mRNA transcription in dsRNA viruses is a highly regulated process but the mechanism of this regulation is not known. Here, by nucleoside triphosphatase (NTPase) assay and comparisons of six high-resolution (2.9–3.1 Å) cryo-electron microscopy structures of cytoplasmic polyhedrosis virus with bound ligands, we show that the large sub-domain of the guanylyltransferase (GTase) domain of the turret protein (TP) also has an ATP-binding site and is likely an ATPase. S-adenosyl-L-methionine (SAM) acts as a signal and binds the methylase-2 domain of TP to induce conformational change of the viral capsid, which in turn activates the putative ATPase. ATP binding/hydrolysis leads to an enlarged capsid for efficient mRNA synthesis, an open GTase domain for His217-mediated guanylyl transfer, and an open methylase-1 domain for SAM binding and methyl transfer. Taken together, our data support a role of the putative ATPase in mediating the activation of mRNA transcription and capping within the confines of the virus.https://doi.org/10.7554/eLife.07901.001
Viruses can only replicate by invading the cells of other organisms, such as plants and animals. Each virus carries genetic material in the form of molecules of DNA or ribonucleic acid (RNA), which are packaged in a shell made of proteins.
The cytoplasmic polyhedrosis virus has a genome made of a type of RNA called double-stranded RNA. Once inside a host cell, sections of the virus genome are copied to make molecules of ‘messenger RNA’ in a process called transcription. Small chemical groups called guanylyl and methyl groups are added to the messenger RNAs before they are used as templates to make the virus proteins.
A small molecule called S-adenosyl-L-methionine (SAM) can activate transcription of the virus genome by binding to a protein called turret in the shell of the virus. The turret protein is involved in adding the guanylyl and methyl groups to the messenger RNA molecules, but it is not clear how the protein activates transcription.
Here, Yu et al. used a technique called cryo electron microscopy to study how the virus binds SAM to activate transcription. The experiments show that the binding of SAM to one region or ‘domain’ of the turret protein leads to changes in the virus shell. This enables another domain of the turret protein to bind a small molecule called ATP and break it down. The energy released from breaking down ATP causes further changes of the shell of the virus to activate transcription and the addition of guanylyl and methyl groups to the newly made messenger RNAs.
In the future, experiments that directly observe the RNA inside each virus shall offer fresh insights as to how the genomes of cytoplasmic polyhedrosis virus and other similar viruses are transcribed.https://doi.org/10.7554/eLife.07901.002
Viral transcription is highly regulated, as demonstrated biochemically in viruses of the Reoviridae (Shatkin and Sipe, 1968; Furuichi, 1974, 1978; Borsa et al., 1981; Farsetta et al., 2000). mRNA transcription in these viruses is activated by external actions, for example, removal of their outer shell in multi-shelled reoviruses (Shatkin and Sipe, 1968; Borsa et al., 1981; Farsetta et al., 2000) and binding of S-adenosyl-L-methionine (SAM) in the single-shelled cytoplasmic polyhedrosis virus (CPV) (Furuichi, 1974, 1978). The outer shell and the binding sites of SAM are far away from the RNA-dependent RNA polymerases (RdRPs) inside the virus. How these external actions regulate viral mRNA transcription has been a mystery.
Viruses in the Reoviridae contain 9–12 segments of dsRNA enclosed within an inner core that is a self-competent molecular machine fully capable of RNA transcription and processing (Mertens et al., 2004; Zhou, 2008). Each of the 9–12 dsRNA segments wraps around an RdRP located underneath an icosahedral vertex and can undergo independent and simultaneous RNA transcription within an intact core (i.e., endogenous RNA transcription) (Smith and Furuichi, 1982). The simplest of these, the single-shelled CPV (Zhou, 2008) has been used as a model system for viral RNA transcription and high-resolution cryo-electron microscopy (cryoEM) studies, as highlighted by the discovery of mRNA cap structures (Furuichi, 1974; Furuichi and Miura, 1975) and the demonstration of near atomic resolution cryoEM (Yu et al., 2008).
To find out how viral mRNA transcription is regulated, we set out to determine a series of structures of CPV in complex with different ligands at resolutions ranging from 2.9 to 3.1 Å. We discovered that the large sub-domain of guanylyltransferase (GTase) domain of CPV turret protein (TP) also has an ATP-binding site and is likely an ATPase that mediates the activation process of viral RNA transcription and capping. This process involves sensing the presence of the signal molecule SAM by methylase −2 (MT-2) domain of CPV TP, activating the putative viral ATPase, enlarging the viral capsid for efficient mRNA syntheses, and opening the GTase and MT-1 to enable guanylyl and methyl transfer.
