Requirements for West Nile Virus (–)- and (+)-Strand Subgenomic RNA Synthesis in Vitro by the Viral RNA-dependent RNA Polymerase Expressed in Escherichia coli*

RNA-dependent RNA polymerases (RdRPs) of the Flaviviridae family catalyze replication of positive (+)- strand viral RNA through synthesis of minus (–)-and progeny (+)-strand RNAs. West Nile virus (WNV), a mosquito-borne member, is a rapidly re-emerging human pathogen in the United States since its first outbreak in 1999. To study the replication of the WNV RNA in vitro, an assay is described here that utilizes the WNV RdRP and subgenomic (–)- and (+)-strand template RNAs containing 5′- and 3′-terminal regions (TR) with the conserved sequence elements. Our results show that both 5′- and 3′-TRs of the (+)-strand RNA template including the wild type cyclization (CYC) motifs are important for RNA synthesis. However, the 3′-TR of the (–)-strand RNA template alone is sufficient for RNA synthesis. Mutational analysis of the CYC motifs revealed that the (+)-strand 5′-CYC motif is critical for (–)-strand RNA synthesis but neither the (–)-strand 5′- nor 3′-CYC motif is important for the (+)-strand RNA synthesis. Moreover, the 5′-cap inhibits the (–)-strand RNA synthesis from the 3′ fold-back structure of (+)-strand RNA template without affecting the de novo synthesis of RNA. These results support a model that “cyclization” of the viral RNA play a role for (–)-strand RNA synthesis but not for (+)-strand RNA synthesis.

West Nile virus (WNV), 1 a mosquito-borne member of Flaviviridae family that consists of more than 70 human pathogens, was first isolated in 1937 from the blood of a female patient in the West Nile district of Uganda (1). Since then, the virus has been isolated in humans, birds, and bird-feeding mosquitoes, Culex univittatus. Serological studies revealed a broad distribution of the virus in Africa, West Asia including India, and the Middle East, but only occasionally isolated in Europe, and is thought to be spread by migratory birds (Ref. 2, for reviews, see Refs. 3 and 4). WNV belongs to the Japanese encephalitis serocomplex group, which includes St. Louis encephalitis, Murray Valley encephalitis, Kunjin virus, and Japanese encephalitis virus (5,6). WNV was not previously isolated in the Western hemisphere. The first outbreak of WNV infections occurred in New York in 1999 (7)(8)(9)(10) in which 62 people were confirmed to be infected, seven of which were fatal. Since then, transmission of WNV is spreading rapidly throughout the United States; 12 states along the eastern seaboard in 2000 to 25 states and the District of Columbia by 2001 (3).
WNV, like other flaviviruses, contain a positive (ϩ)-strand RNA of ϳ11 kb in length that is capped at the 5Ј-end and nonpolyadenylated at the 3Ј-end. WNV RNA contains conserved sequence elements within the 5Ј-and 3Ј-terminal regions (TR), which include two self-complementary cyclization motifs (5Ј-CYC and 3Ј-CYC) of 9 nt in length (11). There is a highly conserved stem-loop structure within the 3Ј-terminal 96 nt of 3Ј-UTR of flaviviral RNAs and a less conserved stem-loop structure within 5Ј-UTR (12)(13)(14) (for reviews, see Refs. 4,15,and 16). In addition, there is a potential pseudoknot tertiary structure within the 3Ј-terminal stem-loop structure of WNV RNA (17). Several host proteins have been shown to bind to the 5Ј-and 3Ј-UTR although their functional role in viral RNA replication has not been clearly established (18 -21) (for a review, see Ref. 22). By mutational analysis, the 3Ј stem-loop region within the 3Ј-UTR was shown to be important for viral replication in vivo (23,24) and in vitro (25).
NS5 is the largest of the flaviviral nonstructural proteins. Flaviviral NS5 contains conserved motifs found in several 5Ј-RNA methyltransferases at its N-terminal region (35,36). The crystal structure of the N-terminal region consisting of 296 amino acid residues of DEN2 NS5 containing the guanylyltransferase/methyltransferase domain was recently reported (37). This domain can catalyze the transfer of methyl group from the S-adenosylmethionine to form the 2Ј-O-methylad-enine but no 7-methylguanosine was formed. The C-terminal region of the flaviviral NS5 contains conserved motifs consistent with those of RNA-dependent RNA polymerases (RdRP) encoded by several positive-strand RNA viruses (38,39). In this study, we expressed the WNV NS5 (from the EG101 strain) with an N-terminal His tag in Escherichia coli and purified the protein using metal affinity chromatography and characterized the biochemical properties of the polymerase in RNA synthesis.
An in vitro viral RdRP assay system for DEN2 was reported that was dependent on addition of exogenous subgenomic viral RNA (ϩ)-strand templates containing conserved sequence motifs within the 5Ј-and 3Ј-TR and purified polymerase in the presence of four nucleoside triphosphatase (40). Two RNA products were formed, a template-sized product and a hairpin product, twice the size of the template RNA. The templatesized RNA (1ϫ) was shown to be double-stranded in nature by its resistance to RNase A digestion under high ionic strength conditions, and it was formed as a result of de novo synthesis from the input (ϩ)-strand RNA (40). The hairpin RNA (2ϫ) was formed by 3Ј-end elongation of the template RNA by a "copyback" mechanism. RNase A digestion of the hairpin RNA also gave rise to a template-sized product upon digestion of the single-stranded "fold-back" region.
In this study, we describe the first in vitro RdRP assay for WNV that utilizes purified NS5 and exogenous subgenomic WNV RNA templates of either (ϩ)-or (Ϫ)-strand polarity; the 5Ј-end of template RNAs was either triphosphorylated or had a 5Ј-cap structure. We show that whereas the uncapped (ϩ)strand RNA template produced both hairpin RNA and the template-sized de novo synthesized RNA, addition of 5Ј-cap on the subgenomic (ϩ)-strand RNA template reduced the formation of hairpin RNA by 48% without affecting the synthesis of the de novo product RNA. The newly synthesized RNA from the (ϩ)-strand template RNA was of (Ϫ)-strand polarity as determined by RNase H mapping. Moreover, we show that the WNV NS5 could also utilize uncapped or 5Ј-capped subgenomic WNV (Ϫ)-strand templates to produce predominantly de novo product that are of (ϩ)-strand polarity. Finally, mutational analysis of the 5Ј-and 3Ј-CYC motifs in both subgenomic WNV (ϩ)-and (Ϫ)-strand templates showed that these motifs play a critical role in (Ϫ)-strand synthesis from (ϩ)-strand RNA templates but do not seem to affect the (ϩ)-strand synthesis from (Ϫ)strand templates.

