Structural Basis for Resistance of the Genotype 2b Hepatitis C Virus NS5B Polymerase to Site A Non-Nucleoside Inhibitors

J. Mol. Biol. (2009) 390, 1048–1059

doi:10.1016/j.jmb.2009.06.012

Available online at www.sciencedirect.com

Structural Basis for Resistance of the Genotype 2b Hepatitis C Virus NS5B Polymerase to Site A Non-Nucleoside Inhibitors Edwin H. Rydberg 1 , Antonella Cellucci 1 , Linda Bartholomew 1 , Marco Mattu 1 , Gaetano Barbato 1 , Steven W. Ludmerer 2 , Donald J. Graham 2 , Sergio Altamura 1 , Giacomo Paonessa 1 , Raffaele De Francesco 1 , Giovanni Migliaccio 1 and Andrea Carfí 1 ⁎ 1

Istituto di Ricerca di Biologia Molecolare, P. Angeletti, Via Pontina Km 30600, I-00040 Pomezia, Rome, Italy 2

Department of Antiviral Research, Merck Research Laboratories, P.O. Box 4, Sumneytown Pike, West Point, PA 19486, USA Received 27 March 2009; received in revised form 29 May 2009; accepted 3 June 2009 Available online 6 June 2009

Hepatitis C virus (HCV) exists in six major genotypes. Compared with the 1b enzyme, genotype 2b HCV polymerase exhibits a more than 100-fold reduction in sensitivity to the indole-N-acetamide class of non-nucleoside inhibitors. These compounds have been shown to bind in a pocket occupied by helix A of the mobile Λ1 loop in the apoenzyme. The three-dimensional structure of the HCV polymerase from genotype 2b was determined to 1.9Å resolution and compared with the genotype 1b enzyme. This structural analysis suggests that genotypic variants result in a different shape of the inhibitor binding site. Mutants of the inhibitor binding pocket were generated in a 1b enzyme and evaluated for their binding affinity and sensitivity to inhibition by indole-N-acetamides. Most of the point mutants showed little variation in activity and IC50, with the exception of 15- and 7fold increases in IC50 for Leu392Ile and Val494Ala mutants (1b→2b), respectively. Furthermore, a 1b replicon with 20-fold resistance to this class of inhibitors was selected and shown to contain the Leu392Ile mutation. Chimeric enzymes, where the 2b fingertip Λ1 loop, pocket or both replaced the corresponding regions of the 1b enzyme, were also generated. The fingertip chimera retained 1b-like inhibitor binding affinity, whereas the other two chimeric constructs and the 2b enzyme displayed between 50and 100-fold reduction in binding affinity. Together, these data suggest that differences in the amino acid composition and shape of the indole-Nacetamide binding pocket are responsible for the resistance of the 2b polymerase to this class of inhibitors. © 2009 Elsevier Ltd. All rights reserved.

Edited by K. Morikawa

Keywords: hepatitis C virus; polymerase; resistance; non-nucleoside inhibitor binding sites

Introduction ⁎Corresponding author. E-mail address: [email protected]. Present address: R. De Francesco, INGM, Genomics and Molecular Biology Unit, Via Francesco Sforza 28, 20122 Milano, Italy. Abbreviations used: 1bdc21, HCV 1b-con1 NS5B ΔC21 containing a C-terminal His6 tag; 1bdc55, HCV 1b-bk NS5B ΔC55; 2adc21, HCV 2a NS5B ΔC21 containing a C-terminal His6 tag; 2bdc21, HCV 2b NS5B ΔC21 containing a C-terminal His6 tag; NNI, non-nucleoside inhibitor; HCV, hepatitis C virus; NS5B, non-structural protein 5B; RdRp, RNA-dependent RNA polymerase.

The hepatitis C virus (HCV) is thought to have infected 180 million people worldwide, an estimated 130 million of whom are at risk of developing chronic liver disease, cirrhosis and hepatocellular carcinoma. Three to four million people are believed to be newly infected each year, and it is estimated that 50%–80% of infections are asymptomatic.1 The currently approved standard of care for chronic HCV infection is based on pegylated interferon-α, either as monotherapy or in combination with ribavirin. However, this therapy is poorly tolerated and of limited efficacy.2

0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.

Structure of HCV Polymerase 2b

HCV exists in six major genotypes, varying by 30%–35% in sequence over the entire genome. In turn, each genotype can be divided in several subtypes, differing between 20% and 25%.3 In addition, the high rate of HCV mutation quickly generates a diverse population of quasi-species from a single subtype.4 Worldwide, genotypes 1 (a and b) and 2 (a and b) are the most prevalent forms, accounting for approximately 13 million cases of disease in the United States and Europe combined. HCV genotype is a major determinant of treatment outcome. HCV belongs to the Flaviviridae family of enveloped viruses; its positive-stranded RNA genome is 9.6 kb long and codes for a single open-reading frame flanked by conserved 3′- and 5′-untranslated regions. A combination of host- and virus-mediated proteolytic processing of the single polyprotein results in at least three structural proteins and seven non-structural proteins (reviewed by Reed and Rice5). Non-structural protein 5B (NS5B), the RNA-dependent RNA polymerase (RdRp) at the core of the HCV replicative complex, has been demonstrated to be crucial for viral infectivity and is a validated target for the development of therapeutics against HCV. The three-dimensional structures of C-terminally truncated forms of NS5B (ΔC21 and ΔC55) from genotypes 1b6–11 and 2a12 have been reported. They share the same general right-handed finger, thumb and palm domain organization and the same catalytic machinery with other single-subunit polynucleotide polymerases (Fig. 1). However, the HCV polymerase has an elongated loop (Λ1), which protrudes from the palm-distal region of the finger domain (the so-called fingertip), also observed in the RdRp's of other viruses, such as bacteriophage Φ6,13 bovine viral diarrhea virus14 and foot-and-mouth disease virus.15 Thus, the fingertip extends from the finger domain to contact the thumb domain, enclosing the active-site cleft and resulting in a relatively closed, spherical appearance for the enzyme. This active-site enclosure also suggests the need for a conformational change to affect the complete catalytic function.7,9 Screening of inhibitor libraries has resulted in the discovery of several classes of non-nucleoside inhibitors (NNIs). Although all these molecules inhibit the polymerase at a stage prior to elongation and non-competitively with RNA template and NTPs, at least three distinct binding sites (Fig. 1) have been identified:16 site A17,18 and site B10,19,20 are located on two opposite sides of helix Q and helix T on the surface of the thumb domain at 30 and 35 Å from the active site, respectively; site C is located in proximity of the enzyme active-site cavity.21,22 Of interest, structural and biochemical studies have shown that site A inhibitors17,18,23 bind to a pocket masked in the apoenzyme by the Λ1 loop helix A (Fig. 1). Further evidence for mobility of this loop was observed for NS5B from the 2a genotype, where both a “closed” form (similar to the apo-1b structure) and an “open” form where helix A has moved away from the thumb domain and adopted a β-hairpin-like structure were observed.12 Moreover,