To reveal the mechanisms of transcriptional regulation of viruses within the Reoviridae family, we determined the cryoEM structures of six CPV/ligand complexes in the presence of magnesium ion: CPV+SAM (i.e., ‘S-CPV’), CPV+SAM+4 nucleoside triphosphates (NTPs) (i.e., transcribing, or ‘t-CPV’), CPV+SAM+GTP+ATP (i.e., ‘SGA-CPV’), CPV+SAM +GTP (i.e., ‘SG-CPV’), CPV+GTP (i.e., ‘G-CPV’), and CPV+ATP (i.e., ‘A-CPV’) at resolutions ranging from 2.9 to 3.1 Å (Figure 1A–C, Table 1, Video 1 and Figure 1—figure supplement 1). This range of resolutions has permitted us to identify side chains of amino acid residues and to define conformations of bound ligands to build atomic models. Like the atomic model of unliganded CPV (Yu et al., 2011), the atomic models of these liganded CPV all contain two conformers of the capsid shell proteins (CSP-A and CSP-B), two conformers of the large protrusion proteins (LPP-3 and LPP-5), and one conformer of TP in each asymmetrical unit (e.g., Figure 1B). We show below that the large sub-domain of GTase domain of TP (Zhou et al., 2003; Yu et al., 2008) also has an additional ATP-binding site and is likely an ATPase (Figure 1D,E). Except for A-CPV, these atomic models also contain ligands revealed in our cryoEM maps (Table 1). In S-CPV, one SAM binds to the MT-2 domain of each TP. In G-CPV, one GTP binds to the GTase site of GTase domain. In SG-CPV, two SAM molecules bind to the MT-1 and MT-2 domains, one Mg2+-GTP to the GTase site and one GTP to the putative ATPase site. In t-CPV and SGA-CPV, two SAM molecules bind to the MT-1 and MT-2 domains (Zhu et al., 2014), one Mg2+-GTP to the GTase site and one ATP to the putative ATPase site. As we will report in detail below, a comparison of these structures and their correlation with the accompanying biochemical results have led to our discovery of a putative ATPase-mediated regulation process for activating viral RNA transcription and capping.
Because SAM is required for efficient mRNA synthesis in CPV in addition to being the methyl donor for mRNA methylation (Furuichi, 1974, 1978), we first asked whether the presence of SAM would have any effect on the structure of CPV. Superposition of the structures (both at 3.1 Å resolution) of S-CPV and the unliganded CPV (Yu et al., 2011) shows that the capsid shell of S-CPV is slightly expanded with a non-uniform outwards movement of all structure proteins (Figure 2A,B, Video 2 and Figure 2—figure supplement 1A–C). (In contrast, the A-CPV structure reported here does not have such capsid expansion and protein movements, see below.) For example, the apical domain of CSP-A, located next to the fivefold axis, has the largest movement of ∼1 Å (RMSD: 0.97 Å); the dimerization domain, located near the twofold axis, has the smallest movement of ∼0.5 Å (RMSD: 0.49 Å); the CPV-unique small protrusion domain, which is located between the apical and the dimerization domain, moves outwards ∼0.8 Å (RMSD: 0.83 Å) (Figure 2A and Video 3). TP, residing on the apical domain of CSP-A, moves outwards ∼1 Å (RMSD: 0.99 Å), which is the same as the displacement of the apical domain of CSP-A (Figure 2B). In CSP-B, located around the threefold axis, the outwards movements of the apical, small protrusion, and dimerization domains are ∼0.85 Å (RMSD: 0.85 Å), 0.6 Å (RMSD: 0.57 Å), and 0.5 Å (RMSD: 0.49 Å), respectively (Figure 2—figure supplement 1A). Accordingly, the movement (RMSD: 0.67 Å) of LPP-5 is slightly larger than that of LPP-3 (RMSD: 0.43 Å) (Figure 2—figure supplement 1B,C).
Both MT-1 and MT-2 domains of TP have the typical structural motif of SAM-dependent methyltransferases with a seven-stranded β-sheet sandwiched by α-helices (Schluckebier et al., 1995; Hodel et al., 1996; Reinisch et al., 2000; Sutton et al., 2007). Unexpectedly, only the MT-2 domain in S-CPV bound SAM (Figure 2C,D and Figure 2—figure supplement 1D). Except for slight displacement due to the outwards movement of TP as described above, the MT-2 domain structure of S-CPV is indistinguishable from that of the unliganded CPV at the current resolution of 3.1 Å (Figure 2E).
The above observed SAM-triggered conformational change correlates with previous biochemical data establishing a role of SAM in inducing mRNA synthesis (Furuichi, 1974, 1978). In order to find out how SAM does this, we obtained a structure of t-CPV at 3.0 Å resolution, that is, virions incubated with SAM, 4 NTPs, and Mg2+. In contrast to that of S-CPV (Figure 3A), the cryoEM image of t-CPV shows string-like densities emanating from the viral particles, which we attribute to newly synthesized mRNA molecules in the process of release from the actively transcribing virions (arrows in Figure 3B). However, no mRNA densities are visible in our icosahedral reconstruction because these RNA molecules are transcripts of different genomic segments at different stages of the dynamic transcription process and are smeared by averaging.
Structural comparison between t-CPV and S-CPV reveals conformational changes of the capsid proteins in t-CPV (Figure 3C,D, Video 4 and Figure 3—figure supplement 1A–C). Among the five protein molecules within each asymmetric unit of CPV, only LPP-3 remains unchanged in both the location and structure (Figure 3—figure supplement 1A) and the other four protein molecules exhibited changes in their locations, their structures, or both. The locations of LPP-5 molecules in t-CPV and S-CPV differ although their structures are the same (Figure 3—figure supplement 1B), indicating a rigid-body type of movement, likely effected by changes of the underlying CSP molecules. By contrast, the other molecules undergo both global domain movements and local conformation changes from S-CPV to t-CPV (Figure 3C,D and Figure 3—figure supplement 1C). All domains of TP undergo global outwards movements (∼9 Å), while only the MT-1 domain and the large sub-domain of GTase domain of TP exhibit local conformation changes (Figure 3C). CSP-A not only rotates outwards (up to 9 Å) around a pivot point near the twofold axis (global domain movements indicated by dashed ellipses in Figure 3D and Video 5), but it also changes conformation in its apical domain (dotted ellipse in Figure 3D). CSP-B also undergoes similar but less obvious changes than CSP-A (Figure 3—figure supplement 1C). These structural changes in CSP molecules result in an enlarged, yet stable capsid of the transcribing CPV. Since viral mRNA synthesis takes place within the confines of intact virus core, an enlarged capsid would facilitate dsRNA template movement, enabling efficient mRNA synthesis.
