Roles of the AX₄GKS and Arginine-Rich Motifs of Hepatitis C Virus RNA Helicase in ATP- and Viral RNA-Binding Activity

Abstract

The nonstructural protein 3 (NS3) of hepatitis C virus (HCV) possesses protease, nucleoside triphosphatase, and helicase activities. Although the enzymatic activities have been extensively studied, the ATP- and RNA-binding domains of the NS3 helicase are not well-characterized. In this study, NS3 proteins with point mutations in the conserved helicase motifs were expressed in Escherichia coli, purified, and analyzed for their effects on ATP binding, RNA binding, ATP hydrolysis, and RNA unwinding. UV cross-linking experiments indicate that the lysine residue in the AX₄GKS motif is directly involved in ATP binding, whereas the NS3(GR1490DT) mutant in which the arginine-rich motif (1486-QRRGRTGR-1493) was changed to QRRDTTGR bound ATP as well as the wild type. The binding activity of HCV NS3 helicase to the viral RNA was drastically reduced with the mutation at Arg1488 (R1488A) and was also affected by the K1236E substitution in the AX₄GKS motif and the R1490A and GR1490DT mutations in the arginine-rich motif. Previously, Arg1490 was suggested, based on the crystal structure of an NS3-deoxyuridine octamer complex, to directly interact with the γ-phosphate group of ATP. Nevertheless, our functional analysis demonstrated the critical roles of Arg1490 in binding to the viral RNA, ATP hydrolysis, and RNA unwinding, but not in ATP binding.

Hepatitis C virus (HCV) is the major causative agent of posttransfusion and sporadic non-A, non-B hepatitis (7, 27). Patients with HCV infection often develop chronic hepatitis that leads to cirrhosis and hepatocellular carcinoma (1, 9, 36). HCV belongs to the family Flaviviridae whose members are enveloped viruses containing a single-stranded positive-sense RNA genome. The HCV genome bears a single open reading frame that encodes a polyprotein encompassing about 3,000 amino acid residues (8, 23, 42). Proteolytic cleavage of the polyprotein by host and viral proteases yields at least 10 products as NH₂-C-E1-E2-p7-NS2-NS3-NS4A-NS4B-NS5A-NS5B-COOH (15, 16, 19, 21, 29, 38). HCV nonstructural protein 3(NS3) is a multifunctional protein that possesses three known enzymatic activities in two independent domains. The N-terminal one-third domain contains a serine protease activity. The NS3 protease and the viral NS2 and NS4A proteins are important for the processing of the HCV polyprotein to generate nonstructural proteins (2, 10, 14, 15, 20, 43). The C-terminal two-thirds domain of the NS3 protein contains several conserved sequences shared by members of the DEAD box family of RNA helicases (12, 13, 37). These include AX₄GKS Walker A nucleotide binding motif (motif I), DECH Walker B nucleoside triphosphate (NTP) binding-hydrolysis motif (motif II), TAT RNA unwinding motif (motif III), and a QRRGRTGR arginine-rich motif (motif VI) thought to be involved in RNA binding. NTPase and RNA helicase activities have been demonstrated for both the NS3 helicase domain (24, 40, 41) and the full-length NS3 protein (11, 18). The kinetics of the ATP hydrolysis and duplex unwinding have been analyzed (32–34). In addition, crystal structures of the NS3 helicase have revealed potential amino acid residues that interact with ATP and RNA substrates (6, 26, 47). However, RNA substrates were located at different clefts in the NS3 structures with or without a bound deoxyuridine octamer. Cocrystal structure of the NS3 helicase and ATP has not been determined, and whether the arginine-rich motif of the NS3 is involved in ATP or RNA binding remained to be functionally determined. In this study, we demonstrated that the lysine residue in the AX₄GKS motif is directly involved in ATP binding and that Arg1488 in the arginine-rich motif (1486-QRRGRTGR-1493) is important for RNA-binding activity. Arg1490 is critical for HCV NS3 protein in binding to the viral RNA, ATP hydrolysis, and RNA unwinding, but not in ATP binding.

Expression and purification of HCV NS3 proteins.

