A Novel Mechanism to Ensure Terminal Initiation by Hepatitis C Virus NS5B Polymerase

Virology 285, 6–11 (2001) doi:10.1006/viro.2001.0948, available online at http://www.idealibrary.com on

RAPID COMMUNICATION

Zhi Hong,*,1 Craig E. Cameron,† Michelle P. Walker,* Christian Castro,† Nanhua Yao,* Johnson Y. N. Lau,* and Weidong Zhong*

*ICN Pharmaceuticals, 3300 Hyland Avenue, Costa Mesa, California 92626; and †Department of Biochemistry and Molecular Biology, Pennsylvania State University, University Park, Pennsylvania 16802 Received January 31, 2001; returned to author for revision February 20, 2001; accepted April 10, 2001

Hepatitis C virus (HCV) nonstructural protein 5B (NS5B) RNA-dependent RNA polymerase (RdRp) has acquired a unique ␤-hairpin in the thumb subdomain which protrudes toward the active site. We report here that this ␤-hairpin plays an important role in positioning the 3Ј terminus of the viral RNA genome for correct initiation of replication. The presence of this ␤-hairpin interferes with polymerase binding to preannealed double-stranded RNA (dsRNA) molecules and allows only the single-stranded 3Ј terminus of an RNA template to bind productively to the active site. We propose that this ␤-hairpin may serve as a “gate” which prevents the 3Ј terminus of the template RNA from slipping through the active site and ensures initiation of replication from the terminus of the genome. This hypothesis is supported by the ability of a ␤-hairpin deletion mutant that utilizes dsRNA substrates and initiates RNA synthesis internally. The proposed terminal initiation mechanism may represent a novel replication strategy adopted by HCV and related viruses. © 2001 Academic Press

Infection by hepatitis C virus (HCV) is a significant known about this class of polymerases until recently. human medical problem with an estimated prevalence of Structural studies of unliganded poliovirus 3Dpol (9) and 170 million cases globally. Up to four million individuals HCV NS5B (4, 13) have revealed unique features and may be infected in the United States alone (1). In most provided useful information for the comparative analysis instances, the virus establishes a chronic and insidious of RdRps in relation to other classes of polymerases. infection that persists for decades. This usually results in This will in turn help to advance our understanding of recurrent and progressively worsening liver inflamma- viral replication at the molecular level for positive- tion, which often leads to more severe disease states stranded RNA viruses. such as cirrhosis and hepatocellular carcinoma. A Unique ␤-Hairpin in the Thumb Subdomain of HCV HCV, a member of the Flaviviridae family, is a positive- NS5B Polymerase. One novel structural component stranded RNA virus. Its life cycle consists of several unique to HCV polymerase is a ␤-hairpin found in the important and interrelated processes that occur primarily thumb subdomain, which protrudes toward the active in the cytoplasm of the host cells (19). Among these site located at the base of the palm subdomain (yellow intricate processes, RNA replication is the centerpiece of ribbon, Fig. 1A). The ␤-hairpin consists of 12 amino acids the viral proliferation cycle. While many viral and/or host (LDCQIYGACYSI) with a tyrosine (Y) and glycine (G) pair factors are suspected to be involved, one certainty is that at the top of the hairpin (position and direction of the the virally encoded polymerase has a key role in the ␤-hairpin is shown at the bottom of Fig. 1A). In stark entire replication process. The interplay between viral contrast, this ␤-hairpin is absent in polioviral polymerase polymerase and its substrates (template, primer, and (3Dpol) (Fig. 1B). This structural difference suggests that incoming nucleotide) sets the initial and minimal frame- the ␤-hairpin may serve a special function important for work for viral replication. For HCV, the nonstructural pro- HCV replication. Poliovirus, a well-studied virus, initiates tein 5B (NS5B) is the RNA-dependent RNA polymerase its genome replication using a protein primer (VPg, prod- (RdRp) required for genome replication (3, 7, 14). Al- uct of 3B) (18). Contrary to the inability of HCV NS5B to though the first RdRp was discovered more than 20 years bind dsRNA substrates productively as reported previ- ago in poliovirus infected cells, relatively little was ously (21), polioviral 3Dpol can efficiently assemble an elongation-competent complex with a symmetrical duplex RNA substrate (sym/sub), suggesting that the 3Dpol 1 To whom correspondence and reprint requests should be ad- apoenzyme can accommodate preannealed double- dressed at ICN Pharmaceuticals, Inc., 3300 Hyland Avenue, Costa stranded template/primer (2). This observation coincides Mesa, CA 92626. Fax: (714) 668-3141. E-mail: [email protected]. with the lack of an analogous ␤-hairpin in 3Dpol (Fig. 1B)

