The predicted stem-loop structure in the 3′-end of the human norovirus antigenomic sequence is required for its genomic RNA synthesis by its RdRp

The norovirus genome consists of a single positive-stranded RNA. The mechanism by which this single-stranded RNA genome is replicated is not well understood. To reveal the mechanism underlying the initiation of the norovirus genomic RNA synthesis by its RNA-dependent RNA polymerase (RdRp), we used an in vitro assay to detect the complementary RNA synthesis activity. Results showed that the purified recombinant RdRp was able to synthesize the complementary positive-sense RNA from a 100-nt template corresponding to the 3′-end of the viral antisense genome sequence, but that the RdRp could not synthesize the antisense genomic RNA from the template corresponding to the 5′-end of the positive-sense genome sequence. We also predicted that the 31 nt region at the 3′-end of the RNA antisense template forms a stem-loop structure. Deletion of this sequence resulted in the loss of complementary RNA synthesis by the RdRp, and connection of the 31 nt to the 3′-end of the inactive positive-sense RNA template resulted in the gain of complementary RNA synthesis by the RdRp. Similarly, an electrophoretic mobility shift assay further revealed that the RdRp bound to the antisense RNA specifically, but was dependent on the 31 nt at the 3′-end. Therefore, based on this observation and further deletion and mutation analyses, we concluded that the predicted stem-loop structure in the 31 nt end and the region close to the antisense viral genomic stem sequences are both important for initiating the positive-sense human norovirus genomic RNA synthesis by its RdRp.

The viral RdRp is responsible for the replication/synthesis of the norovirus genomic RNA. RdRp first synthesizes antisense RNA using the genomic RNA as a template. It then uses these antisense RNAs to synthesize the progeny viral genomic RNAs. During replication, a subgenomic RNA encoding the ORF2 and 3 (3 0 coterminal with the virus genome) is also synthesized in all members of the Caliciviridae (2).
Investigation into the RNA synthesis activity of the norovirus RdRp in vitro or cell-based systems has been reported (3)(4)(5)(6)(7)(8)(9). These papers proposed models of initiation of RNA synthesis by RdRp: when the template RNA has the polyA sequences at the 3 0 -end of the positive-sense genomic RNA, RdRp synthesizes the antisense RNA using the VPg protein primer (4,6). When the RNA does not have the polyA sequences at the 3 0 -end of the antisense genomic RNA, the RdRp synthesizes the sense RNA without primers (3)(4)(5)(7)(8)(9). The importance of RNA stem-loop structures for recognition by the norovirus RdRp has also been reported (10).
To probe the initiation mechanism of the norovirus genomic RNA synthesis by its RdRp more deeply, we investigated here how the RdRp recognizes the antisense genomic RNA, which does not have the polyA sequences at the 3 0 -end, to initiate the synthesis of the complementary positive-sense genomic RNA. In this study, the viral genomic RNA template-specific RNA synthesis by the human norovirus (HuNV) RdRp was demonstrated for the first time in an in vitro assay. Results showed how the RdRp recognizes the antisense genomic RNA to initiate the synthesis of the complementary positive-sense genomic RNA.

Results
HuNV RdRp synthesizes the positive-sense RNA with the antisense viral genome template RNA carrying the 3 0 -end region We used a nonradioactive labeling method to detect RNA synthesis by RdRp in vitro (a schematic of the procedures is shown in Fig. 1A). After the template RNA was incubated with the RdRp, RNA complementary to the template RNA was synthesized, forming a double-stranded RNA (dsRNA) with the template RNA. The dsRNA then remained after incubation with an S1 ribonuclease, to digest the single-stranded RNA into small nucleotides. Subsequently, the synthesis of the dsRNA is then used as an index that the RdRp synthesizes the complementary RNA in the following experiments.
