation of H93A (Fig. 6A) and because 394–395A_2_’s complementation profile against all other alanine block elbow mutants exactly paralleled D692E (Fig. 8). As part of this profile, capping null elbow mutant 389-393A_5_ and Pol null elbow mutant 394-395A_2_ strongly complemented each other, again demonstrating that, though adjacent, their functional asymmetry is indeed reversed and the natures of their complete replication blocks are nonoverlapping.
This dependence on specific elbow domains and their interactions likely limits RNA capping and polymerase functions to particular protein A conformations and crown sites, ensuring that they are only active where and when appropriate for RNA genome replication, and only act on specifically recruited viral RNA templates. For example, since the crown floor is implicated as the site of RNA capping (see Introduction), elbow-mediated floor assembly may position capping-critical elbow aa 389–393 for interactions that activate enzymatic step(s) of RNA capping. The proto-crown protein A structure corresponds to the capping domain’s methyltransferase conformation [13]. The ∼16 Å between the α-phosphate of methyltransferase substrate GTP and acceptor residue H93 for the m^7^GMP product implies that significant GTP movement or local protein A conformational adjustment is required before m^7^GMP is covalently bound and transferred to the viral RNA 5′ end [13]. Since the elbow resides adjacent to the GTP binding site and m^7^GMP acceptor H93 (Figs 1C and 6), elbow aa 389–393 might, e.g. participate in this conformational shift bridging the methyltransferase and guanylyltransferase steps. Such a role could parallel multimerization-induced activation of the structurally related CHIKV nsP1 capping domain in which, after GTP methylation, aa R275 from a neighboring subunit undergoes a conformational change to coordinate the m^7^GTP α-phosphate with its acceptor residue H37, the CHIKV nsP1 equivalent of protein A’s H93 [42].
Similarly, the loss of Pol activity in mutant 394–395A_2_ might be linked to aa 394–395 involvement in transitioning protein A to a polymerase-active conformation. One possibility is that aa 394–395 might participate in releasing the Pol domain on its immediately adjacent flexible linker to “bungee jump” from the upper or lower central turret ring to its central floor active position (Fig. 1 and Introduction). Alternatively or in addition, in such an active conformation, aa 394–395 or adjacent flexible linker sequences might interact with Pol to activate its RNA synthesis functions. This possibility aligns with the need to rearrange elbow interactions for protein A conformation changes and with the strong context dependence of other RNA virus polymerases. Flavivirus RNA polymerase, e.g. binds its N-proximal methyltransferase domain, and various methyltransferase mutations in that interface stimulate or inhibit Pol initiation and elongation [49, 50]. The relative activity of alphavirus nsP4 Pol in negative- versus positive-strand RNA synthesis is strongly altered by changes in the proteolytic processing state of associated RNA replication proteins nsP123 [47]. Rotavirus VP1 Pol only functions inside virion core shells with significant similarities to (+)RNA virus ROs [15, 51]. VP1 Pol interacts with specific peptides in core shell protein VP2 to stimulate or inhibit Pol activity, at least sometimes through induced VP1 conformational changes [52].
In summary, our results establish not only that nodavirus protein A’s elbow is central to crown assembly, but that distinct elbow regions exert highly specific, dramatic and divergent regulatory effects on the activity of flanking RNA capping and Pol domains. Through these regulatory roles, the elbow provides a genetically economical alternative to the use in many other viruses of virally encoded protease processing to activate in turn different RNA synthesis steps. This work also builds on prior results [13, 15], to demonstrate the importance, genetic separability and interdependence of multiple protein A functions and provides key foundations for future studies to define how different protein A segments and interactions coordinate crown state transitions, domain activities, and the overall progression of viral replication.
Supplementary Material
gkag151_Supplemental_File
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Venter PA, Schneemann A. Recent insights into the biology and biomedical applications of flock house virus. Cell Mol Life Sci. 2008;65:2675–87. 10.1007/s 00018-008-8037-y.18516498 PMC 2536769 · doi ↗ · pubmed ↗
- 2Ball LA, Johnson KL. Reverse genetics of nodaviruses. Adv Virus Res. 1999;53:229–44. 10.1016/S 0065-3527(08)60350-4.10582101 · doi ↗ · pubmed ↗
- 3Price BD, Rueckert RR, Ahlquist P. Complete replication of an animal virus and maintenance of expression vectors derived from it in Saccharomyces cerevisiae. Proc Natl Acad Sci USA. 1996;93:9465–70. 10.1073/pnas.93.18.9465.8790353 PMC 38451 · doi ↗ · pubmed ↗
- 4Lu R, Maduro M, Li F et al. Animal virus replication and RN Ai-mediated antiviral silencing in Caenorhabditis elegans. Nature. 2005;436:1040–3. 10.1038/nature 03870.16107851 PMC 1388260 · doi ↗ · pubmed ↗
- 5Cortese M, Goellner S, Acosta EG et al. Ultrastructural characterization of Zika virus replication factories. Cell Rep. 2017;18:2113–23. 10.1016/j.celrep.2017.02.014.28249158 PMC 5340982 · doi ↗ · pubmed ↗
- 6Gillespie LK, Hoenen A, Morgan G et al. The endoplasmic reticulum provides the membrane platform for biogenesis of the flavivirus replication complex. J Virol. 2010;84:10438–47. 10.1128/JVI.00986-10.20686019 PMC 2950591 · doi ↗ · pubmed ↗
- 7Kopek BG, Perkins G, Miller DJ et al. Three-dimensional analysis of a viral RNA replication complex reveals a virus-induced mini-organelle. P Lo S Biol. 2007;5:e 220. 10.1371/journal.pbio.0050220.17696647 PMC 1945040 · doi ↗ · pubmed ↗
- 8Laurent T, Kumar P, Liese S et al. Architecture of the chikungunya virus replication organelle. e Life. 2022;11:e 83042. 10.7554/e Life.83042.36259931 PMC 9633065 · doi ↗ · pubmed ↗