An HCV NS3 protein representing the viral polyprotein from amino acid residues 1043 to 1635 was produced along with its mutant forms, NS3(K1236E) and NS3(GR1490DT), from plasmids pET15b-NS3, pET15b-NS3(K1236E), and pET15b-NS3(GR1490DT), respectively. Nucleotide sequence analysis of both strands of the plasmids indicated that NS3(K1236E) and NS3(GR1490DT) contain amino acid substitutions in conserved motifs I (1230-AX₄GKS-1237) and VI (1486-QRRGRTGR-1493), respectively (Fig. 1A; data not shown), and no additional mutations were generated during the cloning procedures. The overexpressed NS3 proteins were found predominantly in the insoluble fractions of the bacterial lysates (Fig. 1B and C). For biochemical analysis, the NS3 recombinant proteins were purified from sodium dodecyl sulfate (SDS)-polyacrylamide gels and renatured by dialysis (Fig. 1B and C). In addition, to examine possible contaminating activities from host proteins, plasmid pET15b was transformed and expressed in parallel (data not shown). The total cell lysate was prepared, and proteins eluted from the corresponding position of the HCV NS3 protein on SDS-polyacrylamide gels were used as host protein controls in functional analyses.

FIG. 1.

Expression of recombinant HCV NS3 proteins. (A) Schematic representation of the HCV NS3 protein. The bar spanning amino acid residues 1027 to 1657 represents the full-length HCV NS3 protein, and the shaded region from residues 1043 to 1635 represents the NS3 protein used in this study. Amino acid substitutions in the conserved motifs 1230-AX₄GKS-1237 and 1486-QRRGRTGR-1493 of the NS3 mutant proteins are indicated. To obtain the HCV NS3 cDNA spanning amino acid residues 1043 to 1635, total RNA was isolated from serum of an HCV-infected patient, and reverse transcriptase-nested PCR was performed as described previously (49) with C103 (3120-TGGTTGCATCATCACTAGC-3138; nucleotides are numbered starting from the translational initiation site of the HCV polyprotein) and C32 (α4927-TGGTTATGGGGTGCGTGA-α4910; the letters α indicate sequences of the complementary strand) as the first primer set and C104 (3126-CATCATCACTAGCCTCACAGG-3146) and C50 (α4906-TGACCTCATTTTGGACGGCT-α4887) as the second primer set. The HCV NS3 cDNA was first cloned into pGEM-4Z vector and then resubcloned into pET15b to generate plasmid pET15b-NS3 encoding a His-tagged HCV NS3 from amino acid residues 1043 to 1635 of the viral polyprotein. Plasmid pET15b-NS3(K1236E) encodes an NS3 mutant protein, designated NS3(K1236E), in which a Lys-to-Glu substitution was generated at the 1230-AX₄GKS-1237 motif. Plasmid pET15b-NS3(GR1490DT) encodes an NS3 protein with the arginine-rich motif 1486-QRRGRTGR-1493 mutated to QRRDTTGR, designated NS3(GR1490DT). Both plasmids were generated by replacing a subdomain of the wild-type NS3 cDNA-containing plasmid with a cognate PCR fragment bearing the desired mutations. (B) Coomassie blue staining. Total cell lysates (T) were prepared from E. coli BL21(DE3) transformed with plasmids pET15b-NS3 (lanes 1 to 4), pETI5b-NS3(K1236E) (lanes 5 to 8), and pET15b-NS3(GR1490DT) (lanes 9 to 12) and separated into soluble (S) and insoluble pellet (P) fractions, following three freeze-thaw cycles, sonication, and centrifugation. For purification of the recombinant NS3 proteins from insoluble fractions, the insoluble proteins were resuspended and boiled for 5 min prior to SDS-polyacrylamide gel electrophoresis. SDS-polyacrylamide gel electrophoresis was performed in Tricine-Tris buffer. Specific bands representing the isopropyl-β-d-thiogalactoside (IPTG)-induced NS3 proteins were sliced out, recovered through Electro-Eluter (Bio-Rad), and dialyzed against a buffer containing 25 mM Tris and 192 mM glycine. The dialysates were used as the partially purified NS3 proteins (E) in functional analyses. A Bio-Rad protein assay was performed to determine protein concentrations. (C) Western blot analysis. Western blot analysis was performed with the immunoglobulin G fraction of a rabbit serum against the HCV NS3 protein that followed the procedures as previously described (4). Arrowheads in panels B and C indicate the IPTG-induced recombinant NS3 proteins. Lane abbreviations (T, S, P, and E) are defined in the legend to panel B.

ATPase and RNA helicase activities.

The partially purified NS3 proteins were analyzed for ATPase and RNA helicase activities. As shown in Fig. 2, ATPase activity of the wild-type NS3 protein increased in a dose-dependent manner. The possibility of contaminating host ATPase activity in the partially purified NS3 proteins was eliminated, since no activity was detected with the host protein control (Fig. 2A) prepared in parallel as described earlier. ATPase activity was significantly reduced with mutant proteins NS3(K1236E) and NS3(GR1490DT) (Fig. 2). These results indicated that both the AX₄GKS and arginine-rich motifs are involved in ATP hydrolysis.

