Characterization of a Nodavirus Replicase Revealed a de Novo Initiation Mechanism of RNA Synthesis and Terminal Nucleotidyltransferase Activity*

Background: RNA synthesis initiation and 3′-terminal RNA integrity are pivotal for the replication of (+)-RNA viruses. Results: A nodaviral replicase can initiate RNA synthesis in a primer-independent manner and contains a terminal nucleotidyltransferase activity. Conclusion: These activities collaborate to initiate viral RNA synthesis. Significance: This study revealed the initiation mechanism and terminal repair function of the replicase in Nodaviridae. Nodaviruses are a family of positive-stranded RNA viruses with a bipartite genome of RNAs. In nodaviruses, genomic RNA1 encodes protein A, which is recognized as an RNA-dependent RNA polymerase (RdRP) and functions as the sole viral replicase protein responsible for its RNA replication. Although nodaviral RNA replication has been studied in considerable detail, and nodaviruses are well recognized models for investigating viral RNA replication, the mechanism(s) governing the initiation of nodaviral RNA synthesis have not been determined. In this study, we characterized the RdRP activity of Wuhan nodavirus (WhNV) protein A in detail and determined that this nodaviral protein A initiates RNA synthesis via a de novo mechanism, and this RNA synthesis initiation could be independent of other viral or cellular factors. Moreover, we uncovered that WhNV protein A contains a terminal nucleotidyltransferase (TNTase) activity, which is the first time such an activity has been identified in nodaviruses. We subsequently found that the TNTase activity could function in vitro to repair the 3′ initiation site, which may be digested by cellular exonucleases, to ensure the efficiency and accuracy of viral RNA synthesis initiation. Furthermore, we determined the cis-acting elements for RdRP or TNTase activity at the 3′-end of positive or negative strand RNA1. Taken together, our data establish the de novo synthesis initiation mechanism and the TNTase activity of WhNV protein A, and this work represents an important advance toward understanding the mechanism(s) of nodaviral RNA replication.

est group of viruses, and include numerous important pathogens for plants, animals, and humans (1,2). Because no DNA stage is involved in the life cycle of (ϩ)-RNA viruses, which do not include retroviruses that also contain RNA genomes, all (ϩ)-RNA viruses encode their own RNA-dependent RNA polymerases (RdRPs), usually in concert with other viral and host factors, to catalyze RNA-templated viral RNA synthesis (1,3). For viral RNA synthesis, an accurate initiation is required to ensure the viral genome integrity as well as RNA synthesis efficiency (4 -6). Although diverse initiation mechanisms of viral RNA synthesis have been demonstrated for various (ϩ)-RNA viruses, two principally distinct initiation mechanisms, primerdependent and primer-independent (de novo) initiation, are generally employed by viral RdRPs (7,8). For primer-dependent RNA synthesis initiation, either an oligonucleotide or a protein primer is required (9 -12). On the other hand, de novo initiation uses the starting nucleotide to provide the 3Ј-hydroxyl (3Ј-OH) group for adding the next nucleotide, and in this case, RNA synthesis could be initiated at the terminal or an internal site of the template RNA (4,5,7,13,14).
Unlike many other (ϩ)-RNA viruses, such as flavivirus, picornavirus, tombusvirus, and bromovirus, in which a set of viral RNA replicase proteins synthesizes their RNA genomes, nodaviruses encode only one RNA replicase protein, RdRP (protein A), for viral RNA replication (2,25). On the other hand, nodaviral RNA replication is highly parallel with RNA replication of other (ϩ)-RNA viruses in many features (2,25,26). These features make nodaviruses, such as flock house virus (FHV) and Wuhan nodavirus (WhNV), well recognized and simplified models for studying viral RNA replication (20, 21, 23, 26 -33). Previous studies of FHV, the most extensively studied member of the Nodaviridae family, revealed that protein A catalyzes nodaviral RNA synthesis in concert with the mitochondrial outer membrane and other viral or cellular proteins (34 -38). Like the RdRPs of other (ϩ)-RNA viruses, FHV protein A contains several conserved motifs, including the glycine-aspartate-aspartate (GDD) box that is present in all nodaviruses and strictly required for RNA-dependent RNA replication by all known polymerases (39). Moreover, protein A mediates the formation of nodaviral replication complexes and small spherules by inducing significant remodeling of mitochondrial outer membranes (35,37). However, although FHV protein A has been extensively studied, the biochemical features and RNA synthesis initiation mechanism(s) employed by nodaviral RdRPs have never been determined. This obvious gap precludes a better understanding of nodaviral RNA replication and the use of nodaviruses as model systems for studying viral RNA replication.
As a homolog of FHV, WhNV has been well characterized and has provided novel insights about nodaviral subgenomic RNA synthesis (29) and RNA silencing suppression activities (20,21). Our previous study revealed that internal initiation, instead of previously proposed premature termination, is the mechanism governing the subgenomic RNA3 synthesis of WhNV and probably some other nodaviruses (29). Similar to FHV protein A, WhNV protein A contains six of the eight conserved RdRP motifs, including the GDD box, and associates with the mitochondrial outer membrane during viral RNA replication (40).
In this study, we expressed and purified recombinant WhNV protein A in Escherichia coli and characterized its RdRP activity in detail. Our study revealed that this nodaviral protein A can initiate RNA synthesis via a de novo mechanism, and this RNA synthesis initiation could be independent of other viral or cellular factors. We further showed that WhNV protein A can add ribonucleotides to the 3Ј-end of template RNAs, and this terminal nucleotidyltransferase (TNTase) activity of protein A functions to restore one or two 3Ј-proximal nucleotides of template RNA as a potential mechanism for rescuing 3Ј-terminal nucleotide loss. Furthermore, we also determined the cis-acting elements within positive (ϩ)-or negative (Ϫ)-stranded RNA1 template required for RdRP activity or TNTase activity of protein A. Our study identifies a de novo initiation mechanism of nodaviral RNA synthesis and is the first to reveal that nodaviral protein A has TNTase activity, which has been suggested to play a critical role in ensuring efficient and accurate initiation of viral RNA synthesis.