WNV NS5 Expression Plasmid
The sequences corresponding to the WNV NS5 were PCR amplified from the plasmid containing the full-length WNV EG101 strain genome (NCBI number AF260968) using the following primers: the forward primer, 5Ј-TCCCCCGGGTGGGCAAAGGGACGCACCTTG-3Ј and the reverse primer, 5Ј-TCCCCCGGGTTACAGTACTGTGTCCTCAACC-3Ј. The product from this PCR was cloned into the pGEM-T easy vector (Promega) and the sequence was verified. The positive clone was digested with XmaI (underlined) and the fragment was cloned into pQE30 vector (Qiagen), giving rise to pQE30-WNV NS5.

Purification of WNV NS5 from E. coli
For the bacterial expression of WNV NS5, E. coli Top10 FЈ cells (Invitrogen) were transformed with pQE30-WNV NS5 plasmid. Transformed E. coli cells were grown at 37°C in LB medium containing 100 g/ml ampicillin and 0.5% (w/v) glucose to an A 600 nm of 0.6 -0.8. Cells were collected by centrifugation (5000 ϫ g) to remove the glucose from the medium, the pellet was resuspended in LB medium containing 100 g/ml ampicillin and 1 mM isopropyl-1-thio-␤-D-galactopyranoside (Sigma), and incubated for 48 h at 18°C. Bacteria were then collected by centrifugation and stored at Ϫ80°C. For purification of WNV NS5, bacteria were resuspended in a buffer containing 100 mM Tris-HCl, pH 7.5, 300 mM NaCl, 1% Triton X-100, 10 mM 2-mercaptoethanol, and 20% glycerol and were lysed by using a French press. The soluble fraction was loaded onto a Talon column (Clontech). WNV NS5 proteins were eluted with 500 mM imidazole. The peak fractions containing the WNV NS5 protein were pooled and dialyzed against 50 mM Tris-HCl, pH 7.5, 300 mM NaCl, and 40% glycerol. The purified protein was stored at Ϫ20°C.

Plasmid Constructs Encoding Template RNAs
cDNA Constructs Encoding Subgenomic RNA Containing the 5Ј-and 3Ј-TR of WNV Genome-A 281-nt DNA fragment from the 5Ј-TR that includes the 5Ј-untranslated region (UTR) (96 nt) and the 5Ј cyclization motif (CYC, which corresponds to 137-144 nt of the viral genome) was generated by PCR using WNV EG101 plasmid as a template and the forward primer, 5Ј-TAATACGACTCACTATAGAGTAGTTCGCCTGTG-TGAGCTGAC-3Ј (T7 promoter (underlined) was fused to 1-24 nt of the viral genome) and the reverse primer, 5Ј-CACTGCTCGGGTCGGAGC-AAT-3Ј (complementary to 271-291 nt of the viral genome containing the authentic AvaI site (underlined)). An amplified PCR product, WNV5ЈTR 281nt , was cloned into pGEM-T easy vector (pGEM-T easy-WNV5ЈTR). The 3Ј-TR (770 nt) that includes the 3Ј-UTR (631 nt) was amplified using WNV EG101 plasmid as a template and the forward primer, 5Ј-CACAAGAACCCGAGCCACG-3Ј (corresponding to 10251-10269 nt of the viral genome containing the authentic AvaI site (underlined)) and the reverse primer, 5Ј-GCTCTAGAAGATCCTGTGT-TCTCGCACCA-3Ј (complementary to 11009 -11029 nt fused with the XbaI site (underlined)). The plasmid DNA, pGEM-T easy-WNV3ЈTR, was obtained by cloning of 3ЈTR 770nt into the pGEM-T easy vector. The sequences of the 5Ј-and 3Ј-TR cDNAs in the pGEM-T easy plasmids were verified. The 5Ј-and 3Ј-TR encoding fragments were obtained by digestion of the pGEM-T easy plasmids with AvaI and EcoRI (in the vector) and AvaI and XbaI, respectively, and cloned into EcoRI-and XbaI-digested pSP64 vector (Promega) to yield the plasmid, pSP64-WNV1051nt encoding the WNV subgenomic RNA (Fig. 2).
Plasmids Containing Mutant Cyclization Motif (mutCYC)-pSP64-WNV1051nt containing mutCYC motif was obtained using the sitedirected mutagenesis kit (Stratagene). The primers used in the mutagenesis reaction were the following. For the 5Ј mutCYC, the 5Ј-CCGGCAAGAGCCGGGCTGCCTGCAGGCTAAAACGCGGAATGC-CC-3Ј and its complement as primers, and for the 3Ј mutCYC, the 5Ј-CACCACAACAAAACAGCCTGCAGGCACCTGGGATAGA-CTAGGAG-3Ј and its complement as primers were used for PCR. The underlined sequence indicates the replacement of the wild type CYC sequence (wtCYC), TCAATATG, with the mutCYC sequence, CCTG-CAGG. The plasmid containing both 5Ј-and 3Ј-mutCYC motifs was obtained by PCR-based site-directed mutagenesis using primers containing 3Ј mutCYC sequences and the plasmid containing the 5Ј mut-CYC motif as the template (see above).

Preparation of DNA Fragments for Use as Templates in the in
Vitro Transcription by the T7 RNA Polymerase PCR Products Encoding Subgenomic WNV (ϩ)-Strand RNA 1051nt (wt and mutCYC)-To generate the subgenomic WNV (ϩ)RNA 1051nt containing wild type as well as mutCYC motifs, PCR was performed using the appropriate plasmid as the template (Fig. 2) and two primers for PCR as follows: the forward primer, 5Ј-AGCTATGACCATGATTAC-GAATTC-3Ј (pSP64-EcoRI primer) that corresponds to the sequences upstream of the T7 promoter and the reverse primer, 5Ј-AGATCCTGT-GTTCTCGCACCAC-3Ј (WNV-3Ј end primer) that anneals with the 3Ј end region of the WNV 3Ј-UTR.
PCR Products Encoding Subgenomic Wild Type and CYC Mutant WNV (Ϫ)-Strand RNA 1051nt -The first PCR was carried out with the same plasmid construct as above as a template and two primers: the forward primer, 5Ј-TAATACGACTCACTATAGAGATCCTGTGTTCTC-GCACCAC-3Ј that contains T7 promoter (underlined) fused with the 5Ј-end of WNV (Ϫ)-strand, and the reverse primer, 5Ј-AGTAGTTCGC-CTGTGTGAGCTGAC-3Ј (WNV-5ЈUTR (ϩ)-strand primer). A second PCR was carried out with the forward primer containing the T7 promoter, 5Ј-TTACGAATTCGAGCTCGCCCTAATACGACTCACTATAG-3Ј and the same reverse primer consisting of the WNV-5ЈUTR (ϩ)-strand sequence used for the first PCR.
PCR Product Encoding the WNV (ϩ)-Strand 5ЈTR 230nt RNA-PCR was performed with the pSP64-WNV1051nt as the template and two primers: pSP64 (EcoRI) primer as the forward primer and the reverse primer, 5Ј-CGTATTGGTCCCTTGCCGTCGATC-3Ј that corresponds to the sequence complementary to 207-230 nt of the WNV genome.