FIG. 2.

ATPase activity of HCV NS3 proteins. Increasing amounts (0.4 to 160 ng) of the partially purified recombinant HCV NS3 proteins were analyzed for ATPase activity as described previously (46) except that the reaction was performed at pH 7.6 for 30 min. Twenty nanograms of a purified dog pancreatic ATPase (Sigma) and 160 ng of a host protein preparation (H) were analyzed in parallel as positive and negative controls, respectively. An autoradiogram is shown (A). The conversion rates of ATP hydrolysis were calculated with a PhosphorImager (STORM 840; Molecular Dynamics) and plotted as a function of the input proteins (B). Each point represents the average of three independent experiments. Standard deviation bars are shown.

RNA helicase activity of the NS3 mutants was examined by using a partially double-stranded RNA substrate. The wild-type NS3 protein completely unwound the RNA substrate in the presence of ATP, but mutations in NS3(K1236E) and NS3(GR1490DT) completely abrogated the activity (Fig. 3). In addition, consistent with the energy requirement, RNA-unwinding activity of the wild-type NS3 protein was annihilated in the absence of ATP (Fig. 3, lane 4). The effects on ATPase and helicase activities with mutations in the AX₄GKS and arginine-rich motifs have been demonstrated previously (18, 25), but whether the mutated amino acids are directly involved in ATP binding has not been demonstrated. In addition, the role of the conserved motifs involved in RNA binding remained to be elucidated.

FIG. 3.

RNA-unwinding activity of HCV NS3 proteins. An unwinding activity assay was performed with 1 μmole of the NS3 proteins as indicated and 0.2 pmol of a α-³²P-labeled partially double-stranded RNA substrate in the presence (+) or absence (−) of ATP for 30 min at 37°C, in a buffer containing 20 mM Tris [pH 7.0], 1.5 mM MgCl₂, 2 mM dithiothreitol, 4 μg of bovine serum albumin (BSA), 40 U of RNasin, and 2.5 mM ATP. For preparation of the partially double-stranded RNA, plasmid pGEM-4Z was independently digested with HindII and Asp718. The HindII-linearized pGEM-4Z was used as a template to perform in vitro transcription with SP6 RNA polymerase in the absence of [α-³²P]UTP, and the Asp718-linearized template was used to perform transcription with T7 RNA polymerase in the presence of [α-³²P]UTP. These resulted in an unlabeled 42-nucleotide SP6 transcript and a radiolabeled 51-nucleotide T7 transcript. An annealing reaction to generate partially double-stranded RNA substrate essentially followed the procedure as previously described (22) except that the RNA substrate was further purified by resolution in a 15% nondenaturing polyacrylamide gel. The RNA-unwinding reaction was stopped by adding a buffer containing 3% SDS and 150 mM EDTA (pH 8.0). The reaction products were analyzed on a 12% nondenaturing polyacrylamide gel (acrylamide:bisacrylamide weight ratio of 19:1) and visualized by autoradiography. In lane 1, the RNA substrate was heat-denatured to release the ³²P-labeled single-stranded RNA product.

ATP-binding activity of NS3 mutants.

To learn whether the decline of ATPase activity of the NS3 mutant proteins resulted from a reduction in ATP binding, UV cross-linking experiments were carried out with [α-³²P]ATP. Equal amounts of the purified wild-type and mutant NS3 proteins were analyzed together with a purified dog pancreatic ATPase. The amount of [α-³²P]ATP cross-linked to the NS3(GR1490DT) protein was similar to the level of the wild-type NS3 protein but was decreased to less than 30% for NS3(K1236E) as measured by a PhosphorImager (STORM 840; Molecular Dynamics) (Fig. 4). These results indicated that the conserved lysine residue in the AX₄GKS motif is critical for ATP binding of HCV NS3 protein, whereas the Arg1490 residue in the arginine-rich motif (1486-QRRGRTGR-1493) is dispensable for ATP binding.

FIG. 4.

