Isolation and Characterization of Monoclonal Antibodies That Inhibit Hepatitis C Virus NS3 Protease

Abstract

A series of mouse monoclonal antibodies (MAbs) to the nonstructural protein 3 (NS3) of hepatitis C virus was prepared. One of these MAbs, designated 8D4, was found to inhibit NS3 protease activity. This inhibition was competitive with respect to the substrate peptide (K_i = 39 nM) but was significantly decreased by the addition of the NS4A peptide, a coactivator of the NS3 protease. 8D4 also showed marked inhibition of the NS3-dependent cis processing of the NS3/4A polyprotein but had virtually no effect on the succeeding NS3/4A-dependent trans processing of the NS5A/5B polyprotein in vitro. Epitope mapping of 8D4 with a random peptide library revealed a consensus sequence, DxDLV, that matched residues 79 to 83 (DQDLV) of NS3, a region containing the catalytic residue Asp-81. Furthermore, synthetic peptides including this sequence were shown to block the ability of 8D4 to bind to NS3, indicating that 8D4 interacts with the catalytic region of NS3. The data showing decreased inhibition potency of 8D4 against the NS3/4A complex suggest that 8D4 recognizes the conformational state of the protease active site caused by the association of NS4A with the protease.

Nonstructural protein 3 (NS3) of hepatitis C virus (HCV) is a multifunctional virus-specific protein that contains serine protease activity in its N-terminal region and accounts for processing of the viral polyprotein at four cleavage sites, NS3/4A, NS4A/4B, NS4B/5A, and NS5A/5B, whereas helicase and nucleic acid-stimulated nucleoside triphosphatase activities are found in its C-terminal region (see references 2 and 23 for reviews). The NS3 protease requires the NS4A protein as a cofactor for efficient cleavage of the polyprotein (35, 39). Because this enzyme plays an obligatory role in viral replication, it provides a logical target for the development of potentially selective antiviral agents. Development of increasingly specific inhibitors of NS3 requires detailed knowledge of the tertiary structure of the enzyme. X-ray crystallographic analysis (21, 28, 29, 41) and nuclear magnetic resonance (NMR) spectroscopic analysis (1, 6) of the NS3 protease domain with or without the NS4A cofactor have provided a refined picture of the NS3 structure. Those studies show that the overall topology of NS3 protease is similar to that of chymotrypsin-like serine proteases and NS3 forms N-terminal (approximately residues 1 to 93) and C-terminal (residues 94 to 180) six-stranded antiparallel β-barrels that are packed like those of chymotrypsin-like serine proteases (1, 6, 21, 28, 29, 41). The catalytic site of NS3 protease is formed by the triad of residues His-57, Asp-81, and Ser-139 and is found in the crevice between the two barrels. The interaction of NS4A with NS3 was shown to induce conformational changes in NS3 that involve both a structural reorganization of the N-terminal domain and a rearrangement of the protease catalytic site including Asp-81 (1, 21, 29). Although the tertiary structure of NS3 protease has been defined in detail, several loops found in other chymotrypsin family proteases, which play a critical role in defining the shapes of the non-prime-side substrate-binding pockets, are missing from NS3, rendering the substrate-binding groove relatively featureless and therefore making the design of low-molecular-weight inhibitors quite challenging (21). As an alternative approach to the study of structure and for designing inhibitors of NS3 protease, we have prepared a series of monoclonal antibodies (MAbs) for use in probing the tertiary conformation of the enzyme with or without the NS4A cofactor. In the present study, we describe the isolation and characterization of the MAbs designated 7E3, 7E9, and 8D4. One of these MAbs, 8D4, appears to be a competitive inhibitor with respect to the substrate peptide and recognizes a linear surface epitope containing residues 79 to 83 of NS3, a region containing the catalytic residue Asp-81.

MATERIALS AND METHODS

Preparation of MAbs.

Inclusion bodies formed upon overproduction of the N-terminal protease domain of the NS3 protein (NS3_1–160) in Escherichia coli strain SCS1 were dissolved in 8 M urea, 50 mM Tris-HCl (pH 8.5), and 1 mM EDTA. NS3_1–160 protein was then purified by gel filtration chromatography (Sephacryl S-200 HR; Pharmacia) in the presence of 8 M urea, dialyzed against distilled water, and lyophilized. Since we could not obtain any MAb specific for the N-terminal protease domain of the NS3 protein when the full-length NS3 protein had been used to immunize mice, we used NS3_1–160 instead as an immunogen in this study, although this preparation showed no protease activity even after urea was removed by dialysis (data not shown).

Five 6-week-old BALB/c mice were then immunized at 2-week intervals with the NS3_1–160 protein (approximately 100 or 250 μg of protein per injection) emulsified with Freund's complete adjuvant (Difco Laboratories) for the first injection and incomplete adjuvant for the two subsequent injections. Sera were tested by enzyme-linked immunosorbent assay (ELISA) and Western blotting as described below. Spleen cells from the mouse showing the strongest reaction were then fused with the mouse myeloma cell line P3X63Ag8U1. Colony supernatants were screened by ELISA with 96-well microtiter plates that had been coated with approximately 0.1 μg of purified NS3_1–160. The positive supernatants were further tested by Western blotting, and the corresponding hybridomas were cloned by limiting dilution.

