Substrate Complexes of Hepatitis C Virus RNA Polymerase (HC-J4): Structural Evidence for Nucleotide Import and De-Novo Initiation

doi:10.1016/S0022-2836(02)01439-0 J. Mol. Biol. (2003) 326, 1025–1035

Substrate Complexes of Hepatitis C Virus RNA Polymerase (HC-J4): Structural Evidence for Nucleotide Import and De-novo Initiation

Damien O’Farrell, Rachel Trowbridge, David Rowlands and Joachim Ja¨ger*

Astbury Centre of Structural Several crystal structures of the hepatitis C virus NS5B protein (genotype- Molecular Biology, School of 1b, strain J4) complexed with metal ions, single-stranded RNA or nucleo- Biochemistry and Molecular side-triphosphates have been determined. These complexes illustrate Biology, University of Leeds how conserved amino acid side-chains, together with essential structural Leeds LS2 9JT, UK features within the active site, control nucleotide binding and likely mediate de-novo initiation. The incoming nucleotide interacts with several basic residues from an extension on the NS5B ﬁngers domain, a b-hairpin from the NS5B thumb domain and the C-terminal arm. The modular, bi-partite ﬁngers domain carries a long binding groove which guides the template towards the catalytic site. The apo-polymerase structure provides unprecedented insights into potential non-nucleoside inhibitor binding sites located between palm and thumb near motif E, which is unique to RNA polymerases and reverse transcriptases. q 2003 Elsevier Science Ltd. All rights reserved Keywords: hepatitis C virus replication; RNA polymerase; de-novo priming; *Corresponding author crystal structure; substrate complexes

Introduction new treatments three essential viral enzymes, the NS5B polymerase and the NS3 protease–helicase Hepatitis C virus (HCV) is an important human have been selected as potential targets for antiviral pathogen affecting an estimated 200 million indi- therapy.6 The development of effective drugs viduals worldwide.1 The virus is the major cause directed against the reverse transcriptase of of nonA-nonB hepatitis and, despite an apparently human immunodeficiency virus type 1 (HIV-1 RT) active host immune response, establishes a persist- highlights the importance of polymerases as drug ent infection in 80% of the cases.2 Chronic infection targets7–11 and justifies the detailed study of this leads to an increased propensity for the develop- enzyme in HCV. ment of life threatening liver diseases such as The HCV genome was first cloned in 198912 and cirrhosis and hepatocellular carcinoma. Current sequence analysis showed that the virus is related therapies based on pegylated interferon and riba- to the family Flaviviridae. The positive sense RNA virin are expensive, are not well tolerated and are genome has a single open reading frame producing of limited value in the treatment of the common a polyprotein that is processed into three or four HCV genotypes 1a and 1b.3–5 In the search for structural and seven non-structural proteins called NS2, NS3A/B, NS4A/B and NS5A/B. Compara- Abbreviations used: 3Dpol, polymerase protein 3D; tive sequence analysis of NS5B and in vitro activity CNS, crystallography and NMR system; dNTP, studies suggested that it is a RNA dependent RNA deoxyribonucleoside triphosphate; DTT, dithiothreitol; polymerases (RdRp).13 HCV, hepatitis C virus; HIV-1, human immunodeficiency Various crystal forms of NS5B polymerase, all virus type 1; IPTG, isopropyl-D-thiogalactopyranoside; based on the same consensus sequence BK (HC- LB, Luria Bertani; MES, 2-(N-morpholino)-ethane BK, genotype 1b), have been reported recently.14 – 17 sulphonic acid; RDRP, RNA dependent RNA The protein described here differs in sequence polymerase; RT, reverse transcriptase; NNRTI, non- 18 nucleoside RT inhibitor; NS5B, non-structural protein 5B; (HC-J4 ), crystal form, local structure and, most rNTP, ribonucleoside triphosphate. significantly, in biochemical characteristics from E-mail address of the corresponding author: the previously reported BK-derived enzyme. In [email protected] the presence of substrates the HC-J4 RNA

Figure 1. (a) Ribbon diagram showing the overall structures of HC-J4; the domain boundaries and the corresponding colour scheme highlighting the fingers, palm and thumb sub-domains are detailed below. The colour coding is accord- ing to the sub-domain structure of NS5B: fingers in blue, palm in purple, thumb in green and C-terminal arm in gold. The assign- ments of sub-domains and structural motifs L1,L2 and b are in line with those designated in previously published structures.14,15 (b) Ribbon diagrams comparing the overall structures of HC-J4 (left) and HC- BK (right) RNA polymerases. The sequence differences between the two primary structures are as fol- lows: Ala25Pro, Gly46Ser, Val85Ile, Lys114Arg, His120Arg, Lys124Glu, Asn142Ser, Val185Ala, Gly198Lys, Asn213Cys, Asn231Ser, Lys254Arg, Ser300Thr, Ser335Ala, Val338Ala in the fingers and palm sub-domains and Val405Ile, Ser473Thr, Val499Thr, Arg510Lys, Lys523Arg, Lys531Arg, Arg544Gln and Met573- Pro in the thumb. Relevant side- chains have been highlighted on the HC-BK NS5B ribbon model (1QUV14). The diagrams in this figure were generated with Molscript 2.1 and Raster 3D.55,56 polymerase active site structure presented herein domain and two C-terminal regions of NS5B. The reveals interesting spatial constraints from which residues in this region are highly conserved and we have derived a model for de-novo initiation of the interactions appear to play an important role RNA synthesis. In vitro studies have suggested in the initiation step of viral replication by appro- that HCV polymerase is able to initiate replication priately positioning the first rNTP such that an in- de-novo without the requirement for protein or line attack on the alpha-phosphate group on the nucleic acid primers.19 – 22 This initiation mechan- next cognate nucleotide can occur. The refined ism has recently also been described in atomic HC-J4 NS5B structures presented herein provide detail for F6 RNA polymerase.23 Picornaviral 3D insights into potential non-nucleoside inhibitor polymerases,24 – 27 which are among the most binding sites as well as revealing a plausible thoroughly studied polymerases and reverse model for de-novo priming through involvement of transcriptases,28,29 for which full ternary complexes flexible motifs in the thumb and in the C-terminal are available,30 employ different strategies to arm. initiate replication. In structural terms, little is known about the template binding and the formation of the de-novo initiating Watson–Crick base-pair in flavivirus RNA polymerases. Using Results and Discussion our low salt crystal form, soaking experiments with different nucleoside triphosphates reveal that Crystallographic studies the ligands are bound by a network of interactions We have determined, to 2.0 A˚ resolution, the formed by side-chains protruding from the fingers structure of NS5B RdRp from the infectious clone Structural Evidence for De-novo Initiation of HC-J4 Polymerase 1027