PCR Product Encoding the WNV (ϩ)-Strand 3Ј-UTR 631nt RNA-To generate the WNV(ϩ) 3Ј-UTR 631nt RNA, the first PCR was carried out with pSP64-WNV1051nt as the template and two primers: the forward primer, 5Ј-TAATACGACTCACTATAGATACTTTATTAATTGTAAATA-GAC-3Ј that contains the T7 promoter (underlined) fused with the 5Ј terminal sequence of the 3Ј-UTR, and the sequence complementary to the WNV-3Ј end as the reverse primer. The second PCR was performed using the same forward primer containing the T7 promoter used in the second PCR described above and the reverse primer consisting of the sequence complementary to the 3Ј-end of WNV (Ϫ)-strand was the same as that used in the first PCR.

PCR Product Encoding the WNV (Ϫ)-Strand 3Ј-TR 230nt RNA
To generate WNV (Ϫ)-strand 3Ј-TR 230nt RNA, the first PCR was carried out with pSP64-WNV1051nt as the template and the two following primers: the forward primer, 5Ј-TAATACGACTCACTATAGCG-TATTGGTCCCTTGCCGTCGATC-3Ј that contains the T7 promoter (underlined) fused with the sequence complementary to 207-230 nt of the viral genome, and the reverse primer sequence from the beginning of the WNV (ϩ)-strand 5Ј-UTR. The second PCR was performed as described the above.
All PCR were performed with Vent polymerase (New England Biolabs) using the appropriate template and primers. The PCR products were purified using the QIAquick gel extraction kit (Qiagen), and was used in the in vitro transcription reaction.

Preparation of RNA Templates
In vitro transcription was performed with Ampliscribe TM T7 Transcription kit (Epicenter Technologies) as described by the manufacturer using the T7 promoter containing PCR products prepared as described above or using linearized plasmid DNAs (pTM1 vector plasmid digested with BamHI or pSP64 digested with AflIII). In the case of PCR product or linearized plasmid containing SP6 promoter, the Ampliscribe TM SP6 Transcription kit (Epicenter Technologies) was used as described by the manufacturer. 5Ј-Capped RNAs were prepared using AmpliCap TM T7 High Yield Message Maker kit (Epicenter Technologies) as described by the manufacturer. The RNA transcripts were purified as described previously (41) and then used in the RdRP assays. RNA templates with blocked 3Ј-OH groups were prepared using 3Ј-dATP (TriLink BioTechnologies) and poly(A) polymerase (Ambion) as described by Luo et al. (42).

In Vitro RdRP Assay
The in vitro RdRP assays were carried out as described previously (40) except when indicated. Briefly, the reaction mixture (50 M) contained 50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM MgCl 2 , template RNA (0.3 g), 500 M each of ATP, GTP, and UTP, 10 M unlabeled CTP, and 10 Ci of [␣-32 P]CTP along with 270 ng of purified WNV NS5 (2.6 pmol). The incubation was carried out at 25°C, which favored de novo synthesis of RNA over 3Ј-end elongation (40). When the specificity of the RdRP was analyzed using various nonspecific RNA templates, the incubation temperature of the RdRP reaction was 30°C at which the total RNA synthesis was higher than at 25°C. Following extraction with acid phenol/chloroform, the reaction mixture was passed through a P30 gel filtration column (Bio-Rad) to remove the unincorporated nucleotides. RNA was then analyzed by formaldehydeagarose gel electrophoresis and visualized by autoradiography. When indicated, the labeled bands were excised from dried gels and quantified by liquid scintillation counting and density analysis using the imagej program, a free software product available at the National Institutes of Health website. 2 The amount of template RNA, NS5, and incubation time for the RdRP assays were chosen from the linear range of RNA synthesis.

Analysis of RdRP Products by Gel Electrophoresis Systems
The hairpin and the de novo products produced in the RdRP assays were either analyzed directly by formaldehyde-agarose gel electrophoresis or (in the case of wild type and mutCYC (ϩ)-strand RNA templates in the RdRP assays) were first analyzed by native polyacrylamide gel electrophoresis (PAGE; 5%) followed by autoradiography. The bands containing the RNA products were cut out, eluted from the gel (elution buffer: 0.5 M ammonium acetate, 1 mM EDTA and 0.1% SDS), and precipitated with ethanol. The RNA products were further analyzed by formaldehyde-agarose gel, followed by autoradiography as described previously (41).

Analysis of RdRP Products by RNase A Digestion
To analyze the RdRP products, RNase A digestion was performed as described previously (41). Briefly, RdRP products were distributed equally and treated with or without RNase A (Sigma; 200 ng) in 20 l of 2ϫ SSC at 30°C for 30 min. The reactions were stopped by adding 30 l of TES stop buffer (10 mM Tris-HCl, pH 8.0, 50 mM EDTA, and 0.2% SDS), followed by acid phenol/chloroform extraction and ethanol precipitation in the presence of 10 g of yeast tRNA (Ambion). The RNase A-treated products were analyzed on formaldehyde-agarose gel followed by autoradiography.

RNase H Mapping
RNase H mapping was carried out essentially as reported previously (42,43). Following the RdRP assay using either WNV (ϩ)RNA 1051nt or WNV (Ϫ)RNA 1051nt template, RNase A treatment was carried out only for RdRP products synthesized from the WNV (ϩ)RNA 1051nt template to digest the single-stranded loop region of the 2ϫ hairpin product. After RNase A treatment, the reaction mixture was digested with proteinase K (Sigma; 40 g) followed by extraction with acid phenol/chloroform and precipitation with ethanol. The RdRP products were then mixed with 1.5 g of either WNV (ϩ)RNA 1051nt or WNV (Ϫ)RNA 1051nt that was complementary to the template used in the RdRP reaction, denatured at 95°C for 5 min, and reannealed in a hybridization buffer (80% formamide, 40 mM PIPES, pH 6.8, 1 mM EDTA, 400 mM NaCl) at 42°C overnight. RNA samples were then precipitated with ethanol and resuspended with 20 l of RNase H mapping buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 10 units of RNasin (Promega), 4 units of RNase H (Thermostable; Epicenter Technologies), 200 M oligodeoxynucleotide of either (ϩ)-or (Ϫ)-strand polarity). The reaction mixture was incubated at 45°C for 1 h. The reaction was terminated by addition of 2ϫ stop buffer (100 mM Tris-HCl, pH 8.0, 300 mM NaCl, 0.5% SDS). RNA was extracted with acid phenol/chloroform, then precipitated with ethanol, and analyzed by formaldehydeagarose gel.