The t-CPV structure contained two ligands bound to the GTase domain of each TP (Figure 1D, Video 6 and Figure 3—figure supplement 1D). The first is the expected GTP molecule involved in transfer of a guanylyl group catalyzed by GTase and is located at a cleft of the GTase active site (Video 6 and Figure 3—figure supplement 1D). The second is unexpected ligand, bound to the large sub-domain, away from the cleft (Video 6 and Figure 3—figure supplement 1D).
The density of the unexpected ligand is as strong as that of the GTP bound to the GTase site and the surrounding amino acid residues, and it fits very well with the atomic model of an ATP molecule, suggesting that the large sub-domain of GTase domain of TP could also be a viral ATP-binding site (Figure 4A and Video 7). Located outside the turret chamber (Figures 1D, 4A and Figure 4—figure supplement 1), the putative ATP-binding site is inaccessible to both the dsRNA genome and the nascent mRNA, thus rendering it unable to directly participate in the mRNA transcription and the capping reactions but may function as a regulatory protein or enzyme.
To establish the identity of the aforementioned ligand and the function of the putative ATP-binding site, we subsequently performed the following two structural studies.
First, through incubating CPV capsids with SAM, GTP, ATP, and Mg2+, we obtained the SGA-CPV particle which, lacking of UTP and CTP, is incapable of mRNA transcription. Indeed, under cryoEM, SGA-CPV (Figure 4—figure supplement 2A), and S-CPV (Figure 3A) particles look similar and differ from actively transcribing t-CPV particles (Figure 3B). However, the 3.1 Å structure of SGA-CPV shows global movements and local conformational changes of its structure proteins that are indistinguishable from those of t-CPV (Figure 4A,B and Figure 4—figure supplement 2B). Furthermore, the large sub-domain of SGA-CPV GTase domain also contains a ligand similar to that in the t-CPV (Figure 4A,B and Figure 4—figure supplement 2C).
Second, to eliminate the possibility of GTP as the ligand bound to the putative ATP-binding site in t-CPV, we obtained a 3D reconstruction of SG-CPV at 3.1 Å resolution. While the local conformational changes of SG-CPV structure proteins are identical to those of t-CPV (Figure 4C), the global movements of structure proteins of SG-CPV are slightly less than those in t-CPV or SGA-CPV (Figure 4—figure supplement 3A). For example, the movement of GTase domain in SG-CPV is ∼1 Å less than that of t-CPV or SGA-CPV (Figure 4D and Video 8). Most importantly, the density of the bound GTP is not as strong as that of the ligand in t-CPV or SGA-CPV, and its triphosphate group becomes invisible when displayed at the same threshold of 3σ (Figure 4C and Figure 4—figure supplement 3B).
These results indicate that (1) the global movements and local conformational changes of structural proteins observed in t-CPV are not a consequence, but rather a trigger of RNA transcription; (2) the large sub-domain of the GTase domain binds ATP to mediate the conformational changes observed in t-CPV. Consistent with this assignment, only the large sub-domain (the one containing the ATP-binding site) of the GTase domain undergoes significant conformational changes between S-CPV and t-CPV (Figures 3C, 4E). Accompanying these conformational changes, part of the loop connecting α13 and α14 in S-CPV became a helix (αC) in SGA-CPV, SG-CPV, and t-CPV (Figure 4).
Previous biochemical studies have shown that the hydrolysis of ATP is required for mRNA synthesis (Furuichi, 1978) and that efficient synthesis of CPV mRNA depended on the concentrations of SAM and ATP in a synergistic manner (Furuichi, 1981). We reason that the large sub-domain with the ATP-binding site is possibly an ATPase, and the synergy between SAM and ATP reflects a dependence of its activity on the presence of SAM. To test this hypothesis, we first obtained the 3D reconstructions of G-CPV and A-CPV at 2.9 Å and 3.1 Å resolutions, respectively (Figure 1A–C, Table 1, Video 1 and Figure 1—figure supplement 1). Our structures of G-CPV and A-CPV show that, in the absence of SAM, neither GTP nor ATP induced any conformational change (Table 1). While a GTP bound to the GTase active site in G-CPV (Figure 5A and Figure 5—figure supplement 1), neither ATP nor GTP was observed at the newly discovered ATP-binding site in A-CPV and G-CPV (Table 1), indicating that ATP/GTP binding to the large sub-domain of TP GTase domain is directly regulated by SAM, most likely via binding to the MT-2 domain, since the structure of S-CPV revealed only MT-2 domain contained SAM (Figure 2C,D).
We then determined the rates of NTP hydrolysis of CPV in the presence and absence of SAM. Phosphate released upon hydrolysis of NTPs indeed depended on the presence of SAM and the most favorable NTP substrate was ATP, with decreasing rate of hydrolysis of other substrates in the order of GTP > CTP > UTP in the presence of SAM (Figure 5B). Previous biochemical studies have shown that the mRNA transcription of CPV is SAM-dependent and is specifically coupled to ATP hydrolysis (Furuichi, 1978, 1981). Additionally, it has been shown that the removal of turret in orthoreovirus leads loss of mRNA transcription activity (Luongo et al., 2002). Our structural results, when combined with these biochemical data, suggest that the ATP-binding site of TP GTase domain is possibly a SAM-dependent ATPase that mediates the activation of mRNA transcription.