Production and purification of active NS3 protease.

DNA fragments encoding the full length (1 to 631 residues) and N-terminal protease domain (1 to 190 residues) of NS3 of HCV II_J (20) were cloned into the expression vector pMAL-c2 (New England Biolabs) and pMT1, which had been constructed by replacing the tac promoter of pMK2 (17) with the tryptophan promoter, respectively, in order to overproduce a maltose-binding protein–NS3 protease (MBP-NS3_1–631) fusion protein and an N-terminal domain with a hexahistidine tag (His₆-NS3_1–190), respectively. For MBP-NS3_1–631 production, E. coli (strain HB101) cells transformed with the resulting plasmid were cultured overnight in ampicillin (100 μg/ml)-containing 2× YT medium (1.6 g of tryptone, 1 g of yeast extract, and 0.5 g of NaCl per liter of distilled water), further propagated in the same medium (2 liters) to the mid-logarithmic growth phase at 30°C, and then exposed to 1 mM isopropyl-1-thio-β-d-galactoside and cultured for an additional 15 h at 25°C. Cells were then harvested, suspended in buffer (20 mM Tris-HCl [pH 7.4], 0.2 M NaCl, 1 mM EDTA), and disrupted by sonication on ice. MBP-NS3_1–631 was purified by affinity chromatography with amylose resin (New England Biolabs), followed by gel filtration chromatography (Sephacryl S-300 HR; Pharmacia) in order to remove improperly folded species. Fractions containing NS3 protease activity were collected and pooled in a buffer containing 20 mM sodium phosphate, pH 7.2. For His₆-NS3_1–190 production, E. coli (strain JM109) cells transformed with the resulting plasmid were cultured overnight in ampicillin (100 μg/ml)-containing M9 medium (60 g of Na₂HPO₄, 30 g of KH₂PO₄, 10 g of NH₄Cl, 5 g of NaCl per liter of distilled water, pH 7.4) supplemented with 0.2% Casamino Acids, 0.2% glucose, 1 mM MgSO₄, 0.1 mM CaCl₂, and 1 mM thiamine, further propagated in the same medium (3.6 liters) for 15 h, and then exposed to 25 μg of 3-indoleacrylic acid/ml and cultured for an additional 8 h at 20°C. Cells were then harvested, suspended in buffer (50 mM Tris-HCl [pH 7.5], 0.3 M NaCl, 20 mM imidazole, 10% glycerol), and disrupted by sonication on ice. His₆-NS3_1–190 was then purified by metal chelate affinity (Ni-NTA Superflow; Qiagen) and ion exchange chromatographies (S-Sepharose Fastflow; Pharmacia). Fractions containing NS3 protease activity were collected and pooled in a buffer containing 50 mM Tris-HCl (pH 7.5), 10% glycerol, and 0.5% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS). Both NS3 proteases thus prepared showed >90% purity as assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (data not shown) and were stored at −70°C after the addition of an equal volume of glycerol.

Enzymatic assays for NS3 protease activity.

When a synthetic peptide carrying the NS5A/5B cleavage site and having a fluorescent moiety at its N terminus (2-aminobenzoyl [Abz]-EDVVECSMSY-NH₂) was used as a substrate, the reaction mixture contained 50 mM Tris-HCl (pH 8.5), 30 mM NaCl, 5 mM CaCl₂, and 10% glycerol or 50 mM Tris-HCl (pH 8.5), 30 mM NaCl, 2% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate and 5% glycerol for MBP-NS3_1–631 or His₆-NS3_1–190, respectively. Either of the NS3 proteases was preincubated in the presence or absence of MAbs and the NS4A peptide (H-LTTGSVVIVGRIILSGRPAVVPD-OH [Pep4A_18–40]) (35) for 10 min at room temperature in a total volume of 80 μl. The reaction was initiated by the addition of 20 μl of the substrate peptide dissolved in 50 mM Tris-HCl (pH 8.5), 30 mM NaCl, 5 mM CaCl₂, and 25 mM dithiothreitol. The final concentration of the peptide substrate in the reaction mixture was 0.025 or 0.2 mM, unless otherwise indicated, when in the presence or absence of Pep4A_18–40, respectively. It was continued at 37°C for 15 or 60 min with or without Pep4A_18–40, respectively, and was quenched by the addition of 100 μl of 0.5% trifluoroacetic acid (TFA). The fluorescent reaction product (Abz-EDVVEC-OH) was then separated on a reversed-phase high-pressure liquid chromatography (HPLC) apparatus equipped with a C₄ column (4.6 by 50 mm; Vydac) by elution at a flow rate of 2 ml/min with an aqueous solution containing 14% acetonitrile and 0.1% TFA, and the fluorescence signal was detected by a spectrofluorometric detector (Shimadzu, Kyoto, Japan) with excitation and emission wavelengths at 320 and 425 nm, respectively. The product was quantified by integration of the chromatogram with respect to the chemically synthesized peptide standard with a structure identical to that of the cleavage product. Steady-state kinetic parameters and the K_i value were determined by Lineweaver-Burk plot and Dixon plot analyses, respectively. All peptides used for enzymatic assays were synthesized and purified by HPLC by the Peptide Institute Inc. (Osaka, Japan).