RESULTS
Purification of Full-length WNV NS5-We found that E. coli TOP10FЈ cells transformed by WNV NS5 yielded higher amounts of NS5 protein in the soluble form compared with E. coli XL1-BLUE, which was used for DEN2 NS5. Moreover, addition of glucose (0.5%) to the LB growth medium during the initial phase of E. coli growth was necessary as observed for E. coli XL1-BLUE transformed by DEN2 NS5 plasmid presumably because of toxicity of the leaky expression of the NS5 protein from the lac promoter (40) (data not shown). After the growth of E. coli cells reached A 600 nm of 0.6 -0.8, glucose was removed and expression of WNV NS5 was induced by the addition of isopropyl-␤-D-thiogalactopyranoside and the bacterial culture was incubated as described under ''Experimental Procedures.'' The NS5 protein, subsequent to elution from the metal affinity (Talon TM resin) column and dialysis, was analyzed by SDS-PAGE (8%) and Coomassie Blue staining (Fig. 1). The NS5 protein purified in this manner had some minor contaminants and its purity was estimated to be 80% by density analysis. The purified WNV NS5 protein was free of any RNase contamination as judged from the results of incubation with the subgenomic WNV (ϩ)RNA 1051nt template at 37°C for 2 h followed by analysis by electrophoresis on a 4% acrylamide, 8 M urea PAGE gel and staining with ethidium bromide. No contamination of any RNase was detected (data not shown). Therefore, this WNV NS5 protein was used for subsequent experiments (described below).
RdRP Assay Using WNV Subgenomic RNA Templates-First, we determined whether the purified WNV NS5 was enzymatically active in RNA synthesis on the WNV (ϩ)RNA 1051nt as the template used in the RdRP assay. This template RNA contains the authentic regulatory elements within 5Ј-and 3Ј-TRs including the two self-complementary 9-nucleotide long CYC motifs located within the 5Ј-and 3Јconserved sequence elements reported previously (11) (see Fig.  2B). RdRP assay was carried out in a reaction mixture containing the four NTPs in which the ␣-32 P-labeled CTP was included along with the template RNA and the purified NS5 polymerase. The RNA products synthesized in the reaction were analyzed by a formaldehyde-agarose gel followed by autoradiography. In this in vitro assay, only the newly synthesized RNA would be detected as a labeled product. As shown in Fig. 3, RNA synthesis was observed only when the WNV (ϩ)RNA 1051nt template and the purified NS5 were added in the reaction (lane 1). Two major products were visualized on a formaldehyde-agarose gel followed by autoradiography. No RNA synthesis occurred when the template RNA was either omitted or replaced with a DNA template (obtained by PCR), as well as when the NS5 was omitted or heat-inactivated (Fig. 3, lanes 2-5). Moreover, to test whether the polymerase activity of WNV NS5 shows preference for the viral template RNA, unrelated RNA templates were used in the RdRP assay. Unrelated RNA templates were synthesized by T7 or SP6 polymerase-catalyzed in vitro transcription from the T7 promoter-containing pTM1 plasmid vector DNA (44), linearized by BamHI or the SP6 promoter-contain-ing pSP64 plasmid DNA linearized by AflIII, giving rise to either a 645-or 407-nt RNA, respectively. WNV NS5 was barely active on these unrelated RNA templates (Fig. 3, lanes 6  and 7). These results indicate that the purified WNV NS5 is a true RdRP, showing preference for the WNV subgenomic RNA template rather than for unrelated RNA templates. The nature of the two RdRP products (lane 1) was characterized by RNase A treatment and RNase H mapping (see below).
Analysis of RdRP Products from (ϩ)-Strand RNA Templates--The two products formed in the WNV NS5 polymerase assay (Fig. 3, lane 1) are similar to those synthesized by DEN2 NS5 polymerase described previously (40). In that study, it was shown that the template-sized product was synthesized by de novo initiation at the 3Ј-end of template RNA and was resistant to RNase A treatment in a high ionic strength buffer because of its double-stranded nature. On the other hand, the product twice the size (2ϫ, hairpin) of the template was shown to be because of 3Ј-end elongation of the template RNA and was sensitive to RNase A digestion because of the single-stranded loop region and gave rise to a template-sized product (40). In this study, we sought to determine the polarity of the RNA products synthesized by WNV NS5 by RNase H mapping. The experimental scheme for RNase H mapping is shown in Fig.  4A. Following the RdRP assays, the mixtures were treated with RNase A to remove excess unlabeled (ϩ)-strand template RNA present in the RdRP reaction (step 2). The newly synthesized template-sized RNA product annealed to the unlabeled template (ϩ) RNA would be resistant to RNase A treatment. However, if the template-sized product was synthesized by a putative terminal transferase activity of WNV NS5 similar to that of the hepatitis C virus NS5B polymerase reported previously (45), then the product would be digested by RNase A. To analyze the RdRP products by RNase A digestion, the RdRP reac- A, schematic representation of the WNV RNA genome is shown. CYC denotes cyclization motif; in C and F, CYC represents complementary sequences as the RNAs are of (Ϫ)-strand polarity. B, and C, plasmid constructs encoding subgenomic RNA WT of (ϩ) -and (Ϫ)-strand polarity, respectively, containing the authentic 5Јand 3Ј-terminal regions of the viral genome are shown. For generating RNA templates from the plasmid constructs, PCRs were carried out using appropriate primers in which the forward primer consisted of T7 promoter. PCR products were used for in vitro transcription catalyzed by T7 RNA polymerase as described under ''Experimental Procedures.'' Inverted T7 promoter (in C and F) denotes DNA template oriented for the synthesis of RNA of (Ϫ)-strand polarity. D and E represent plasmid constructs encoding the 5Ј-TR 230nt and 3Ј-UTR 631nt RNAs of (ϩ)-strand polarity. The plasmid constructs encoding subgenomic RNAs with mutations within the CYC motifs of (ϩ)-or (Ϫ)-strand polarity were derived from the corresponding plasmid construct for B or C as described under ''Experimental Procedures.'' Construct F encodes 3Ј-TR 230nt RNA of (Ϫ)-strand polarity.  6 and 7). The reactions were incubated at 30°C at which the total RNA synthesis was optimal. In lane 6, the pTM1 vector, linearized by BamHI, was used for in vitro transcription catalyzed by T7 RNA polymerase to produce the 645-nt RNA. In lane 7, the pSP64 plasmid vector linearized by AflIII was used to produce a 407-nucleotide transcript by SP6 RNA polymerase. Incubation of the RdRP assay was carried out at 30°C. The products of RdRP assays were analyzed by formaldehyde-agarose gel followed by autoradiography. The mobility of RNA size markers are indicated on the left-hand side of the gel.  1-10). The products were analyzed by formaldehyde-agarose gel electrophoresis and visualized by autoradiography. D, the reaction conditions were same as in B. Lanes 1-4, the RdRP reactions were carried out using WNV (Ϫ)RNA 1051nt as the template (step 1 in A). RNase A (step 2) treatment was omitted. Next, annealing with the unlabeled (ϩ)-strand RNA 1051nt was carried out to remove the template (Ϫ)-strand RNA 1051nt (lanes 2-4). DNA oligomer of either (ϩ)-or (Ϫ)-strand polarity was hybridized (lanes 3 and 4, respectively) prior to RNase H digestion. The products were analyzed by formaldehyde-agarose gel electrophoresis followed by autoradiography. The mobility of RNA size markers are indicated on the left-hand side of the gel. tion mixtures were extracted with acid phenol/chloroform and precipitated with ethanol. The RNA products were denatured and annealed with excess unlabeled (Ϫ)-strand template RNA (step 3) to displace the (ϩ)-template strand from the newly synthesized labeled RNA. The mixture was aliquoted equally and annealed to single-stranded oligodeoxynucleotide of either (ϩ) (nucleotide position number 704 -726) or (Ϫ) (nucleotide position number 726 -704) strand polarity. The RNA-DNA hybrid was digested with RNase H (step 4) and the products were analyzed by formaldehyde-agarose gel electrophoresis, followed by autoradiography. If the newly synthesized RdRP product is of template length, then step 4 would give rise to RNA fragments of 700 and 325 nt in length (Fig. 4A).