The HuNV GII. P3 RdRp was tagged with six histidines at its N-terminus, expressed in Escherichia coli cells, purified, and analyzed by SDS-PAGE. The mobility of the major band was calculated as 57 kDa according to the molecular weight markers (lane 1 in Fig. 1B), and the band was detected by anti-RdRp and anti-His antibodies (lanes 2 and 3 in Fig. 1B, respectively). Finally, the purity of the RdRp was calculated as 96% (lane 1 in Fig. 1B).
The 5 0 terminal nt 1 to 100 of viral genomic RNA of the HuNV GII. P3 strain (100S [sense]) or its complementary sequence (100AS [antisense]) was then used as template RNAs (see Fig. 1A). All the template RNAs used here carry GGG and CCC sequences at their 5 0 -and 3 0 -ends, respectively, to efficiently synthesize template RNAs in vitro and form stable dsRNAs. When increasing amounts of RdRp were incubated with 100S RNA, almost no RNA was synthesized (lanes 4-7 on the upper column in Fig. 2A). However, when increasing amounts of RdRp were incubated with 100AS RNA, dsRNA was synthesized in increasing amounts (lanes 8-11 in the upper column in Fig. 2A). To confirm that these synthesized RNAs were dsRNAs, the RNAs were treated with S1 ribonucleases. The amounts of the synthesized dsRNA were then increased according to the amounts of the RdRp (lanes 8-11 in the lower column in Fig. 2A and the graph in Fig. 2B). The efficient activity of S1 ribonucleases was confirmed by the fact that the marker ssRNAs (100S and 100AS) were digested well, but the dsRNA (ds100) markers were resisted with S1 ribonuclease treatment (lanes 1, 2, and 3 on the lower column in Fig. 2A, respectively). These results therefore indicate that the RdRp specifically recognizes the 3 0 -end of the antisense viral genome RNA and indicate also that this region is important to initiate the synthesis of the positive-strand RNA by the RdRp.
The predicted stem-loop structure and the region close to the stem are required for initiating the synthesis of the positivesense RNA by the RdRp We examined the secondary structure in the 3 0 -end regions of the antisense genomic RNA of HuNV GII.P3 strain. We found that nt 74 to 87 (located at the 3 0 -terminus) in the 100AS RNA forms a predicted stem-loop structure (see Experimental procedures and Fig. 3A). The sequence of nt 74 to 100 was highly conserved among the norovirus genogroups I, II, IV, and VII (Fig. 3B). To determine whether this stemloop (nt 74-87) was important for initiating RNA synthesis by the RdRp, a deletion mutant of 100AS RNA that lacks the 31 nt (nt 70-100) at the 3 0 -terminus of the 100AS RNA was constructed (designated as Δ31AS RNA, Fig. 4A). Its complementary template RNA was also constructed as a control (designated as Δ31S RNA). When Δ31AS RNA and Δ31S RNA were used as template RNAs, the RdRp did not synthesize the  Figure 1. Schematic illustration of a nonradioactive labeling method to detect RNA synthesis by RdRp in vitro and characterization of the purified RdRp. A, as shown by red or blue arrows, respectively, the region of the RNA (positive sense or antisense (complementary) sequences of the 5 0 -or 3 0 -end 100 nt of the human norovirus genomic RNA) used as templates in this study is shown. Template single-stranded RNA (ssRNA) and norovirus RNAdependent RNA polymerase (RdRp) were mixed and incubated to synthesize complementary RNA. The newly synthesized ssRNAs formed doublestranded RNA (dsRNA) with an excess of ssRNA templates. The ssRNAs in the reaction mixture were digested with S1 ribonuclease to visualize dsRNA only. RNA molecules before and after S1 ribonuclease treatment were separated by nondenaturing polyacrylamide gel electrophoresis (native PAGE) and stained with SYBR Green II. B, 31.7 nmol (lane 1) or 2.3 pmol (lanes 2 and 3) of the purified RdRp was performed using SDS-PAGE. After SDS-PAGE, the gel was visualized using Coomassie dye staining (CDS; lane 1) or detected with Western blotting using anti-RdRp antiserum (lane 2) or using anti-His-tag antibody (lane 3). M, molecular weight markers.