Table 1. Summary of data collection and reﬁnement statistics No. of atoms of Average protein/ redundancy ligand/ ˚ b Soaking Resolution Rsym ,I/ Completeness of water Rmsd bonds (A)/ Rcryst / a c conditions range (%) sI . (%) measurements molecules angles (deg.) Rfree (%)

Apo N/A 25.0–2.0 10.3 9.1 98.6 5.4 8774/0/ 0.014/1.84 20.5/ 1014 24.6 UTP 12.5 mM UTP, 25.0–2.6 12.8 6.2 97.9 8.0 8774/58/ 0.013/1.87 17.3/

15 mM MnCl2 793 24.8 rU5 250 mMrU5, 25.0–2.9 13.5 4.5 95.2 8.0 8774/ 0.008/1.59 22.9/ 15 mM MnCl2 154/0 29.5 P P P P a l l Rsym ¼ Phkl j Ij;hkl 2kIhkllP= hkl jIj;hkl where Ij;hkl is the intensity of the jth reﬂection hkl and kIl is the average intensity. b l l Rcryst ¼ hkl Fobs 2 Fcalc = hklFobs: c Rfree calculated as for Rcryst but on the 10% of the data excluded from the reﬁnement calculation.

HC-J4 derived consensus sequence of HCV (Figure herein, the complex of F6 RNA polymerase with 1(a)). Additionally, we have soaked into the apo- DNA23 and molecular modelling studies using the polymerase crystals in the presence of manganese HIV-1 RT:DNA complex30 docked into the HCV a short single-stranded RNA oligomer (rU5). polymerase active site cleft together provide indirect Finally, HC-J4 NS5B complexes with rNTPs and evidence that this part of the fingers domain might manganese have been established. The structures function to recognise the HCV helicase and so con- of these complexes have been determined to 2.9 A˚ tribute to the formation of an active replicase and 2.5 A˚ resolution, respectively. Details of the complex. Presumably, specific binding of the NS3 crystallographic data for three HC-J4 NS5B struc- helicase is important32 for efficient separation of tures are collated in Table 1. The protein (NS5B double stranded regions in the HCV genome shortly D21) used in these studies contains a C-terminal before insertion of the template into the polymerase deletion of 21 amino acid residues and a C-terminal active site. hexa-histidine tag. All residues, excepting those in a The differences in primary structure between the “finger-tips” surface loop (149–151) and those near BK protein and the truncated bacterially expressed the C terminus 567–570 and the hexa-histidine tag, or the full-length baculovirus expressed HC-J4 are well ordered and show strong electron density polymerase also affect the overall enzymatic in the final difference Fourier maps. activity of full-length and C-terminally truncated versions (NS5B D21 and D55) of HCV RNA polymerase. Using a 407 nt HCV RNA fragment, corre- Primary structure and kinetic properties of sponding to the 30end of the positive strand HC-J4 NS5B genome, as a template, the activity of HC-J4 polymerase is maximal at 37 8C but rapidly declines at There are 24 differences in primary structure 42 8C(Figure 2(a)). At room temperature the between the consensus sequences of the NS5B pro- activity of HC-J4 NS5B is at least 25-fold lower teins of the J4 isolate and the BK isolate, which had than at 37 8C, while at 30 8C the activity is fourfold been previously used for structural studies. The lower than at 37 8C. In our hands, HC-BK NS5B positions of sequence differences (PFAM code D21 purified and analysed in parallel with HC-J4 PF00998†) are located on the protein surface away NS5B D21 shows a similar temperature profile from the active site and are clustered in four with the optimum at 37 8C(Figure 2(a)). Previously regions: the fingers domain, the back of the published studies on HC-BK NS5B, however, thumb, the “flap” region, and near the C terminus reported the optimum temperature for activity to (Figure 1(b)). The differences closest to the active be between 22 8C13,33 – 35 and 30–32 8C.21,36 This site map to positions 142 and 405, but both residues observed discrepancy in the optimum temperature aremorethan16A˚ away from the conserved cataly- for HC-BK NS5B protein from previously pub- tic residues in sequence motifs A and C.31 More lished work and from the studies presented here importantly, a cluster of six differences between the may in part be due to differences in the purifi- J4 and BK sequences are found on the fingers cation procedures as well as buffer composition of domain (positions 85, 114, 120, 124, 131, 142). This the RdRp assays. Both homopolymeric and virus cluster is located adjacent to an area of conserved derived RNA substrates were used to study the et al 16 residues identified by Bressanelli . Modelling divalent cation requirements of HC-J4 NS5B studies suggest that it would be topologically plaus- (Figure 2(b)). As for other DNA and RNA poly- ible that the region is involved in controlling access merases, magnesium and manganese ions support of template RNA to the active site (data not shown). HC-J4 RNA polymerase activity at concentrations The HC-J4 NS5B template–RNA complex presented of 12.5 mM and 2.5 mM, respectively. In contrast to data published on poliovirus 3Dpol,37 incorpor- † http://pfam.wustl.edu ation and catalytic turnover by HC-J4 NS5B has 1028 Structural Evidence for De-novo Initiation of HC-J4 Polymerase