1049

Fig. 1. Location of NNI binding sites on HCV NS5B ΔC55-1b polymerase. The three sites are localized in pockets formed by helices O, Q, S and T. Sites A and B are each located on the surface of the thumb domain. Site A is occupied by helix A, the tip of the fingertip Λ1 loop, in the inhibitor-free NS5B polymerase structures. Site C is located near the enzyme active site.

in the same study, a site B inhibitor was reported to bind only to the closed form of the enzyme and to cause a transition to the open form. These data are evidence for crucial interdomain communication within the polymerase that can be interfered at least by both site A and site B inhibitors. In addition to resistance mutations at each inhibitor binding site,17,24,25 it has been found that inherent genotypic differences are often responsible for dramatically different sensitivities to various inhibitors (reviewed by De Francesco & Carfi,16 Beaulieu & Tsantrizos,26 Li et al.27 and Ma et al.28). During the characterization of HCV polymerase NNIs, we have discovered that the genotype 2b polymerase is more than 100-fold less sensitive to inhibition by the indoleN-acetamides than is the genotype 1b enzyme (Table 1). To understand the molecular basis for this genotypic heterogeneity in NS5B inhibition, we have determined the three-dimensional structure of the C-terminally truncated (ΔC21) HCV 2b.2 NS5B.29 Furthermore, a series of 1b NS5B variants of the indoleN-acetamides binding site, site A, were produced, and their binding and inhibition by this class of molecules were assessed. Our structural, biochemical and binding data suggest that differences in the shape of the site A binding pocket are in a large part responsible for the resistance of the 2b polymerase to the indole-Nacetamide inhibitors. Furthermore, our data show that variations at positions 392 and 494 in genotype 2b NS5B have the most significant effect on the sensitivity of the HCV NS5B to this class of compounds.

Results and Discussion Protein expression and purification The genotype 2b ΔC21 HCV polymerase (2bdc21) was expressed and purified in soluble form with

Structure of HCV Polymerase 2b

1050 Table 1. Genotypic sensitivity to indole-N-acetamides

benzothiadiazines, 22 phenyl-phenoxy acrylic acids21,22,31 and the reversible covalent arylsulfonyl rhodanines.32 X-ray structures of 1bdc21 with inhibitors from each class have been reported.21,22,31–33 Most of the residues forming this pocket are invariant across HCV genotypes, with the exception of Met414 (Gln in genotype 2, Val in genotype 4a) and Tyr415 (the conservative Phe mutation is present in several genotypes, such as 1a/c and 6a). In addition, two other mutations are located in proximity of this site and may affect inhibitor binding. Cys316 and Ile447 in genotype 1b are mutated in Asn and Met, respectively, in genotype 2. The Cys316Asn mutation may partially affect the position of Arg200 side chain, whereas the larger side chain of Met447 in the β-flap may constrain the position of the adjacent Gln414 side chain, preventing optimal positioning of the inhibitor (Fig. 3). Previous studies have shown that the mutation Met414Thr is sufficient to confer resistance to the benzothiadiazines in both the isolated polymerase and the replicon system,34 demonstrating that the amino acid in this position plays an important role for binding of site C inhibitors. NNI site B

truncation of the 21-amino-acid C-terminal region, corresponding to the highly hydrophobic membrane anchor. The expression level was substantially lower than that of the corresponding NS5B ΔC21 enzyme from genotype 1b-bk (1bdc21), and the purified yield of 2bdc21 was 2.5 mg/L. In addition, upon concentration, it was apparent that the solubility of 2bdc21 was also significantly lower (ca 7 mg/mL) than that of the 1bdc21 (greater than 20 mg/mL). Nevertheless, the purified 2bdc21 protein could be crystallized, and its structure was solved by molecular replacement and refined to 1.9-Å resolution (Table 2). Overall structure of 2bdc21 The structure of 2bdc21 is highly similar to that of the HCV NS5B ΔC21 enzymes from genotypes 1b (1bdc21)6,7,9 and 2a (2adc21) in the “closed” form.12 If the palm domains are superimposed, the overall Cα r. m.s. deviations between 2bdc21 and 1bdc21 and between 2bdc12 and 2adc21 are 0.42 and 0.25 Å, respectively. The largest variations were observed in the N-terminal region of the fingertip Λ1 loop (residues 21–27) and in the C-terminal tail, suggesting some flexibility in these regions of the protein (Fig. 2a and b). However, despite shifts of 1.5–3.2 Å in the Cα atom positions of the C-terminal tail regions, the side chains of the hydrophobic L547W550F551 motif30 are generally spatially conserved (Fig. 2b). NNI site C The internal allosteric NNI pocket is located in the active-site cavity, 8.0 Å away from both the active site and helix O (Fig. 1). Several classes of compounds are known to bind this site, including

Site B is on the surface of the thumb and is formed by helices Q and T and the beginning of the Cterminal tail (Fig. 1). It is a long, narrow cleft 35 Å

Table 2. Crystallographic data and refinement statistics for HCV 2b NS5B 2b Space group Unit cell parameters a (Å) b (Å) c (Å) β (°) Data collection Beamline Wavelength (Å) Resolution (Å) High-resolution shell (Å) Total observations Unique reflections Completeness (%) Refinement Refinement resolution Unique reflections (working/test) Rworking/Rfree Number of atomsb r.m.s. deviation bond length (Å) r.m.s. deviation bond angle (Å) Mean B-factor (overall, Å2) Ramachandran analysis (%) Most favored Additionally favored a

C2 153.25 64.50 135.58 90.16 European Synchrotron Radiation Facility, ID14-2 0.933 1.90 1.97–1.90 328,676 104,470 97.3 (88.3)a 20–1.90 99,343/5127 16.6/21.0 8745/—/830 0.024 1.87 21.0 81.5 6.5

The values in parentheses refer to the highest-resolution shell. Non-hydrogen protein atoms/non-hydrogen inhibitor atoms/water molecules. b