The conserved Lys1236 residue, but not Arg1490, is critical for the ATP-binding activity of HCV NS3 protein. To determine the ATP-binding activity, UV cross-linking experiments were performed at 4°C with [α-³²P]ATP (3,000 Ci/mmol) and 120 μg each of the partially purified NS3 proteins as indicated. In addition, 2 μg of a purified dog pancreatic ATPase (Sigma) was analyzed in parallel as a positive control (lane 1). The reaction mixtures in a buffer containing 20 mM morpholinepropanesulfonic acid (MOPS)-KOH [pH 7.0], 25 mM NaCl, 2.5 mM MgCl₂, and 25% glycerol were irradiated at 254 nm from a distance of 3 cm for 20 min and subjected to SDS–8% polyacrylamide gel electrophoresis. Coomassie blue staining (A) and autoradiography (B) are shown. Arrowheads indicate the purified dog pancreatic ATPase.

The AX₄GKS motif is the Walker A nucleotide binding motif (motif I) of RNA helicases (37, 45). Mutation at the lysine residue in the conserved motif of eIF-4A abrogated nucleotide binding, ATP hydrolysis, and RNA helicase activities (31, 35). ATP serves as an energy source and is essential for RNA-unwinding activity. The effect of the lysine substitution on the ATP-binding activity is likely to be the major cause that the RNA helicase activity of NS3(K1236E) was abolished. Recent studies of the crystal structure of the HCV NS3 helicase indicated that the AX₄GKS motif forms a phosphate-binding loop for binding to the β-phosphate group of ATP (26). In addition, the AX₄GKS motif is located at the N terminus of an α-helix structure and in close proximity to the conserved aspartic acid residue of the DEXH motif (47). This allows the lysine residue of the AX₄GKS motif to make an additional contact with the aspartic acid residue of the DEXH motif (6, 26, 47). DEXH is the Walker B NTP binding-hydrolysis motif (motif II) that interacts with the magnesium ion of Mg-ATP. The reduced ATPase activity of NS3(K1236E) (Fig. 2) may reflect a combined effect of the single amino acid substitution in the AX₄GKS motif on binding of ATP and interaction with the conserved DEXH motif.

So far, the crystal structure of the HCV NS3 helicase-ATP complex has not been determined. A recent study of the cocrystal structure of PcrA DNA helicase and a nonhydrolyzable ATP analog (adenylyl imidodiphosphate [ADPNP]) indicated that Arg610 in motif VI (599-EERRLAYVGITRA-611) (39) of PcrA contacts with the γ-phosphate group of ADPNP (44). In the “inchworm” unwinding model (48) of HCV NS3 helicase, Arg1490 in motif VI (1486-QRRGRTGR-1493) was proposed to directly contact the γ-phosphate group of a bound molecule of ATP (26, 28). Nevertheless, our functional analysis with the NS3(GR1490DT) mutant protein demonstrated that Arg1490 is not critical for ATP binding to the NS3 protein (Fig. 4). However, the reduction in the ATPase activity of NS3(GR1490DT) (Fig. 2) did indicate that the arginine-rich motif is involved in ATP hydrolysis.

RNA-binding activity of NS3 mutants.

RNA-binding activity of the NS3 mutant proteins was examined by filter binding assay and Northwestern analysis with an HCV 3′-end RNA (3′CNU RNA) (5) as the probe. The relative RNA-binding activity of NS3(GR1490DT) to the wild-type NS3 protein was about 10% by filter binding assay (Fig. 5A) and 26% by Northwestern analysis (Fig. 5B). Although the values of 10 versus 26% in the two assays differed by two- to threefold, the results were reproducible and clearly demonstrated that Arg1490 in the arginine-rich motif is important for HCV NS3 protein to bind to the viral RNA. In addition, amino acid substitution at Lys1236 [NS3(K1236E)] decreased RNA binding to about 30% (by the filter binding assay) to 40% (by Northwestern analysis) of the wild-type level (Fig. 5). These results suggested a sequential binding mechanism of ATP and RNA to the HCV NS3 helicase. Substitution at Lys1236 drastically reduced the ATP-binding activity and rendered NS3 helicase in a conformation not suitable for RNA binding. This is similar to the mechanisms proposed for eIF4A (30) and PcrA (44); binding of ATP results in a conformational change in the helicases that allows RNA binding and induces ATP hydrolysis.

FIG. 5.