In vitro initiation of genomic RNA synthesis by HuNV RdRp complementary RNA from either template RNAs (lanes 11 and 13 at the upper column in Fig. 4A). To remove the singlestranded template RNA to confirm the synthesis of the dsRNA, the reaction products shown on the upper column in Figure 4A were treated with S1 ribonucleases. Absent or faint RNAs were detected (lanes 11 and 13 in the lower column in Fig. 4A, respectively). These results therefore indicate that nt 70 to 100 of 100AS RNA containing the predicted stem-loop structure is necessary for initiating the complementary positive-strand genomic RNA synthesis by the RdRp.
To confirm that the 31 nt RNA, carrying the nt 70 to 100 of 100AS RNA (31AS RNA, shown in Fig. S1), was sufficient for RNA synthesis by the RdRp, two chimeric RNA templates were further prepared: (i) the 5 0 -end of 31AS RNA was connected to the 3 0 -end of the Δ31S RNA (designated as add31AS-S), and (ii) the 3 0 -end of 31S RNA was connected to the 5 0 -end of the Δ31AS RNA (designated as add31S-AS and complementary to add31AS-S RNA, Fig. S1). The RdRp also synthesized dsRNA with add31AS-S RNA as the template, but a few amounts of dsRNA were detected when the complementary add31S-AS RNA was used as a template (lanes 11 and 13, respectively, in Fig. 4B). These results therefore confirmed the importance of the 31 nt region at the 3 0 -end of antisense genomic RNA for the RdRp activity.
Similarly, to narrow down the regions in nt 70 to 100 of the 100AS RNA responsible for initiating RNA synthesis by the RdRp, the following deletion mutants were constructed: Δ6AS, Δ9AS, and Δ12AS RNA, in which nt 95 to 100, nt 92 to 100, or nt 89 to 100 were deleted, respectively (Fig. 4C). The result showed that the RdRp synthesized dsRNAs derived from Δ6AS and Δ9AS, respectively, to the same extent as that of the 100AS RNA, whereas the synthesis of dsRNA from the Δ12AS RNA template was much less efficient (Fig. 4C), indicating that the nt 90 to 100 is not important for initiating dsRNA synthesis.
To know the importance of the stem-loop region (nt 74-87), three kinds of mutant template RNAs were also constructed: (i) Δsl  (Fig. 4F). When the ΔslAS and swAS RNAs were used as templates for RdRp, small amount of the dsRNAs was synthesized (lane 13 in Fig. 4, D and E, respectively). Alternatively, dsRNA was synthesized as well with lpAS RNA (lane 13 in Fig. 4F). These results, therefore, also indicate that the stemloop region is important, and the stem region (nt 74-78 and nt 83-87) is more important than the top-loop region (nt 79-82) (see Fig. 3). All these results of the activities of 100S, 100AS, and its mutated template RNAs are summarized in Fig. S2.

Interaction of the 3 0 -end of 100 nt antigenomic RNA with the RdRp
The direct interaction of 100AS RNA with the RdRp was examined by the electrophoretic mobility shift assay (EMSA) technique using RNAs labeled with 32 Phosphate at the 5 0 -end. When 100AS RNA was incubated with the RdRp, the amount of the RNA-RdRp complex increased as the amount of the RdRp increased (lanes 1-4 in Fig. 5A). However, when Δ31AS RNA lacking in the 31 nt of the 3 0 -end of the 100AS RNA was incubated  [8][9][10][11], and the purified HuNV RdRp (0-32 μM) were mixed. Half of the total volume of samples was mixed with 6× gel-loading buffer and loaded on native PAGE (the upper column). S1 ribonuclease was added to the remaining half of the samples, and the samples were incubated and loaded on native PAGE (the lower column). The dsRNA products separated in native PAGE before and after S1 ribonuclease treatment are shown in the upper and lower columns, respectively. Red arrow indicates newly synthesized RNA products, which had formed dsRNAs with an excess of template RNA and identically migrated to dsRNA markers (lanes 4-11). B, the relative amounts of the newly synthesized dsRNA products (the transcriptional activity) were analyzed using ImageJ v.2.1.0/1.53c computer software and shown as the graph. The amounts shown were the averages of three experiments. The standard deviations are also shown in the graph. Each result of the three independent experiments is also shown by the three kinds of shapes (circles, rectangles, and hexagons).