EXPERIMENTAL PROCEDURES
Construction of Expression Plasmids-To construct the bacterial expression vector for MBP-Prot A or its derivative, the WhNV protein A coding sequence (GenBank TM code AY962576) was PCR-amplified from the cDNA of the WhNV RNA1 genome pWh1(G,0) plasmid constructed as described previously (29) with the primers ProtA-For (GCGGATCCATGGTGTCAG-TAATCAAGAC) and ProtA-Rev (GCGTCGACTTAGCTTA-AGGAACTATTC). The BamHI and SalI nuclease sites were introduced during PCR amplification. An internal BamHI nuclease site at position 1333 in the protein A coding sequence was removed through introduction of a synonymous mutation (G1332A) using PCR-mediated mutagenesis according to our standard procedures (20,21,29,41) (primers CTGAaGATCC-GAAAATGCCTGGCCAG and CTGGCCAGGCATTTTCG-GATCtTCAG, where the lowercase characters indicate the nucleotides replaced for mutagenesis). The polymerase active site mutagenesis GDD to GAA, containing the double substitution of both Asp-726 and Asp-727 to alanine, was also carried out by PCR-mediated mutagenesis (primers GGGATAAGGTAGGG-GCAGTGTTTGGTGCTGCTAGCCTAAACGCTAATCAC-CAAGGAGAG and CTCTCCTTGGTGATTAGCGTTTAG-GCTAGCAGCACCAAACACTGCCCCTACCTTATCCC). After digestion with BamHI and SalI (Takara Bio, Shiga, Japan), the PCR-amplified fragments were cloned into the BamHI-SalI site of the pMAL-c2X vector (New England Biolabs, Ipswich, MA). To construct the bacterial expression vector for Prot A⌬N(1-200)-His 6 or its derivative, an approach similar to that described above was utilized. The EocRI and SalI nuclease sites were introduced during the PCR amplification, and the primers are GCGAATTCGATCCAATATCCAA-CAACCACGCATTAG and GCGTCGACTTAGTGGTGGT GGTGGTGGTGGCTTAAGGAACTATTCTTAAAGAC. The PCR-amplified fragments were cloned into the EcoRI-SalI site of the pET-28a vector (Novagen, Germany). Transformants were analyzed by restriction enzyme mapping and confirmed by sequencing (BioDev-Tech. Co., Beijing, China).
Expression and Purification of Recombinant Proteins-The recombinant protein MBP-Prot A, its mutant, and MBP alone were expressed in E. coli TB1 strain. Cells were grown at 37°C in LB broth with ampicillin (100 mg/liter) and glucose (20 mg/liter). Upon reaching an optical density at 600 nm of 0.6, protein expression was induced by adding 1 mM isopropyl-␤-Dthiogalactopyranoside. Cells were harvested and resuspended using the lysis buffer (20 mM Tris-HCl (pH 7.5), 200 mM NaCl, 1 mM EDTA, 1 mM sodium azide, 10 mM ␤-mercaptoethanol, and 1 mM DTT) after incubation at 25°C for 6 h. Cells were lysed using freezing and thawing, enzyme lysis, and ultrasonication, and the soluble recombinant proteins were purified using amylose resin (New England BioLabs) according to the manufacturer's protocol. The recombinant His 6 -tagged proteins were expressed and purified according to our standard protocol (19,20). Purified proteins were concentrated using Amicon Ultra-15 filters (Millipore, Schwalbach, Germany) and stored in 25 mM HEPES (pH 7.5) at Ϫ20°C according to our standard procedures (20,21,42). Proteins were confirmed by 10% SDS-PAGE and Western blot. All proteins were subjected to 10% SDS-PAGE and quantified with Bio-Rad Quantity One software by using known amounts of bovine serum albumin as controls.
Construction and Purification of RNA Templates-The DNA fragment of the 3Ј-end of the (ϩ)-RNA1, (Ϫ)-RNA1, or (Ϫ)-EGFP ORF was amplified from the cDNA of the WhNV RNA1 genome pWh1(G,0) plasmid constructed as described previously (29) or pAC1-EGFP plasmid constructed as described previously (40). Following the manufacturer's protocol (Promega, Madison, WI), each pair of the primers was designed to include one primer, which contains a portion of the 5Ј-end of the expected RNA product plus the T7 promoter region (TAATACGACTCACTATAGG), and the other primer, whose sequence is reverse-complementary to the 3Ј-end of the expected RNA product. RNA was transcribed in vitro from the PCR products using T7 RNA polymerase (Promega) for 4 h, respectively. DNA templates were removed by RQ1 RNase-free DNase I (Promega) at 37°C for 1 h. The RNAs were then extracted using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. For the kinetic studies and TNTase activity assay, to eliminate T7 RNA polymerase and NTP, RNAs were electrophoresed in a 6% polyacrylamide, 7 M urea gel and further purified by the Poly-Gel RNA extraction kit (Omega Bio-Tek, Norcross, GA) according to the manufacturer's instructions.
The blocking of RNA template was conducted by oxidating the 3Ј-hydroxyl group (3Ј-OH), using sodium metaperiodate (NaIO 4 ) (10,11,43). In brief, the RNA templates were placed in 100 mM sodium acetate (pH 5.0) and 100 mM NaIO 4 . The reactions were incubated at 25°C for 90 min, and then the treated RNAs were purified by the Poly-Gel RNA extraction kit (Omega Bio-Tek) according to the manufacturer's instructions. All purified RNAs were quantified with Bio-Rad Quantity One software by using known amounts of RNA as controls after Northern blot analysis.
For S1 nuclease analysis, the RdRP reaction products were incubated with S1 nuclease (Takara Bio), 10ϫ S1 nuclease buffer (300 mM sodium acetate (pH 4.5), 10 mM ZnSO 4 , 2800 mM NaCl), and RNase-DNase free water to a final volume of 50 l. The reaction mixture was incubated at 37°C for 10 or 60 min and then stopped by adding 0.1 M EDTA. The reaction mixture was supplemented with RNA loading mix (containing 20% 10ϫ loading buffer (Takara Bio), 50% formamide, and 20% formaldehyde) of equal volume and then heated at 65°C for 5 min before Northern blot analysis. The 10ϫ loading buffer contains 1% SDS, 50% glycerol, and 0.05% bromphenol blue.
The kinetic analyses of WhNV protein A were performed in a 25-l reaction volume containing 600 nM purified MBP-Prot A, 500 M each ribonucleotide except the limiting ribonucleotide (other components were similar with standard RdRP reaction). Reactions were incubated at 20°C for 40 min. The limiting ribonucleotide concentrations were 2.0, 2.5, 3.3, 6.0, 7.0, 8.0, 9.0, and 10.0 M of each NTP. Lineweaver-Burk plots were used to measure the kinetic parameters K m and k cat.