To examine the NS3 protease activity toward in vitro-translated substrates, we synthesized radiolabeled NS3/4A (for cis processing) and NS5A/5B (for trans processing) polyproteins by using a coupled transcription-translation system (TNT rabbit reticulocyte lysate system; Promega) according to the manufacturer's protocol. DNA fragments encoding NS3/4A and NS5A/5B cleavage sites, which encompassed residues 1027 to 1711 and 2320 to 3011, respectively, of the HCV-II_J polyprotein, were cloned into the plasmid pTZ18U (Pharmacia) downstream of the T7 promoter to generate pTZNS3-4K and pTZNS5, respectively. For the cis-processing assay, 1 μg of pTZNS3-4K was added to 50 μl of the TNT reaction mixture containing 40 μCi of [³⁵S]methionine (1,000 Ci/mmol; Amersham) in the presence or absence of MAbs and incubated at 30°C for 90 min. For the trans-processing assay, NS3/4A protease, which had been translated without radiolabel in vitro, was first preincubated at 30°C for 10 min in the presence or absence of MAbs. The reaction was then initiated by the addition of a radiolabeled polyprotein containing the NS5A/5B cleavage site and was incubated at 30°C for 2 h. For both assays, reactions were quenched by the addition of an equal volume of 2× SDS-sample buffer (62.5 mM Tris-HCl, 2% SDS, 25% glycerol, 0.3 M 2-mercaptoethanol, 0.01% bromophenol blue) and the products were separated by SDS-PAGE. The amounts of the products were integrated and analyzed with a phosphoimager instrument (BAS 2000; Fuji Photo Film, Tokyo, Japan).

Interaction of antibodies with NS3 protein or epitope peptides. (i) Western blot.

After bacterial cell lysates or purified NS3 proteins had been separated by SDS-PAGE, the proteins were electrically transferred to a polyvinylidene difluoride membrane (Bio-Rad Laboratories), and the membrane was preblocked with phosphate-buffered saline (PBS) containing 5% nonfat dry milk. The resultant membrane was incubated with 10 μg of MAbs/ml for 2 to 8 h at room temperature, followed by labeling with biotinylated goat anti-mouse immunoglobulin G (IgG) antibody (Life Technologies Inc.) and streptavidin-peroxidase conjugate (Tago Inc.). Reactive bands were stained with a 3,3′-diaminobenzidine peroxidase substrate tablet set (Sigma). Band intensities were quantified by densitometric analysis (Atto Corp., Tokyo, Japan) of the blots.

(ii) ELISA.

Ninety-six-well microtiter plates were coated with an appropriate amount (0.01 to 0.1 μg) of MBP-NS3 protein dissolved in PBS containing 0.05% Tween 20 and incubated overnight at 4°C. Serially diluted MAbs were then incubated for 1 h at room temperature. After a thorough washing with PBS-0.05% Tween 20, MAbs bound to the target antigen were labeled with biotinylated goat anti-mouse IgG antibody and streptavidin-peroxidase conjugate and identified with ortho-phenylenediamine as a substrate for the peroxidase reaction.

(iii) Surface plasmon resonance.

Kinetic analysis of the interactions of MAbs with NS3 protease was done with an IAsys Auto+ Optical Biosensor instrument (Affinity Sensors, Cambridge, United Kingdom). Approximately 1.5 μg of MBP-NS3_1–631 was immobilized on the carboxymethylated dextran layer via N-hydroxysuccinimide and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide chemistry. A coupling efficiency of 2,800 arc was obtained. For the determination of the kinetic constants, a diluted solution of MAbs ranging from 30 to 700 nM was added to the cuvette. Binding traces were recorded for at least four different concentrations. Association and dissociation rate constants were calculated by the computer program FASTFIT (Affinity Sensors) using the association phase.

Epitope mapping.

Random peptide libraries inserted within the thioredoxin active site loop and displayed on bacterial flagella (FliTrx; Invitrogen) were used for the determination of epitopes recognized by these MAbs, as described by Lu et al. (30) with modifications. A portion (∼10¹⁰ cells) of overnight-cultured FliTrx Library in IMC medium (1× M9 salts, 0.2% Casamino Acids, 0.5% glucose, 1 mM MgCl₂, 100-μg/ml ampicillin) at 25°C was further propagated in 50 ml of fresh IMC medium supplemented with 100 μg of tryptophan/ml for an additional 6 h to induce expression of the gene encoding flagellin-thioredoxin fusion proteins containing the peptide library. The library-displaying E. coli cells, which had been washed and resuspended in 0.5 ml of buffer (IMC medium, 150 mM NaCl, 0.2 mg of bovine serum albumin/ml, 0.4% α-methyl mannoside) were mixed with 20 μg of each of the MAbs. After a 30-min incubation, paramagnetic microbeads conjugated with goat-anti-mouse IgG (Miltenyi Biotec, Bergisch Gladbach, Germany) were added. E. coli cells that bound primary and secondary antibodies with magnetic beads were separated by magnetic cell sorting technology (32). Cells enriched by binding to either of three MAbs were cultured overnight, followed by two additional panning cycles. After the third panning, approximately 20 clones were subjected to the Western blot analysis in order to identify positive clones.