The two molecules forming the HC-J4 NS5B asymmetric unit can be superimposed onto other high-resolution models (e.g. 1C2P15) with rms deviations in Ca positions of 0.549 A˚ and 0.556 A˚ , respectively. As expected, the palm domains show the best agreement among all HCV polymerase models. Alpha carbon atom positions in motif C (residues 287–348) can be superimposed with 0.346 A˚ rms distance. More significant variations are found between the models of HC-J4-D21 and HC-BK-D55 NS5B (1CSJ16), in which 55 amino acid residues have been deleted from the C terminus. Here, deviations in backbone atom positions of up to 1.4 A˚ are observed around the b-ribbon motif in the flap of the thumb domain. The movement of this region shows some internal flexing and does not appear to be concerted. The overall comparison of the two HCV polymerases (HC-J4-D21and HC-BK-D21), however, indicates that NS5B may have evolved a more rigid structure than other viral polymerases. In the absence of viral RNA template, HCV polymerase appears to adopt an energetically more favourable, closed conformation. The average temperature factors of all independently determined NS5B structures (either BK or J4 derived) are consistently lower than those found for many HIV-1 reverse Figure 2. (a) Temperature profile for HC-J4 and HC-BK transcriptase structures,11,41,42 poliovirus 3Dpol38 RNA polymerase activity assayed by TCA precipitation. and Escherichia coli pol I Klenow fragment.43 Inter- Optimum activity is achieved at 37 8C for both proteins, estingly, many bacterial and other viral poly- at least five degrees higher21,36 and more than 10 8C 13,33 – 35 merases have been found to display an unusual higher than previously reported for the BK derived degree of internal flexibility.38,44 protein, as well as at least 10 8C higher than the optimum reported for other HCV 1b derived NS5B proteins.19,22,50,57,58 (b) Metal ions that support RNA The fingers domain directs the RNA template 2þ 2þ polymerase activity. Only Mg and Mn have been into the active site cleft found to mediate catalytic turnover, whereas other divalent cations did not confer any activity. This is in In the HC-J4 NS5B/rU5 complex, the short RNA contrast to reports on poliovirus 3D RNA polymerase.37 oligomer makes numerous interactions with residues on the fingers domain and binds in a groove running from the N terminus towards the active not been observed in the presence of other divalent site cleft (Figure 3(a) and (b)). The groove, which metal ions such as Ca2þ,Fe2þ,Co2þ,Ni2þ and Cu2þ. in the HC-J4 apo-polymerase structure is filled with water molecules that presumably mimic the oxygen atoms in the phosphodiester backbone of Structural comparison of HC-J4 NS5B with the template, directs the single stranded RNA other RNA polymerases towards the active site cavity. The positions of the phosphate groups are in Following the determination of the structures of good agreement with models derived from the poliovirus 3D polymerase,38 HCV polymerase,14 – 17 superposition of HIV-1 RT/dsDNA complex onto F623 and calici virus RNA polymerase39 a detailed NS5B. The uracil bases are rotated towards the picture of enzyme function has begun to emerge, solvent, away from the binding groove, thereby together with the recognition of common structural making few interactions with functional groups themes within the core of the viral replication on the protein surface. The last base, however, machinery. All of these RNA polymerases contain interacts with several residues extending from the three major subdomains termed fingers, palm and fingertips, the b-flap and the C-terminal arm thumb, in analogy to a closed right hand. The over- (Figure 3(b)). In the absence of an incoming nucleo- all features of the J4 derived RNA polymerase are tide complementary to the template base the RNA in good agreement with these other RdRps. The overshoots its binding site predicted by the HIV-1 correlation coefficient of 0.68 in the molecular RT/dsDNA model. Consequently, it partly replacement procedure40 using a HC-BK polymer- overlaps with the binding site of the incoming ase model published by Ago and co-workers14 nucleotide. In contrast to the F6 RNA polymerase (1QUV) against HC-J4 apo-polymerase data initiation complexes, however, there is no supports this finding. specific binding pocket S to accommodate the Structural Evidence for De-novo Initiation of HC-J4 Polymerase 1029

terminal base of a template RNA. Our structure of HCV RNA polymerase suggests a slightly different mode of initiation (see below) and may not be fully compatible with the ratcheting mechanism put forward by Stuart and co-workers.23 The extent of structural overlap in the fingers domain in all crystallographically independent models of NS5B and the apparent lack of flexibility of this domain in the presence and absence of template RNA is unprecedented and atypical for RdRp or reverse transcriptase structures. The a-carbon atoms in the fingers domain can be superimposed with rms deviations of 0.491 A˚ or better. The apparent rigidity of the fingers domain is enhanced by a prominent structural feature common to all recently determined RdRp models. This conserved feature appears in the form of extensive loops on the fingers domains of the HCV and the F6bac- teriophage polymerases. Two long extensions on the fingers domain form numerous hydrophobic and hydrophilic interactions with the adjacent thumb domain. The surface areas buried within this region are 1360 A˚ 2 and 2940 A˚ 2 in HC-J4 and F6 RdRps, respectively. It is conceivable that, with the exception of the thumb domain, the enzyme undergoes very little conformational change during the structural step of the catalytic cycle. A rolling motion of helices in the fingers domain and a tightening of the active site around bound primer/template has been observed in a number of ternary complexes, for example polymerase beta and T7 DNA polymerase. In HC-J4 NS5B, however, a tightening of the fingers domain due to substrate binding is unlikely to occur as modelling studies suggest that the volume of the active site would decrease to the extent that double- stranded RNA could not be accommodated. Thus, a flexing of the fingers in HCV RNA polymerase may not be required for efficient turnover. These findings are supported by the results of soaking short RNA and DNA oligomers into pre-grown HC-J4 NS5B crystals. The fingers domain, which is rU5 apo Figure 3. (a) A sA weighted (Fobs 2 Fcalc) omit map not involved in any restricting lattice contacts, contoured at 3.0 rmsd with the refined coordinates of shows no major structural changes when com- the oligo (rU)5 HC-J4 NS5B complex superimposed. The pared to the 2.0 A˚ structure of apo HC-J4 NS5B. HC-J4 polymerase crystals were soaked with the RNA Furthermore, the catalytically relevant binding m oligomer at a concentration of 250 M for 30 min and mode of a single incoming NTP (see below) and then transferred to a cryoprotectant solution containing mother liquor, 25% glycerol, 20% (w/v) sucrose, 10% (w/v) xylitol, 250 mM oligo(U)5 and 15 mM MnCl2 for a further minute. The average temperature factor of the nucleotides (kBl ¼ 57 A˚ 2) is slightly above the average B The thumb domain has been removed for clarity. Using of the two NS5B molecules in the asymmetric unit indi- the coordinates of the 2.5 A˚ UTP complex (see Figure 4), cating that the RNA shows some movement within the the rNTP has been superimposed onto the RNA template binding cleft. (b) Ribbon diagram showing the fingers complex. The two ligands fit tightly in their respective and thumb domains of HC-J4 NS5B and indicating the access channels. Unexpectedly, the template appears to general location of the template binding site. The RNA overshoot the expected site that would allow the for- oligomer interacts directly with residues 14, 93, 95, 97, mation of a Watson–Crick base-pair and protrudes into 98, 139, 141, and 160. Note the close proximity between the rNTP binding pocket. Note, that the active site of the residues in the “b-flap” (thumb) and rU5 and rU4 in HC-J4 RNA polymerase does not contain a specificity the template strand. (c) A composite diagram showing a pocket S as seen in the F6 RNA polymerase structure23 surface representation of the HC-J4 RNA polymerase Clearly, there is little room to accommodate a second active site and the access channels for the template incoming nucleotide. The figure was generated with RNA (in atom colours) and incoming rNTP (in white). SPOCK.59 1030 Structural Evidence for De-novo Initiation of HC-J4 Polymerase