For studying the effects of RdRP inhibitors, different concentrations of gliotoxin or phosphonoacetic acid (PAA) were added into the standard RdRP reaction mixtures. Gliotoxin was purchased from EMD Chemicals (Billerica, MA), and PAA was purchased from Sigma. To explore the effects of primers in the RdRP reaction, oligo(A) 18 , oligo(U) 18 , oligo(C) 18 , oligo(G) 18 , and the specific RNA primer were commercially synthesized by Takara Bio, whereas oligo(dT) 18 and the specific DNA primer used were commercially synthesized by Invitrogen.
All reactions were stopped by adding equal volumes of the RNA loading mix. The samples were then heated at 65°C for 5 min and flash-chilled on ice, followed by Northern blot analysis according to our standard procedures (19,29). The sequences of 1-100 nucleotides at the 5Ј-end of (ϩ)-RNA1 and 3Ј-end of (Ϫ)-RNA1 were used as the probes to separately hybridize the template RNA and the RdRP products when the 3Ј-end of (Ϫ)-RNA1 was used as a template. Similarly, the sequences of 1-100 nucleotides at the 5Ј-end of (Ϫ)-RNA1 and 3Ј-end of (ϩ)-RNA1 were used as the probes to separately hybridize the template RNA and the RdRp products when the 3Ј-end of (ϩ)-RNA1 was used as template. All probes were labeled with digoxigenin (DIG)-UTP (Roche Applied Science) by in vitro transcription according to our standard procedures (19,29). After hybridization, the nylon membrane was incubated with anti-DIG-AP antibody (Roche Applied Science) and then incubated with CDP-Star TM (Roche Applied Science), a kind of chemiluminescent alkaline phosphatase substrate, at 37°C for 10 -15 min. The signal is then visualized by radiography on an x-ray film. The RNAs were quantified with Bio-Rad Quantity One software by using known amounts of expected RNA products as controls.
TNTase Activity Assay-For the standard TNTase reaction, 250 mM DIG-labeled UTP mix (65% DIG-labeled UTP together with 35% UTP) was used instead of 250 mM (each) NTP in standard RdRP reaction. Other components were similar with standard RdRP reaction. Other NTPs in a concentration of 250 mM were added into the standard TNTase reaction mixtures in the same conditions. The reactions were stopped by adding equal volumes of the RNA loading mix and heated at 65°C for 5 min. The products of the TNTase reaction were transferred onto Nϩ nylon membrane (Millipore) and fixed. The nylon membrane was then treated by the antibody anti-DIG-AP (Roche Applied Science).
Then we incubated the nylon membrane with CDP-Star TM (Roche Applied Science) at 37°C for 10 -15 min. The signal was then visualized by radiography on x-ray film.

WhNV Protein A Possesses RdRP
Activity-To characterize the RdRP activity of WhNV protein A, wild-type and replication-defective mutant forms of soluble recombinant protein A are required. For this purpose, we expressed wild-type MBP fusion protein A (MBP-Prot A) and GDD-to-GAA mutant protein A (MBP-mProt A) in E. coli and purified the protein using amylose resins. Of note, mutating the conserved RNA replication GDD motif to GAA would normally disrupt the RdRP activity, thereby resulting in a replication-defective mutant (39, 44 -46). Although the solubility of the recombinant proteins was quite limited, both MBP-Prot A and MBP-mProt A were successfully expressed at their expected molecular masses (around 158 kDa; Fig. 1A). (1-100)pA, was incubated with the indicated proteins. The templates and reaction products were analyzed on a denaturing formaldehyde-agarose gel and detected via Northern blot analysis. D, the template, (ϩ)-RNA1(1-100) or (ϩ)-RNA1(1-100)pA, was incubated with the indicated proteins. The templates and reaction products were analyzed and detected as in C. E, the RNA products synthesized in C and D were subjected to RT-PCR. RT was conducted in the presence or absence of specific RT primers, followed by PCR amplification. PCR products were electrophoresed through an agarose gel and visualized via ethidium bromide staining. Lane M, DNA ladder. F, the RNA products synthesized in C were subjected to S1 nuclease digestion analyses under the indicated conditions. The templates and reaction products were analyzed and detected as in C. G, the template, (Ϫ)-RNA1(1-400) was incubated with the indicated proteins and all four NTPs. After RdRP reaction, the products were analyzed by electrophoresis on denaturing formaldehyde-agarose gel and detected via Northern blot analysis.
To determine whether WhNV protein A could direct RNA1specific sequences for RNA synthesis initiation in vitro, we used the 3Ј-end 100 nucleotides of (ϩ)-RNA1 and (Ϫ)-RNA1, (ϩ)-RNA1(1-100), and (Ϫ)-RNA1(1-100) as the templates for RNA synthesis (Fig. 1B). Moreover, a poly(A) tail (pA) was added to each template to assess the sensitivity of protein A to the 3Ј-end addition of extra nucleotides (Fig. 1B). The RNA products synthesized by protein A were subjected to electrophoresis with denaturing formaldehyde-agarose gels and detected via Northern blot analysis. Protein A efficiently initiated (ϩ)-RNA1 synthesis from the (Ϫ)-RNA1(1-100) template (Fig. 1C, lane 6), whereas in the presence of the poly-A tail, the ability of protein A to synthesize (ϩ)-RNA1 was much lower (Fig. 1C, lane 10). As expected, the negative controls, MBP alone, and GDD-to-GAA mutant protein A (MBP-mProt A) could not initiate RNA synthesis (Fig. 1C, lanes 5, 7, 9, and 11). Similar results were obtained when (ϩ)-RNA1(1-100) or (ϩ)-RNA1(1-100)pA was used as a template for (Ϫ)-RNA synthesis (Fig. 1D). Interestingly, our results showed that purified recombinant protein A did not require the presence of primer or other viral or cellular factors for RNA synthesis (Fig. 1, C and D), suggesting that other viral or cellular factors are not required for the replication initiation ability of protein A.
To further confirm that the RNA products were synthesized from the provided templates, we performed RT-PCR to reverse transcribe the RNA products to DNA using specific primers. The RNA products were reverse transcribed into DNA fragments at the expected size (100 nt) when specific RT-PCR primers were used (Fig. 1E, lanes 2, 4, 6, and 8). On the other hand, when specific primers were not supplemented in the reverse transcription step of RT-PCR, DNA fragments were not detected (Fig. 1E, lanes 1, 3, 5, and 7), thereby excluding the possibility that some DNA contaminants existed in the RNA1 templates. Moreover, these reverse transcribed DNA fragments were further analyzed by DNA sequencing, which confirmed that the RNA products were synthesized specifically from the provided RNA templates (data not shown). Furthermore, when (ϩ)-and (Ϫ)-RNA1(1-100)pA were used as templates, the DNA fragments could not be detected when oligo(dT) 18 was used as a primer to amplify RT-synthesized cDNAs, thus indicating that the poly(A) tail was not recognized by the RdRP (protein A) and that the initiation sites were upstream of the poly(A) tails (data not shown).