For determining the nucleotide sequence of the positive clones, DNA fragments spanning the random-peptide-encoding region were first amplified by PCR using primers FliTrx-2 (5′-TCACCGGTGGTGATAACGAT-3′) and RSR-2 (5′-CGATGTTCAGTTTTGCAACG-3′). The amplified fragment was then directly sequenced with a BigDye terminator kit (PE Biosystems) primed by FliTrx-1 (5′-TTATTCACCTGACTGACGAC-3′) for the sense strand or RSR-1 (5′-TTGCCCTGATATTCGTCAGCG-3′) for the antisense strand, and the DNA sequence was analyzed by a Genetic Analyzer 310 (PE Biosystems).

RESULTS

Isolation of anti-NS3 MAbs.

Three stable hybridomas producing high-affinity monoclonal IgG antibodies, designated as 7E3, 7E9, and 8D4, specific for the protease domain of NS3, were obtained by immunization of mice with purified NS3_1–160 of HCV-II_J (20). Antibody subclasses of these MAbs determined by use of a Mouse Mono AB ID kit (Zymed Laboratories) were IgG1κ.

Binding affinities of these antibodies to the full-length NS3 protease (MBP-NS3_1–631) were first analyzed by monitoring the real-time binding with the surface plasmon resonance device (IAsys Biosensor). The equilibrium dissociation constants (K_d) of MAbs were calculated from the ratio of the kinetic rate constants (Table 1). 8D4 showed the most profound binding affinity (K_d = 0.043 μM), whereas 7E3 showed moderate affinity (K_d = 0.192 μM) to MBP-NS3_1–631 and 7E9 showed the weakest binding affinity (K_d = 0.658 μM) among the three MAbs (Table 1).

TABLE 1.

Kinetic analysis of interactions of MAbs with MBP-NS3_1–631^a

MAb	kon (103 M−1s−1)	koff (10−3s−1)	Kd (10−6 M)
7E3	13.2 ± 0.57	2.54 ± 0.20	0.192
7E9	5.97 ± 1.1	3.93 ± 0.24	0.658
8D4	24.9 ± 1.7	1.07 ± 0.30	0.043

^aThe rate constants of association (k_on) and dissociation (k_off) of each of the MAbs with MBP-NS3_1–631 were analyzed with an IAsys Auto+ Optical Biosensor (Affinity Sensors) and were calculated by the computer program FASTFIT using the association phase, as described in Materials and Methods. Values are given as means plus or minus standard deviations. The equilibrium dissociation constants (K_d) were calculated by the equation K_d = k_off/k_on. 

Effect of MAbs on NS3-dependent cleavage of polyproteins.

When the NS3/4A polyprotein was produced in vitro by use of a coupled transcription-translation system, intramolecular cleavage (cis processing) of the polyprotein at the NS3/4A junction mediated by the NS3 protease was observed as assessed by SDS-PAGE (Fig. 1A). The effect of MAbs on this NS3-dependent cis-processing step was then examined to compare the amounts of the cleavage products formed in the presence and absence of 100 μg of each of the MAbs per ml. 8D4 appeared to be a potent inhibitor of the cis-processing activity of NS3 protease (Fig. 1A, lane 5), but the other two MAbs (7E3 and 7E9) did not show a significant effect at this concentration (Fig. 1A, lanes 3 and 4). The extent of 8D4 inhibition of cis-processing activity of NS3 was proportional to the concentration of 8D4 in the reaction, and a 50% inhibitory concentration (IC₅₀) of 10 μg/ml (67 nM) was obtained (Fig. 1B).

FIG. 1.

Effect of MAbs on polyprotein processing activity of NS3 protease. (A) NS3 protein (lane 1) or NS3/4A polyprotein in the absence (lane 2) or presence of 100 μg of 7E3 (lane 3), 7E9 (lane 4), or 8D4 (lane 5)/ml was synthesized in vitro by a coupled transcription-translation system for 2 h at 30°C, and the NS3-dependent intramolecular cleavage of the polyprotein (cis processing) was analyzed by SDS-PAGE. (B) Various amounts of 8D4 were then added to the reaction, and the 8D4-mediated inhibition of the cis-processing step of NS3 was evaluated. The concentrations of 8D4 used were 0, 100, 50, 25, 12.5, 6.25, 3.13, and 1.56 μg/ml for lanes 1, 2, 3, 4, 5, 6, 7, and 8, respectively. (C) The ability of MAbs to inhibit the NS3/4A protease-mediated intermolecular cleavage (trans processing) of the polyprotein at the NS5A/5B site was also examined. The polyprotein containing the NS5A/5B cleavage site (NS5AΔN/5B), which encompassed residues 2320 to 3011 of the HCV-II_J polyprotein, was first synthesized (lane 1). NS3/4A protease, which had been translated without radiolabel in vitro, was preincubated at 30°C for 10 min in the absence (lane 2) or presence of 7E3 (lane 3), 7E9 (lane 4), or 8D4 (lane 5) at an antibody concentration of 100 μg/ml. The reaction was then initiated by the addition of a radiolabeled NS5AΔN/5B and was incubated for 2 h at 30°C. Numbers on the left indicate the apparent molecular masses that were determined by use of the prestained (Bio-Rad Laboratories) or ¹⁴C-methylated (Amersham Pharmacia Biotech) protein-molecular-weight markers in panels A and B or C, respectively. Note that three or four additional bands seen in the gels were degraded and/or incompletely translated polyproteins.