When the WNV (ϩ)RNA 1051nt template was used in the RdRP reaction, both hairpin and template-sized products were observed (Fig. 4B, lane 1). RNase A treatment converted the entire hairpin (2ϫ) product into an RNase A-resistant template-sized (1ϫ) product. The RNA band migrating as 1ϫ template-sized product appears broad compared with the one in lane 1 (no RNase A treatment) because it consists of two species: the RNase A-resistant product produced from the 2ϫ hairpin RNA and the de novo synthesis product. However, these products were not separable by either shorter exposure of the gel (data not shown) or by shorter incubation periods of the RdRP reactions (see Fig. 4C). In addition, the RNA products shorter than the template length (ϳ530 nt) (Fig. 4B, lane 2) were generated only upon RNase A treatment suggesting that these were produced from the RNase A-sensitive shorter hairpin molecules generated by RdRP. RNase H digestion without adding the strand-specific oligodeoxynucleotide primers did not cleave these products (Fig. 4B, lanes 1 and 2 versus lane 3). However, after RNase A treatment, denaturation and reannealing with unlabeled WNV (Ϫ)RNA 1051nt was carried out followed by hybridization with the oligodeoxynucleotides of the (ϩ)-or the (Ϫ)-strand polarity (steps 3 and 4 in Fig. 4A). The expected RNase H products of 700 and 325 nt were generated only with the oligodeoxynucleotide of (ϩ) polarity but not with that of the (Ϫ) polarity (Fig. 4B, lanes 4 and 5, respectively). The shorter RNAs produced upon RNase A digestion were not sensitive to RNase H digestion in the absence of any oligodeoxynucleotide primer or in the presence of the primer of (Ϫ) polarity (lanes 3 and 5, respectively), but was sensitive to annealing with the primer of (ϩ)-strand polarity indicating that it contained shorter newly synthesized RNA of (Ϫ) polarity. The percent of these shorter RNA products is much less in the total RNA synthesized (sum of 2ϫ hairpin and de novo product; see lane 2). The pathway for generation of these shorter hairpin RNAs was not pursued further in this study. These results, taken together, indicated that the newly synthesized RNA present in the template-sized and hairpin products were of (Ϫ) polarity.
RNA Synthesis by WNV NS5 on WNV (Ϫ)RNA 1051nt Template-Next, we examined whether the WNV NS5 could utilize exogenous subgenomic WNV (Ϫ)RNA 1051nt template (obtained by in vitro transcription of the PCR product from the construct in Fig. 2C) and synthesize RNA of (ϩ) polarity in vitro. The results indicated that the polymerase could efficiently utilize the WNV (Ϫ) RNA template and synthesized predominantly a template-sized RNA product and very little of the hairpin product. The template-sized product was resistant to RNase A digestion indicating that the newly synthesized RNA annealed to the template RNA was double-stranded in nature (data not shown; see also Fig. 5). The polarity of the RNA product was determined by RNase H mapping following a similar scheme illustrated in Fig. 4A except that RNase A digestion (step 2) was omitted because of predominance of the template-sized product over the hairpin product. The unlabeled WNV (ϩ)RNA 1051nt was used after denaturation (step 3 in Fig. 4A) to hybridize with the excess WNV (Ϫ)RNA 1051nt template used in the RdRP reaction. After this annealing step, the template RNA would no longer interfere with hybridization of the oligodeoxynucleotide primer. The results shown in Fig. 4D indicated that only when the oligodeoxynucleotide of (Ϫ) polarity was annealed, the RNase H generated the expected size of 700and 325-nt RNA fragments (Fig. 4D, lane 4) but not when the oligodeoxynucleotide of (ϩ) polarity was annealed (lane 3). RNase H digestion of the RNA-oligodeoxynucleotide hybrid was not complete because the presence of excess WNV (Ϫ)RNA 1051nt template used in the RdRP reaction presumably reannealed to the labeled product RNA as the RNase A digestion was omitted in this experiment. When the RdRP products were digested with RNase A prior to annealing with the oligodeoxynucleotide of (Ϫ) polarity, RNase H was able to digest the RNA-DNA oligomer hybrid completely similar to the experiment shown in Fig. 4B (lane 4). Lanes 1 and 2 represent the negative controls in which annealing of the unlabeled WNV (ϩ)RNA 1051nt and the addition of DNA oligomers were omitted, respectively, prior to RNase H digestion. These results indicated that the WNV (Ϫ)RNA 1051nt serves as an efficient template for RNA synthesis by WNV NS5 and that the product RNA was predominantly of template-sized and of (ϩ) polarity.