FIG. 1. Structural comparison between HCV NS5B RdRp and poliovirus 3Dpol RdRp. Molecular surface structures were constructed for both polymerases (turquoise color for HCV and blue color for poliovirus) and oriented with the palm at the base, the fingers to the left, and the thumb subdomain to the right. The essential GDD motifs (polymerase motif C; part of the active site) are colored red. In HCV NS5B, the ␤-hairpin comprising residues 443 through 454 is colored yellow and protrudes toward the active site (4). The sequence of the ␤-hairpin is shown at the bottom of (A). In both structures, the unique molecular shape of the RdRps with the fully encircled NTP binding site is clear. This encircled active site likely exists in 3Dpol because the extreme aminoterminal residues in the crystal structure contact the thumb subdomain, just as in HCV NS5B, before proceeding to form the fingers subdomain. This figure was produced using the program INSIGHT II (Molecular Simulation Inc., San Diego, CA). In the crystal structure of poliovirus 3Dpol (9), approximately 30% of the protein residues were unobserved. The availability of HCV NS5B RdRp structure (4) allows a model of the complete poliovirus 3Dpol structure to be constructed. Based on the similar polypeptide fold in regions common to both crystallographic models, the missing portions of the 3Dpol structure were simply transferred from NS5B and the correct side chains were assigned with the most common rotamer. The junctions between crystallographically observed and inferred residues were adjusted according to local similarity to a database of polypeptide fragments derived from high-resolution crystal structures. which, in HCV NS5B polymerase, imposes steric hin- shorten the ␤-hairpin without disrupting the structural drance against docking the dsRNA (13). For HCV, assem- integrity and the catalytic activity of HCV NS5B. bly of a catalytically competent complex requires the Shortened ␤-Hairpin Allows Internal Initiation from RNA template to be single-stranded at the 3Ј terminus Preannealed Duplex RNA. Guided by the crystal struc- (10, 21). Upon template binding to the polymerase, short ture of HCV NS5B polymerase (4), four amino acids from RNA primers (i.e., the initiating nucleotide, one to three each strand were deleted simultaneously to shorten the nucleotides in length) basepair with the terminal tem- ␤-hairpin without potential global structural perturbation plate bases in the active site of the polymerase and that may affect the RdRp activity. The resulting deletion prime RNA synthesis from the 3Ј terminus of the template mutant, BL⌬8, retained two amino acids at the top of the (21). A structural model for the quaternary complex of ␤-hairpin, which were changed from YG to GG to favor HCV polymerase/template/primer/nucleotide has been the formation of the observed type IЈ ␤-turn between the proposed, which suggests that the unique ␤-hairpin two ␤-strands (from LDCQIYGACYSItoLGGI). This dele- plays an important role in positioning the 3Ј terminus of tion mutant as well as its parent wild-type NS5B protein viral RNA properly in the polymerase active site for cor- were produced in bacterial cells and purified as de- rect initiation of RNA replication (21). To test this hypoth- scribed previously (9, 13). The purity of the wild-type and esis, a deletion mutant was constructed to significantly BL⌬8 NS5B is above 95% and comparable to that of the 8 RAPID COMMUNICATION