We also examined the ability of the MAbs to inhibit the NS3/4A protease-mediated intermolecular cleavage (trans processing) of the polyprotein at the NS5A/5B junction. None of the MAbs tested, however, showed any effect on trans processing of the polyprotein at an antibody concentration of 100 μg/ml (Fig. 1C), suggesting the decreased sensitivity of NS3 to 8D4 in the presence of the NS4A cofactor.

Effect of MAbs on NS3-dependent cleavage of a peptide substrate.

We next assessed the abilities of these MAbs to inhibit the NS3 protease activity toward a synthetic peptide (Abz-EDVVECSMSY-NH₂) containing the NS5A/5B cleavage site as a substrate. Purified full-length NS3 protease (MBP-NS3_1–631) was mixed with various amounts of MAbs, and the residual protease activity was determined. All three MAbs appeared to inhibit the cleavage of the substrate peptide under this assay condition (Fig. 2A). The addition of the NS4A peptide (Pep4A_18–40), which comprised residues 18 to 40 of the NS4A protein and is known to increase the catalytic efficiency of NS3 protease (3, 13, 26, 35, 39), to the reaction mixture decreased the inhibition potencies of all three MAbs (Fig. 2A).

FIG. 2.

Effect of MAbs on NS3 protease activity toward a peptide substrate. (A) Proteolytic activity of purified MBP-NS3_1–631 was evaluated with a synthetic peptide (2-Abz-EDVVECSMSY-NH₂) that served as a substrate in the presence or absence of MAbs. MBP-NS3_1–631 (22 nM) was mixed with 0 (white bars), 11 (hatched bars), and 22 (gray bars) nM concentrations of the indicated MAbs or a 22 nM concentration of the indicated MAbs in the presence of 10 μM PepNS4A_18–40 (H-LTTGSVVIVGRIILSGRPAVVPD-OH) (black bars). (B) The dose-dependent inhibition by 8D4 of the proteolytic activity of the protease domain of NS3 was also evaluated. His₆-NS3_1–190 (38 nM) was mixed with the indicated amounts of 8D4 in the absence (■) or presence (●) of 15 μM PepNS4A_18–40, and the residual protease activities were determined. Note that the IC₅₀s of 8D4 for His₆-NS3_1–190 were 2.6 and 0.028 μM in the presence and absence of PepNS4A_18–40, respectively, as obtained from the graph.

We further examined the extent of inhibition potencies of 8D4 in the presence and absence of Pep4A_18–40 by using the protease domain of NS3 (His₆-NS3_1–190) as an enzyme source (Fig. 2B). The IC₅₀s of 8D4 for the His₆-NS3_1–190 protease were 0.028 and 2.6 μM in the absence and presence of Pep4A_18–40, respectively (Fig. 2B). This 90-fold-decreased sensitivity of NS3 to 8D4 by the addition of Pep4A_18–40 appeared to be consistent with the data indicating that 8D4 had virtually no effect on the NS3/4A-dependent trans processing of the NS5A/5B polyprotein (Fig. 1C) and may have been caused by the competition between 8D4 and Pep4A_18–40 for the identical binding site in NS3 or by decreased binding affinities as a result of the conformational changes in NS3 induced by its association with Pep4A_18–40.

Steady-state kinetic analysis of 8D4 inhibition of NS3 protease.

The cleavage of the substrate peptide mediated by the full-length NS3 protease (MBP-NS3_1–631) was linear with respect to time for at least 60 min in the absence of the NS4A cofactor after the initiation of the reaction (data not shown). Steady-state kinetic constants determined under this assay condition (K_m = 382 ± 14 μM; k_cat = 10.8 ± 1.6 min⁻¹) (Fig. 3A) were comparable to those determined by others (35, 38). We then conducted the kinetic experiments to analyze the inhibition mechanism of 8D4 in the absence of Pep4A_18–40. As shown in Fig. 3A, linear competitive inhibition by 8D4 was observed for the proteolytic cleavage of the peptide substrate by MBP-NS3_1–631, indicating that 8D4 directly blocked the binding of the substrate peptide in the absence of the NS4A cofactor. The K_i value of 8D4 inhibition determined by the Dixon plot analysis was 39 nM (Fig. 3B), in good agreement with the K_d value of 8D4 (Table 1).