Because S1 nuclease catalyzes the specific degradation of single-stranded RNA to mononucleotides, we incubated the RNA products with S1 nuclease. In contrast to single-stranded RNA, the RNA products synthesized by protein A should complement the templates, thereby resulting in a double-stranded form of RNA that protects the RNA products from S1 nuclease degradation. The RNA products became much less susceptible to S1 nuclease treatment (Fig. 1F, lanes 3-5), whereas separating the double strands into single strands by heating at 95°C resulted in the complete degradation of the RNA products (Fig.  1F, lanes 1 and 2).
Moreover, we tried to examine the ability of protein A to initiate replication using a full-length (Ϫ)-RNA1 template; however, full-length RNA product(s) could not be observed, but products of ϳ300 -500 nt were readily detectable (data not shown). This incomplete replication of full-length RNA1 may be due to the fact that the in vitro RdRP reaction condition is not sufficient for replicating such a long RNA. Indeed, membrane and other cellular factors are not present in this reaction but are required for proper viral RNA replication, as suggested in previous studies of FHV (47,48). On the other hand, the 400-nt (Ϫ)-RNA1 template was fully replicated by protein A (Fig. 1G).
Furthermore, to exclude the potential interference of the MBP fusion tag with the RdRP activity, we expressed a recombinant protein A that carries a His 6 tag at its C terminus and has its N-terminal 200 residues (hydrophobic transmembrane region) (40) deleted (Prot A⌬N(1-200)-His 6 ). We also introduced the GDD-to-GAA mutation into this protein to generate mProt A⌬N(1-200)-His 6 ( Fig. 2A) . Our results showed that His 6 -tagged Prot A⌬N(1-200) also possesses RdRP activity, like full-length MBP-Prot A, and His 6 -tagged mutant Prot A could not initiate RNA synthesis (Fig. 2B). Taken together, our results show that WhNV protein A possesses the RdRP activity to initiate the RNA synthesis from nodaviral RNA1-specific templates.
WhNV Protein A Initiates RNA Synthesis via a de Novo Mechanism-After determining that protein A possesses RdRP activity, we sought to identify the mechanism by which protein A initiates RNA synthesis. To this end, (Ϫ)-RNA1(1-100) and (Ϫ)-RNA1(1-100)pA were used as templates, and specific DNA or RNA primers were added to the RdRP-catalyzed RNA synthesis reaction to determine whether these primers could promote (ϩ)-RNA synthesis. DNA primers exhibited little effect on the RdRP reaction (Fig. 3, A (lanes 4 and 5) and B (lanes 4 and 5)), whereas the presence of RNA primers dramatically inhibited the RdRP reaction (Fig. 3, A (lanes 3 and 5) and B (lanes 3 and 5)). To confirm these results, we added a poly(U) (pU), poly(C) (pC), or poly(G) tail (pG) to the 3Ј-end of the (Ϫ)-RNA1(1-100) template to generate (Ϫ)-RNA1(1-100)pU, (Ϫ)-RNA1(1-100)pC, and (Ϫ)-RNA1(1-100)pG together with previously generated (Ϫ)-RNA1(1-100)pA as templates, and different concentrations (0.05-1.0 M) of primers, as indicated, were added to the RdRP reaction together with protein A (Fig.  3C). Our results further confirmed that RNA primers negatively affected RdRP-catalyzed RNA synthesis, and the inhibitory effects were sensitive to the dose of RNA primers (Fig. 3C). This inhibitory effect of RNA primers could be caused by preventing the RdRP from accessing the initiation site on the single-stranded template or by hampering the formation of the proper RNA structure around the initiation site. Based on these and previous results, the RdRP-catalyzed RNA synthesis does not require an exogenous primer.
In the absence of an exogenous primer, the 3Ј-end of some RNA templates can fold back to form a hairpin structure, and then the folded back endogenous sequence can serve as a primer to initiate RNA synthesis. This mechanism is called "copy-back" initiation, and it is also primer-dependent (11)(12)(13). If such a mechanism is used, a product, whose size is the double of the RNA template, should be observed because the synthesized strand is covalently linked to the template through the hairpin. Our previous results (Figs. 1 and 3) showed that no double-sized product (ϳ200 or ϳ236 nt) was detected. To further exclude the possibility that a copy-back mechanism is used by protein A, we designed templates, (Ϫ)-RNA1(1-100) and (Ϫ)-RNA1(1-100)pA, in which the 3Ј-OHs were oxidized. The oxidized 3Ј-OH cannot be used to add extra nucleotides for oligonucleotide elongation and thus would completely block RNA synthesis via a copy-back mechanism (10,11,43). Our results showed that the templates with oxidized 3Ј-OH exhibited the same ability to initiate RNA synthesis (Fig. 3, A (lanes  5-7) and B (lanes 5-7)), thereby showing that a copy-back mechanism was not involved in the RdRP reaction. Altogether, our findings show that WhNV protein A initiates RNA synthesis generally via a de novo mechanism.
Biochemical Characterization and the Effects of RdRP Inhibitors on the RdRP Activity of Protein A-To characterize the biochemical properties of WhNV protein A, we examined its RdRP activity under varying reaction conditions, such as time, enzyme concentration, temperature, pH, and divalent metal ion concentration. Because protein A initiates RNA synthesis FIGURE 3. WhNV protein A initiates RNA synthesis via a de novo mechanism. A, the indicated template, intact or with its 3Ј-end blocked via oxidation, was incubated with the indicated protein and specific DNA or RNA primer. The templates and reaction products were analyzed by electrophoresis on a denaturing formaldehyde-agarose gel and detected via Northern blot analysis. Lane 1, synthesized RNA at the designated size (100 nt) generated by T7 polymerasemediated in vitro transcription. B, the indicated template with a pA, intact or with its 3Ј-end blocked via oxidation, was incubated with the indicated protein and oligo(dT) 18 or oligo(U) 18 primer. The templates and reaction products were analyzed and detected as in A. Lane 1, RNA at the designated size (118 nt) generated by T7 polymerase. C, the templates in the presence or absence of pA, pU, pC, or pG were incubated with MBP-Prot A together with the indicated specific primers at different concentrations. The synthesized RNA products were measured via Bio-Rad Quantity One software, and the relative RdRP activities were determined by comparing the RNA product level in the presence of the indicated primer with the RNA product level without the primer. Error bars, S.D. values from at least three independently repeated experiments.
generally via a de novo mechanism, we determined the optimal conditions for de novo RNA synthesis using (Ϫ)-RNA1(1-100) as the template. The RdRP products were quantified with Bio-Rad Quantity One software using known amounts of expected RNA as controls after Northern blot analysis. A standard curve exhibiting the linear relationship between RNA quantity and optical density is shown in Fig. 4A.