Figure 4. Stereo diagram showing electron density corresponding to UTP and manganese. The sA weighted UTP apo (Fobs 2 Fcalc) omit map is contoured at 3.0 rmsd. The ligand has been soaked into the nucleotide-binding pocket at a concentration of 12.5 mM for 5 min. The average temperature factor of the UTP is kBl ¼ 46 A˚ 2. Interestingly, all three conserved aspartates (Asp220, Asp318 and Asp319) interact with the metal ions. The g-phosphate forms two close interactions with guanidinium groups of Arg41 and Arg222. The base interacts with Ser556, Lys 141 and Arg158, a conserved residue in “sequence motif F”.15 A water molecule is bound 3.8 A˚ away from the metal ion and about 4.1 A˚ away from the a-phosphorous atom in line with the bridging phosphodiester oxygen and appears to mimic the attacking 30O2 of the primer terminus. This juxtaposition of functional groups and solvent molecules suggests that the UTP is bound in a productive manner. The close contacts between the uracil and the C-terminal residues and the unprecendented interactions of the g-phosphate corroborate the results obtained by Shim et al.,60 whereby the triphosphate is not strictly required for initiation of RNA synthesis. The ﬁgures were generated with SPOCK.59 the proximity of conserved residues Arg48, Lys51, ternary complex of HIV-1 RT.30 Soaks with dNTPs Arg158, Ser282, Thr283 and Asn291 in the ﬁngers and metal ions into apo-DNA polymerase crystals domain may indicate that the active site structure of E. coli Klenow Fragment, for example, produce is fully “assembled” and ready to carry out the complexes in which the nucleotide is bound at a phosphoryl transfer step. remote, non-catalytic site on the polymerase domain.45 By contrast, in HC-J4 NS5B, UTP forms numerous interactions with polar and charged Nucleotide import side-chains and with two metal ions bound to the

Soaking experiments using UTP and MnCl2 into catalytic aspartates (Figure 4). The a- and b-phos- pre-grown HC-J4 NS5B crystals have revealed, in phates are in direct contact with both manganese detail, the location of a single NTP molecule in the ions. Upon superposition of motifs A and C in polymerase active site and the interactions HIV-1 RT and HC-J4 NS5B, the carboxylate involved in productive NTP binding (Figure 4). residues as well as the a-phosphate overlap closely. The predominantly positively charged access The g-phosphate from the superimposed arrested tunnel is funnel shaped and narrows towards the ternary complex of HIV-1 RT closely overlaps the metal binding site. The primer grip and the active b-phosphate in the HC-J4 NS5B UTP complex. site b-turn formed by residues in motif C are at Interestingly, the g-phosphate of UTP points away the bottom end of the funnel. The structures of the from the two metal ions and, instead, one of the NS5B/rNTP complexes indicate how incoming g-oxygens forms a charged interaction with NH1 nucleoside triphosphates might access the metals of Arg48. Lys51 and Arg222 are less than 3.5 A˚ by travelling along the funnel and how, sub- away from O2g. Thus, the pattern of interactions sequently, pyrophosphate might exit the active site involving the b- and g-phosphates is different without necessitating large structural changes. In from those observed in other polymerase contrast to NTP complexes reported with HC-BK complexes.30,44,46,47 In the dsDNA complexes most D55 NS5B,17 only a single NTP binding site can be of the oxygens are in contact with either of the Mg observed in our crystal form. The C terminus and ions, thereby “curling” around both metals and the b-flap are in close proximity to the HC-J4 NTP compensating the negative charges. In HC-J4, both binding site. The removal of the C terminus in long protrusions (or “finger tips”) are populated HC-BK D55 RNA polymerase causes a shift in the with arginine and lysine residues. These side- positions of the b-flap backbone atoms of approxi- chains attract nucleoside triphosphates to the mately 2.0 A˚ . The increase in active site volume of active site and likely facilitate the expulsion of this NS5B variant affects the rNTP interaction pyrophosphates and, more importantly, the pattern and presumably allows the binding of an productive and accurate incorporation of correct additional triphosphate as described by Bressanelli nucleotides. and co-workers.17 Residual electron density in the vicinity of In the absence of template RNA, the UTP binds Ser367, the primer grip, adjacent to Asp319 and near the conserved sequence motifs A, C and E in the metal binding site has been interpreted as a a position similar to that observed in the arrested water molecule (Figure 4), which is 3.8 A˚ away Structural Evidence for De-novo Initiation of HC-J4 Polymerase 1031 from the metal ion and about 4.1 A˚ away from the secondary and tertiary structure elements in the a-phosphorous atom. This position closely palm and thumb domains supports the view that resembles the predicted position of the 30OH seen picornavirus, flavivirus and retrovirus poly- in the ternary complexes of T7 DNA polymerase47 merases may have evolved from a common ances- and HIV-1 RT.30 tor, a question that could not be answered by comparative sequence analyses.31 The fingers domains within these polymerases, however, differ Internal cavities significantly and highlight the need for adaptations Surface calculations using the high-resolution to cope with structurally diverse substrates and HC-J4 NS5B models and analyses of internal facilitate the different initiation mechanisms.49 volumes have revealed the presence of two notable This comparison also suggests certain confor- cavities in the domain interface between palm and mational changes that might take place in NS5B thumb. The cavities are located at the base of the upon binding RNA. The close interactions of thumb domain between the primer grip in motif E fingertips and thumb (via helices M, O, A and and the two beta strands of motif C (Figure 5(a) loop L2) might explain the slow on-rate observed and (b)). Metal A and the GDD polymerase motif for double-stranded RNA substrates.50 Further- are approximately 15 A˚ away from the cavity. The more, it is possible that a concerted movement of interior surface consists of predominantly hydro- the two domains necessary for the translocation of phobic amino acid side-chains (Figure 5(a) and the nascent RNA molecule during the catalytic (b)). cycle is restricted and occurs on a smaller scale The general location of the cavities (Figure 5(a) than that observed in HIV RT.30,41,44,51 The thumb/ and (b)), their shape and size (total volume of fingertips interaction could play an important role 320 A˚ 3 for both cavities) suggests that these site in maintaining the fingers domain in the closed could serve as binding pockets for NS5B inhibitors form in the absence of ligands. The thumb itself analogous to the non-competitive, non-nucleoside appears to be in a closed conformation and would inhibitors used against HIV-1 RT (NNRTI). Inter- have to open upon formation of the first two estingly, such a pocket had been identified pre- Watson–Crick base-pairs. The position of the viously in HIV-1 RT through a combination of divalent metal ions relative to the trio of catalytic screening with non-nucleoside inhibitors9 and aspartate residues and the single binding site X-ray crystallography.10,11 The three-dimensional observed for the incoming ribonucleotide are in structure of HIV-1 RT complexed with Nevirapine good overall agreement with the arrested ternary provided detailed insights into how NNRTIs open complex of HIV-1 RT. The active site cavity in the up the protein and insert into a hydrophobic HC-J4 or HC-BK polymerases (see Figure 3(c)) pocket.11,48 Similar to the NNRTI binding site, the appears smaller than in F6 RNA polymerase. The pocket described here for HC-J4 NS5B is located specific binding pocket (site S) described in F6 far enough away from the surface to exclude bulk polymerase23 is not present in HCV polymerases solvent from the hydrophobic pocket, yet the bind- and, thus, in the absence of rNTPs the terminal ing site is readily accessible for non-nucleoside nucleotide (T1 in F6 polymerase) reaches beyond drugs which then could prevent the closure of the site P approaching the catalytic aspartates. HCV thumb domain. NS5B may not be able to accommodate two ribo- This particular domain interface is unique to nucleoside triphosphates (see Figure 3(c)) as seen RNA polymerases and reverse transcriptases as in the initiation complex described by Butcher only these enzymes contain motif E (termed the et al.23 The RNA polymerase complexes and “primer group”), that is conserved in sequence models presented herein provide a basis for and structure. Such a structural motif is absent in designing more decisive experiments to examine DNA polymerases and, therefore, equivalent questions relating to initiation and elongation in cavities do not occur in the thumb/palm domain single-strand RNA viruses. interface of this class of enzymes. In RdRps and RTs, however, this region might act as a “cushion” to absorb motions of the thumb domain that presumably occur in the translocation step of the Materials and Methods polymerase catalytic cycle. Expression and purification of recombinant NS5B-D21 Conclusion Commercially available plasmids (pET23a, Novagen) The different structures of HC-J4 RNA polymer- were used to clone and express recombinant HC-J4 D ase complexed either with template or nucleoside NS5B- 21 (Accession number: AF054250) in E. coli BL21 (DE3). The truncated mutant of NS5B was subcloned triphosphates give important clues to the processes from HC-J4 cDNA, harbouring the NS5B gene, by PCR. of nucleotide import and substrate recognition, the The primers used 5B-S5 (50 CTA GCT AGC ATG TCC mode of initiation and the switch to elongation TAT ACG TGG ACA 30) and 5B-AS4 (30 AGA GCA CGG mode in this class of RNA replicating enzymes. GCT GGG GCG GAG CTC GCC 50) have a NheI site The overall structure and the conservation of and a XhoI site, respectively, which were used to ligate 1032 Structural Evidence for De-novo Initiation of HC-J4 Polymerase