FIG. 3.

Steady-state kinetic analysis of 8D4-mediated inhibition. The reaction mixture containing 50 mM Tris-HCl (pH 8.5), 30 mM NaCl, 5 mM CaCl₂, 10% glycerol, and 22 nM MBP-NS3_1–631 protease was preincubated in the presence or absence of various amounts of 8D4 for 10 min at room temperature, and the reaction was then initiated by the addition of the substrate peptide (2-Abz-EDVVECSMSY-NH₂). After a 60-min incubation at 37°C, the reaction was quenched by the addition of 0.5% TFA, followed by separation and quantification of the cleavage product by HPLC as described in Materials and Methods. Concentrations of the substrate used were 0.2, 0.3, 0.4, 0.6, 0.8, and 1 mM; those of 8D4 were 0, 20, 40, and 60 nM. Data obtained were then analyzed by both Lineweaver-Burk plot (A) and Dixon plot (B) analysis. The K_m and k_cat values of MBP-NS3_1–631 determined by the Lineweaver-Burk plot in the absence of 8D4 were 382 ± 14 μM and 10.8 ± 1.6 min⁻¹, respectively, and the K_i value of 8D4 determined by the Dixon plot was 39 nM.

Determination of epitopes recognized by MAbs.

We next attempted to identify epitopes recognized by these MAbs, using the random peptide library displayed on the bacterial flagella. After three rounds of panning cycles against each of the three MAbs, individual colonies that were specifically stained by the corresponding antibody in the Western blots were isolated (Fig. 4). Because the growth rates of the bacteria and amounts of peptide-inserted thioredoxin-flagellin fusion proteins produced in each of the isolated bacterial clones were different, we used Anti-Thio antibody (Invitrogen), which specifically recognizes thioredoxin, to normalize the amounts of the peptide-fusion proteins. Binding affinities between the peptide fusion and 8D4 were thus calculated as the relative band intensities of the Western blot by dividing the intensities of bands stained with 8D4 by those of bands stained with Anti-Thio antibody (Fig. 4B and C): values of relative band intensities were 4.5, 0.58, 0.37, >22, and 0.0032 for clones 35, 38, 53 and 121 and the original library, respectively. The 121 clone showed the most profound binding affinity with 8D4 in semiquantitative Western blot analysis (Fig. 4C, lane 4), whereas virtually no band was seen in the original library (Fig. 4C, lane 5).

FIG. 4.

Western blot analysis of E. coli clones selected for MAb 8D4. Cell lysates of E. coli clones selected by binding to MAb 8D4 and that without panning (original library) were analyzed on 12.5% acrylamide gels. Lanes 1, clone 35; lanes 2, clone 38; lanes 3, clone 53; lanes 4, clone 121; lanes 5, original library. The gels stained with Coomassie brilliant blue (A) and identical Western blots probed with anti-thioredoxin MAb (B) and 8D4 (C) are shown. The arrow indicates the peptide-inserted thioredoxin-flagellin fusion proteins. Because the peptide library was expressed as thioredoxin fusion proteins, intensities of bands detected by the anti-thioredoxin MAb (Anti-Thio Antibody; Invitrogen) indicate the amounts of peptide libraries produced in each of the isolated clones. Relative binding affinities to 8D4 were thus determined to be 4.5, 0.58, 0.37, >22, and 0.0032 for clones 35, 38, 53, and 121 and the original library, respectively, by dividing the band intensities obtained with 8D4 by those obtained with anti-thioredoxin MAb.

Alignment of the deduced amino acid sequence derived from the nucleotide sequence of the random-peptide-encoding region of positive clones allowed us to determine the consensus sequences, GWP, RRRG, and DxDLV, for MAbs 7E3, 7E9, and 8D4, respectively (Fig. 5). These sequences strongly matched GWP, RRRG, and DQDLV, which correspond to residues 84 to 86, 117 to 120, and 79 to 83, respectively, of the NS3 protease (Fig. 5), indicating that each of the MAbs recognized a sequential linear peptide chain as its epitope. It is noteworthy that one of the catalytic residues of NS3 protease, Asp-81, is included in the candidate recognition site of 8D4 (Fig. 5).

FIG. 5.

Amino acid sequences of the peptides displayed on E. coli clones that were selected for MAbs. The deduced amino acid sequences derived from the nucleotide sequences of the random-peptide-encoding regions of isolated clones selected with 7E3, 7E9, and 8D4 are shown. Consensus amino acid residues are indicated in bold, and the amino acid sequence of NS3 protease of HCV-II_J (20) that corresponds to the predicted recognition site of each of the MAbs is given at the top.