Structure of HCV Polymerase 2b

1051 cleft. However, these mutations are in solventexposed positions of the structure and do not appear to result in important structural differences between the two NS5B pockets. Site A: the indole-N-acetamide NNI binding site Site A, where the NNIs with the related benzimidazole and indole-N-acetamide scaffolds bind, is located in a pocket in the thumb domain that in the apoenzyme is filled by helix A of the mobile Λ1 fingertip loop. The site consists of two components, the fingertip Λ1 loop and the pocket. The structure of the complex between compound 1 and the 1bdc5518 revealed that the inhibitor establishes several contacts with amino acids forming the site A pocket (Fig. 4). Resistant replicons have been isolated only with mutations of Pro495, Pro496 or, to a lesser extent, Val499 to Ala or Ser.17,34,35 Both Pro495 and Pro496 are strictly conserved across all genotypes. There are few mutations between genotypes 1b and 2b/a at site A (Fig. 4), in addition to the “nearby” mutation Val499Ala. Superposition of the 2bdc21 structure here reported with that of the 1bdc55 [Protein Data Bank (PDB) ID 2BRL] shows that these mutations significantly alter the shape of the inhibitor binding site (compare Fig. 5a and b). Ala393Thr and Leu392Ile would affect the region in proximity of the inhibitor phenyl ring. In the 1b enzyme, Ala393 provides a lipophilic environment for the para-position of the inhibitor phenyl moiety. This is more favorable than that generated by the polar Thr393 oxydryl atom, which would also be in

Fig. 2. Structural comparison between NS5B ΔC21-1b (PDB ID 2FVC) and ΔC21-2b. (a) The Cα traces of the ΔC21-1b and ΔC21-2b fingertip Λ1 loops are shown as coils in yellow and cyan, respectively. The side chains of some of the 1b and 2b amino acids are shown in stick representation. A transparent gray surface covers the ΔC21-2b polymerase thumb domain. (b) The Cα trace of the ΔC21-2b polymerase, except the C-terminal residues (540–561), is shown as a ribbon in gray. The C-termini of the two ΔC21-2b molecules in the asymmetric unit of the crystal and the C-terminal residues of the ΔC21-1b are shown in blue, red and yellow, respectively. Regions immediately up- and downstream the C-terminal L547W550F551 motif are poorly ordered in most of the HCV ΔC21 polymerase structures determined so far, but the motif itself is well defined and makes the same interactions with the thumb domain in all known structures. Figures 2, 3 and 5 were produced and rendered in PyMOL (DeLano Scientific).

from the catalytic site. Known inhibitors of this site are dihydropyranone, 20 thiophene carboxylic acids 10,12 and some cyclopentylethyl-phenoxy acrylic acids.31 In the 2b enzyme, there are four conservative variations with respect to genotype 1b (this site is identical in genotypes 2a and 2b) in the

Fig. 3. Structural comparison between ΔC21-1b (PDB ID 2FVC)22 and ΔC21-2b at the internal site C. The Cα trace is shown only for the 1b enzyme in a ribbon representation. The side chains of some of the key residues forming the inhibitor binding site C are shown in stick representation. The Cα position of F551, from the L547W550F551 motif, is indicated by a small black circle. Amino acids from the NS5B ΔC21-2b enzyme are shown in cyan, whereas the equivalent residues from the NS5B ΔC21-1b enzyme are in green. The ΔC21-2b polymerase is resistant to compounds that inhibit the 1b enzyme by binding to site C due to the Met414Gln mutation.

1052

Structure of HCV Polymerase 2b

Fig. 4. Schematic representation of the interaction between compound 1 and the ΔC55-1b polymerase (PDB ID 2BRL).18 Red, black and blue spheres represent oxygen, carbon and nitrogen atoms, respectively. The letter R on the inhibitor indicates the piperidine substituent (Table 1). Residues from the 1b polymerase involved in hydrophobic contacts with the inhibitor are represented by red semicircles. Blue semicircles indicate the location of amino acid variations between the 1b and 2b polymerases. The same mutations, with the exception of A393T, are also present in the genotype 2a polymerase. Hydrogen bonds are shown as dotted black lines. With the exception of the piperidine substituent, all the atoms of the inhibitor are within 4.0-Å distance from ΔC55-1b site A pocket residues.

closer contact to the phenyl ring para-C atom (Fig. 5c). Interestingly, genotypes 2a and 2b have most of the residues in the indole binding site in common, with the exception of position 393, which is Ala in 2a as it is in 1b (Fig. 4). This suggests that Ala393Thr mutation could be responsible for the difference in resistance to these compounds between genotypes 2a and 2b. Another change that likely would affect the interaction with the inhibitor is Leu392Ile. In the 2bdc21 structure, the β-methyl group of Ile392 is oriented toward the pocket, thus reducing the space available for the inhibitor (Fig. 5a). In particular, the Ile392 Cβ2 would likely form several unfavorable contacts with both the inhibitor phenyl and cyclohexyl moieties (Fig. 5a and c). Leu425Ile, at the “bottom” of the cyclohexyl pocket, where Leu29 interacts in the apo structure, has the most subtle effect of the five variants, producing a slightly deeper pocket in the 2bdc21 enzyme. Finally, Leu424Val and Val494Ala interact with the indole moiety, and these concomitant variations result in a deeper pocket in the 2b structure, significantly reducing the favorable van der Waals contacts between the indole and the protein (Fig. 5a–c). The influence of Val499 in resistance to the related benzimidazole inhibitors (see above) is likely played out through its role in orienting Arg503, which may interact with substituents on the carboxylic acid, and in part constraining the Pro493–Gly496 loop (Fig. 5d). Thus, the influence of the Ala variant on inhibitor binding would be expected to be variable and dependent on the specific indole substituent.

Enzymatic and kinetic evaluation of variants To clarify the contribution of the pocket mutations to resistance to site A NNI, we generated a series of variants with the 2b mutations introduced into the 1b-con1 scaffold. Point variants for each of the genotypic differences in the pocket (Fig. 4) and Val499Ala in addition to three chimeras, fingertip (Ala15Gly, Ala16Pro, Ser19Glu, Ala25Pro, Leu31Met, His33Phe, Met36Lys), pocket (Leu392Ile, Ala393Thr, Ile424Val, Leu425Ile, Val494Ala, Val499Ala) and full site A (fingertip + pocket), were produced. The fingertip Λ1 loop, although disordered in the structure of the 1bdc55 in complex with the indole-like inhibitor18 and thus unlikely to make any strong contacts with the inhibitor, could conceivably attenuate inhibitor accessibility to the binding site via altered flexibility in different genotypes. The high degree of sequence and structure conservation of the region surrounding site A suggests that the six variants in the pocket chimera should provide a strong approximation of the complete 2b pocket. The only other mutations in the region surrounding the site A pocket are the “second sphere” variants Ala400Val, Ala421Val and Cys488Thr. However, superposition of the 1bdc21 and 2bdc21 structures suggests that these mutations do not affect the position of the main and side chains of residues directly contributing to the inhibitor binding pocket. All point variants had enzyme activities within 3fold of the wild-type enzymes, with the exception of Ala393Thr (10-fold decrease in activity). On the