In vitro initiation of genomic RNA synthesis by HuNV RdRp with increasing amounts of the RdRp, Δ31AS RNA formed a complex with the RdRp much less efficiently than 100AS RNA even at the highest concentration (lanes 5-8 in Fig. 5A). These results indicate that the 31 nt is recognized by the RdRp with high affinity and implies that this recognition is required for the initiation of genome positive-strand RNA synthesis.
To confirm that this interaction was specific, two nonlabeled RNAs, 100AS RNA and Δ31AS RNA, were used as competitors in EMSA (Fig. 5B). When the 32 P-labeled 100AS RNA was incubated with increasing amounts of the same nonlabeled 100AS RNA, the amount of the labeled RNA-RdRp complex decreased in correspondence with the increase in nonlabeled 100AS RNA. The amount of the shifted RNA-RdRp complex decreased to 13.5% (lanes 1-5 in Fig. 5B). In contrast, when 32 Plabeled 100AS RNA was mixed with increasing amounts of nonlabeled Δ31AS RNA lacking in nt 70 to 100 at the 3 0 -end of 100AS RNA, the 32 P-labeled 100AS RNA-RdRp complex competed with nonlabeled Δ31AS RNA, but the efficiency of competition with nonlabeled Δ31AS RNA (decreased to 35.8%) was lower than that with the nonlabeled 100AS RNA (decreased to 13.5%). These results thus indicate that the 100AS RNA and the RdRp interact specifically and imply that the 31 nt RNA (31AS RNA) is required for this interaction.
Similarly, since the 31AS RNA (the nt 70-100 of 100AS RNA, see Fig. S1) was important for initiating genomic RNA synthesis by the RdRp as demonstrated above (lanes 13 and 11 in Fig. 4, A and B, respectively); EMSA was used to examine the interaction of the 31AS RNA with the RdRp. The 32 Plabeled 31AS RNA was incubated with increasing amounts of the RdRp. However, no 31AS RNA-RdRp complex was detected even at the highest concentration (600 pmol) of the RdRp (lanes 9-12 and the graph in Fig. 5A). Also, when the nonlabeled 31AS RNA was used as a competitor for the complex of 100AS RNA and RdRp, the competition was less efficient than when using the nonlabeled 100AS RNA as a competitor (decreased to 47.3% versus 13.5%, lanes 11-15 versus lanes 1-5 and the graph in Fig. 5B). It is possible that the 31AS RNA was too short to form the stable complex with the RdRp efficiently. Therefore, to determine whether the 31AS RNA region interacted with the RdRp, the interaction of the RdRp with add31AS-S RNA (the structure formed by connecting the 5 0 -end of 31AS RNA to the 3 0 -end of Δ31S RNA (see Fig. S1)) was examined using EMSA. When the 32 Plabeled add31AS-S RNA was incubated with the RdRp, the amount of the RNA-RdRp complex increased in correspondence with the amounts of the RdRp (lanes 5-8 and the graph In vitro initiation of genomic RNA synthesis by HuNV RdRp in Fig. 5C). However, the Δ31S RNA itself (nt 31-100 of the positive-sense RNA) did not have an activity to synthesize the complementary RNA by the RdRp (lane 11 in Fig. 4A). It also did not form a complex with the RdRp (lanes 9-12 and the graph in Fig. 5C). Finally, these results show that 31AS RNA is required for interacting with the RdRp and required for initiating RNA synthesis as well.