In order to confirm that DQDLV residues of NS3 comprise the epitope recognized by 8D4, we synthesized various 20-mer peptides (Table 2) and tested their ability to compete with MBP-NS3_1–631 in the binding reaction with 8D4. 8D4 was first preincubated with each of the peptides or MBP-NS3_1–631 for 1 h at room temperature and then transferred to a microtiter plate that had been coated with MBP-NS3_1–631, after which normal ELISA analysis was conducted. The peptide carrying the sequence identical to amino acid residues 76 to 87 of NS3 (PepNS3) (Table 2) most significantly blocked the ability of 8D4 to bind to MBP-NS3_1–631 (Fig. 6), whereas blockage occurred only at much higher concentrations of the peptides such as Pep8D4-35, -38, -53, and -121 that carried the sequence derived from respective 8D4-reactive bacterial clones (Table 2; Fig. 5). The extent of this competition was proportional to the concentration of the peptides used in the preincubation, indicating that both the peptides and MBP-NS3_1–631 competed for the same binding site of 8D4 (Fig. 6). It is of interest that Pep8D4-35 showed poor ability for this competition, although this peptide contained the DQDLV motif, suggesting that neighboring residues may also be involved in the recognition by 8D4.

TABLE 2.

Peptide sequences used for competitive ELISA

Name	Sequence
PepNS3	H- WCGP TNVDQDLVGWPA GLCL -OH
Pep8D4-35	H- WCGP GNLKADQDLVAG GLCL -OH
Pep8D4-38	H- WCGP NNVDSDLLVPHG GLCL -OH
Pep8D4-53	H- WCGP MVMCESGVDLVP GLCL -OH
Pep8D4-121	H- WCGP AEMQKADRDLVV GLCL -OH

FIG. 6.

Competitive ELISA analysis of the ability of synthetic peptide to prevent binding between 8D4 and NS3 protease. Approximately 100 ng of 8D4/ml was first preincubated with the indicated concentrations of PepNS3 (●), Pep8D4-35 (○), Pep8D4-38 (▵), Pep8D4-53 (□), Pep8D4-121 (▴), or MBP-NS3_1–631 (■) for 1 h at room temperature and then transferred to a microtiter plate that had been coated with 10 ng of MBP-NS3_1–631. After thorough washing with phosphate-buffered saline containing 0.05% Tween 20, MAbs bound to the target antigen were labeled with biotinylated goat anti-mouse IgG antibody and streptavidin-peroxidase conjugate and detected with ortho-phenylenediamine as a substrate for the peroxidase reaction. Peptide sequences used are given in Table 2.

DISCUSSION

HCV NS3 protease is an attractive target for the development of anti-HCV agents, since NS3 is one of the virus-encoded proteases and has several properties that are clearly distinct from those of related cellular enzymes (4, 11, 16, 18, 40). A number of groups have described different approaches to investigating NS3 protease inhibitors, such as substrate or product mimetics (27, 37), high-throughput screening of chemical compounds (19), and a selection of affinity ligands from a large pool of macromolecular random libraries (8, 9, 22, 31). Structure-based drug designs are a logical strategy for developing inhibitors, especially because the three-dimensional and solution structures of NS3 protease have been solved by X-ray crystallography (21, 28) and NMR spectroscopy (1, 6), respectively. It is, however, reported that the relatively featureless substrate-binding groove of NS3 protease would render the design of low-molecular-weight inhibitors extremely challenging (21). Immunochemical approaches employing MAbs are a useful alternative method for providing structural templates for designing low-molecular-weight lead compounds based on the complementarity-determining regions (CDR) (10, 36). However, the catalytic amino acids are mostly buried inside a cleft on the enzyme's surface (25), and this part of the molecule is regarded as having low immunogenicity (33, 34). In addition, the N-terminal portion of the NS3 protein has been shown to have low immunogenicity relative to that of the C terminus in both humans and mice (5). Preparation of antibody with inhibition potency toward NS3 protease thus seems a daunting task. In fact, we isolated only several MAbs that recognized the C-terminal region of NS3 when full-length NS3 protein (NS3_1–631) had been used to immunize mice (Y. Ohba, unpublished results). Martin et al. (31) described a camelized V_H domain antibody inhibitor of NS3 protease isolated from a phage-displayed synthetic repertoire, but the isolated proteins dimerized upon antigen interaction, bringing into question their potential and the single-domain nature of these proteins. Dimasi et al. (8) also described a minibody (“minimized” antibody-like protein) inhibitor of NS3 protease, but isolated minibodies showed relatively low affinity for NS3 protease.

In the present study, we isolated a series of MAbs by screening of hybridomas prepared from mice immunized with NS3_1–160 and extensively characterized one of these MAbs, 8D4. This MAb showed potent affinities both for binding to MBP-NS3_1–631 and for competitive inhibition of its protease activity with K_d and K_i values of 43 and 39 nM, respectively. 8D4 also displayed substantial inhibition potency toward polyprotein processing activity of NS3 in cis. Furthermore, detailed epitope mapping demonstrated that the reactive site of 8D4 included the catalytic residue Asp-81, clearly indicating the mechanism of inhibition of NS3 protease mediated by 8D4. It is thus interesting that 8D4 serves as a structural template for designing low-molecular-weight lead compounds that target the active site of NS3. In this regard, we have isolated genes for MAb 8D4 to determine the deduced amino acid sequences of CDRs, and a study focusing on the design of peptide analogs derived from the CDR residues of 8D4 is now in progress.