Effect of the 5Ј-Cap Structure on RNA Synthesis by the WNV NS5-According to the current model for flaviviral RNA replication, the (Ϫ)-strand synthesized from the (ϩ)-strand exists predominantly in the double-stranded replicative form and RNA templates were prepared as either 5Ј-capped or uncapped (u.c.). The reaction mixtures were incubated at 25°C for 1.5 h. The products of RdRP reactions were either treated with RNase A (as indicated by ϩ) or untreated (as indicated by Ϫ). Hairpins, twice the size of the template RNA, and the same size as the template RNA were produced (see text for details). Lanes 1 and 2, 5Ј-uncapped WNV (ϩ)-strand RNA template without or with RNase A treatment, respectively; lanes 3 and 4, 5Јcapped WNV(ϩ)-strand RNA template (without or with RNase A treatment, respectively); lanes 5 and 6, 5Ј-uncapped WNV (Ϫ)-strand RNA template (without or with RNase A treatment, respectively); lanes 7 and 8, 5Ј-capped WNV (Ϫ)-strand RNA template (without or with RNase A treatment, respectively). The products were analyzed by formaldehyde-agarose gel followed by autoradiography. The mobility of RNA size markers are indicated on the left-hand side of the gel. serves as the recycling template for (ϩ)-strand genomic RNA synthesis (for a review, see Ref. 46). We sought to determine whether the 5Ј-cap structure influences RNA synthesis by the WNV NS5 and the nature of the products formed. To address these questions, subgenomic RNA 1051nt templates of (ϩ) and (Ϫ) polarity were synthesized with or without cap structure (m 7 G(5Ј)ppp(5Ј)G) by T7 RNA polymerase-catalyzed in vitro transcription and used in the RdRP reaction. At the same time, to determine the nature of RdRP products synthesized by WNV NS5 (hairpin or template size product), equal aliquots of the RdRP products were incubated with or without RNase A in a high ionic strength buffer as described under ''Experimental Procedures.'' When uncapped WNV (ϩ)RNA 1051nt was used in the RdRP assay, WNV NS5 produced the two products, hairpin (2ϫ) and template-sized (1ϫ) RNAs, as observed before and RNase A treatment converted the hairpin RNA into a templatesized product (Fig. 5, lanes 1 and 2). However, most of the RdRP products synthesized from the capped WNV(ϩ)RNA 1051nt template migrated at the position of the template-sized product (Fig. 5, lane 3). In fact, the addition of 5Ј-cap structure reduced the synthesis of the hairpin product twice the size of the template RNA, although shorter hairpin RNA synthesis was increased as evident from the density of the band migrating faster than the template-sized product after RNase A treatment (Fig. 5, lane 4 versus lane 2). One possible explanation is that there is a strong pause site within the 3Ј-UTR for WNV NS5 polymerase that produces shorter hairpin RNAs through an unknown mechanism that migrates as (1ϫ) template-sized RNAs on a completely denaturing formaldehyde-agarose gel. When the uncapped-and capped-WNV(ϩ)RNA 1051nt transcripts synthesized by T7 RNA polymerase were analyzed by native 5% polyacrylamide gel electrophoresis and stained with ethidium bromide, there was no detectable amount of shorter RNA molecules (data not shown). Switching the promoter from T7 to SP6 for producing the in vitro transcripts for use as templates in the RdRP assays also did not alleviate the production of short hairpin RNAs (data not shown). Because we did not observe these shorter hairpin RNAs in the DEN2 RdRP assays or when WNV (Ϫ)RNA 1051nt templates were used, we conclude that this result is unique to the subgenomic WNV (ϩ)RNA 1051nt templates and WNV RdRP. When the uncapped or capped WNV (Ϫ)RNA 1051nt was used in the in vitro RdRP reaction, a template-sized product was predominantly synthesized by WNV NS5 consistent with the result shown in Fig. 4D for the uncapped RNA template. This RNA was RNase A-resistant indicating that it is a double-stranded RNA (Fig. 5, lane 6).
RNA Synthesis by WNV NS5 on 3Ј-OH-blocked Templates-The results suggest that the template-sized product synthesized from the WNV (ϩ)-or (Ϫ)-strand RNAs arises by de novo initiation by the WNV polymerase at the 3Ј end of template RNA. To confirm this notion, the RNA templates, synthesized by T7 RNA polymerase-catalyzed in vitro transcription, were blocked at the 3Ј-OH moiety by incorporation of 3Ј-dAMP catalyzed by E. coli poly(A) polymerase as described for hepatitis C virus NS5B polymerase (42). Both capped and uncapped WNV (ϩ)-or (Ϫ)-strand RNA 1051nt templates blocked at the 3Ј-OH were used in the in vitro RdRP assays. When the uncapped WNV (ϩ)RNA 1051nt with the free 3Ј-OH was used in the reaction, both hairpin and template-sized products were produced (Fig. 6, lane 4) as observed before. However, the 3Ј-OHblocked RNA templates of either (ϩ) or (Ϫ) polarity gave rise to only template-sized products (Fig. 6, lanes 1-3). Because no hairpin RNAs were produced in these reactions, the blocking of the 3Ј-OH ends of the template RNAs was presumably efficient. Synthesis of template-sized product by the putative terminal transferase activity of WNV NS5 would also be blocked in this reaction. These results suggest that template-sized RNA produced by WNV NS5 is a product of de novo initiation at the 3Ј-end of the template RNA.
The Effect of Mutations within the CYC Motifs on Template Efficiency of WNV Subgenomic RNAs-Next, we sought to determine whether the highly conserved 5Ј-and 3Ј-CYC motifs present within the terminal regions of WNV genomic RNA influence RNA synthesis from either WNV (ϩ)-or (Ϫ)RNA 1051nt templates with or without 5Ј-cap structure. We had previously reported that the mutation of the 5Ј-CYC motif reduced (Ϫ)-strand RNA synthesis to less than 5% of the level achieved from the wild type DEN2 (ϩ)-strand subgenomic RNA template, whereas mutation of the 3Ј-CYC was tolerated as the activity was reduced to only about 50% using either DEN2-infected mosquito cell lysates (25) or purified DEN2 NS5 (40) as the source of the viral polymerase. In those studies, the wild type and mutant templates used were 5Ј-uncapped.
In this study, we sought to determine whether CYC motifs are required on the WNV template RNAs for either (ϩ)-or (Ϫ)-strand RNA synthesis or for both by the WNV polymerase in vitro as this question has not been addressed thus far for any flavivirus using either the replicon system or the in vitro RdRP assay system. To test the requirement of CYC motifs for RNA synthesis, the wild type CYC sequence, UCAAUAUG, was replaced with a mutant CYC sequence, CCUGCAGG. Both (ϩ)and (Ϫ)-strand RNA templates were produced either with or without 5Ј cap structure by in vitro transcription catalyzed by T7 RNA polymerase. First, the effects of mutation of the CYC in the WNV (ϩ)RNA 1051nt template for (Ϫ)-strand RNA synthesis were examined. As described earlier, the RNA products synthesized from WNV (ϩ)RNA 1051nt template RNA contained short hairpin product(s) that migrated at the same position of the template-sized de novo product (Fig. 5, lanes 2 and 4) in the completely denaturing formaldehyde-agarose gel. However, the FIG. 6. RdRP assays using 3-OH blocked RNA templates. RNA templates were blocked at the 3Ј-OH by incorporation of 3Ј dATP and used in the RdRP assays as described under ''Experimental Procedures.'' RdRP reactions were carried out and the products were analyzed by formaldehyde-agarose gel followed by autoradiography. Lane 1, 3Ј-OH blocked 5Ј-uncapped (u.c.) WNV (ϩ)-strand RNA template; lane 2, 3Ј-OH blocked 5Ј-capped WNV (ϩ)-strand RNA template; lane 3, 3Ј-OH blocked 5Ј-uncapped (u.c.) WNV (Ϫ)-strand RNA template; lane 4, 5Ј-uncapped (u.c.) WNV (ϩ)-strand RNA template as the control. The mobility of RNA size markers are indicated on the left-hand side of the gel. highly structured hairpin RNAs (2ϫ) as well as shorter hairpin RNAs were separable from the template-sized double-stranded RNA synthesized by de novo initiation by electrophoresis on a non-denaturing polyacrylamide gel (5%) and visualized by autoradiography. The bands containing 2ϫ hairpin and 1ϫ de novo products were eluted from the gel and analyzed by formaldehyde-agarose gel electrophoresis followed by autoradiography. The labeled products were quantified by scintillation counting and density analysis. As shown in Fig. 7, the hairpin product was synthesized abundantly compared with the de novo product from the uncapped wild type WNV (ϩ)RNA 1051nt template (Fig. 7A, lane 1). Mutation of the 5Ј-CYC motif in the uncapped WNV (ϩ)RNA template dramatically diminished the synthesis of the hairpin product and nearly abolished the de novo product (Fig. 7A, lane 2). Total RNA synthesis (hairpin plus de novo products) of uncapped WNV(ϩ)5ЈmutCYC RNA 1051nt was about 3% of the total RNA synthesized from the uncapped wild type WNV (ϩ)RNA 1051nt . However, the mutation of the 3Ј-CYC motif in the uncapped WNV(ϩ)3ЈmutCYC RNA reduced the template efficiency to only about 30% for RNA synthesis (Fig. 7A, lane  3). However, if the 5Ј-CYC mutation was complementary to the 3Ј-CYC mutation as in the double mutant, then the template activity of WNV(ϩ)5Ј3ЈmutCYC RNA 1051nt was restored to the same level as that of the uncapped template containing the mutant 3Ј-CYC motif (Fig. 7A, lane 4). These results suggest that the 5Ј-CYC plays an important role in viral (Ϫ)-strand synthesis because mutation in this motif severely affected the template activity. However, the activity can be partially restored by a complementary mutation of the 3-CYC motif.