(isoproyl-b-D-thiogalactopyranoside) and grown for 16–18 hours at 27 8C. Cells were harvested by centrifugation. The cell pellet was resuspended in solubilisation buffer (100 mM Tris

pH 8, 300 mM NaCl, 0.1% Triton X100, 10 mM MgCl2, 5mg/ml DNAse I (Sigma) and 1 £ complete proteinase inhibitor tablet (Roche Diagnostics) and sonicated on ice. The sample was centrifuged, the supernatant collected and filtered using a 0.22 mm filter and the pH adjusted to 8. The sample was applied to a XK16 column (Pharmacia) containing 12 ml of Ni-NTA superflow resin (Qiagen) pre-equilibrated with buffer A (500 mM NaCl, 20 mM imidazole, 50 mM NaH2PO4, pH 8). After loading the sample the column was washed with buffer A until the absorbance (OD280) returned to base level. A linear imidazole gradient between buffer A and buffer B (50 mM NaH2PO4, 500 mM NaCl, pH 8, 250 mM imidazole) was applied and HC-J4 NS5B eluted at ,130– 140 mM imidazole. Fractions were checked for purity by SDS-PAGE on a 12.5% acrylamide gel. Fractions of .95% purity were pooled and dialysed against storage buffer (100 mM Tris pH 8, 300 mM NaCl, 30% Glycerol, 5 mM DTT, and pH 8) overnight.