The amount of RNA products increased with the reaction time up to 60 min (Fig. 4B). Moreover, when the reaction time was 60 min, the amount of RNA products strongly and positively correlated with the RdRP concentrations from 0 to 0.7 M (Fig. 4C). The RdRP activity of protein A was also strongly dependent on temperature and pH, and it was highest at around 20°C (Fig. 4D) and pH 7.5 (Fig. 4E), consistent with the internal environment of lepidopterous larvae enterocytes. Furthermore, RdRP activity was measured in the presence of increasing con-centrations of divalent metal ions, Mg 2ϩ or Mn 2ϩ . Our results showed that the RdRP activity of WhNV protein A prefers Mn 2ϩ over Mg 2ϩ as a cofactor in vitro (Fig. 4F). The maximum RdRP activity was observed at 1.5 mM Mn 2ϩ , and higher concentrations of Mn 2ϩ negatively affected RdRP-catalyzed RNA synthesis (Fig. 4F). Such a preference for Mn 2ϩ is similar to that of several other viral RdRPs, including poliovirus 3D pol , norovirus 3D, hepatitis C virus NS5B, etc. (13, 49 -52).
We also measured the kinetic parameters K m (the Michaelis constant that describes the amount of substrate needed for an enzyme to obtain half of its maximum rate of reaction) and k cat (a constant that describes the turnover rate of an enzyme-substrate complex to product and enzyme) and indicated the RdRP activity of WhNV protein A as k cat /K m (the rate constant that measures catalytic efficiency) in the presence of different limiting ribonucleotides. The results derived from these experi-  Table 1. Our results showed that WhNV protein A was able to incorporate all kinds of NTPs on the RNA template. The catalytic efficiency (k cat / K m ) of this activity with ATP, UTP, CTP, and GTP ranged from 13.0 to 28.2 (liters⅐mol Ϫ1 ⅐s Ϫ1 ), and the elongation rate (k cat ) with ATP, UTP, CTP, and GTP ranged from 7.12 ϫ 10 Ϫ5 to 22.19 ϫ 10 Ϫ5 nucleotides incorporated/s (Table 1). Considering the limited solubility of the recombinant protein, the kinetic parameters K m and k cat of the protein in the soluble fraction may not accurately reflect the corresponding properties of the protein in its native state.
Others have reported previously that NTP preincubation with RNA templates enhanced viral RdRP activity to synthesize RNA in vitro (53,54). To determine whether NTP preincubation also enhances nodaviral RdRP activity, we first preincubated MBP-Prot A and the RNA template, (Ϫ)-RNA1(1-100), with a 0.5 or 1 mM concentration of each NTP for 30 min and then incubated them for an additional 60 min with 0.25 mM total NTP mix as substrates. The amounts of RNA products with the preincubation step were then compared with those without the preincubation step (Fig. 6A). Our results showed that NTP preincubation significantly enhanced RdRP activity (Fig. 6A). Moreover, we evaluated the sensitivity of RdRP activity to heat, EDTA, and SDS, and our results showed that any one of these conditions could almost abolish RdRP-catalyzed RNA synthesis (Fig. 6B).
Last, we examined the effects of polymerase-specific inhibitors on protein A. Gliotoxin and PAA are well known specific inhibitors to several viral polymerases (50,(55)(56)(57)(58)(59)(60)(61), and we supplemented either inhibitor into the RdRP reaction. Just as it did for norovirus (50), PAA obviously inhibited the RdRP activity of WhNV in a dose-response manner (Fig. 6C, lanes 2 and 6 -8), and the half-inhibitory concentration of PAA for WhNV RdRP activity was no more than 25 mM, whereas gliotoxin exhibited little inhibitory effect, and its half inhibitory concentration was higher than 250 mM (Fig. 6C, lanes 2-5).
WhNV Protein A Also Possesses TNTase Activity-TNTase activity has been reported in several viral RNA polymerases to add nucleotides to the 3Ј-ends of RNAs (11,49,62,63). This activity could be used by RNA viruses as a mechanism to restore the 3Ј initiation site for RNA synthesis and is also one of the prerequisites for a copy-back mechanism (63). To determine whether WhNV protein A also contains TNTase activity, we incubated MBP-Prot A with (Ϫ)-RNA1(1-100) in a reaction mixture containing DIG-labeled UTP mix (65% DIG-labeled UTP together with 35% non-labeled UTP) and/or other NTPs. Our results showed that protein A exhibited the TNTase activity, that can add DIG-labeled UTP to the 3Ј-end of the (Ϫ)-RNA1 substrate in the presence of the indicated NTPs (Fig. 7A,  lanes 2-9). Of note, in the presence of all NTPs, protein A possessed both RdRP and TNTase activities, which could incorporate DIG-UTP into the RdRP product, resulting in much higher level of DIG-labeled products (Fig. 7A, lane 2). Moreover, because the presence of one or two other NTPs did not  appear to interfere with the addition of DIG-UTP, the TNTase activity may have some preference for UTP.
Others have reported previously that for viral RdRPs that also contain TNTase activity, the GDD motif is normally required for both RdRP and TNTase activities (63). Thus, we sought to determine whether the GDD-to-GAA mutation also abolishes the TNTase activity of protein A. As shown in Fig. 7B, the GDD-to-GAA mutant protein A (MBP-mProt A) lost both its TNTase (lanes 2, 4, 6, and 8) and RdRP (lanes 4 and 8) activities using either (Ϫ)-RNA1(1-100) or (ϩ)-RNA1(1-100) as a substrate/template. In addition, His 6 -tagged Prot A⌬N  possessed the same TNTase activity that full-length MBPtagged Prot A possessed, and the GDD-to-GAA mutation also abolished this activity (Fig. 7E). These results confirmed that the GDD motif is required for both activities of protein A and further eliminated the possibility that some unknown cellular TNTase contaminated the purified protein A preparation.