Structure of HCV Polymerase 2b

1053

Fig. 5. Differences in the shape of the site A pocket between the HCV NS5B 1b (PDB ID 2BRL) and 2b polymerases. (a) Surface of the ΔC55-1b site A pocket with 1b and 2b amino acid side chains underneath in yellow and cyan, respectively. The Cδ of Ile392 side chain and the hydroxyl group of Thr393 side chain in the 2b enzyme stick out from the 1b site A surface, thus reducing the space available for the inhibitor. (b) Surface of the NS5B ΔC21-2b site A pocket with 2b and 1b amino acid side chains underneath represented as in (a). The Cγ of the NS5B 1b Val494 side chain sticks out from the 2b site A surface, indicating a larger 2b site A pocket. In addition, two pockets can be observed in proximity of the Leu392 side chain, which are absent in the 1b polymerase site A surface, pointing to a deeper site A pocket in the 2b polymerase [see panel (a)]. (c) Compound 1 modeled in the ΔC21-2b site A pocket: amino acids that may affect the binding affinity of indole-N-acetamide inhibitors are shown in red. (d) Val499 in the 1b polymerase may constrain the position of Arg503 and the 493–496 loop. The Cα positions of Pro496 and Gly493 are indicated by black circles.

contrary, the chimera enzymes were more than 500fold less active (Fig. 6a). It is likely that the mutations introduced in these molecules perturbed the interaction between the fingertip loop and the thumb domain, resulting in catalytically defective

polymerases. The IC50 value for compound 1 was determined on each of the point variants and the 1bcon1 and 2b enzymes. The results, as shown in Fig. 6b, are normalized to the IC50 of 1b-con1. Three of the point variants caused an increase in resistance,

Structure of HCV Polymerase 2b

1054

Fig. 6. Characterization of site A 1b → 2b mutants. (a) Enzymatic activity of the ΔC21-1b site A point mutants and fingertip, thumb and entire site A chimera mutants. (b) Resistance of the 1b → 2b mutants to inhibition by compound 1. Red asterisks indicate the NS5B 1b polymerase point mutants for which compound 1 has an increased IC50 compared with the wild-type 1b enzyme.

and three others caused an increase in susceptibility. Leu392Ile and Val494Ala were 15- and 7-fold more resistant, respectively, while Val499Ala was approximately the same as wild-type enzyme (1.5-fold more resistant). Ile424Val and Leu425Ile were equivalent to wild type (1.5- and 2-fold more sensitive), whereas Ala393Thr was 4-fold more sensitive to inhibition. With the notable exception of the Ala393Thr mutants, these results are in agreement with the predictions from the structural analysis. Using the above data, we constructed a triple mutant incorporating the mutations that desensitized the enzyme to the inhibitor (Leu392Ile, Val494Ala and Val499Ala). Although the increase in IC50 for Val499Ala against compound 1 was marginal, the point variant was included as a significant contribution to resistance due to the observation of different effects on resistance against alternatively substituted indoles.17 The activity of the triple mutant was within 3-fold of the wild type, and the IC50 was 52-fold higher than that of 1b-con1 (Fig. 6a and b). Thus, these three mutations were sufficient to account for more than half of the resistance of the 2b enzyme. Inhibitor binding studies To gain further insights into the resistance of 2b NS5B to indole-N-acetamides, we characterized wild-type 1b and 2b enzymes as well as the chimeric 1b → 2b proteins for binding to compound 1 using surface plasmon resonance analysis. Determination of the kinetic binding constants kon and koff was not possible for some of the enzymes (1b, 2b, Leu392Ile

and Val494Ala) because the sensorgrams exhibited a multiexponential character, indicating that more than a single kinetic event was being monitored; for these mutants, only the equilibrium measurements were used to determine the Kd values. For Ala393Thr, Leu425Ile and Val499Ala, kinetic and equilibrium evaluations were performed, resulting in very similar Kd values (Table 3). The binding studies showed that the 2b enzyme has a much lower affinity for the inhibitor than the 1b enzyme. Importantly, this difference appeared to be due mainly to an exceedingly fast dissociation while the sensorgram association phase remains comparable (Fig. 7), demonstrating that resistance is not caused by an impaired accessibility to the inhibitor binding pocket. Similar behavior was observed for the “pocket” and “entire site A” chimera, which showed a fast dissociation as for wild-type 2b. Finally, the “fingertip” chimera bound the inhibitor as Table 3. Binding constants of compound 1 for the NS5B 1b, 2b, and 1b → 2b mutant enzymes Variant

kon (1/ms)

koff (1/s)

A393T L425I V499A 1b L392I V494A 2b “Fingertip” “Pocket” “Full site A”

2.51e + 04 1.27e + 04 9.81e + 03

2.69e − 03 5.86e − 03 6.78e − 03

KD (μM)

Half-life (min)

0.11 0.46 0.69 0.18 0.85 1.60 9.40 0.08 3.20 8.60

4.29 1.97 1.70

Structure of HCV Polymerase 2b

1055 Some of the mutations that were detected matched previously reported mutations, namely in V494, P495 and P496. However, one of the resistant clones harbored the new mutation Leu392Ile in NS5B. The role of this mutation in HCV 1b polymerase resistance to the indole-N-acetamide inhibitors was confirmed by a reverse genetic experiment where the single Leu392Ile mutation was introduced in the wild-type RNA replicon and transiently transfected in Huh-7 cells. This mutant replicon resulted in 20fold more resistance to compound 1 than the wildtype replicon clone.

Fig. 7. Indole-N-acetamide inhibitor binding to the 1b, 2b and 1b/2b polymerase chimeras measured by Biacore. Biacore sensorgrams show the response over time (resonance units) during the binding of compound 1 to wild-type NS5B ΔC21-1b (green), ΔC21-1b finger chimera (blue), ΔC21-1b thumb chimera (red), ΔC21-1b pocket chimera (dark gray) and wild-type ΔC21-2b (cyan) enzymes. Binding of compound 1 to the ΔC21-2b and ΔC21-1b pocket chimera enzymes shows a very fast koff. Instead, binding of compound 1 to the NS5B ΔC21-1b fingertip chimera and wild-type 1b polymerase shows very similar koff and kon values.

effectively as did the wild-type 1b enzyme (Fig. 7), confirming that the difference in the shape of the indole binding pocket rather than accessibility to the site or structure of the fingertip is responsible for the reduced affinity and consequently for the resistance of the 2b enzyme to this class of inhibitors. Analysis of the Kd values for the single-pocket mutants is in agreement with the structural analysis and the initial inhibition studies, with Val494Ala and Leu392Ile being the mutants most impaired in binding (Table 3). Selection of Leu392Ile-resistant 1b replicon Our biochemical data suggest that HCV 1b polymerase with mutations in the site A pocket might be resistant to the indole-N-acetamide NNIs. Thus, we used the replicon system to identify HCV 1b polymerase mutants conferring resistance to indole-N-acetamide compounds. Replicon-harboring cells were incubated with G418 and escalating concentrations of a cell-permeable indole-N-acetamide structural analogue of compound 1 (see also Materials and Methods). This protocol allowed the selection of about 30 drug-resistant clones that were then isolated, expanded and further characterized. Most of these clones exhibited reduced susceptibility to compound 1, with EC50 values that were at least three to four times higher than those of the wild-type replicon cells. The RNA of the clones with at least a four-fold shift in inhibitor susceptibility was isolated, and a PCR fragment corresponding to the NS5B coding region was sequenced. Sequence analyses showed that replicons isolated from resistant clones contained mutations that were not present in the replicon of cells not treated with the inhibitor.