was unable to utilize sym/sub or preannealed dsRNA efficiently to direct RNA synthesis (Fig. 2B, lane 2). In contrast, the BL⌬8 mutant was able to utilize sym/sub as efficiently as poliovirus 3Dpol (Fig. 2B, compare lanes 3 and 4) under the same assay conditions (described in Fig. 3 legend). Interestingly, the overall RdRp activity of BL⌬8 was increased approximately fivefold compared to the wild-type NS5B (Fig. 4). These results support the notion that the ␤-hairpin prevents HCV polymerase from binding to a double-stranded template/primer duplex in a productive manner, while polioviral polymerase lacking the ␤-hairpin binds to dsRNA and efficiently extends the labeled primer. This observation argues that the accom- modation of the nascent dsRNA (a product of elongation FIG. 2. A ␤-hairpin deletion mutant allows productive binding to dsRNA substrate to initiate RNA synthesis. (A) Expression and purification of HCV during replication) by HCV polymerase must be accom- NS5B and poliovirus 3Dpol. NS5B cDNA was isolated from a subgenomic panied by a conformational change from initiation to Ј replicon clone (pFKI389/NS3-3 /wt) kindly provided by Dr. Ralf Barten- elongation in the polymerase active site, perhaps by schlager (Johannes-Gutenberg University, Mainz, Germany). The region moving the ␤-hairpin away from the active site. encoding the C-terminal 21 amino acids was removed to improve the ⌬ HCV NS5B initiates RNA synthesis from the initiating solubility of NS5B as described previously (7). The BL 8 deletion mutant Ј was created from the C-terminally deleted NS5B by using the Quick- nucleotide (de novo) or a dinucleotide at the 3 terminus Change Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The of the template (21). The dinucleotide-mediated terminal poliovirus 3Dpol was produced as described (8). The purity of each poly- initiation assay was selected to characterize the ␤-hair- merase was above 95%. (B) Use of duplex RNA substrate. Single-nucle- pin deletion mutant. The initiating dinucleotide, believed otide incorporation assay was performed using end-labeled sym/sub RNA to be the product of abortive initiation, was shown to as the template/primer pair. AMP incorporation was compared among wild-type NS5B (lane 2), ␤-hairpin deletion mutant BL⌬8 (lane 3), poliovirus prime RNA synthesis efficiently by HCV NS5B (21), a 3Dpol (lane 4). No template control is in lane 1. A measure of 2 ␮M of each phenomenon that was also demonstrated in bacterio- enzyme was employed, respectively. phage T7 RNA polymerase-mediated RNA synthesis (6). Terminal initiation is in accordance with the quaternary poliovirus 3Dpol (Fig. 2A). The BL⌬8 mutant was first structural model wherein the 3Ј terminus of the RNA tested for its ability to utilize the dsRNA substrate, sym/ template sterically clashes with the tip of the ␤-hairpin sub. The wild-type NS5B, as shown previously (5, 21), loop and is unable to slip through the active site (21).

FIG. 3. The ␤-hairpin deletion mutant initiates RNA synthesis internally. (A) Internal initiation from a dinucleotide primer. “TER” and “INT” template RNAs were employed to distinguish terminal vs internal initiation from the primer 33pGpG. RdRp activities of wild-type NS5B (lanes 2 and 3), deletion mutant BL⌬8 (lanes 4 and 5) were compared. “Ϫ” in lane 1 represents a control without the template RNA. “P” represents the dinucleotide pGpG and “P ϩ 1” is the trinucleotide product as a result of single nucleotide addition. (B) Internal initiation from a preannealed primer. A preannealed template and primer (P, an end-labeled primer 33pCGGGCC) were tested for single-nucleotide (AMP) incorporation. Lane 1, no template control “Ϫ”; lane 2, wild-type NS5B; lane 3, BL⌬8 mutant; lane 4, poliovirus 3Dpol. The standard nucleotide incorporation reaction was carried out in a volume of 20 ␮l containing 50 mM HEPES (pH 7.3), 10 mM ␤-mercaptoethanol, 50 mM NaCl, 5 mM MgCl2,2␮M of end-labeled sym/sub RNA, or 10 ␮M of “TER” or “INT” template RNA and 10 ␮M of end-labeled dinucleotide, 2 ␮M of polymerase (HCV NS5B or poliovirus 3Dpol), and 100 ␮M of CTP or ATP (2, 21). The product was separated on a 23–25% PAGE/6 M urea/1ϫ TBE gel and exposed to X-ray film. Individual bands were quantified using a Phosphorimager (Typhoon 8600, Molecular Dynamics, Sunnyvale, CA). All synthetic RNA templates and primers were chemically synthesized (Oligos, Etc., Wilsonville, OR). RAPID COMMUNICATION 9