Moreover, we sought to determine whether the TNTase activity of protein A requires the presence of 3Ј-OH on the RNA substrates, (Ϫ)-RNA1(1-100) and (ϩ)-RNA1(1-100). Our results showed that when the 3Ј-OHs of the RNA substrates were oxidized, protein A lost its ability to add DIG-UTP to the substrates (Fig. 7C, lanes 2, 4, 6, and 8). On the other hand, protein A still exhibited the RdRP activity to catalyze RNA synthesis even when the 3Ј-OHs of RNAs were oxidized (Fig. 7C,  lanes 3, 5, 7, and 9).
We also measured the TNTase activity in the presence of Mg 2ϩ or Mn 2ϩ . Our results showed that both Mg 2ϩ and Mn 2ϩ can support the TNTase activity of WhNV protein A (Fig. 7D). Altogether, our findings demonstrate that WhNV protein A also possesses the TNTase activity to add extra nucleotides to the 3Ј end of (ϩ)-RNA or (Ϫ)-RNA substrates.
The RdRP and TNTase Activities of WhNV Protein A Depend on the 3Ј-Proximal Nucleotides of RNA1-Others have reported previously that the 3Ј-regions of RNA templates/substrates are critical for RdRP and TNTase activities of several RNA viruses (4 -6, 63, 64). To determine whether this is also the case for WhNV protein A, we evaluated the impact of mutations or deletions at the 3Ј-end of (Ϫ)-RNA1(1-100) on RNA synthesis (Fig. 8A).
First, we evaluated the effects of 3Ј-nucleotide deletions of (Ϫ)-RNA1(1-100) on RdRP-catalyzed RNA synthesis. In this experiment, 3Ј-OH oxidation, which blocks 3Ј-nucleotide addition, was also used to investigate the role of TNTase in initiating RNA synthesis. When the TNTase-catalyzed nucleotide addition was blocked by 3Ј-OH oxidation, even a single nucleotide loss in the 3Ј-end of the (Ϫ)-RNA1(1-100) template resulted in the complete abolishment of RNA synthesis (Fig. 8B, bottom,  lanes 3-7). On the other hand, in the absence of 3Ј-OH oxidation, RNA synthesis initiation could tolerate the loss of one or two nucleotides in the 3Ј-end (Fig. 8B, top, lanes 3 and 4), whereas the loss of additional nucleotide(s) completely blocked RNA synthesis (Fig. 8B, top, lanes 5-7).
Moreover, we evaluated the effects of 3Ј-nucleotide deletions of (Ϫ)-RNA1(1-100) on the TNTase-catalyzed nucleotide addition. When one or two nucleotides were lost at the 3Ј-end of the (Ϫ)-RNA1(1-100) substrate, the TNTase-catalyzed nucleotide addition could be detected using a standard TNTase assay (Fig. 8C, top, lanes 3 and 4), whereas the loss of additional nucleotide(s) completely blocked the TNTase reaction (Fig. 8C,  top, lanes 5-7). As negative controls, 3Ј-OH oxidation abolished the TNTase reactions under any conditions (Fig. 8C, bottom). Similar results were obtained when (ϩ)-RNA1(1-100) and its 1-5-nt deletion mutants were used as the RdRP template and TNTase substrate (data not shown). These results show that the integrity of the 3Ј-end of the RNA template is essential for initiating RNA synthesis and imply that the TNTase activity of protein A can restore the loss of one or two nucleotides to rescue RdRP-catalyzed RNA synthesis.
Furthermore, we evaluated the effects of extra nucleotides at the 3Ј-end of templates, and our results showed that in the presence of the indicated extra nucleotides (1-4 nucleotides), protein A retained its ability to initiate RNA synthesis (Fig. 8D). These results show that the RdRP reaction is sensitive to terminal losses or substitutions but insensitive to terminal additions, provided that not too many nucleotides are added like the addition of an 18-nt poly(A) tail, as shown previously (Fig. 1, B-D).
On the other hand, when the last nucleotide of the template was substituted, the U-to-A or U-to-C substitution completely blocked RNA synthesis, whereas the U-to-G substitution was well tolerated (Fig. 8E). Moreover, we evaluated the impacts of the last two 3Ј-end nucleotides on RNA synthesis initiation. In this set of experiments, the 3Ј-OH groups of the indicated (Ϫ)-RNA1(1-100) templates were oxidized to exclude the possible interference of TNTase activity. As shown in Fig. 8F, the (Ϫ)-RNA1 template, which has its last 2 nt (AU), replaced with UU, AG, GG, or UG, can still be replicated, and the AU-to-UU substitution resulted in even more efficient replication compared with the wild-type 3Ј-terminus (AU).
TNTase Activity Can Function to Repair the 3Ј-Terminal Nucleotide(s) Loss of the Template and Restore Replication Initiation in Vitro-Because the RdRP and TNTase activities of a viral replicase share the same catalytic site (GDD), it was difficult to study the role of one activity in the absence of the other. To determine whether the TNTase activity of protein A can restore the replicability of the template with loss of 3Ј-terminal nucleotide(s), we designed a set of experiments to deliberately separate the TNTase and RdRP reactions. As illustrated in Fig.  9A, in the first step, we supplemented a single type of NTP to only allow the TNTase reaction, and then, in the second step,  2 and 4 -9). B, the (ϩ)-RNA1(1-100) (lanes 2-5) or (Ϫ)-RNA1(1-100) (lanes 6 -9) substrates as well as DIG-labeled UTP mix were reacted with MBP-Prot A or MBP-mProt A, as indicated, in the absence (lanes 2, 3, 6, and 7) or presence (lanes 4, 5, 8, and 9) of ATP, CTP, and GTP mix. C, the (Ϫ)-RNA1(1-100) (lanes 2-5) or (ϩ)-RNA1(1-100) (lanes 6 -9) substrates were intact (lanes 2, 3, 6, and 7) or with its 3Ј-end blocked via oxidation (lanes 4, 5, 8, and 9). The indicated substrates were incubated with DIG-labeled UTP mix in the presence or absence of ATP, CTP, and GTP mix. D, the (Ϫ)-RNA1(1-100) substrate was reacted with MBP-Prot A with DIG-labeled UTP mix in the absence (lane 1) or presence of a 2 mM concentration of the indicated divalent metal ion (lane 2, Mg 2ϩ ; lane 3, Mn 2ϩ ). E, the (Ϫ)-RNA1(1-100) substrate was reacted with the indicated proteins together with DIG-labeled UTP mix.