Inhibitory effects of nucleoside analogs on the RNA synthesis by the HuNV RdRp
We examined whether our in vitro assay detected RdRp inhibitory effects by compounds. We tested 2 0 -C-methylcytidine (2 0 CM) and its ribonucleotide triphosphate form, 2 0 -C-methylcytidine triphosphate (2 0 CM-CTP), as known inhibitors of the HuNV RdRp. When increasing concentrations of 2 0 CM were added to the reaction mixture, no inhibitory effect on the RNA synthesis was observed (lanes 3-7 at the lower column and the graph in Fig. S3). However, when increasing concentrations of the converted form, 2 0 CM-CTP, were added to the reaction mixture, the amounts of synthesized dsRNA were decreased from 100% to 0.5% (lanes 8-12 and the graph in Fig. S3). Our result was also consistent with previous studies and reasonable because it is known that 2 0 CM has an inhibitory effect after it is converted to the ribonucleotide triphosphate form, 2 0 -Cmethylcytidine triphosphate (2 0 CM-CTP) by cellular kinases (11,12). The IC 50 (half of the maximal inhibitory concentration) of 2 0 CM-CTP was calculated as 57 ± 20 μM.

Discussion
We determined here that the HuNV RdRp initiates the template-specific initiation of RNA synthesis, and that the RdRp interacts with the 3 0 -end region of the antisense genomic RNA. First, we demonstrated that the HuNV RdRp synthesized complementary (positive-strand genomic) RNA from the 100 nt of the 3 0 -end region of the antisense genomic template RNA (100AS RNA), but did not synthesize the antisense RNA from the 100 nt of the 5 0 -end region of the genomic template RNA (100S RNA) ( Fig. 2A). Similarly, the introduction of deletions and mutations in the 31 nt region revealed that the predicted stem-loop region and the region close to the stem sequences are required as a template for RNA synthesis by the RdRp. There are reports on the activity of the HuNV RdRp in vitro, and the template RNAs used in these reports are those corresponding to partial 3 0 -terminal genomic or sense/antisense subgenomic sequences of HuNV (3,4,7,9) or a nonviral template (5,7,8). The HuNV RdRp has the activity to synthesize the complementary RNAs using any sequences of the template RNAs used in these reports. In contrast, we demonstrated the genomic sequence specificity of template RNA. The genome specificity would come from the concentrations of the RdRp in the reaction mixture; the concentration of the RdRp used in these reports was about 30 to 200 times higher than that used in this study.
Second, we also demonstrated the interaction of HuNV RdRp with antisense genomic RNA by using EMSA. Our results showed that the RdRp forms a complex with 100AS RNA, and the specific interaction was shown by competition with the same nonlabeled 100AS RNA. The deletion of nt 70 to 100 in 100AS (Δ31AS RNA) also caused a loss in the interaction with the RdRp in EMSA, and the competition with Δ31AS RNA was less efficient than that with 100AS for the 100AS RNA-RdRp binding. These results indicate that the nt 70 to 100 (31 nt) was required for the interaction with the RdRp (Fig. 5, A and  B). The add31AS-S RNA, which is composed of the connection between the 5 0 -end of 31AS RNA and the 3 0 -end of Δ31S RNA (shown in Fig. S1), formed the complex with the RdRp by EMSA (Fig. 5C). These results also show that the nt 70 to 100 in 100AS RNA (31AS RNA) is required for interaction with the RdRp. As shown in Figure 4B, the add31AS-S RNA was active as a template for the RdRp. Finally, these results imply that the 3 0 -end (the nt 70-100) of the antisense genomic RNA is required for recognition and initiation by the RdRp for synthesis of positive-sense genomic RNA.