3), indicating that the shortened ␤-hairpin allowed internal initiation of RNA synthesis by permitting the 3Ј terminus of the template to slip through the active site. Poliovirus 3Dpol, which lacks the ␤-hairpin, allowed internal initiation (Fig. 3B, lane 4) from a primer (P) preannealed to the middle of a template, an activity that could be duplicated by the BL⌬8 mutant but not by the wild- type NS5B polymerase (Fig. 3B, compare lanes 2 and 3). The above results suggest that the apo form of the HCV polymerase is unable to initiate efficient RNA synthesis internally due to the presence of the ␤-hairpin, whereas poliovirus, although capable of internal initiation, has acquired a discrete mechanism to ensure terminal initiation from the poly(A) stretch at the end of the viral genome (18). The BL⌬8 Mutant NS5B Produces Predominantly Trun- cated RNA Products. To compare the RdRp activity of the wild-type and the BL⌬8 mutant NS5Bs on a natural HCV template, an HCV minigenome was constructed that con- tains authentic 5Ј and 3Ј termini and portions of the coding sequence (pMinigenome) (Fig. 4A) (15). In vitro transcribed minigenome RNAs were used as templates in a standard RdRp assay. Products larger than the input RNA, particularly of dimer size, were apparent for both wild-type and BL⌬8 mutant NS5Bs (Fig. 4B, lanes 3–7 for the wild-type NS5B; lanes 8–12 for the BL⌬8 mutant NS5B). This is consistent with the observation that NS5B ⌬ FIG. 4. RdRp activity of the wild-type and the BL 8 mutant NS5Bs on is capable of initiating RNA synthesis via the copy-back in vitro transcribed HCV RNA. (A) Structure of the minigenome RNA Ј used for RdRp assay. Numbers refer to nucleotide positions within the mechanism at the 3 terminus (3, 14). Also in agreement HCV subgenomic replicon (15). An internal deletion between two KpnI with previous reports (14, 17), the near dimer-size hairpin sites at positions 1684 and 7477 reduced the size of the minigenome product appeared 15–20 min after the start of the reac- RNA to about 2.1 kb. (B) Comparison of the RNA products from the HCV tion, corresponding to an incorporation rate of approxi- minigenome RNA template. RdRp reactions were initiated by adding 1 ␮ ⌬ mately 100–140 nucleotides per minute. Interestingly, the g of each protein: lanes 2–7, wild-type NS5B; lanes 8–13, BL 8 NS5B. ⌬ Reactions were terminated by the addition of SDS-containing buffer BL 8 mutant produced RNA products varying in size, and proteinase K at 10 (lanes 3 and 8), 20 (lanes 4 and 9), 30 (lanes 5 which