For A-C, lane 1 represents synthesized DIG-labeled RNA at the designated size (100 nt) generated by T7 polymerase-mediated in vitro transcription. For A-E, the substrates and TNTase reaction products were analyzed by electrophoresis on denaturing formaldehyde-agarose gel and detected as described under "Experimental Procedures." the TNTase-reacted RNAs were purified and further oxidized on their 3Ј-end OHs to block further TNTase reaction, followed by RdRP reaction in the presence of all four types of NTPs. This experimental design allowed us to better evaluate the potential terminal repair function of the protein A TNTase activity without the interference of its RdRP activity.
As shown in Fig. 9B, when UTP or GTP was supplemented in the TNTase reaction (the first step), the reacted (Ϫ)-RNA1 templates, which had a 1-2-nt deletion at their 3Ј-termini, were efficiently replicated in the RdRP reaction (the second step) (Fig. 9B, lanes 4, 6, 9, and 11). Interestingly, this result is consistent with the previous observation shown in Fig. 8F. Taken together, these data indicate that the TNTase activity of protein A can restore the replicability of the RNA1 template by repairing the 3Ј-terminal nucleotide loss.
cis-Acting Elements at the 3Ј-End of RNA1 Govern the RdRP and TNTase Activities-Because protein A could use (ϩ)-RNA1(1-100) or (Ϫ)-RNA1(1-100) as a template for RNA syn- thesis initiation, these 3Ј-proximal 100 nucleotides of (ϩ)-RNA1 and (Ϫ)-RNA1 must contain cis-acting elements for initiating viral RNA synthesis. To identify the cis-acting elements required for the RdRP and TNTase activities, we designed a set of RNA templates/substrates that included the 3Ј-proximal 100 nucleotides of the (Ϫ)-strand EGFP RNA sequence followed by the indicated RNA1 sequences, as illustrated in Fig. 10A.

DISCUSSION
Like all (ϩ)-RNA viruses, nodaviruses encode an RdRP that functions as the catalytic subunit of the viral replication complex. Protein A, the nodaviral RdRP, directs viral RNA synthesis in concert with other viral or cellular factors. Although nodaviral RNA replication has been studied in considerable detail (20, 21, 26 -33, 35-38, 65-67), and nodaviruses are well recognized models for investigating viral RNA replication, the mechanism(s) governing the initiation of nodaviral RNA synthesis has not been determined. RNA-templated viral RNA synthesis can be initiated by two principally different mechanisms that are either primer-independent (de novo) or primer-dependent (7,8). Herein, we report that WhNV protein A can initiate RNA synthesis in the absence of other viral or cellular factors (Fig. 1). Moreover, based on our findings that protein A-catalyzed RNA synthesis did not depend on the primer or the 3Ј-OH of the RNA template (Fig. 3), we conclude that WhNV protein A initiates RNA synthesis mainly, if not only, via a de novo mechanism, and this initiation mechanism may generally apply to other nodaviruses. Furthermore, we found that a nodaviral protein A has TNTase activity, which could repair the 3Ј initiation site to restore replication initiation in vitro. In addition, we also identified the cis-acting elements for RdRP or TNTase activity at the 3Ј-end of positive or negative strand RNA1.
Both the genomic RNA1 and RNA2 of nodaviruses are capped but not polyadenylated (17,18). Moreover, the presence of a poly(A) tail reduced the level of RNA synthesis from either the (ϩ)-RNA or (Ϫ)-RNA template (Fig. 1). Interestingly, when (ϩ)-RNA or (Ϫ)-RNA with a poly-A tail was used as a template, the DNA fragments (RT-PCR products) could not be observed when oligo(dT) 18 was used as the primer to amplify the cDNA reverse-transcribed from RdRP-synthesized RNA (data not shown), thereby indicating that poly(A) tails were not recognized by this nodaviral RdRP. Thus, WhNV protein A directly recognizes the cis-acting elements at the 3Ј-ends of (ϩ)-RNA1  1 (lanes 2-6), or 2 (lanes 7-11) nucleotides were reacted with MBP-Prot A as indicated. Either no type (lanes 1, 2, and 7) or one type (lanes 3-6 and 8 -11) of the NTPs was added in the first step, as indicated. The templates and RdRP reaction products were analyzed by electrophoresis on denaturing formaldehyde-agarose gels and then detected via Northern blot analysis.
and (Ϫ)-RNA1 to initiate viral RNA synthesis. For de novo initiation, viral RdRPs normally synthesize dinucleotides using the 3Ј-end of RNA as a template (68,69). Considering the primerindependent activity of WhNV protein A, this mechanism should also apply to this nodaviral RdRP.
We have found that the RdRP activity of protein A was sensitive to RdRP inhibitor PAA but not to gliotoxin (Fig. 6C). Normally, PAA is an inhibitor of the replication of DNA viruses, and gliotoxin is an inhibitor of the replication of RNA viruses. Similarly, the RdRP of another (ϩ)-RNA virus, norovirus, is also sensitive to PAA but not to gliotoxin (50). On the other hand, RdRPs of poliovirus and hepatitis C virus show sensitivity to gliotoxin (56,60). The differences in the inhibitor sensitivities may be attributed to the primer-and poly(A)-independent mechanism of nodaviral protein A or noroviral RdRP (50) and may also be attributed to the special amino acid sequence of WhNV RdRP that is partly similar to DNA polymerase by sequence analysis (data not shown).
Previous studies of FHV genomic RNA replication revealed that its RNA synthesis depends on cis-acting elements at the 3Ј-end (26,64,70,71), and the cis-acting element of (ϩ)-RNA1 is around 7.5 times longer than that of (Ϫ)-RNA1 (26). Moreover, the last 2 nucleotides of the 3Ј-end of (Ϫ)-RNA1 have been found to be critical for (ϩ)-RNA1 replication initiation (64,71). Consistently, our findings revealed that both the RdRP and TNTase activities of WhNV protein A required cis-acting elements at the 3Ј-ends of (ϩ)-RNA1 and (Ϫ)-RNA1 (Fig. 8), and the RdRP activity required the 3Ј-proximal 60 nt of (ϩ)-RNA1 as a cis-acting element, which is 6 times longer than that of (Ϫ)-RNA1 (the 3Ј-proximal 10 nt) (Fig. 10). On the other hand, we found that the last 2 nucleotides of the 3Ј-end of either (Ϫ)-RNA1 or (ϩ)-RNA1 are required for the replication initi- FIGURE 10. Determination of cis-acting elements at the 3-end of RNA1 for the RdRP and TNTase activities. A, a schematic illustration of the RNA templates that used part of the EGFP sequence and the indicated (ϩ)-RNA1 or (Ϫ)-RNA1 sequences at the 3Ј-end. B and C, MBP-Prot A was reacted with the indicated RNA templates. The templates and RdRP reaction products were analyzed on denaturing formaldehyde-agarose gels and detected via Northern blot analysis. Lane 1, synthesized RNA at the designated size (180 or 125 nt) generated by T7 polymerase-mediated in vitro transcription. For B, the RNA templates were analyzed via Northern blots with shorter (top) or longer (middle) electrophoresis time. D, MBP-Prot A was reacted with the indicated RNA substrates. The substrates and TNTase reaction products were analyzed and detected as described under "Experimental Procedures." Lane 1, synthesized DIG-labeled RNA at the designated size (100 nt) generated by T7 polymerase-mediated in vitro transcription. ation ( Fig. 8B) (data not shown). Based on these findings, although FHV and WhNV have quite limited homology in the amino acid sequences of their respective protein A as well as the nucleotide sequences of their 3Ј-end cis-acting elements, similar strategies may be utilized by their respective protein A to replicate nodaviral genomic RNAs.