Polymerase activity assays Homopolymeric template/primer (polyC/oligoG) or heteropolymeric 407 nt 30 HCV RNA templates were used to determine the activity of the purified J4-NS5B- D21. The assay was carried out in a 25 ml reaction volume containing 20 mM Tris pH 7, 12 mM NaCl, Figure 5. (a) Well-defined sA weighted electron den- 10 mM MgCl2, 1 mM DTT, 0.25 mg Poly (C), 0.125 mg 0 32 sity near motifs C and E (“primer grip”) in HC-J4 NS5B oligo (G)12 or 50 ng of 3 HCV RNA, 2 mCi P–GTP, and contoured at 1.5 rmsd. Note the gap in electron density ,100 ng of purified protein (as calculated by Bradford between the predominantly hydrophobic residues assay using BSA as a standard). Reactions were carried (Arg200, Leu204, Phe203, Thr207, Trp208, Thr312, out for 2 hours at 37 8C. Reaction product was precipi- Met313, Leu314, Cys324, Ile323, Val322, Val321, Leu320, tated with 40 mg of Salmon sperm DNA and 1 ml of Leu360, Glu361, Ile363, Thr364, Ser365, Ser368, Asn369, 10% TCA for 1 hour on ice. Product was collected on and Val370) in the two motifs. The gap results in the for- GF/C filter papers (Whatman) and washed three times mation of a small cavity adjacent to the thumb domain. with 0.1 M sodium pyrophosphate in 0.1N HCl, followed (b) Internal cavities are found between motifs C and E by two washes with 95% EtOH. Filters were left to air in the active sites of HC-J4 NS5B. The location of these dry before counting dry in a scintillation counter. cavities is preserved in all HCV NS5B models and is essentially equivalent to the location of the non-nucleo- Crystallisation side inhibitor binding site first described in HIV-1 RT/ 10,11 Nevirapine complex. These findings support the HC-J4 NS5B protein was stored in 10 mM Tris, pH 7.5, notion that this conserved structural element (motif E) 600 mM sodium chloride, 10% (w/v) glycerol, 5 mM plays an important role in the structural step during the DTT. Crystals were grown at 18 8C using the microbatch polymerase cycle and, thus, presents a potential target method in microtitre plates. Protein was mixed in equal for antiviral chemotherapy. The figures were generated volume with the mother liquor (50 mM MES, pH 5.0, 59 with SPOCK. 20% PEG 4000, 10% glycerol and 5 mM DTT) and incubated at 18 8C. Small needle-shaped crystals appeared after 3–4 days and grow to 0.07 £ 0.07 £ 0.43 mm within ,2 weeks; occasionally, the gene into pET23a. The vector was then transformed crystals up to 0.15 £ 0.15 £ 0.83 mm were obtained. The into the E. coli expression cell line, BL21 (DE3). HC-J4 NS5B crystals belong to the orthorhombic space Starter cultures of 200 ml LB medium were prepared group P212121 with unit cell dimensions typically of ˚ ˚ ˚ from frozen glycerol stocks of BL21 (DE3) containing a ¼ 104 A, b ¼ 105 A, c ¼ 134 A. the recombinant plasmid, which were streaked out on LB agar plates containing 100 mg/ml Ampicillin. These Data collection, structure determination, modelling small cultures were grown to an OD ¼ 0.6 and cells 600 and refinement were collected by centrifugation. The supernatant was removed and the cell pellet was resuspended in 200 ml Crystals of D21 NS5B were flash frozen in liquid pro- of fresh LB (Amp 100 mg/ml) broth. 20 ml of the sub- pane or liquid nitrogen after soaking them at 20 8C for culture were added to each 2 l baffle flasks containing 1–10 min in cryo-protectant solution containing 300 mM 500 ml of LB broth (Amp 100 mg/ml). The cultures were sodium chloride pH 6.9, 25% (W/v) glycerol, 20% (w/v) grown to an OD600 ¼ 0.6 at 37 8C at which point they sucrose, 10% (w/v) xylitol. The nucleoside triphosphates were induced with a final concentration of 0.4 mM IPTG were soaked into HC-J4 polymerase crystals at 12.5 mM Structural Evidence for De-novo Initiation of HC-J4 Polymerase 1033 for 5 min and the crystals were transferred from the well 4. Kanazawa, Y., Hayashi, N., Mita, E., Li, T., Hagiwara, to cryo-protectant solution (see above) also containing H., Kasahara, A. et al. (1994). Influence of viral quasi- 12.5 mM rNTP and 15 mM MnCl2 for a further minute. species on effectiveness of interferon therapy in The U5 RNA oligomer was soaked into native crystals at chronic hepatitis C patients. Hepatology, 20, a concentration of 250 mM for 30 min. The crystals were 1121–1130. subsequently transferred to a cryo-protectant solution 5. Lindsay, K. L., Trepo, C., Heintges, T., Shiffman, also containing 250 mM oligo(U)5 and 15 mM MnCl2 for M. L., Gordon, S. C., Hoefs, J. C. et al. (2001). A ran- a further minute. The best freezing results were obtained domized, double-blind trial comparing pegylated when crystals were transferred in a single step into large interferon alfa-2b to interferon alfa-2b as initial treat- volumes of cryo-protectant (1 ml/crystal). Microdialysis ment for chronic hepatitis C. Hepatology, 34, 395–403. techniques or longer soaking times to exchange mother 6. Bartenschlager, R. (1997). Candidate targets for hepa- liquor with cryo-protectant were not successful. X-ray titis C virus-specific antiviral therapy. Intervirology, data were collected at 100 K. The crystals contain two 40, 378–393. molecules per asymmetric unit, resulting in a 7. Mitsuya, H., Yarchoan, R. & Broder, S. (1990). ˚ 3 52 VM ¼ 2.83 A /Da and a solvent content of about 55%. Molecular targets for AIDS therapy. Science, 249, Data were indexed, scaled and reduced using the pro- 1533–1544. grams DENZO and SCALEPACK.53 The three data sets 8. Larder, B. A. & Stammers, D. K. (1999). Closing in on used for structural analyses and refinement of native HIV drug resistance. Nature Struct. Biol. 6, 103–106. HC-J4 polymerase, rNTP and oligo(rU)5 NS5B com- 9. Merluzzi, V. J., Hargrave, K. D., Labadia, M., plexes are collated in Table 1. Grozinger, K., Skoog, M., Wu, J. C. et al. (1990). Inhi- Phases for the parent HC-J4 NS5B structure factor bition of HIV-1 replication by a non-nucleoside amplitudes were derived by molecular replacement reverse transcriptase inhibitor. Science, 250, (AMoRe40) using 1QUV as a trial model.14 The solution 1411–1413. was confirmed by the phased translation function 10. Kohlstaedt, L. A., Wang, J., Friedman, J. M., Rice, P. A. routines implemented in CNS54 using the data from a & Steitz, T. A. (1992). Crystal structure at 3.5 A˚ resol- soaking experiment with 15 mM manganese present in ution of HIV-1 reverse transcriptase complexed with the cryo-solution. Further improvement of the initial an inhibitor. Science, 256, 1783–1790. maps was achieved by rigid-body refinement and two- 11. Smerdon, S. J., Ja¨ger, J., Wang, J., Kohlstaedt, L. A., fold non-crystallographic density averaging using the Chirino, A. J., Friedman, J. M. et al. (1994). Structure program CNS. Maps for the HC-J4 apo-polymerase of the binding site for nonnucleoside inhibitors of were calculated with bulk solvent correction and sigma the reverse transcriptase of human immuno- weighting including all data between 25 A˚ and 1.96 A˚ .54 deficiency virus type 1. Proc. Natl Acad. Sci USA, 91, Initial electron density maps for the data derived from 3911–3915. rNTP soaks and the (rU)5 RNA complex were calculated 12. Choo, Q.-L., Kuo, G., Weiner, A. J., Overby, L. R., by truncating the low resolution terms at 6 A˚ . Bradley, D. W. & Houghton, M. (1989). Isolation of a cDNA clone derived from a blood-born Non-A, Non-B viral hepatitis genome. Science, 244, 359–362. 13. Behrens, S. E., Tomei, L. & De Francesco, R. (1996). Identification and properties of the RNA-dependent RNA polymerase of hepatitis C virus. EMBO J. 15, Acknowledgements 12–22. 14. Ago, H., Adachi, T., Yoshida, A., Yamamoto, M., We thank Janice Pata, Susan Firbank and David Habuka, N., Yatsunami, K. & Miyano, M. (1999). Sangar for critical reading of the manuscript. We Crystal structure of the RNA-dependent RNA poly- are grateful to Simon Phillips for his support and merase of hepatitis C virus. Struct. Fold. Des. 7, access to facilities. The clone pCV-J4L6S was kindly 1417–1426. provided by Jens Bukh and the National Institutes 15. Lesburg, C. A., Cable, M. B., Ferrari, E., Hong, Z., of Health. We gratefully acknowledge SRS Mannarino, A. F. & Weber, P. C. (1999). Crystal structure of the RNA-dependent RNA polymerase from Daresbury and EMBL Hamburg for beamline hepatitis C virus reveals a fully encircled active site. access and support during X-ray data collection. Nature Struct. Biol. 6, 937–943. D.O’F. was supported by a studentship award by 16. Bressanelli, S., Tomei, L., Roussel, A., Incitti, I., Vitale, Microsciences Ltd and the BBSRC. This research R. L., Mathieu, M. et al. (1999). Crystal structure of was funded in part by grants from Action Research the RNA-dependent RNA polymerase of hepatitis C S/P3206 and Yorkshire Cancer Research L253 to J.J. virus. Proc. Natl Acad. Sci. USA, 96, 13034–13039. 17. Bressanelli, S., Tomei, L., Rey, F. A. & De Francesco, R. (2002). Structural analysis of the hepatitis C virus References RNA polymerase in complex with ribonucleotides. J. Virol. 76, 3482–3492. 1. WHO (1999). Global surveillance and control of 18. Yanagi, M., St Claire, M., Shapiro, M., Emerson, S. U., hepatitis C. J. Viral Hepat. 6, 35–47. Purcell, R. H. & Bukh, J. (1998). Transcripts of a 2. Clarke, B. (1997). Molecular virology of hepatitis C chimeric cDNA clone of hepatitis C virus genotype virus. J. Gen. Virol. 78, 2397–2410. 1b are infectious in vivo. Virology, 244, 161–172. 3. Okada, S., Akahane, Y., Suzuki, H., Okamoto, H. & 19. Oh, J-W., Ito, T. & Lai, M. M. C. (1999). A recombi- Mishiro, S. (1992). The degree of variability in the nant hepatitis C virus RNA-dependent RNA poly- amino terminal region of the E2/NS1 protein of merase capable of copying the full-length viral hepatiis C virus correlates with the responsiveness RNA. J. Virol. 73, 7694–7702. to interferon therapy in viremic patients. Hepatology, 20. Zhong, W., Uss, A. S., Ferrari, E., Lau, J. Y. N. & 16, 619–624. Hong, Z. (2000). De novo initiation of RNA synthesis 1034 Structural Evidence for De-novo Initiation of HC-J4 Polymerase