Conclusions In this article, we report the X-ray crystal structure of the ΔC21 HCV polymerase from genotype 2b and that of a panel of 1b → 2b single mutants and chimera enzymes that were characterized for their susceptibility to inhibition and their binding affinities to an indole-N-acetamide NNI. The structure was found in a closed conformation, as previously observed for the genotype 1b polymerase and for the 2a polymerase. Slight changes were observed in the conformation of the fingertip and in the C-terminal region, confirming the flexibility of these regions of the molecule. No large rearrangement of α-helix A of the Λ1 loop, as observed in the 2a polymerase in crystal form 2, was observed although comparison of the two independent molecules present in the asymmetric unit of the crystal and that with other 1b polymerase structures suggest that this region is indeed conformationally flexible. The structure of the HCV 2b polymerase provides a plausible explanation for the observed resistance of the 2b enzyme to the indole-based NNIs. Comparison of the 1b and 2b indole-N-acetamide NNI binding pockets reveals that, together, the conservative mutations in the amino acids forming the binding site result in an altered pocket shape. In particular, mutations of Leu392Ile and Ala393Thr cause a narrowing of the pocket in proximity of the phenyl ring of the inhibitor, whereas mutations of Val494Ala and Leu425Ile result in a deeper pocket, thus decreasing the number of van der Waals contacts with the indole core of the inhibitor. The structural analysis is supported by the results of our mutagenesis and resistance-to-inhibition studies. Among all the mutants, Leu392Ile and Val494Ala NS5B enzymes were found to be the most resistant, 15- and 7-fold, respectively. Furthermore, we found that the Leu392Ile mutation in the 1b replicon confers 20-fold resistance to inhibition by an indole-N-acetamide NNI without affecting NS5B replication efficiency. Combination of the Leu392Ile and Val494Ala mutations with Val499Ala, a mutation that only marginally affected resistance in vitro, resulted in an NS5B enzyme with a 50-fold reduction in sensitivity to inhibition by the indole-N-acetamide NNIs. Thus, these three mutations together may account for a large fraction of the resistance of the HCV 2b polymerase to this class of molecules.

1056

Structure of HCV Polymerase 2b

Notably, Thr393Ala resulted in a 4-fold greater susceptibility to inhibition instead of being resistant, as suggested by the structural analysis. However, it has to be noted that this mutant was 10-fold less active in our activity assay. It is plausible that the increased susceptibility to the indole inhibitors of this mutant may be due to changes in its enzymatic mechanism. For instance, this mutation might disfavor the elongation step compared with the initiation step, resulting in an overall increased susceptibility to these inhibitors. The affinities of wild type, mutants and chimera enzymes for the inhibitor were then measured by surface plasmon resonance. This study confirmed that the decrease in inhibitor susceptibility was associated with a decrease in binding affinity. Analysis of the binding curves showed that, with the exception of Leu392Ile, the kon of binding lies within 1 order of magnitude for most of the mutants tested and for both the 1b and 2b enzymes, demonstrating that resistance is not caused by a decreased mobility of the fingertip α-helix A and/or a lower accessibility to the binding pocket. Rather, the binding studies show that dissociation (i.e., stability of the complex) is faster for the 2b enzyme and for the “pocket” and “entire site A” chimera polymerases, lowering the complex half-life (Fig. 7). In conclusion, the present study suggests that the conservative mutations in the NNI site A binding pocket are at the basis of the observed resistance of the 2b HCV polymerase to the indole-N-acetamide class of inhibitors. Although we identified residues in the inhibitor binding site that contribute to resistance, we cannot exclude that other mutations in other regions of the enzyme, farther away from the binding pocket, might also partially contribute to resistance (e.g., by affecting the mobility of the helices containing residues that participate in the formation of the pocket). Finally, it is important to note that the mutations we identified can arise in the 1b polymerase by simply one base change in the HCV 1b genome and therefore may easily emerge upon treatment with this class of inhibitors.

NaCl, 10% (v/v) glycerol, 10 mM β-mercaptoethanol and protease inhibitor cocktail (Roche)], the cells were lysed by a microfluidizer. The suspension was then stirred with DNase I (5 U/mL) and MgCl2 (10 mM) for 30 min at 4 °C before centrifuging at 18,000g for 45 min. The clarified solution was then subjected to Ni-nitrilotriacetic acid chromatography (5 mL; Qiagen). Protein was eluted with a step gradient of imidazole (20, 40, 200 and 500 mM), and fractions containing NS5B, as judged by SDS-PAGE and Western blotting, were pooled and dialyzed overnight against buffer B [10 mM Hepes, pH 8.0, 400 mM NaCl, 10% (v/v) glycerol, 5 mM DTT and 1 mM ethylenediaminetetraacetic acid]. The dialyzed protein was further subjected to heparin chromatography (5-mL HiTrap Heparin column, Amersham). The bound protein was eluted with a linear NaCl gradient (400 mM to 1.0 M NaCl over 5 column volumes), and the fractions containing 2bdc21 (as judged by SDS-PAGE and Western blotting) were pooled and concentrated before applying to a Superdex-200 gel-filtration column (Amersham) equilibrated in buffer B containing 600 mM NaCl. The peak fractions, now N95% pure as judged by SDS-PAGE, were pooled and concentrated to 6.5 mg/mL. Expression of 1Bcon1 NS5B ΔC21⋅His6 wild type and variants was performed in a 3-L fermentor (BioFlo-3000, New Brunswick Scientific) under batch conditions. An overnight culture was used to inoculate (1:100 v/v) a minimal medium [50 mM KPi, 25 mM (NH3)2SO4, 0.5 mM CaCl2, 1 mM MgSO4, 30 g/L of glucose, supplemented with trace metals]. The fermentor culture was grown at 30 °C to an optical density at 600 nm of 10 before expression was induced at 18 °C with 0.4 mM IPTG. The cells were further incubated for 23 h before harvesting. Purification was performed similarly to the 2bdc21 NS5B with the following exceptions: lysis was performed by cell disruptor after freeze-thawing, DNase I digestion was carried out for 2 h and no gel filtration was performed. Expression levels of the 1b-con1 variants varied greatly in yield; for example, the Ftip and Pocket chimeras yielded 2 mg of purified protein from a 3-L fermentor culture, while V494A had a 10-fold higher yield. Solubility of most variants was not a concern; however, during purification of the Ftip chimera, precipitate was obvious in the peak fractions from Ni-nitrilotriacetic acid chromatography. Nevertheless, no difference in aggregation was observed between the clarified Ftip chimera and wild-type 2b as judged by dynamic light scattering (data not shown).