appeared as a smear by denaturing agarose gel and 10), 50 (lanes 2, 6, 11, and 13), or 70 (lanes 7 and 12) min after the electrophoresis (Fig. 4B, lanes 8–12). Many of these reactions started. Control reactions contained no nucleotides (lanes 2 heterogeneous RNA products could occur by internal and 13) for 50 min. Samples were analyzed by denaturing guanidinium- initiation as a result of the ␤-hairpin deletion, consistent isothiocyanate-1% agarose gel electrophoresis. Minigenome monomer ␤ size marker was obtained by in vitro transcription of the HCV mini- with the notion that the -hairpin is necessary to ensure genome in the presence of ␣-[33P]UTP (lane 1). (C) Comparison of the terminal initiation. The overall RNA synthesis activity of RNA products from the homopolymeric RNA template/primer, poly(A)/ the BL⌬8 mutant was at least five times greater than that oligo(dT), using increasing enzyme concentrations. Reactions were of the wild-type NS5B (Fig. 4B, compare lanes 8–12 with initiated by the addition of 0.5 (lanes 1 and 4), 1.0 (lanes 2 and 5), or 2.0 ␮ lanes 3–7). In concert, enhanced RdRp activity by the (lanes 3 and 6) g of enzyme and terminated after 45 min at 30°C by ⌬ the addition of phenol/chloroform. BL 8 mutant was observed when using the homopolymeric RNA template/primer, poly(A)/oligo(dT), in a standard RdRp assay (Fig. 4C). Much of the enhanced RdRp Indeed, our results (Fig. 3A) demonstrate that HCV NS5B activity resulted from initiation from internally annealed is unable to utilize the “INT” template (lacking the termi- primers, leading to smaller than template-sized products nal dicytidylate bases) to initiate RNA synthesis from the (Fig. 4C, compare lanes 1–3 with lanes 4–6). Again, these diguanylate primer (33pGpG; lane 4), whereas the enzyme results support the proposed function of the ␤-hairpin. utilized the “TER” template with the terminal dicytidylate Identification of a Conserved Arginine Critical for De bases very efficiently (lane 5). Clearly, the wild-type HCV Novo Initiation from the 3Ј Terminus. A structural model NS5B could not initiate nucleotide incorporation from the was constructed to further define the molecular require- internal dicytidylate (CC) template bases. However, the ments at the active site for efficient assembly of the ␤-hairpin deletion mutant BL⌬8 could utilize the INT template RNA and the short initiating dinucleotide (21). template to initiate RNA synthesis (Fig. 3A, lanes 2 and Based on this model, a conserved arginine (R386) was 10 RAPID COMMUNICATION