by hepatitis C virus nonstructural protein 5B poly- virus NS5B protein: characterization of its RNA polymerase. J. Virol. 74, 2017–2022. merase activity and RNA binding. Hepatology, 29, 21. Luo, G., Hamatake, R. K., Mathis, D. M., Racela, J., 1227–1235. Rigat, K. L., Lemm, J. & Colonno, R. J. (2000). De 37. Arnold, J. J., Ghosh, S. K. & Cameron, C. E. (1999). novo initiation of RNA synthesis by the RNA-depen- Poliovirus RNA-dependent RNA polymerase dent RNA polymerase (NS5B) of hepatitis C virus. (3D(pol)). Divalent cation modulation of primer, tem- J. Virol. 74, 851–863. plate, and nucleotide selection. J. Biol. Chem. 274, 22. Sun, X.-L., Johnson, R. B., Hockinan, M. A. & Wang, 37060–37069. Q. M. (2000). De novo RNA synthesis catalyzed by 38. Hansen, J. L., Long, A. M. & Schultz, S. C. (1997). HCV RNA-dependent RNA polymerase. Biochem. Structure of the RNA-dependent RNA polymerase Biophys. Res. Commun. 268, 798–803. of poliovirus. Structure, 5, 1109–1122. 23. Butcher, S. J., Grimes, J. M., Makeyev, E. V., Bamford, 39. Ng, K. K. S., Cherney, M. M., Vazquez, A. L., Machin, D. H. & Stuart, D. I. (2001). A mechanism for initiat- A., Alonso, J. M. M., Parra, F. & James, M. N. G. ing RNA-dependent RNA polymerization. Nature, (2002). Crystal structures of active and inactive con- 410, 235–240. formations of a caliciviral RNA-dependent RNA 24. Lee, Y. F., Nomoto, A., Detjen, B. M. & Wimmer, E. polymerase. J. Biol. Chem. 277, 1381–1387. (1977). A protein covalently linked to poliovirus 40. Navaza, J. (2001). Implementation of molecular genome RNA. Proc. Natl Acad. Sci. USA, 74, 59–63. replacement in AMoRe. Acta. Crystallog. sect. D, 57, 25. Paul, A. V., Rieder, E., Kim, D. W., van Boom, J. H. & 1367–1372. Wimmer, E. (2000). Identification of an RNA hairpin 41. Rodgers, D. W., Gamblin, S. J., Harris, B. A., Ray, S., in poliovirus RNA that serves as the primary Culp, J. S., Hellmig, B. et al. (1995). The structure of template in the in vitro uridylylation of VPg. J. Virol. unliganded reverse transcriptase from the human 75, 10359–10370. immunodeficiency virus type 1. Proc. Natl Acad. Sci. 26. Gerber, K., Wimmer, E. & Paul, A. V. (2001). Bio- USA, 92, 1222–1226. chemical and genetic studies of the initation of 42. Esnouf, R., Ren, J., Ross, C., Jones, Y., Stammers, D. & human rhinovirus 2 RNA replication: purification Stuart, D. (1995). Mechanism of inhibition of HIV-1 and enzymatic analysis of the RNA-dependent RNA reverse transcriptase by non-nucleoside inhibitors. polymerase 3D(pol). J. Virol. 75, 10969–10978. Nature Struct. Biol. 2, 303–308. 27. Gerber, K., Wimmer, E. & Paul, A. V. (2001). Bio- 43. Ollis, D. L., Brick, P., Hamlin, R., Xuong, N. G. & chemical and genetic studies of the initation of Steitz, T. A. (1985). Structure of large fragment of human rhinovirus 2 RNA replication: identification Escherichia coli DNA polymerase I complexed with of a cis-replicating element in the coding sequence dTMP. Nature, 313, 762–766. of 2A(pro). J. Virol. 75, 10979–10990. 44. Ja¨ger, J., Smerdon, S. J., Wang, J., Boisvert, D. C. & 28. Gilboa, E., Mitra, S. W., Goff, S. & Baltimore, D. Steitz, T. A. (1994). Comparison of three different (1979). A detailed model of reverse transcription crystal forms shows HIV-1 reverse transcriptase dis- and tests of crucial aspects. Cell, 18, 93–100. plays an internal swivel motion. Structure, 2, 29. Goff, S. P. (1990). Retroviral reverse transcriptase: 869–876. synthesis, structure and function. J. Acquir. Immun. 45. Beese, L. S., Friedman, J. M. & Steitz, T. A. (1993). Defic. Syndr. 3, 817–831. Crystal structures of the Klenow fragment of DNA 30. Huang, H., Chopra, R., Verdine, G. L. & Harrison, polymerase I complexed with deoxynucleoside tri- S. C. (1998). Structure of a covalently trapped phosphate and pyrophosphate. Biochemistry, 32, catalytic complex of HIV-1 reverse transcriptase: 14095–14101. implications for drug resistance. Science, 282, 46. Sawaya, M. R., Prasad, R., Wilson, S. H., Kraut, J. & 1669–1675. Pelletier, H. (1997). Crystal structures of human 31. Koonin, E. V. (1991). The phylogeny of RNA-depen- DNA polymerase beta complexed with gapped and dent RNA polymerases of positive-strand RNA nicked DNA: evidence for an induced fit mechanism. viruses. J. Gen. Virol. 72, 2197–2206. Biochemistry, 36, 11205–11215. 32. Kolykhalov, A. A., Agapov, E. V., Blight, K. J., 47. Doublie´, S., Tabor, S., Long, A. M., Richardson, C. C. Mihalik, K., Feinstone, S. M. & Rice, C. M. (1997). & Ellenberger, T. (1998). Crystal structure of a bac- Transmission of hepatitis C by intrahepatic inocu- teriophage T7 DNA replication complex at 2.2 A˚ lation with transcribed RNA. Science, 277, 570–574. resolution. Nature, 391, 251–258. 33. Ferrari, E., Wright-Minogue, J., Fang, J. W., Baroudy, 48. Ren, J., Esnouf, R. M., Hopkins, A. L., Stuart, D. I. & B. M., Lau, J. Y. & Hong, Z. (1999). Characterization Stammers, D. K. (1999). Crystallographic analysis of of soluble hepatitis C virus RNA-dependent RNA the binding modes of thiazoloisoindolinone non- polymerase expressed in Escherichia coli. J. Virol. 73, nucleoside inhibitors to HIV-1 reverse transcriptase 1649–1654. and comparison with modeling studies. J. Med. 34. Carroll, S. S., Sardana, V., Yang, Z., Jacobs, A. R., Chem. 42, 3845–3851. Mizenko, C., Hall, D. et al. (2000). Only a small 49. Hobson, S. D., Rosenblum, E. S., Richards, O. C., fraction of purified hepatitis C virus RNA-dependent Richmond, K., Kirkegaard, K. & Schultz, S. C. RNA polymerase is catalytically competent: (2001). Oligomeric structures of poliovirus polymer- implications for viral replication and in vitro assays. ase are important for function. EMBO J. 20, Biochemistry, 39, 8243–8249. 1153–1163. 35. Uchiyama, Y., Huang, Y., Kanamori, H., Uchida, M., 50. Lohmann, V., Roos, A., Ko¨rner, F., Koch, J. O. & Doi, T., Takamizawa, A. et al. (2002). Measurement Bartenschlager, R. (1998). Biochemical and kinetic of HCV RdRp activity with C-terminal 21 aa trun- analyses of NS5B RNA-dependent RNA polymerase cated NS5b protein: optimization of assay conditions. of the hepatitis C virus. Virology, 249, 108–118. Hepatol. Res. 23, 90–97. 51. Kensch, O., Restle, T., Wo¨hrl, B. M., Goody, R. S. & 36. Ishii, K., Tanaka, Y., Yap, C.-C., Aizaki, H., Matsuura, Steinhoff, H. J. (2000). Temperature-dependent equi- Y. & Miyamura, T. (1999). Expression of hepatitis C librium between the open and closed conformation Structural Evidence for De-novo Initiation of HC-J4 Polymerase 1035