Materials and Methods

RdRp assay and IC50 determinations

Cloning and mutagenesis The cloning of 1Bcon1 and 2B.2 NS5B ΔC21⋅His6-pT7 (2bdc21) has been described previously.29,35 All point variants and chimeras were generated using either QuikChange XLII or QuikChange Multi kits (Stratagene) as per the manufacturer's instructions. Protein expression and purification The 2bdc21 enzyme was expressed from Escherichia coli strain BL21 Codon Plus (Novagen). After growth to an optical density at 600 nm of 0.8, the cells were induced with 0.4 mM IPTG and incubated at 18 °C for 23 h before harvesting by centrifugation. Following resuspension of the pellet in buffer A [10 mM Tris–HCl, pH 8.0, 300 mM

The standard HCV RdRp assay used here has been previously reported.24,36 Briefly, between 2 and 100 nM enzyme (dependent on the activity of the variant) was preincubated with template/primer (polyA/oligoU) for 10 min at room temperature prior to addition of 3H-UTP. The 50-μL reaction was further incubated for 90 min at room temperature with shaking. An equal volume of 20% trichloroacetic acid/20 mM sodium pyrophosphate was added, followed by 5-min incubation on ice, to stop the reaction. The samples were then filtered using Unifilter GF/B plates (Packard), and the scintillation was read in a TOPCOUNT (Packard) after addition of 50 μL of Microscint (Packard). For inhibition assays, the same protocol was followed with the addition of a 10-min preincubation of enzyme with inhibitor (compound 1 in Table 1). Synthesis of compound 1 has been previously described.37 Data for IC50 curves were produced by plotting the percentage

Structure of HCV Polymerase 2b inhibition from serial dilutions of compound 1 against compound concentration and fitting to a four-parameter logistic equation using KaleidaGraph software. Protein crystallization and data collection The 2bdc21 protein was crystallized by the sitting-drop vapor-diffusion method combining 2 μL of enzyme solution (6.5 mg/mL) and 2 μL of well solution (100 mM Na citrate, pH 6.1, 200 mM NaCl, 10% polyethylene glycol 4000 and 5 mM DTT). Large, thin plates (300 μM × 200 μM × 20 μM) grew in 1 day. For data collection, crystals were transferred to cryo-protectant solution (mother liquor containing 20% glycerol) and were flash frozen in liquid nitrogen. X-ray diffraction data to 1.9-Å resolution were collected at the European Synchrotron Radiation Facility (Grenoble, France) at 100 K with an ADSC Q4 CCD detector. The crystals belonged to the monoclinic C2 space group with a = 153.25 Å, b = 64.49 Å, c = 135.58 Å, β = 90.16° and two molecules per asymmetric unit. DENZO and SCALEPACK38 were used to integrate and scale the data, respectively. The CCP4 suite of programs39 was used for further data processing. Final data collection statistics are summarized in Table 3. Structure determination and refinement The structure of 2bdc21 was solved by molecular replacement with MOLREP.40 A monomer of the 1b His6-HCV NS5B ΔC219 was used as search model (PDB ID 1C2P). All non-conserved residues and the residues of NNI binding site A were replaced by alanine and the entire Λ1 loop (amino acids 20–40) was omitted from the initial model to reduce model bias for initial electron density map calculation. Model building and correction were performed in O41 using σA-weighted electron density maps as calculated by Arp/wArp in Molrep mode.42 Positional refinement was performed in Crystallography & NMR System43 using a maximum-likelihood target.44 A bulk solvent correction and anisotropic B-factor tensor were applied throughout the refinement. Near the end of refinement, water molecules were added and individual B-factors were refined. The final model includes 1116 of the total 1152 residues in the asymmetric unit and 1300 water molecules. Weak or no electron density was observed for residues 149–153, 564–570 and the C-terminal His6 tag in both molecules of the asymmetric unit. No residue was located in disallowed regions of the Ramachandran plot as assessed with PROCHECK.45 Refinement statistics are presented in Table 3.

1057 negligible. All measurements were made at 25 °C in 10 mM Hepes, pH 7.4, containing 150 mM NaCl, 3.4 mM ethylenediaminetetraacetic acid and 0.005% surfactant P-20. Inhibitor concentrations used were between 10 nM and 10 μM, dependent on the polymerase mutant immobilized. Experiments used for kinetic constant evaluation were performed allowing 180 s of association-phase contact time and 700 s of dissociation-phase contact time, while experiments performed at equilibrium times were performed allowing 360 and 700 s, respectively. Dissociation–regeneration steps were obtained using short pulses of 15 s of NaCl at 500 mM. Parameter evaluation was obtained using the Biacore Biaevaluation software. Baseline correction was obtained using the double-reference method, whereas the reference-corrected curves were further subtracted by a blank injection of buffer. The mathematical model used in the fitting was the 1:1 model with no mass transfer; Kd values were obtained either by simultaneous fit of both kon and koff values or by equilibrium analysis. Selection of indole-N-acetamide-resistant replicons Replicon-harboring HBI10A cells were cultured and clones resistant to indole-N-acetamide were selected as described previously.46 An indole-N-acetamide inhibitor was dissolved in dimethyl sulfoxide and serially diluted such that the final dimethyl sulfoxide concentration was below 1%. HBI10A cells were plated in 15-cm tissue culture dishes at a density of 3 ×103/cm2 and cultured in the presence of 0.8 mg/mL of G418 and inhibitor at concentrations increasing from 0.2 to 1 μM. Approximately 15 days after the beginning of selection, small colonies of cells resistant to inhibitor and antibiotic became visible and were isolated. The sequence of HCV replicons present in drug-resistant clones was obtained following extraction of replicon RNA from resistant clones, reverse-transcriptase PCR and direct sequencing of PCR products.46 The effect of indole-N-acetamide on viral replication and the replication proficiency of the mutant replicons were estimated by monitoring expression of the NS3 protein by cell ELISA with the anti-NS3 Mab 10E5/24 as described previously.46 For reverse genetic experiments, the selected mutation was introduced in the wild-type replicon and RNA transfected by electroporation in 10AIFN cells as described previously.46 Cells were supplemented with the compounds between 1 and 4 h after transfection. Accession number Coordinates and the structure factors have been deposited in the PDB with accession number 3GSZ.