The density for the bound GTP in the GTase site is strong in t-CPV (Figure 6A, Video 6 and Figure 6—figure supplement 1A) but weak in G-CPV, particularly at the triphosphate group (Figure 5A and Figure 5—figure supplement 1).
Compared to that of G-CPV, the opening (or ‘gate’) leading to the GTase site and coupled to the putative mRNA releasing hole (Yu et al., 2011) is widened from 13 to 15 Å in t-CPV, likely to accommodate nascent mRNA (Figures 5A, 6A,B). Following the nomenclature of PBCV-1 GTase (Hakansson et al., 1997), we designate the states of GTase domain in G-CPV and t-CPV as closed and open states, respectively. In the open state of GTase domain in t-CPV, we observed extra density next to the β, γ phosphates of the bound GTP (Figure 6A and Video 6), which we interpret as coordinated Mg2+ for two reasons. First, Mg2+ was the only divalent cation in our reaction mixture. Second, the GTase domains in SGA-CPV and SG-CPV are also in the open state with prominent densities attributable to Mg2+ (Figure 6C,D and Video 9). Biochemical data have shown that the Mg2+ is required for GTase activity of viruses in Reoviridae (Yamakawa et al., 1982; Le Blois et al., 1992; Martinez-Costas et al., 1995; Qiu and Luongo, 2003; Mohd Jaafar et al., 2005). Therefore, only the open state (with the putative Mg2+) is catalytically active. Remarkably, the gate opening is achieved through the displacement of α14 (i.e., the gate helix), one of the three helices comprising the active site of the putative viral ATPase (Figures 4A–C, 6B). However, even though the gate helix α14 controls the open and closed state, it is not part of the GTase active site. In fact, the active sites of the putative ATPase and GTase do not share any amino acids or secondary elements (Figure 1E). Therefore, the putative ATPase regulates GTase activity allosterically.
The bound GTP molecules in the GTase open and closed states exhibit differences in their conformations and interactions with the GTase active site (Figures 5A, 6). The triphosphate moiety forms only one hydrogen bond (between the β phosphate and Tyr59) in the closed conformation (Figure 5A) but forms three or four more hydrogen bonds in the open conformation including the two formed by the α phosphate with His212 and Arg255 (Figure 6A,C,D). The more extensive hydrogen bonds observed in the open conformation is consistent with our assignment of it as the active state.
Catalysis of guanylyl transfer occurs in two steps: reaction with GTP to form a covalent enzyme–GMP intermediate (enzyme guanylylation) and transfer of GMP onto the 5′-diphosphorylated acceptor. Previous loss-of-function mutagenesis study of lysine residues in mammalian reovirus suggested that Lys190 of GTase domain is responsible for guanylylation of GTases (Luongo et al., 2000). Lys190 is located in a 28-aa segment (residues 168–195) that connects two structurally conserved β strands. Surprisingly, the conserved β strands (β3 and β4) in CPV GTPase is connected by a segment of only 13 aa (a loop from residues 166–178), which contains no lysine (Figure 1D,E). The connecting segments do not have sequence or structural similarities. Within the vicinity of the bound GTP, the only lysine residue in CPV GTase domain is Lys234, which maps to a non-conserved residue (Ser259) in mammalian reovirus GTase domain. Moreover, during GTase transition from its closed to open state, the α phosphorus of the GTP moves towards a histidine-rich segment and away from Lys234 (Figures 6, 7 and Figure 6—figure supplement 1). Therefore, our structures indicate that Lys234 cannot directly participate in guanylylation of GTase, a conclusion that is contrary to a previous suggestion based on a likely incorrect placement of a GMP molecule in the active site in a poorer resolution map (Yang et al., 2012).
By contrast, the histidine-rich segment contains two histidines (His208 and His217) that are either hydrogen bonded to or in proximity with the α- or β-phosphate of the bound GTP in the open state of CPV GTase (Figure 6A,C,D) and conserved in both the orthoreovirus (Reinisch et al., 2000) and aquareovirus (Zhang et al., 2010b) (Figure 6—figure supplement 2). Located on the same side of the leaving group (i.e., β, γ-diphosphate), His208 is hydrogen bonded to the β-phosphate of the GTP in SGA-CPV and SG-CPV. Therefore, His208 is not the active site residue; rather, it stabilizes the charge built up on the β-phosphate in the transition state during the catalysis process. Instead, His217 is likely the active site residue. His217 and the leaving group are on opposite sides of the α phosphorus, a geometry suitable for in-line nucleophilic attack of the α phosphorus by His217 (Figures 6, 7 and Figure 6—figure supplement 2B). Indeed, these two conserved histidine residues in orthoreovirus are required for the GTase activities (Qiu and Luongo, 2003). Furthermore, unlike KxDG GTases that have maximum activity at high pH, GTases of viruses in the Reoviridae family have maximum activity at pH about or lower than the pKa value (∼6.0) of histidine (Qiu and Luongo, 2003). Because we observed only the pre guanylylation state of GTase in all three CPV structures (t-CPV, SGA-CPV, and SG-CPV), we reason that enzyme guanylylation mediated by His217 is likely the rate-limiting step in the process of guanylyl transfer.