FIG. 5. A conserved arginine critical for de novo initiation. A critical arginine at position 386 was identified to play a role in stabilizing the initiating nucleotide(s). (A) Effect on RNA synthesis from the dinucleotide primer, pGpG. Lane 1, no template control “Ϫ”; lane 2, wild-type NS5B; lane 3, NS5B with R386Q mutation; lane 4, BL⌬8 mutant; lane 5, BL⌬8 with R386Q mutation. “P” represents the end-labeled primer and “P ϩ 1” represents the extended primer. (B) Effect on RNA synthesis from the preannealed primer, sym/sub. The lane assignment is the same as (A). identified and proposed to play a role in stabilizing the servation that the 3Ј terminal poly(A) region of the polio- initiating nucleotide (for de novo synthesis) or the dinu- virus genome is the origin of RNA replication (20). HCV, cleotide (for pGpG-mediated synthesis), which would an RNA virus in the Flaviviridae family, appears to have otherwise be unable to basepair efficiently with the tem- adopted a different strategy for de novo RNA replication plate base(s). Interestingly, our previous mutational anal- (16, 22). This study shows that an intrinsic structural ysis had demonstrated that the equivalent Arg residue element is employed by HCV for precise terminal initia-

(R518) in bovine viral diarrhea virus (BVDV) NS5B was tion of the viral genome replication. important for de novo RNA synthesis (12). Indeed, site- A recent study demonstrated that BVDV NS5B poly- directed mutagenesis revealed that R386 was also impor- merase specifically recognizes the terminal template tant for the dinucleotide-initiated RNA synthesis, since base(s) for de novo RNA synthesis (11), providing evi- mutation to glutamine (R386Q) significantly reduced the dence that terminal initiation of RNA replication is also ability to utilize the dinucleotide for both the wild-type required by this related virus. Thus, the identification of a and BL⌬8 mutant NS5B (Fig. 5A, compare lanes 2 vs 3 novel and intrinsic structural element (such as the ␤-hair- and 4 vs 5). However, this mutation had little effect on pin in HCV NS5B) for terminal initiation may have a RNA synthesis from a preannealed duplex RNA (sym/ broader impact on our understanding of viral replication, sub) catalyzed by the deletion mutant BL⌬8 (Fig. 5B, as it may represent a replication strategy employed by lanes 4 and 5). These observations support the hypoth- related RNA viruses. esis that R386 plays a critical role in stabilizing the inter- Ј action between the 3 terminus of the RNA template and ACKNOWLEDGMENTS the initiating nucleotide(s). We believe that the ␤-hairpin C.E.C. is a recipient of a Howard Temin Award (CA75118) and is and R386 coordinate RNA synthesis from an initiating nucleotide (de novo) or the initiating dinucleotide. The supported in part by NIAID Grant AI47350. We thank Ralf Barten- schlager for providing the valuable subgenomic HCV replicon DNA. We dinucleotide is probably an abortive transcript of de novo also acknowledge the technical assistance of Shannon Dempsey. initiation that can prime efficient RNA synthesis when annealed to a bound template in the active site. REFERENCES Concluding Remarks. Terminal initiation is a critical first step for many positive-stranded RNA virus replica- 1. Alter, M. J., Kruszon-Moran, D., Nainan, O. V., McQuillan, G. M., Gao, tions. In the absence of repetitive sequence motifs, strict F., Moyer, L. A., Kaslow, R. A., and Margolis, H. S. (1999). The prevalence of hepatitis C virus infection in the United States, terminal initiation ensures the integrity of a viral genome. 1988 through 1994. N. Engl. J. Med. 341, 556–562. For poliovirus, a prototype virus representing the Picor- 2. Arnold, J. J., and Cameron, C. E. (2000). Poliovirus RNA-dependent naviridae family, precise terminal initiation is achieved by RNA polymerase (3Dpol): Assembly of stable, elongation-compe- a protein primer which assembles the replication com- tent complexes by using a symmetrical primer/template sub- plex at the 3Ј poly(A) region and initiates RNA synthesis strate (sym/sub). J. Biol. Chem. 275(8), 5329–5336. 3. Behrens, S.-E., Tomei, L., and De Francesco, R. (1996). Identification via a uridylylated protein primer (VPg) (18). This uridyly- and properties of the RNA-dependent RNA polymerase of hep- pol lation is highly specific as it requires specifically 3D atitis C virus. EMBO J. 15(1), 12–22. activity, UTP, and poly(A), which coincides with the ob- 4. Bressanelli, S., Tomei, L., Roussel, A., Incitti, I., Vitale, R. L., Mathieu, RAPID COMMUNICATION 11