Next, we examined the effects of 5Ј-cap structure on template efficiencies of RNAs for de novo versus hairpin RNA synthesis as well as on the effects of mutations in the 5Ј-and/or 3Ј-CYC motifs. When the 5Ј-capped WNV (ϩ)RNA 1051nt template was used in the RdRP assay, synthesis of the hairpin product was reduced to about 50% (Fig. 7A, lane 5 versus lane 1) consistent with the result shown in Fig. 5. The differences in the amounts of the template-sized product (1ϫ) relative to the hairpin (2ϫ) product in Figs. 5 (lane 3) and 7A (lane 5) are because of the removal of short hairpin RNAs from the pool of template-sized RNAs by prior electrophoresis on a native PAGE (5%) in the latter experiment to be able to compare only the synthesis of de novo product versus the hairpin (2ϫ) product. The total RNA synthesized from the 5Ј-capped WNV(ϩ)5Ј-mutCYC RNA 1051nt template was severely reduced (to about 6%) and the de novo product was affected much more than the hairpin product similar to that observed for the uncapped template containing the 5Ј-CYC mutation. However, it was restored albeit partially in the double mutant in which complementary mutation was introduced in the 3Ј-CYC motif that was about 50% of the level obtained with the uncapped template (lane 4).
Next, we examined whether RNA synthesis from the WNV (Ϫ)RNA 1051nt template was influenced by mutations of CYC motifs. RdRP assays were performed using subgenomic RNAs containing 5Ј-and 3Ј-wtCYC motifs, 5Ј-mutCYC/3Ј-wtCYC, 5Ј-wtCYC/3Ј-mutCYC, or 5Ј-mutCYC/3Ј-mutCYC motifs. The RdRP products were analyzed by formaldehyde-agarose gel followed by autoradiography. Interestingly, mutations of either 3Ј-or 5Ј-CYC motifs of (Ϫ)-strand polarity (complement of FIG. 7. Role of CYC motifs in (؊)-and (؉)-strand RNA synthesis in vitro by WNV NS5. A, RdRP assays were carried out using WNV (ϩ)RNA 1051nt template containing wild type (WT), 5Ј-mutCYC, 3Ј-mut-CYC, or 5Ј,3ЈmutCYC. The templates were either 5Ј-uncapped (lanes 1-4) or 5Ј-capped (lanes [5][6][7][8]. After RdRP reactions, the products were separated on 5% PAGE followed by autoradiography; then the bands containing the hairpin products and de novo products were cut out and the products were eluted from the gel. The 2ϫ hairpin products and the template-sized de novo products were analyzed by formaldehyde-agarose gel followed by autoradiography. The total RNAs and the hairpin and de novo products were quantified by scintillation counting and densitometry. The mobility of RNA size markers are indicated on the left-hand side of the gel. B, RdRP assays were carried out using WNV (Ϫ)RNA 1051nt templates containing WT and mutant CYC motifs as described in A. RdRP products were analyzed by formaldehyde-agarose gel followed by autoradiography and quantified as described in A. either 5Ј-mutCYC or 3Ј-mutCYC motifs in the (ϩ)-strand template, respectively) did not affect (ϩ)-strand RNA synthesis compared with the wild type WNV (Ϫ)-strand RNA template (Fig. 7B, lanes 2 and 3, compared with lane 1). Moreover, the addition of the 5Ј-cap structure to any of the WNV (Ϫ)RNA templates also did not seem to affect (ϩ)-strand RNA synthesis (data not shown). The estimation of the total RNA synthesis also revealed no significant reduction of RdRP activity between wild type and CYC mutant RNA templates in which either motif or both motifs were mutated. These results indicate that the mutation of CYC motifs in WNV (Ϫ)-strand RNA template had no effect for (ϩ)-strand RNA synthesis; that is, neither the complementarity nor the sequence of CYC was found to be necessary for (ϩ)-strand RNA synthesis in vitro. Taken together, these results support the conclusion that the two complementary CYC motifs play an important role only for (Ϫ)strand RNA synthesis from WNV (ϩ) template RNA but not for (ϩ)-strand RNA synthesis by the WNV NS5 polymerase in vitro.
The Template Efficiency of 3Ј-UTR of Either (ϩ)-or (Ϫ)-Strand RNA for in Vitro RdRP Activity-It was reported previously that the in vitro RdRP assay using DEN2 virus-infected cell extracts or the purified DEN2 NS5 polymerase, the 5Јterminal region of DEN2 (ϩ)RNA 230nt (TR 230nt ) that includes the 5Ј-UTR 96nt and the 5Ј-CYC motif was quite active for RNA synthesis, whereas the 3Ј-UTR (ϩ)RNA 454nt alone was not active. The interaction of the 5Ј-TR 230nt with 3Ј-UTR 454nt containing either wild type or complementary mutant CYC motifs was necessary for activation of the DEN2 3Ј-UTR(ϩ) RNA template (40,41). On the other hand, hepatitis C virus NS5B polymerase initiates RNA synthesis from the 3Ј-TR of the (Ϫ)-strand viral RNA more efficiently than the 3Ј-TR of the (ϩ)-strand polarity (47). Therefore, we sought to examine the template efficiency of 3Ј-TR 230nt WNV RNA of (Ϫ)-strand polarity in comparison with the 3Ј-UTR 631nt RNA of (ϩ)-strand polarity.