It has been proposed that MAbs be used as conformation-dependent reagents for investigating the structure of a protein, changes in its conformation, and its folding mechanism (15). We found in this study that the addition of the NS4A peptide, Pep4A_18–40, to the reaction significantly reduced the inhibition potency of 8D4, indicating decreased binding affinity of 8D4 to the NS3/Pep4A_18–40 complex. This is most likely due to the repositioning or conformational rearrangements of the 8D4-binding site containing Asp-81 of the NS3 protease induced by association with NS4A, consistent with the recent X-ray crystallographic studies showing that interaction with NS4A involves both a structural reorganization of the N-terminal domain and a rearrangement of the catalytic triad of NS3 protease (21, 29, 41). Specifically, the Asp-81 side chain is oriented away from His-57 in the NS3 crystal structure in the absence of the NS4A cofactor (28), whereas Asp-81 moves closer to Ser-139 and their side chains adopt a more chymotrypsin-like orientation in the presence of the NS4A peptide (21). On the other hand, the average solution structure of NS3 with or without the NS4A cofactor determined by NMR spectroscopy showed that the strands E1 to F1 bearing the Asp-81 residue are similarly positioned despite the lack of experimental constraints of this region (1). Moreover, Landro et al. (24) have demonstrated that the pH dependence of the NS3 hydrolysis reaction is not affected by the presence of NS4A, indicating that the presence of NS4A has no substantial effect on the pK_a of the catalytic residues. In any event, it is evident that structural configurations of Asp-81 and neighboring residues, whether in the presence or absence of the NS4A cofactor, play a critical role in the catalytic function of the NS3 protease. Kinetic and structural analysis of the binding of the protein complex (NS3, 8D4, and/or NS4A) may allow us to elucidate the detailed nature of the function of Asp-81 and neighboring residues in the NS3 protease catalytic activity, as well as that of the inhibitory activity of 8D4.

It has been shown that NS3 protease-mediated processing of HCV polyprotein at the NS3/4A, NS4A/4B, and NS4B/5A sites is dependent on the NS4A cofactor, whereas processing at the NS5A/5B site occurs in its absence but is significantly increased by this cofactor (3, 13, 26, 39). Among these sites, processing at the NS3/4A site is an intramolecular reaction, as evidenced by the absence of detectable NS3/4A precursors after a brief pulse-labeling period, the inability to cleave the NS3/4A site in trans, and the insensitivity of cleavage kinetics at this site to dilution (3, 4, 26, 40). Although Failla et al. (12) reported that cis-processing activity of the NS3 protease at the NS3/4A site is not affected even when the NS4A-interaction site of NS3 is removed by the truncation of 28 N-terminal residues, it is not clear yet whether the NS4A cofactor is required for the cis-processing reaction of NS3 at this site, since the cofactor is covalently linked to the protease, compensating for a defective protein-protein surface interaction. Our finding, however, that only the cis-processing activity of the NS3 protease at the NS3/4A site is inhibited by 8D4, which specifically recognizes NS3 but not the NS3/4A complex, strongly suggests that the NS3 protease cleaves this site prior to forming a complex with NS4A. It is thus likely that different conformational states of NS3 protease are responsible for the HCV polyprotein processing at different sites. This MAb 8D4 represents an interesting tool for investigating the mechanism and pathway of NS3-dependent HCV polyprotein processing in HCV-infected cells.

It is noteworthy that both 7E3 and 7E9 inhibited the protease activity of NS3 as assessed with a peptide but not with in vitro-translated polyproteins as substrates. It may be possible that the NS3 protease fused with MBP used in the assay has a specific conformation enabling both MAbs to bind, although binding affinities of those MAbs to MBP-NS3_1–631 appeared to be somewhat weaker than the affinity of 8D4. Alternatively, it may also be likely that both MAbs recognized the NS3 protease when it was complexed with the peptide substrate but not with the polyproteins. In this regard, it is worth noting that ¹¹⁷RRRG (amino acid residues that comprise the epitope recognized by 7E9) has been shown to form a loop between the β strands B2 and C2 (21, 28). This loop has been thought to play a role in maintaining a well-defined substrate-binding channel of NS3 as observed in alpha-lytic protease (14, 21) and has been predicted to contain the zinc ion coordination site (Cys-145 and His-149) that plays a role in structural stabilization of NS3 (7). This loop of NS3 has also been reported to be 14 to 15 residues shorter than that of chymotrypsin or alpha-lytic protease, resulting in a relatively solvent-exposed substrate-binding channel (21, 28). Indeed, kinetic analysis of inhibition by 7E9 of MBP-NS3_1–631 showed a competitive manner with respect to the substrate peptide (M. Misawa, unpublished results), indicating that 7E9 has an ability to directly block the binding of the substrate peptide to the NS3 protease. It will thus be interesting to elucidate the functions of the loop and three consecutive arginine residues located at the β strands B2 and C2 in the substrate specificity of NS3 protease and its interactions with 7E9.