Both mutations at Lys1236 and Arg1490 affected the RNA-binding activity of HCV NS3 protein. (A) Filter binding assay. An RNA-binding reaction was performed as previously described (17) with 1 μmole of ³²P-labeled HCV 3′CNU RNA (5) and 10 nmole of proteins NS3 (sample 1), NS3(K1236E) (sample 2), NS3(GR1490DT) (sample 3), BSA (sample 4), and a host protein preparation (Host) (sample 5) as indicated. Input RNA (10%), 10% of the total reaction mixture spotted directly onto a nitrocellulose membrane; Bound RNA, 90% of the reaction mixture from which the unbound RNA has been removed. Radioactivity on the nitrocellulose membranes was counted with a scintillation counter (Beckman LS6000TA). The percentage of HCV 3′CNU RNA bound to the wild-type NS3 protein was normalized to 100%, and the relative RNA-binding activity of each of the NS3 mutants and control proteins was calculated. Each column represents the average of three independent experiments. Standard deviation bars are shown. (B) Northwestern analysis. Two identical sets of the indicated proteins were resolved on an SDS-polyacrylamide gel in Tricine-Tris buffer and analyzed, respectively, by Coomassie blue staining (top) and Northwestern analysis with the [α-³²P]UTP-labeled HCV 3′CNU RNA (bottom). Northwestern analysis was performed as previously described (3). Radioactivity of the bound RNA was measured by a PhosphorImager (STORM 840; Molecular Dynamics), and the relative RNA-binding activity of the mutant and control proteins to the wild-type NS3 protein was calculated as indicated.

In agreement with our results of the roles of the arginine-rich motif of HCV NS3, previous studies have demonstrated that the conserved motif of eIF4A is required for ATP hydrolysis and RNA binding (30). In addition, crystal structure analysis and computer graphics modeling indicated that the guanidinium groups of Arg1487, -1490, and -1493 in the arginine-rich motif of the HCV NS3 helicase are ideally situated to bind the phosphate groups of an RNA substrate (6, 47). However, an electrophoretic mobility shift assay performed in a previous study with the C-terminal three-fourths of the HCV NS3 protein and a nonviral RNA probe in vitro transcribed from a PvuII-linearized pGEM3 suggested that Arg1488 is the only arginine residue in motif VI (1486-QRRGRTGR-1493) involved in RNA binding (25).

To understand the possible role of Arg1488 in binding to HCV RNA, plasmids pET15b-NS3(R1488A) and pET15b-NS3(R1490A) encoding NS3 proteins with single amino acid substitutions at Arg1488 [designated NS3(R1488A)] and -1490 [designated NS3(R1490A)], respectively, were further generated and expressed in Escherichia coli BL21(DE3) (data not shown). Mutant proteins NS3(R1488A) and NS3(R1490A) were purified from the insoluble fractions of the bacterial lysates (Fig. 6A and B) and analyzed for binding activity to the HCV 3′CNU RNA (5). When compared to the wild-type NS3 protein, NS3(R1490A) demonstrated a moderate binding activity whereas the Arg1488 mutation reduced the viral RNA-binding activity to 15% of that of the wild type (Fig. 6C).

FIG. 6.

The Arg1488 residue is important for the RNA-binding activity of HCV NS3 protein. HCV NS3 mutant proteins NS3(R1488A) and NS3(R1490A) were produced from plasmids pET15b-NS3(R1488A) and pET15b-NS3(R1490A), respectively. Expression and purification of the His-tagged mutant proteins followed the procedures described in the legend to Fig. 1B. Three identical sets of the BSA control and purified wild-type and mutant NS3 proteins as indicated were resolved on an SDS-polyacrylamide gel and analyzed by Coomassie blue staining (A), Western blot analysis with a monoclonal antibody against the 6XHis-tag (CLONTECH) (B), and Northwestern analysis with the [α-³²P]UTP-labeled HCV 3′CNU RNA probe described in the legend to Fig. 5B (C).

Previous analyses of crystal structure and computer graphics modeling have suggested that Arg1490 is involved in RNA binding (6, 47). However, the studies did not reveal the important role of Arg1488 in interacting with RNA substrate, which conflicted with the functional study indicating that Arg1488 was the only arginine residue in the motif binding to a nonviral RNA (25). Although nonspecific RNAs have been used to study RNA-binding characteristics of many viral proteins, the use of a specific viral RNA probe should provide a condition better mimicking the natural environment. Our results indicate that both Arg1488 and 1490 are involved in binding to the HCV RNA. Taken together, the present study clearly demonstrated that the conserved lysine residues in the AX₄GKS motif is critical for ATP binding, whereas Arg1488 and -1490 in the arginine-rich motif (1486-QRRGRTGR-1493) are important for the HCV NS3 protein to interact with the viral RNA.

Nucleotide sequence accession number.

The cDNA sequence encoding the HCV NS3 protein has been deposited in EMBL under accession no. AJ238652.