Our structures also indicate that the putative viral ATPase regulates the methyl transfer activity of MT-1 through a long-range allosteric effect (Figures 1D, 8 and Figure 8—figure supplement 1). First of all, active sites of MT-1 and the putative ATPase are spatially separated from each other (Figure 1D and Figure 8—figure supplement 1) as the distance from the putative ATPase site to the MT-1 in the same molecule is ∼80 Å, while that to a neighboring MT-1 is ∼40 Å (Figure 8—figure supplement 1A). Second, even if the putative ATPase is activated by SAM but lacks ATP for binding/hydrolysis, MT-1 remains incapable of SAM binding as was observed in S-CPV. In t-CPV where ATP is available, MT-1 becomes SAM bound (Figure 8 and Figure 8—figure supplement 2A). The structures of MT-2 in S-CPV and t-CPV are essentially identical (Figure 8A,B), but their MT-1 structures differ. In particular, two loops lining one side of the un-occupied MT-1 active site in S-CPV shifted up to 4 Å in t-CPV, resulting in an enlarged active site to accommodate the SAM molecule required for methyl transfer (Figure 8C,D). In SGA-CPV and SG-CPV, the structures of MT-1 domains with bound SAM are essentially identical to those in t-CPV (Figure 8—figure supplement 2), though the outwards movement of MT-1 in SG-CPV is ∼1.1 Å less (Figure 8E), likely due to the lower rate of GTP hydrolysis by the ATPase (Figure 5B).
In this study, we discovered that the large sub-domain of CPV GTase domain has an ATP-binding site and is likely an ATPase. This putative viral ATPase has the conserved structural motif for recognition of the adenine base of ATP: hydrogen bonds between the side chains of Ser301, Asp302 and Tyr316 and the adenine ring, pi–pi stacking between the side chain of Tyr305 with the adenine base, and the cation–pi stacking between the side chain of Arg271 and the adenine base (Figure 4A,B). However, this putative viral ATPase lacks the structural motifs of canonical nucleoside triphosphatases (NTPases) (including cellular kinases), most notably the P loop, for binding the phosphoryl moiety of NTP (Saraste et al., 1990; Smith and Rayment, 1996; Snider and Houry, 2008). Instead, the phosphoryl group of the bound ATP is stabilized through hydrogen bonds with the side chain of Arg271 or Tyr268 from α13 helix (Figure 4A,B). Although the large sub-domain has an α helices/β sheet fold, the putative viral ATPase active site is composed of three consecutive helices of α13, α14, and α15 (Figure 4A,B). Structurally, the putative ATPase is different from all ATPase known to date, and it may thus represent a new type of ATPase.
Although both GTP and ATP can bind at the putative viral ATPase site, their interactions with the protein have some differences (Figure 4A–C). The base and ribose of the bound GTP are less hydrogen bounded to the active site than those of the bound ATP. More importantly, while the triphosphate group of the bound GTP forms only one hydrogen bond through its β phosphate with Tyr268 of the active site (Figure 4C), the triphosphate moiety of the bound ATP is better stabilized by forming two more hydrogen bonds with the protein (Figure 4A,B). Conceivably, the more hydrogen bonded ATP is a more efficient substrate of the putative viral ATPase for hydrolysis than GTP, consistent with the colorimetric assay of CPV NTPase activity (Figure 5B).
By integrating the atomic structures of six different CPV particles and correlation with NTPase assay results, we propose a viral ATPase-mediated activation of mRNA transcription and capping, as illustrated in Figure 9. As a virus must rely on host cell for replication, it is to the best interest of the virus to remain quiescent outside host cells (Figure 9A). CPV senses the entrance into host cytoplasm by detecting the presence of SAM. SAM, acting as a signal and binding to its receptor of MT-2, causes initial conformational change of the virus capsid, which activates the putative viral ATPase (Figure 9B). The activated viral ATPase then binds and hydrolyzes ATP to cause three major structural transformations, leading to mRNA transcription and capping (Figure 9C). First, as a result of the translocation of CSP, the viral capsid is enlarged, facilitating dsRNA template movement and enabling efficient mRNA synthesis (Figure 9C). Second, the GTase domain transforms from its closed to open state (Figure 9C). Although GTP binds to the GTase active site in both states of GTase domain, only in the open state can the GTase bind Mg2+ and catalyze His217-mediated guanylyl transfer. Third, the MT-1 domain transforms from its closed to open conformation (Figure 9C). Only in its open conformation can MT-1 bind SAM. While the MT-1 and GTase domains from the same molecule are separated by the bridge domain and have no direct contact with each other, the MT-1 from a neighboring TP sits atop the putative ATPase sub-domain of GTase domain (Figure 8—figure supplement 1). We, therefore propose that the putative ATPase regulates the activity of MT-1 in a neighboring TP, probably through the conformational changes of the putative ATPase sub-domain upon ATP binding/hydrolysis. Notably, from S-CPV to t-CPV, the C-terminal loop of GTase domain exhibits significant movement towards the MT-1 domain of its neighboring TP molecule, presumably contributing to open the active site of the MT-1 (Figure 8—figure supplement 1).
Several viruses within the Reoviridae, such as rotaviruses and blue-tongue viruses, cause wide spread diseases in human and live stocks. Some of these multi-shelled viruses in the Reoviridae, including the animal reovirus and blue-tongue virus, have been shown to have ATPase activity (Noble and Nibert, 1997; Ramadevi and Roy, 1998). The remarkable parallel between the ATPase activity and the transcription activity indicated that they may have also employed the ATPase-mediated activation for mRNA transcription and capping identified here. For example, reovirus cores had high level of ATPase activity (Noble and Nibert, 1997) and could synthesize mRNA (Shatkin and Sipe, 1968; Banerjee and Shatkin, 1970; Drayna and Fields, 1982; Farsetta et al., 2000), but virions had little ATPase activity (Noble and Nibert, 1997) and could not synthesize mRNA (Shatkin and Sipe, 1968; Farsetta et al., 2000). Thus, it is the removal of outer shell other than SAM that triggers the activation process in multi-shelled reoviruses.