As shown in Figure 3, the regions containing the predicted stem-loop structure are important for recognizing and initiating the synthesis of the positive-sense genomic RNA by the norovirus RdRp. Simmonds et al. have reported predicted secondary structures at three locations: the 5 0 -end, 3 0 -end, and the upstream of the subgenomic RNA in the whole HuNV and murine norovirus (MNV) genomic and antigenomic RNAs. They also showed that all of these regions are important for MNV genome replication (10). Additionally, the upstream subgenomic region was shown as a promoter of the antisense subgenomic RNA (10,13,14). Likewise, the importance of the stem-loop structure located close to the end of the genomic or antisense genomic RNA in the synthesis of viral genomic or antigenomic RNA by viral RdRp has also been reported in other nonsegmented single-stranded positive-sense RNA viruses (e.g., dengue virus [DENV], hepatitis C virus [HCV], and classical swine fever virus [CSFV]) (15)(16)(17).
The nt 74 to 100 region of GI, GII, GIV, and GVII noroviruses antisense genomic RNA is highly conserved among genogroups, although the GIII, GV, and GVI genogroups are different from this "conserved sequence" (Fig. 3B). This result also indicates that this mechanism of initiation of viral genomic RNA synthesis by the RdRp is common among many genogroups, especially the GI, GII, GIV, and GVII genogroups, which infect humans and canines, respectively.
As we reported here, the 31 nt RNA located at the 3 0 -end region of the antisense HuNV genomic RNA, which contains the predicted stem-loop structure, is required to bind to the RdRp and to initiate the RNA synthesis by the RdRp. In GI, GII, GIV, and GVII noroviruses, the nt 74 to 100 regions corresponding to the 3 0 -end regions of the antisense subgenomic RNA are also highly conserved and homologous to those of the antisense genomic RNA (Fig. S4 and Fig. 3B). This implies that nt 74 to 100 of the antisense subgenomic RNA also forms the stem-loop structure the same as the antisense genomic RNA; also, the mechanism of the synthesis of the subgenomic RNA by the RdRp is probably the same.
The RdRp also interacts with other stem-loop RNA regions (promoter for synthesizing the subgenomic RNA), the complementary sequences of which are located upstream of the antisense genomic RNA, and initiates the RNA synthesis (10,13,14).
These results imply that the norovirus RdRp binds to at least three regions of stem-loop structures in the antisense genomic RNA (Fig. S5); as we reported here, the one (indicated as 1 in Fig. S5) is located close to the 3 0 -end of the antisense genomic RNA to synthesize the genomic RNA. The second one (indicated as 2 in Fig. S5) is located at the 3 0 -end of the antisense subgenomic RNA. The third one (indicated as 3 in Fig. S5) is the promoter sequence synthesizing the subgenomic RNA from the antisense genomic RNA (13,14). The sequences and structures have a high similarity between regions 1 and 2 (Fig. S4). Therefore, it is important to know how the RdRp distinguishes these three regions (1, 2, and 3) to bind and initiate RNA synthesis in the next stage. We proposed one model for the RdRp recognition of region 2 and region 3: the RdRp binds to region 3 (promoter) in the antisense genomic RNA preferentially, but once region 2 becomes the 3 0 -end after the antisense subgenomic RNA is synthesized, the RdRp binds to region 2 preferentially. Host cell factors, other viral proteins, and/or the conformational changes after RdRp binds to these regions might control which regions the RdRp binds preferentially to initiate the syntheses of genomic or subgenomic RNAs.
The nucleoside analog inhibitor of RdRp, 2 0 -C-methylcytidine (2 0 CM), was initially an HCV inhibitor (12). It works as a chain terminator and competes with CTP during the synthesis of genome RNA by viral RdRp. We also showed that 2 0 CM-CTP inhibited the synthesis of the viral genomic RNA by the HuNV RdRp (Fig. S3). The IC 50 was 57 ± 20 μM. This inhibitory efficiency on HuNV was similar to that reported by Jin et al. (7) (IC 50 is 33.6 ± 5.8 μM). Similarly, several antiviral drugs have been developed so far, and most of them are directacting antivirals (DAAs) that selectively target viral components (e.g., RdRp, helicase, or protease) without affecting cellular function, thus minimizing side effects (18,19). Our in vitro RdRp assay system will therefore be a valuable tool to screen seed compounds for developing DAAs against HuNV RdRp.