DISCUSSION
The RNA-dependent RNA polymerase activity has been demonstrated for purified recombinant NS5 from DEN-1, DEN-2, Kunjin, and hepatitis C virus NS5B (40, 42, 48 -51). In the study of WNV NS5, no de novo synthesis was demonstrated and the polymerase was strictly primer-dependent on 3Ј-OHblocked template RNA (52), whereas on unblocked RNA, it produced a hairpin product by a copy-back mechanism. In this study, we describe the first characterization of a viral RNA template-specific and template-dependent in vitro RdRP assay system for WNV (EG101 strain) using the E. coli-expressed and purified WNV NS5. The results shown here indicate that the WNV NS5 produced both hairpin product (2ϫ) and de novo product (1ϫ de novo) RNAs from the (ϩ)-and (Ϫ)-strand RNA templates because of 3Ј-end elongation and de novo initiation, respectively. Moreover, shorter hairpin RNAs were also produced from the (ϩ)-strand RNA template that migrate as the template-sized RNAs on the completely denaturing gel electrophoresis system (1ϫ hairpin). Addition of 5Ј-cap structure to the (ϩ)-template inhibited the synthesis of 2ϫ hairpin by about 50%, such that the ratio of the de novo product to the hairpin product was increased (0.34 to 1.1, Fig. 7A). However, 5Ј-capping did not influence this ratio from (Ϫ)-strand RNA template (data not shown).
It was shown previously that DEN2 NS5 polymerase switches the mode of RNA synthesis from de novo product at low temperatures to that of hairpin product at higher temperatures. The ratio of the de novo to the hairpin product was reduced from 2 to 0.25 by a temperature shift from 20 to 40°C. The de novo initiation events required incubation of the components, the (ϩ)-strand template RNA, NS5, ATP, and GTP at lower temperatures. Once the heparin-resistant, de novo initiation complex is formed, then 3Ј-elongation was insensitive to temperature variations and the de novo product was preferentially formed (40,53). In the context of viral infection, only the de novo synthesized RNA by the viral replicase is physiologically important to faithfully replicate viral RNA through multiple rounds. Using the in vitro system described here for WNV and for DEN2 described in previous studies (25,40,41,53), it may be possible to delineate other parameters that modulate the de novo synthesis of viral RNA in vitro.
We had shown previously that the (Ϫ)-strand RNA synthesis requires interaction between the two terminal regions of the (ϩ)-strand RNA template (25,40,41). This functional interaction is affected by deletion of the 5Ј-UTR, mutations of the 5Ј- RNAs, as noted below, that correspond to 3Ј-TR and/or 5Ј-TR of either WNV (ϩ)-or WNV (Ϫ)-strand polarity were synthesized. 0.2 g of 230-nt template RNA and 0.5 g of 631-nt template RNAs were used in the in vitro RdRP assay. RdRP products were analyzed by the formaldehyde-agarose gel followed by autoradiography. Lane 1, WNV (Ϫ)-3ЈTR 230nt ; lane 2, WNV (ϩ)3ЈUTR 631nt ; lane 3, WNV (ϩ)3ЈUTR 631nt ϩ WNV (ϩ)5ЈTR 230nt ; lane 4, WNV (ϩ)5ЈTR 230nt . The mobility of RNA size markers are indicated on the left-hand side of the gel. and 3Ј-CYC motif, and mutations within the 3Ј-terminal stemloop structures. The 3Ј stem-loop structure has also been shown to be important for DEN2 viral replication in cultured cells, as shown by mutagenesis of infectious DEN2 RNA (24). A single base pair that would abolish the predicted pseudoknot structure was shown to affect the viral (Ϫ)-strand RNA synthesis significantly in vitro (25) and in vivo. 3 The substitution mutations of the 5Ј-CYC motif had a more drastic effect than the mutation of the 3Ј-CYC motif, whereas complementary mutations in both motifs restored the (Ϫ)-strand RNA synthesis by either infected cytoplasmic extracts or the purified polymerase (25,40). In those studies, uncapped DEN2 RNA templates were used for the in vitro RdRP assays.
From this study, the requirement for WNV (Ϫ)-strand RNA synthesis also follows the trend previously established for other flaviviruses (41, 54 -56) and for poliovirus with regard to the requirement for interaction between the 5Ј and 3Ј ends (57)(58)(59). We examined the role of 5Ј-cap as well as CYC motifs for synthesis of WNV RNA of not only (Ϫ)-strand but also of (ϩ)-strand polarity. We show here that the CYC motifs play a role only in WNV (Ϫ)-strand RNA synthesis. Because the 3Ј-UTR of (ϩ)-strand RNA alone was not active and only the subgenomic RNA containing the 5Ј-TR 230nt RNA of (ϩ)-strand polarity was active, the results suggest that a functional interaction between the terminal regions of the viral (ϩ)-strand RNA is necessary for (Ϫ)-strand RNA synthesis and that the role of the CYC motifs may be to facilitate this interaction. Once the (Ϫ)-strand is synthesized, according to the current model for flaviviral replication, the viral (Ϫ)-strand exists in the double-stranded replicative form and serves as a recycling template for synthesis of viral (ϩ)-strand RNAs. Our results suggest that mutations in the CYC motifs of (Ϫ)-strand template have no apparent effect on the synthesis of viral (ϩ)strands in vitro.
The mutations in the 5Ј-or 3Ј-CYC motif have also been shown to affect replication of yellow fever viral RNA, shown by using an infectious clone (56) or yellow fever viral replicon (55) as well as Kunjin and WNV RNA replication using the replicon systems (54,60). By using a psoralen-UV cross-linking method, we showed that the CYC motifs play a role in bringing the two ends together for physical interaction (25). These observations, taken together, suggest that interaction between the two terminal regions through secondary and tertiary structural elements including CYC motifs, perhaps stabilized by host and viral proteins in the replicase complex, provide an active (ϩ)strand RNA template for (Ϫ)-strand RNA synthesis. Our results show that mutations of either 5Ј-or 3Ј-CYC motif in both uncapped and 5Ј-capped (ϩ)-strand RNA templates significantly affect (Ϫ)-strand RNA synthesis. The de novo synthesis was particularly affected 5-10-fold more with the 5Ј-mutCYC motif than with the 3Ј-mutCYC motif. Furthermore, mutations of both 5Ј-and 3Ј-CYC motifs, even though complementary to each other, did not restore RNA synthesis completely. However, mutations of CYC motifs in the (Ϫ)-strand RNA template (complementary sequences of CYC motifs in the (ϩ)-strand RNA templates) did not affect total RNA synthesis or de novo synthesis of (ϩ)-strand RNA. In this regard, it is noteworthy that the 3Ј-TR RNA 230nt of WNV (Ϫ)-strand polarity itself is an active template for the WNV NS5 in the RdRP assay, whereas the 3Ј-UTR of the (ϩ)-strand RNA genome was a poor template unless the 5Ј-TR 230nt with the wild type 5Ј-CYC is added in the RdRP assay. These results reveal for the first time that CYC motifs play an important role only for (Ϫ)-strand RNA synthesis from viral genomic (ϩ)-strand RNA template. The in vitro system described here may be useful to decipher the parameters that modulate de novo synthesis of both (Ϫ)-and (ϩ)strand RNA synthesis in vitro.