CPV virions were purified as described (Yu et al., 2008). Briefly, purified polyhedra were treated with an alkaline solution of 0.2 M Na2CO3-NaHCO3 (pH 10.8) for 1 hr. The suspension was centrifuged at 10,000×g for 40 min. The resulting supernatant was collected and then centrifuged again at 80,000×g for 60 min at 4°C to pellet the CPV virions. The final pellet was re-suspended in a reaction buffer (70 mM pH 8.0 Tris-Cl, 10 mM MgCl2, and 100 mM NaCl) and used for the following experiments.
We prepared six different CPV samples using a protocol modified from a previously described CPV transcription essay (Smith and Furuichi, 1980). Reaction mixtures (30 μl) contained purified CPV, 70 mM Tris-Cl (pH 8.0), 10 mM MgCl2, 100 mM NaCl, and 1 mM SAM (S-CPV), or 1 mM SAM+2 mM GTP+2 mM UTP+2 mM CTP+4 mM ATP (t-CPV), or 1 mM SAM+2 mM GTP+ 2 mM ATP (SGA-CPV), or 1 mM SAM+2 mM GTP (SG-CPV), or 2 mM GTP (G-CPV), or 2 mM ATP (A-CPV). All reactions were incubated at 31°C for 15 min and stopped by quenching the reaction tubes on ice.
Each of the six different CPV particles mentioned above was embedded in a thin layer of vitreous ice suspended across the holes of holey carbon films by plunge-freezing into liquid ethane. Before data collection, beam tilt was carefully minimized by coma-free alignment. Viral particle samples were kept at liquid-nitrogen temperature. CryoEM images were recorded on Kodak SO163 films at a dosage of ∼25 electrons/Å2 on an FEI Titan Krios cryo-electron microscope operated at 300 kV and 59,000× nominal magnification with parallel beam illumination. The films were digitized with a Nikon scanner at a step size of 6.35 µm/pixel, corresponding to 1.076 Å/pixel at the sample level. Individual particle images (960 × 960 pixels) were first boxed out automatically by the autoBox program in the IMIRS package (Liang et al., 2002) and then followed by manual screening using the EMAN boxer program (Ludtke et al., 1999) to keep only the well-separated, contamination-free, intact RNA-containing particles.
The program CTFFIND (Mindell and Grigorieff, 2003) was used to determine the defocus value and astigmatism parameters for each micrograph. We determined particle orientation, center parameters with the IMIRS package running in MPI-enabled Windows workstations (Liang et al., 2002). 3D reconstruction was performed by eLite3D using graphical processing units (Zhang et al., 2010a). We considered astigmatism during CTF correction in the orientation/center refinement and 3D reconstruction steps.
Effective resolutions of the final reconstructions were estimated to be 2.9–3.1 Å (Figure 1A–C, Table 1 and Figure 1—figure supplement 1), based on the structural features revealed in the cryoEM density maps, R-factors (Wolf et al., 2010), and Fourier shell correlation coefficient (FSC) criterion as defined by Rosenthal and Henderson (Rosenthal and Henderson, 2003). We have previously shown that our common-lines-based programs do not suffer from the problems of over-fitting or model bias (Zhou et al., 2014). To provide further validation, we took advantage of the existence of identical structures of t-CPV and SGA-CPV (both independently determined) and calculated the FSC curves between the SGA-CPV map and SGA-CPV model and that between the SGA-CPV map and t-CPV model. Because these structures were independently determined, they are essentially the same as ‘gold-standard’ FSC (Scheres and Chen, 2012). These analyses further support our conclusion that our reconstructions do not have over-fitting (Figure 1—figure supplement 1B).
Rebuilding the model to fit the EM map was done manually with COOT (Emsley and Cowtan, 2004) with the help of REMO (Li and Zhang, 2009). The ‘regularize zone’ utility of COOT was used to improve model stereochemistry.
These coarse full-atom models were then refined in a pseudocrystallographic manner using Phenix (Adams et al., 2010). This procedure only improves atomic models and does not modify the cryoEM density map. Densities for individual proteins were segmented, put in artificial crystal lattices, and then used to calculate their structural factors. The amplitudes and phases of these structural factors were used as pseudo-experimental diffraction data for model refinement in Phenix. To improve the areas of interaction between different protein subunits, we put the refined structures of all five subunits from an asymmetric unit into a single coordinate file and pseudo-crystallographically refined them simultaneously with their non-crystallographic symmetry. This refinement process uses pseudo-experimental diffraction data generated from the cryoEM map of an asymmetric unit.
CryoEM reconstruction was visualized and segmented using Chimera (Pettersen et al., 2004). All figures were prepared with Chimera and COOT.