Conclusively, we identified the regions in the antigenomic HuNV RNA that are required for the interaction and initiation of the genomic RNA synthesis by its RdRp in vitro.

Plasmid architecture
The plasmid for expressing the RdRp protein coding for the cDNA (nt 3581-5113) of the human norovirus GII. P3_GII.3 U201 strain (accession no. AB039782) (20) was amplified by PCR using the specific primers containing the NdeI restriction sites at the 5 0 -end, the stop codon (TAA), and the XbaI restriction site at the 3 0 -end (shown in the table in Supporting information). Thereafter, it was cloned into the NdeI-XhoI site of the pET28b vector (Merck Millipore) (designated pET28b-RdRp).
All plasmids for the template RNAs were prepared by PCR using nt 1 to 100 cDNA corresponding to the 5 0 -terminal of the genomic RNA of the human norovirus GII.P3_GII.3 U201 strain (corresponding to the positive strand) or its complementary sequence (corresponding to the antisense strand), specific primers containing the XbaI restriction site-T7 promoter sequences-three Gs (GGG) at the 5 0 -end, or three Cs (CCC) at the BsaI-EcoRI sites at the 3 0 -end (shown in the table in Supporting information), and Ex Taq DNA polymerase (Takara). The amplified DNA fragments were then cloned into the same restriction sites of pUC19 vector (Takara). When the plasmids, pUCadd31AS-S and pUCadd31S-AS, were prepared, pUCΔ31S or pUCΔ31AS was used as the DNA template for PCR. Similarly, all the plasmids for the template RNA were designed to carry GGG and CCC sequences at the 5 0 -and 3 0ends, respectively. This plasmid was then used to synthesize template RNAs efficiently by T7 polymerase and to form the stable dsRNAs.
Preparation of the HuNV RNA-dependent RNA polymerase (RdRp) E. coli cells, ArcticExpress (DE3) (Agilent Technologies, Inc) carrying the pET28b-RdRp were cultured in 1 l of Luria Broth Base medium (Invitrogen) at 30 C until the turbidity at the optical density (OD) 590 was approximately 0.6 (for up to 3 h) and then placed in ice-cold water for 20 min. Next, isopropyl β-D-1-thiogalactopyranoside (final concentration, 1 mM; Wako) was added, and the cells were cultured for an additional 24 h at 13 C. To purify RdRp (His-RdRp) containing a His tag at the N-terminus, the culture was then centrifuged at 4400g on a JLA-10.500 rotor (Beckman Coulter) for 5 min. Afterward, 20 ml of buffer (50-mM Tris-HCl (pH 7.5), 150-mM NaCl, and 5-mM β-mercaptoethanol) was added to the bacterial pellets, and the pellets were suspended in the buffer and sonicated by five rounds of a 30 s on ¾-in. chip at an intensity level of 5 with subsequent cooling in iced water (Astrason Ultrasonic Processor XL-2020; Misonix, Inc). The sample was then centrifuged at 22,700g on a JA-20 rotor (Beckman Coulter) for 15 min, after which the supernatants were loaded onto a 2.5-ml buffer-equilibrated Ni-NTA column (Qiagen). After washing the column first with 200-ml buffer, and then with 15-ml His-A buffer (50-mM Tris-HCl (pH 7.5), 150-mM NaCl, 5-mM β-mercaptoethanol, 10-mM imidazole), His-RdRp was eluted with 10-ml elution buffer (50-mM Tris-HCl (pH 7.5), 150-mM NaCl, 5-mM β-mercaptoethanol， 200-mM imidazole). The His-RdRp was subsequently purified further with a buffer-equilibrated gel filtration column (HiLoad 16/60 Superdex 200; GE Healthcare).