Surface plasmon resonance measurements Surface plasmon resonance analysis was performed using BIACORE 3000 optical biosensor with researchgrade CM5 sensor chips (Biacore AB, Uppsala, Sweden). The protein was immobilized using standard aminecoupling chemistry as described by the chip supplier (Biacore, Inc.). The protein at a concentration of 50 nM in 10 mM 4-morpholineethanesulfonic acid, pH 6.4, was injected for 7 min, resulting in immobilized densities averaging 1000 ± 300 RU (resonance units), which was sufficient to bind a range of 10–100 RU of inhibitors. In all cases, the surfaces were blocked with a 7-min injection of 1.0 M ethanolamine, pH 8.0. Flow rates were 40 μL/min during the association–dissociation phase. For each mutant, tests were performed to verify that mass transfer was

Acknowledgements We thank Stefania Di Marco, Licia Tomei and Matthew Bottomley for discussions and the staff at beamline ID14-2 of the European Synchrotron Radiation Facility for help during data collection. This paper is dedicated to the memory of our beloved colleague and friend Giovanni Migliaccio. Dedication. This paper is dedicated to the memory of our beloved colleague and friend Giovanni Migliaccio.

1058

References 1. WHO (January 2006, posting date). Initiative for Vaccine Research (IVR): hepatitis C [Online]. http://www.who. int/vaccine_research/diseases/viral_cancers/en/ index2.html. 2. Feld, J. J. & Hoofnagle, J. H. (2005). Mechanism of action of interferon and ribavirin in treatment of hepatitis C. Nature, 436, 967–972. 3. Simmonds, P. (2004). Genetic diversity and evolution of hepatitis C virus—15 years on. J. Gen. Virol. 85, 3173–3188. 4. Bowen, D. G. & Walker, C. M. (2005). The origin of quasispecies: cause or consequence of chronic hepatitis C viral infection? J. Hepatol. 42, 408–417. 5. Reed, K. E. & Rice, C. M. (2000). Overview of hepatitis C virus genome structure, polyprotein processing, and protein properties. Curr. Top. Microbiol. Immunol. 242, 55–84. 6. Ago, H., Adachi, T., Yoshida, A., Yamamoto, M., Habuka, N., Yatsunami, K. & Miyano, M. (1999). Crystal structure of the RNA-dependent RNA polymerase of hepatitis C virus. Structure, 7, 1417–1426. 7. Bressanelli, S., Tomei, L., Roussel, A., Incitti, I., Vitale, R. L., Mathieu, M. et al. (1999). Crystal structure of the RNA-dependent RNA polymerase of hepatitis C virus. Proc. Natl Acad. Sci. USA, 96, 13034–13039. 8. Bressanelli, S., Tomei, L., Rey, F. A. & De Francesco, R. (2002). Structural analysis of the hepatitis C virus RNA polymerase in complex with ribonucleotides. J. Virol. 76, 3482–3492. 9. Lesburg, C. A., Cable, M. B., Ferrari, E., Hong, Z., Mannarino, A. F. & Weber, P. C. (1999). Crystal structure of the RNA-dependent RNA polymerase from hepatitis C virus reveals a fully encircled active site. Nat. Struct. Biol. 6, 937–943. 10. Wang, M., Ng, K. K., Cherney, M. M., Chan, L., Yannopoulos, C. G., Bedard, J. et al. (2003). Nonnucleoside analogue inhibitors bind to an allosteric site on HCV NS5B polymerase. Crystal structures and mechanism of inhibition. J. Biol. Chem. 278, 9489–9495. 11. O'Farrell, D., Trowbridge, R., Rowlands, D. & Jager, J. (2003). Substrate complexes of hepatitis C virus RNA polymerase (HC-J4): structural evidence for nucleotide import and de-novo initiation. J. Mol. Biol. 326, 1025–1035. 12. Biswal, B. K., Cherney, M. M., Wang, M., Chan, L., Yannopoulos, C. G., Bilimoria, D. et al. (2005). Crystal structures of the RNA dependent RNA polymerase genotype 2a of hepatitis C virus reveal two conformations and suggest mechanisms of inhibition by nonnucleoside inhibitors. J. Biol. Chem. 280, 18202–18210. 13. Butcher, S. J., Grimes, J. M., Makeyev, E. V., Bamford, D. H. & Stuart, D. I. (2001). A mechanism for initiating RNA-dependent RNA polymerization. Nature, 410, 235–240. 14. Choi, K. H., Groarke, J. M., Young, D. C., Kuhn, R. J., Smith, J. L., Pevear, D. C. & Rossmann, M. G. (2004). The structure of the RNA-dependent RNA polymerase from bovine viral diarrhea virus establishes the role of GTP in de novo initiation. Proc. Natl Acad. Sci. USA, 101, 4425–4430. 15. Ferrer-Orta, C., Arias, A., Perez-Luque, R., Escarmis, C., Domingo, E. & Verdaguer, N. (2004). Structure of foot-and-mouth disease virus RNA-dependent RNA polymerase and its complex with a template–primer RNA. J. Biol. Chem. 279, 47212–47221.