The NTPase reactions and colorimetric assay were performed as described by Noble and Nibert (1997) in 1.5-ml eppendorf tubes. NTPase reaction mixtures contained 100 mM Tris-Cl (pH 8.0), 100 mM NaCl, 10 mM MgCl2, without or with 1 mM SAM, 6 × 1011 CPV particles per ml, and 1 mM of one of the 4 NTPs in a total volume of 60 µl. Reaction components were mixed on ice, incubated at 31°C for 30 min and then returned to ice. Termination of each reaction was ensured by the addition of an equal volume of 10% trichloroacetic acid. To measure the amount of phosphate ion in each sample, the stopped reaction mixture was mixed with an equal volume of colorimetric reagent (3 vol of 0.8% ammonium molybdate, 1 vol of 6 N sulfuric acid, 1 vol of 10% [wt/vol] ascorbic acid). After all samples in the experiment were added, the eppendorf tubes were incubated in a water bath at 31°C for 30 min. During development, a reduced phosphomolybdate complex was formed, which was blue in color and quantifiable by A655 nm. In each experiment, samples containing NTP but no CPV were included to permit correction for background.
The cryoEM density maps and atomic coordinates reported here are deposited in the EM Data Bank and the Protein Data Bank with accession codes EMD-6371 (A-CPV), EMD-6374 (G-CPV), EMD-6375 (S-CPV), EMD-6376 (SG-CPV), EMD-6377 (SGA-CPV), EMD-6378 (t-CPV) (Yu et al., 2015a; 2015b; 2015c; 2015d; 2015e; 2015f) and 3JAZ (A-CPV), 3JB0 (G-CPV), 3JB1 (S-CPV), 3JB2 (SG-CPV), 3JB3 (SGA-CPV), 3JAY (t-CPV) (Yu et al., 2015g; 2015h; 2015i; 2015j; 2015k; 2015l), respectively.
PHENIX: a comprehensive Python-based system for macromolecular structure solutionActa Crystallographica. Section D, Biological Crystallography 66:213–221.https://doi.org/10.1107/S0907444909052925
Transcription in vitro by reovirus-associated ribonucleic acid-dependent polymeraseJournal of Virology 6:1–11.
Activation and characterization of the reovirus transcriptase: genetic analysisJournal of Virology 41:110–118.
Transcriptional activities of reovirus RNA polymerase in recoated cores. Initiation and elongation are regulated by separate mechanismsThe Journal of Biological Chemistry 275:39693–39701.https://doi.org/10.1074/jbc.M004562200
‘Pretranscriptional capping’ in the biosynthesis of cytoplasmic polyhedrosis virus mRNAProceedings of the National Academy of Sciences of USA 75:1086–1090.https://doi.org/10.1073/pnas.75.3.1086
Allosteric stimulatory effect of S-adenosylmethionine on the RNA polymerase in cytoplasmic polyhedrosis virus. A model for the positive control of eukaryotic transcriptionThe Journal of Biological Chemistry 256:483–493.
IMIRS: a high-resolution 3D reconstruction package integrated with a relational image databaseJournal of Structural Biology 137:292–304.https://doi.org/10.1016/S1047-8477(02)00014-X
EMAN: semi-automated software for high resolution single particle reconstructionsJournal of Structural Biology 128:82–97.https://doi.org/10.1006/jsbi.1999.4174
Identification of the guanylyltransferase region and active site in reovirus mRNA capping protein lambda2The Journal of Biological Chemistry 275:2804–2810.https://doi.org/10.1074/jbc.275.4.2804
Cypovirus, reoviridaeIn: CM Fauquet, MA Mayo, J Maniloff, U Desselberger, LA Ball, editors. Virus taxonomy, VIIIth report of the ICTV. London: Elsevier/Academic Press. pp. 522–533.
Accurate determination of local defocus and specimen tilt in electron microscopyJournal of Structural Biology 142:334–347.https://doi.org/10.1016/S1047-8477(03)00069-8
Identification and functional analysis of VP3, the guanylyltransferase of Banna virus (genus Seadornavirus, family Reoviridae)The Journal of General Virology 86:1141–1146.https://doi.org/10.1099/vir.0.80579-0
Characterization of an ATPase activity in reovirus cores and its genetic association with core-shell protein lambda1Journal of Virology 71:2182–2191.
UCSF Chimera–a visualization system for exploratory research and analysisJournal of Computational Chemistry 25:1605–1612.https://doi.org/10.1002/jcc.20084
Bluetongue virus core protein VP4 has nucleoside triphosphate phosphohydrolase activityThe Journal of General Virology 79:2475–2480.
The P-loop–a common motif in ATP- and GTP-binding proteinsTrends in Biochemical Sciences 15:430–434.https://doi.org/10.1016/0968-0004(90)90281-F
Prevention of overfitting in cryo-EM structure determinationNature Methods 9:853–854.https://doi.org/10.1038/nmeth.2115
Universal catalytic domain structure of AdoMet-dependent methyltransferasesJournal of Molecular Biology 247:16–20.https://doi.org/10.1006/jmbi.1994.0117
The double-stranded RNA genome segments of cytoplasmic polyhedrosis virus are independently transcribedJournal of Virology 41:326–329.
AAA+ proteins: diversity in function, similarity in structureBiochemical Society Transactions 36:72–77.https://doi.org/10.1042/BST0360072
Bluetongue virus VP4 is an RNA-capping assembly lineNature Structural & Molecular Biology 14:449–451.https://doi.org/10.1038/nsmb1225
Low cost, high performance GPU computing solution for atomic resolution cryoEM single-particle reconstructionJournal of Structural Biology 172:400–406.https://doi.org/10.1016/j.jsb.2010.05.006
CypovirusIn: JT Patton, editors. Segmented double-stranded RNA viruses: structure and molecular biology. Norfolk, UK: Caister Academic Press. pp. 27–43.