of the p66 subunit of HIV-1 reverse transcriptase 57. Lohmann, V., Ko¨rner, F., Herian, U. & Bartenschlager, revealed by site-directed spin labelling. J. Mol. Biol. R. (1997). Biochemical properties of hepatitis C virus 301, 1029–1039. NS5B RNA-dependent RNA polymerase and identi- 52. Matthews, B. W. (1968). Solvent content of protein ﬁcation of amino acid sequence motifs essential for crystals. J. Mol. Biol. 33, 491–497. enzymatic activity. J. Virol. 71, 8416–8428. 53. Otwinowski, Z. & Minor, W. (1997). Processing of 58. Yamashita, T., Kaneko, S., Shirota, Y., Qin, W., X-ray diffraction data collected in oscillation mode. Nomura, T., Kobayashi, K. & Murakami, S. (1998). Methods Enzymol. 276, 307–326. RNA-dependent RNA polymerase activity of the 54. Bru¨ nger, A. T., Adams, P. D., Clore, G. M., DeLano, soluble recombinant hepatitis C virus NS5B protein W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). truncated at the C-terminal region. J. Biol. Chem. 273, Crystallography & NMR system: a new software 15479–15486. suite for macromolecular structure determination. 59. Christopher, J. A. (1998). SPOCK: The Structural Acta. Crystallog., sect. D, 54, 905–921. Properties Observation and Calculation Kit (Program 55. Bacon, D. & Anderson, W. F. (1988). A fast algorithm Manual), The Center for Macromolecular Design, for rendering space-ﬁlling molecule pictures. J. Mol. Texas A&M University, College Station, TX. Graph. 6, 219–220. 60. Shim, J. H., Larson, G., Wu, Z. W. & Hong, Z. (2002). 56. Merritt, E. A. & Murphy, M. E. P. (1994). Raster3D Selection of 30-template bases and initiating nucleo- Version 2.0. A program for photorealistic molecular tides by hepatitis C virus NS5B RNA-dependent graphics. Acta. Crystallog., sect. D, 50, 869–873. RNA polymerase. J. Virol. 76, 7030–7039.

Edited by J. Karn

(Received 14 August 2002; received in revised form 6 December 2002; accepted 9 December 2002) 本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP 图书馆。图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具