The 3Ј-end regions of template RNAs are of great importance for de novo initiation of RNA synthesis (4 -6, 63), and the 3Ј-ends of viral RNA genomes may be subjected to degradation by cellular exonucleases (63,(72)(73)(74). Others have reported previously that several viral RdRPs also have TNTase activity to add non-templated nucleotides to the 3Ј-end of viral RNAs (11,49,62,63). Although these TNTase activities play important roles in viral RNA synthesis (63), until now, only a few viral RdRPs have been firmly demonstrated to also possess TNTase activity, and the repair of the 3Ј-terminal loss of viral RNAs was often suggested as a potential role of TNTases in viral RNA replication (11,49,62,63). However, because the RdRP and TNTase activities of a viral replicase share the same catalytic site (GDD), it is difficult to study the role of one activity without involving the other. In the case of protein A, we designed a set of experiments to deliberately separate the TNTase and RdRP reactions, allowing us to better evaluate its potential terminal repair function without the interference of its RdRP activity. Our results clearly show that adding U or G to the 3Ј-end of (Ϫ)-RNA1 template, which loses 1 or 2 nt, can restore the replication initiation (Fig. 9). Although the TNTase of protein A may randomly add nucleotides to the 3Ј-end of RNA, our evaluation of the impact of the last two nucleotides on RNA synthesis initiation showed that the RNA1 template, which has its last 2 nt (AU) replaced with UU, AG, GG, or UG, can still be replicated, and the AU-to-UU replacement even resulted in more replication efficiency than wild type (AU) (Fig. 8F). Thus, considering that all four NTPs are present in cells, replicationcompetent 3Ј-termini should be partly recreated by the TNTase activity of protein A in cells.
Interestingly, the TNTase activity of protein A did not add any nucleotide to the (ϩ)-RNA1 or (Ϫ)-RNA1 substrate losing more than two nucleotides, and any such loss of more than two nucleotides resulted in a complete block of RNA synthesis (Fig.  8B). Moreover, we found that WhNV protein A also added nucleotides to intact (ϩ)-RNA1 or (Ϫ)-RNA1 that had no nucleotide loss. The reason that such a nucleotide addition happens is still unclear, but this kind of nucleotide addition FIGURE 11. Hypothesized model for de novo initiation mechanism and terminal repair function of nodaviral protein A. Protein A recognizes the 3Ј-proximal 60 nt of the (ϩ)-RNA1 template (A) or the 3Ј-proximal 10 nt of the (Ϫ)-RNA1 template (B) and initiates RNA synthesis via a de novo mechanism, as indicated by solid arrows. Under certain conditions, the 3Ј-end of the viral RNAs may be degraded by some cellular exonucleases, which would damage the 3Ј-end integrity and inhibit the initiation of viral RNAs synthesis. In the case that one or two nucleotides at the 3Ј-ends of the (ϩ)-RNA1 (A) or (Ϫ)-RNA1 (B) template are lost, protein A will recognize the 3Ј-proximal 3-60 nt of the (ϩ)-RNA1 template (A) or the 3Ј-proximal 3-30 nt of the (Ϫ)-RNA1 template (B) and add nucleotides to the 3Ј-end using its TNTase activity to restore the initiation of RNA synthesis.
to non-digested RNA may serve as an active protection strategy for potential degradation in the 3Ј-end of RNA1. On the other hand, the TNTase activity required the 3Ј-proximal 3-60 nt of (ϩ)-RNA1 and the 3Ј-proximal 3-30 nt of (Ϫ)-RNA1 (Fig. 10), showing that the RdRP and TNTase activities require different cis-acting elements. Considering the significant difference in the size of these cis-acting elements, the two activities should recognize the 3Ј-end of RNA1 via different mechanisms.
The typical architecture for the RdRP of RNA viruses usually includes three tunnels: the first for RNA template entry, a second for NTP/pyrophosphate (PP i ) exchange, and a third for dsRNA product exit (49,68,69). For the TNTase activity of the WhNV RdRP, it is interesting to envision which tunnel is responsible for the (ϩ)-or (Ϫ)-RNA strand to gain access to the catalytic site of the RdRP, thereby allowing GDD-dependent NTP addition to the 3Ј-end of the RNAs. Considering RNA polarity issues, the most reasonable speculation is that the TNTase RNA substrate enters the catalytic site by backing into the dsRNA exit tunnel (49). Moreover, blocking the dsRNA exit tunnel to prevent the reentry of RNAs to the catalytic site by host proteins or antiviral compounds may be a strategy to inhibit the replication of (ϩ)-RNA viruses.
In this study, we characterized the RdRP activity of a nodaviral protein A in detail and demonstrated that de novo initiation is the mechanism by which this nodaviral protein A initiates RNA synthesis. To our knowledge, this is the first study to demonstrate TNTase activity in a nodavirus and to characterize it in detail. Although the exact role of TNTase is very difficult to determine in cells because of the technical limitations, our findings indicate that the TNTase activity of protein A can repair the 3Ј initiation site to restore replication initiation at least in vitro. In summary, we have proposed a model in which the terminal repair function of protein A TNTase activity together with its RdRP activity ensure the efficient and accurate initiation of viral RNA synthesis (Fig. 11), and this model may generally apply to other nodaviruses. Altogether, our work represents not only an important step toward a better understanding of nodaviral RNA replication, but also progress in understanding the role of this kind of RdRP-associated enzymatic activity in viral RNA replication.