Structure of HCV Polymerase 2b 16. De Francesco, R. & Carfi, A. (2007). Advances in the development of new therapeutic agents targeting the NS3-4A serine protease or the NS5B RNA-dependent RNA polymerase of the hepatitis C virus. Adv. Drug Delivery Rev. 59, 1242–1262. 17. Kukolj, G., McGibbon, G. A., McKercher, G., Marquis, M., Lefebvre, S., Thauvette, L. et al. (2005). Binding site characterization and resistance to a class of nonnucleoside inhibitors of the hepatitis C virus NS5B polymerase. J. Biol. Chem. 280, 39260–39267. 18. Di Marco, S., Volpari, C., Tomei, L., Altamura, S., Harper, S., Narjes, F. et al. (2005). Interdomain communication in hepatitis C virus polymerase abolished by small molecule inhibitors bound to a novel allosteric site. J. Biol. Chem. 280, 29765–29770. 19. Biswal, B. K., Wang, M., Cherney, M. M., Chan, L., Yannopoulos, C. G., Bilimoria, D. et al. (2006). Nonnucleoside inhibitors binding to hepatitis C virus NS5B polymerase reveal a novel mechanism of inhibition. J. Mol. Biol. 361, 33–45. 20. Love, R. A., Parge, H. E., Yu, X., Hickey, M. J., Diehl, W., Gao, J. et al. (2003). Crystallographic identification of a noncompetitive inhibitor binding site on the hepatitis C virus NS5B RNA polymerase enzyme. J. Virol. 77, 7575–7581. 21. Pfefferkorn, J. A., Greene, M. L., Nugent, R. A., Gross, R. J., Mitchell, M. A., Finzel, B. C. et al. (2005). Inhibitors of HCV NS5B polymerase: Part 1. Evaluation of the southern region of (2Z)-2-(benzoylamino)3-(5-phenyl-2-furyl)acrylic acid. Bioorg. Med. Chem. Lett. 15, 2481–2486. 22. Tedesco, R., Shaw, A. N., Bambal, R., Chai, D., Concha, N. O., Darcy, M. G. et al. (2006). 3-(1,1dioxo-2H-(1,2,4)-benzothiadiazin-3-yl)-4-hydroxy-2 (1H)-quinolinones, potent inhibitors of hepatitis C virus RNA-dependent RNA polymerase. J. Med. Chem. 49, 971–983. 23. Harper, S., Pacini, B., Avolio, S., Di Filippo, M., Migliaccio, G., Laufer, R. et al. (2005). Development and preliminary optimization of indole-N-acetamide inhibitors of hepatitis C virus NS5B polymerase. J. Med. Chem. 48, 1314–1317. 24. Tomei, L., Altamura, S., Bartholomew, L., Biroccio, A., Ceccacci, A., Pacini, L. et al. (2003). Mechanism of action and antiviral activity of benzimidazolebased allosteric inhibitors of the hepatitis C virus RNA-dependent RNA polymerase. J. Virol. 77, 13225–13231. 25. Tomei, L., Altamura, S., Bartholomew, L., Bisbocci, M., Bailey, C., Bosserman, M. et al. (2004). Characterization of the inhibition of hepatitis C virus RNA replication by nonnucleosides. J. Virol. 78, 938–946. 26. Beaulieu, P. L. & Tsantrizos, Y. S. (2004). Inhibitors of the HCV NS5B polymerase: new hope for the treatment of hepatitis C infections. Curr. Opin. Invest. Drugs, 5, 838–850. 27. Li, H., Tatlock, J., Linton, A., Gonzalez, J., Borchardt, A., Dragovich, P. et al. (2006). Identification and structure-based optimization of novel dihydropyrones as potent HCV RNA polymerase inhibitors. Bioorg. Med. Chem. Lett. 16, 4834–4838. 28. Ma, H., Leveque, V., De Witte, A., Li, W., Hendricks, T., Clausen, S. M. et al. (2005). Inhibition of native hepatitis C virus replicase by nucleotide and nonnucleoside inhibitors. Virology, 332, 8–15. 29. Graham, D. J., Stahlhut, M., Flores, O., Olsen, D. B., Hazuda, D. J., Lafemina, R. L. & Ludmerer, S. W. (2006). A genotype 2b NS5B polymerase with novel

Structure of HCV Polymerase 2b

30.

31.

32.

33.

34.

35.

36.

37.

substitutions supports replication of a chimeric HCV 1b:2b replicon containing a genotype 1b NS3-5A background. Antiviral Res. 69, 24–30. Adachi, T., Ago, H., Habuka, N., Okuda, K., Komatsu, M., Ikeda, S. & Yatsunami, K. (2002). The essential role of C-terminal residues in regulating the activity of hepatitis C virus RNA-dependent RNA polymerase. Biochim. Biophys. Acta, 1601, 38–48. Pfefferkorn, J. A., Nugent, R., Gross, R. J., Greene, M., Mitchell, M. A., Reding, M. T. et al. (2005). Inhibitors of HCV NS5B polymerase: Part 2. Evaluation of the northern region of (2Z)-2-benzoylamino-3-(4-phenoxy-phenyl)-acrylic acid. Bioorg. Med. Chem. Lett. 15, 2812–2818. Powers, J. P., Piper, D. E., Li, Y., Mayorga, V., Anzola, J., Chen, J. M. et al. (2006). SAR and mode of action of novel non-nucleoside inhibitors of hepatitis C NS5b RNA polymerase. J. Med. Chem. 49, 1034–1046. Lee, G., Piper, D. E., Wang, Z., Anzola, J., Powers, J., Walker, N. & Li, Y. (2006). Novel inhibitors of hepatitis C virus RNA-dependent RNA polymerases. J. Mol. Biol. 357, 1051–1057. Nguyen, T. T., Gates, A. T., Gutshall, L. L., Johnston, V. K., Gu, B., Duffy, K. J. et al. (2003). Resistance profile of a hepatitis C virus RNA-dependent RNA polymerase benzothiadiazine inhibitor. Antimicrob. Agents Chemother. 47, 3525–3530. Ludmerer, S. W., Graham, D. J., Boots, E., Murray, E. M., Simcoe, A., Markel, E. J. et al. (2005). Replication fitness and NS5B drug sensitivity of diverse hepatitis C virus isolates characterized by using a transient replication assay. Antimicrob. Agents Chemother. 49, 2059–2069. Tomei, L., Vitale, R. L., Incitti, I., Serafini, S., Altamura, S., Vitelli, A. & De Francesco, R. (2000). Biochemical characterization of a hepatitis C virus RNA-dependent RNA polymerase mutant lacking the C-terminal hydrophobic sequence. J. Gen. Virol. 81, 759–767. Harper, S., Avolio, S., Pacini, B., Di Filippo, M., Altamura, S., Tomei, L. et al. (2005). Potent inhibitors of subgenomic hepatitis C virus RNA replication

1059

38. 39.

40. 41.

42.

43.

44.

45.

46.

through optimization of indole-N-acetamide allosteric inhibitors of the viral NS5B polymerase. J. Med. Chem. 48, 4547–4557. Otwinowski, Z. & Minor, W. (1997). Processing of Xray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. Collaborative Computational Project No. 4. (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 50, 760–763. Vagin, A. & Teplyakov, A. (1997). MOLREP: an automated program for molecular replacement. J. Appl. Crystallogr. 30, 1022–1025. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr., Sect. A: Found. Crystallogr. 47, 110–119. Perrakis, A., Morris, R. & Lamzin, V. S. (1999). Automated protein model building combined with iterative structure refinement. Nat. Struct. Biol. 6, 458–463. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography & NMR System: a new software suite for macromolecular structure determination. Acta Crystallogr., Sect. D: Biol. Crystallogr. 54, 905–921. Brünger, A. T., Adams, P. D. & Rice, L. M. (1998). Recent developments for the efficient crystallographic refinement of macromolecular structures. Curr. Opin. Struct. Biol. 8, 606–611. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291. Trozzi, C., Bartholomew, L., Ceccacci, A., Biasiol, G., Pacini, L., Altamura, S. et al. (2003). In vitro selection and characterization of hepatitis C virus serine protease variants resistant to an active-site peptide inhibitor. J. Virol. 77, 3669–3679.