M., De Francesco, R., and Rey, F. A. (1999). Crystal structure of 13. Lesburg, C. A., Cable, M. B., Ferrari, E., Hong, Z., Mannarino, A. F., the RNA-dependent RNA polymerase of hepatitis C virus. Proc. and Weber, P. C. (1999). Crystal structure of the RNA-dependent Natl. Acad. Sci. USA 96(23), 13034–13039. RNA polymerase from hepatitis C virus reveals a fully encircled 5. Carroll, S. S., Sardana, V., Yang, Z., Jacobs, A. R., Mizenko, C., Hall, active site. Nat. Struct. Biol. 6(10), 937–943. D., Hill, L., Zugay-Murphy, J., and Kuo, L. C. (2000). Only a small 14. Lohmann, V., Korner, F., Herian, U., and Bartenschlager, R. (1997). fraction of purified hepatitis C RNA-dependent RNA polymerase Biochemical properties of hepatitis C virus NS5B RNA-depen- is catalytically competent: Implications for viral replication and dent RNA polymerase and identification of amino acid sequence in vitro assays. Biochemistry 39(28), 8243–8249. motifs essential for enzymatic activity. J. Virol. 71(11), 8416–8428. 6. Cheetham, G. M., and Steitz, T. A. (1999). Structure of a transcribing 15. Lohmann, V., Korner, F., Koch, J.-O., Herian, U., Theilmann, L., and T7 RNA polymerase initiation complex. Science 286(5448), Bartenschlager, R. (1999). Replication of subgenomic hepatitis C 2305–2309. virus RNAs in a hepatoma cell line. Science 285, 110–113. 7. Ferrari, E., Wright-Minogue, J., Fang, J. W. S., Baroudy, B. M., Lau, 16. Luo, G., Hamatake, R. K., Mathis, D. M., Racela, J., Rigat, K. L., J. Y. N., and Hong, Z. (1999). Characterization of soluble hepatitis Lemm, J., and Colonno, R. J. (2000). De novo initiation of RNA C virus RNA-dependent RNA polymerase expressed in Esche- synthesis by the RNA-dependent RNA polymerase (NS5B) of richia coli. J. Virol. 73(2), 1649–1654. hepatitis C virus. J. Virol. 74(2), 851–863. 8. Gohara, D. W., Ha, C. S., Ghosh, S. K. B., Arnold, J. J., Wisniewski, 17. Oh, J. W., Ito, T., and Lai, M. M. (1999). A recombinant hepatitis C T. J., and Cameron, C. E. (1999). Production of “authentic” poliovirus RNA-dependent RNA polymerase capable of copying the virus RNA-dependent RNA polymerase (3Dpol) by ubiquitin- full-length viral RNA. J. Virol. 73(9), 7694–7702. protease-mediated cleavage in Escherichia coli. Protein Expr. 18. Paul, A. V., van Boom, J. H., Filippov, D., and Wimmer, E. (1998). Purif. 17, 128–138. Protein-primed RNA synthesis by purified poliovirus RNA poly- 9. Hansen, J. L., Long, A. M., and Schultz, S. C. (1997). Structure of the RNA-dependent RNA polymerase of poliovirus. Structure 5, merase. Nature 393, 280–284. 1109–1122. 19. Rice, C. M. (1996). Flaviviridae: The viruses and their replication. In 10. Kao, C. C., Yang, X., Kline, A., Wang, Q. M., Barket, D., and Heinz, “Virology” (B. N. Fields, D. M. Knipe, and P. M. Howley, Eds.), 3rd B. A. (2000). Template requirements for RNA synthesis by a ed., pp. 931–960. Raven Press, New York. recombinant hepatitis C virus RNA-dependent RNA polymerase. 20. Wimmer, E. (1982). Genome-linked proteins of viruses. Cell 28, J. Virol. 74(23), 11121–11128. 199–201. 11. Kim, M. J., Zhong, W., Hong, Z., and Kao, C. C. (2000). Template 21. Zhong, W., Ferrari, E., Lesburg, C. A., Maag, D., Ghosh, S. K., nucleotide moieties required for de novo initiation of RNA syn- Cameron, C. E., Lau, J. Y. N., and Hong, Z. (2000). Template/ thesis by a recombinant viral RNA-dependent RNA polymerase. primer requirements and single nucleotide incorporation by J. Virol. 74(22), 10312–10322. hepatitis C virus nonstructural protein 5B polymerase. J. Virol. 12. Lai, V. C. H., Kao, C. C., Ferrari, E., Park, J., Uss, A. S., Wright- 74(19), 9134–9143. Minogue, J., Hong, Z., and Lau, J. Y. N. (1999). Mutational anal- 22. Zhong, W., Uss, A. S., Ferrari, E., Lau, J. Y., and Hong, Z. (2000). De ysis of bovine viral diarrhea virus RNA-dependent RNA polymer- novo initiation of RNA synthesis by hepatitis C virus nonstruc- ase. J. Virol. 73, 10129–10136. tural protein 5B polymerase. J. Virol. 74(4), 2017–2022.