Identification of the Active Sites in the Methyltransferases of a Transcribing dsRNA VirusJournal of Molecular Biology 426:2167–2174.https://doi.org/10.1016/j.jmb.2014.03.013
Sjors HW ScheresReviewing Editor; Medical Research Council Laboratory of Molecular Biology, United Kingdom
eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.
Thank you for sending your work entitled “ATP-binding mediates RNA transcription and capping in a dsRNA virus” for consideration at eLife. Your article has been favorably evaluated by Michael Marletta (Senior editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors. The following individuals responsible for the peer review of your submission have agreed to reveal their identity: Sjors Scheres (Reviewing editor) and Venkatar Prasad (peer reviewer). A further reviewer remains anonymous.
All three reviewers thought that the structures presented were of excellent quality and represented a tour-de-force use of near atomic resolution cryo-EM on fairly complicated viral system to derive the mechanistic basis of an intrinsic feature that is common to turreted reoviruses. In particular, the mechanism of SAM activation of the transcriptase was considered to be a nice contribution. Still, a few concerns were raised that would require your careful attention in a revised manuscript. These are (in order of importance):
1) The suggested ATPase activity at the observed site and its postulated role during transcription are not convincing. The presence of ATP bound there where they see it does not mean that this is the ATPase active site. Normally, to catch ATP bound at the active site of an ATPase, it is necessary to use a non-hydrolysable analog. Your manuscript shows also that the particle hydrolyses ATP in the presence of SAM, so why would this ATP molecule be intact in the ATPase active site? In addition, ATP is hydrolyzed usually to recover the energy for some other function, like in motor proteins. It is not obvious here how is the energy of ATP hydrolysis used during transcription. Would its hydrolysis lead to a further change of conformation of the TP? Or how do you envision the use of the ATPase to help in transcription? Currently, the data do not justify the title “ATP-binding mediates RNA transcription and capping in a dsRNA virus”.
2) The description of the actual conformational changes is difficult to follow, and at points appears contradictory. The figures are also not very helpful in showing the conformational change. Morphings of the particle from one state to the others, provided as supplementary animations, would help a lot.
3) Also, the observed conformational differences among some of these maps (e.g. S-CPV and CPV) are very small to support the hypothesis that 'the enlarged capsid may facilitate dsRNA template movement'. Could small differences in magnification have occurred among the different data sets and provide an alternative explanation for the enlarged capsid size? Could you also provide an estimate of the internal cavity in cubic Angstroms for each of the relevant reconstructions?
4) It is not clear how the location of the magnesium location is identified. Whether the typical magnesium coordination geometry is satisfied, and how tentative is the assignment needs to be addressed.
5) In the t-CPV structure: it looks like the transcripts are getting extruded, but it appears there is no density that is observed in the turrets that can be assigned to exiting RNA molecules. Considering that endogenous transcription is a dynamic process, one would expect some of the structural features in the exit pathway should be smeared; however, there is no discussion of this in the manuscript.
The density in the presented figures and movie looks quite good, so the following comments may not be that relevant to this particular paper. However, it is good practice to show clarity in all manuscripts about how potential overfitting was assessed in both cryo-EM map refinement and the atomic model refinement inside that map.
6) You state that 'We have previously shown that our common-lines based programs do not suffer from the problems of over-fitting or model bias (Zhou et al., 2014)'. However, the Zhou 2014 paper doesn't explicitly mention overfitting, although it does show a gold-standard FSC curve in the supplementary information. Does that mean that you used gold-standard refinement procedures in the calculation of these maps? If so, you should mention this explicitly instead of citing the Zhou 2014 paper. If you didn't, then refinements should be redone in the gold-standard manner (or by limiting the resolution used in refinement). Also, it would be good to show (gold-standard) FSC curves for each reconstruction in Figure 1–figure supplement 1.
7) R-factors are reported for the refined atomic models. Also in this refinement overfitting can be an issue. Without proper cross-validation, the R-factor curves in Figure 1supplement 1 cannot be trusted. You could either choose to limit the resolution used in the model refinement (and then assess the falloff of the R-factor or the FSC between the model and the map beyond that resolution, i.e. the predictive power of the model), or you could refine a scrambled version of the model in one of the gold-standard half-maps and then calculate FSC/R-factors between that model and that same half-map as well as FSC/R-factors between that model and the other half-map. The difference between these 2 curves is an indication of the amount of overfitting in the model. In the absence of overfitting, the 2 curves should overlap.https://doi.org/10.7554/eLife.07901.033
- Z Hong Zhou
- Jingchen Sun
- Z Hong Zhou
- Jingchen Sun
- Jingchen Sun
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank Xing Zhang for help in model refinement, Jonathan Jih for assistance in model building, and Ke Ding for making Figure 9. This research is supported in part by the National Institute of Health (GM071940 and AI094386 to ZHZ), the National Natural Science Foundation of China (31172263 to JS), the Natural Science Foundation of Guangdong Province, China (S2013010016750 to JS), and Specialized Research Fund for the Doctoral Program of Higher Education, China (20134404110023 to JS). We acknowledge the use of instruments at the Electron Imaging Center for Nanomachines supported by UCLA and by instrumentation grants from NIH (1S10RR23057) and NSF (DBI-1338135).
- Sjors HW Scheres, Reviewing Editor, Medical Research Council Laboratory of Molecular Biology, United Kingdom
- Received: April 4, 2015
- Accepted: June 19, 2015
- Version of Record published: August 4, 2015 (version 1)
© 2015, Yu et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.