Preparation of template RNAs
Fifty micrograms of each plasmid was digested using BsaI (New England Biolabs) to exclude the nonviral sequences at the 3 0 -end. RNA was then synthesized from a template of the linearized DNA using T7 polymerase and purified with a gel filtration column (HiLoad 16/60 Superdex 75, or 200) (the purity of RNAs; at least 90%) as described by Mckenna et al. (21) and Shimoike et al. (22).
The 5 0 -end 32 P-labeled template RNAs were prepared as follows: the calf intestinal alkaline phosphatase (Takara) was used to dephosphorylate the 5 0 -end of the purified template RNAs. Subsequently, the 32 Phosphate was bound to the 5 0 -end of the dephosphorylated template RNA by the T4 polynucleotide kinase (Takara), followed by phenol/chloroform, and precipitated by ethanol. The unreacted γ-32P-ATP in the RNA was then removed using Sephadex G-50 fine resin (GE Healthcare). The 32 P-labeling efficiencies of RNAs were found to be 49% to 51%. To the other portion, 7 μl of ten times diluted S1 nuclease (89 U/μl; Promega) and 3 μl of S1 ribonuclease buffer were added and incubated for 30 min at 30 C to degrade the ssRNA and detect double-stranded RNA (dsRNA) comprising the template RNA and the newly synthesized complementary strands. After this step, the 6× gelloading buffer was then added, and the native PAGE was performed as described above. After native PAGE, the gel was then stained with SYBR Green II (LONZA) and detected using a LAS-3000 imaging analyzer (Fuji Film).

Assay for RNA synthesis by RdRp
Prediction of the RNA secondary structure and alignments of NV antisense genome sequences The nucleotide secondary structure of the 100AS RNA (nt 1-100, GII.P3 U201 strain) was predicted using the Genetyx-Mac software v.19.0.3 (Genetyx Corporation) by "RNA secondary structure prediction" (23) using default settings. The alignments of the 3 0 -end 100 nt regions of the norovirus antisense genomic or antisense subgenomic RNA sequences were then calculated using the Genetyx-Mac software v. 19

Electrophoretic mobility shift assay (EMSA)
The 32 P-labeled 5 0 -end-labeled RNA (1 pmol) and the RdRp (200, 400, 600 pmol) were incubated in the reaction buffer (50-mM Tris-HCl pH 7.4, 1.7% glycerol, 5-mM β-mercaptoethanol, and 0.5 μg/μl of yeast RNA [Thermo Fisher Scientific]) at 30 C for 10 min (total volume, 20 μl). A 4 μl of 6× gel-loading buffer was then added to the reaction mixture. The RNA and RdRp complex was separated with native PAGE at a constant voltage of 60 V for 90 min at 4 C in the 1/2 × running buffer (12.5-mM Tris, 86-mM glycine, pH 8.3) after the gel was prerun at a constant voltage of 60 V for 30 min at 4 C and analyzed by Typhoon FLA 7000 image analyzer (GE Healthcare). In the case of the competition assay, the procedures are the same as above except that the nonlabeled RNA (0, 2.4, 4.8, 12.0, or 24.0 pmol) was incubated with 32 P-labeled 100AS RNA (0.2 pmol) and the RdRp (600 pmol), respectively.
Assay for inhibitory effects of the nucleoside analog on RNA synthesis 2 0 -C-methylcytidine (2 0 CM, Merck) and 2 0 -C-methylcytidine triphosphate triethylammonium salts (2 0 CM-CTP, Carbosynth Limited) were added (final concentration of 0, 1, 10, 100, or 500 μM) to the reaction mixture that is the same content described in the Assay for RNA synthesis by RdRp except for the concentration of NTPs (800 μM each) and incubated at 30 C for 60 min, after which the dsRNA signal was detected as described above. The chemical compound concentrations resulting in a 50% reduction in dsRNA drugfree control production were determined based on the doseresponse curve and defined as the mean 50% inhibitory concentration (IC 50 ) values of the RdRp activity.

Data availability
All data are contained within the manuscript Supporting information-This article contains supporting information (10,13,14,24).