Inhibition of hepatitis C virus NS2/3 processing by NS4A peptides. Implications for control of viral processing.

The NS2/3 protease of hepatitis C virus is responsible for a single cleavage in the viral polyprotein between the nonstructural proteins NS2 and NS3. The minimal protein region necessary to catalyze this cleavage includes most of NS2 and the N-terminal one-third of NS3. Autocleavage reactions using NS2/3 protein translated in vitro are used here to investigate the inhibitory potential of peptides likely to affect the reaction. Peptides representing the cleaved sequence have no effect upon reaction rates, and the reaction rate is insensitive to dilution. Both results are consistent with prior suggestions that the NS2/3 cleavage is an intramolecular reaction. Surprisingly, peptides containing the 12-amino acid region of NS4A responsible for binding to NS3 inhibit the NS2/3 reaction with K(i) values as low as 3 microM. Unrelated peptide sequences of similar composition are not inhibitory, and neither are peptides containing incomplete segments of the NS4A region that binds to NS3. Inhibition of NS2/3 by NS4A peptides can be rationalized from the organizing effect of NS4A on the N terminus of NS3 (the NS2/3 cleavage point) as suggested by the known three-dimensional structure of the NS3 protease domain (Yan, Y., Li, Y., Munshi, S., Sardana, V., Cole, J. L., Sardana, M., Steinkuhler, C., Tomei, L., De Francesco, R., Kuo, L. C., and Chen, Z. (1998) Protein Sci. 7, 837-847). These findings may imply a sequential order to proteolytic maturation events in hepatitis C virus.

Hepatitis C virus (HCV) 1 is a positive-strand RNA virus that is the major cause of non-A, non-B hepatitis (1,2). The HCV genome encodes a single polyprotein of approximately 3000 amino acids containing the viral proteins in the following order: C-E1-E2-p7-NS2-NS3-NS4A-NS4B-NS5A-NS5B. The NS proteins are thought to be nonstructural and are involved with the enzymatic functions of viral replication and processing of the viral polyprotein. Release of the individual proteins from the polyprotein precursor is mediated by both cellular and viral proteases (3)(4)(5)(6) (see also reviews in Refs. [7][8][9]. The proteolytic release of mature NS4A, NS4B, NS5A, and NS5B is catalyzed by the chymotrypsin-like serine protease contained within the N-terminal domain of NS3 (termed "NS3 protease"), whereas host cell proteases release C, E1, E2, and p7, creating the N terminus of NS2 at amino acid 810 (10,11). The cleavage between amino acids 1026 and 1027 of the HCV open reading frame that separates NS2 from NS3 is dependent upon protein regions of both NS2 and NS3 flanking the cleaved site; this autocleaving moiety is termed the NS2/3 protease (4,12). The cleavage is independent of the catalytic activity of the NS3 protease, as demonstrated with mutational studies (4,13).
The NS2/3 cleavage reaction has been studied in bacterial, mammalian, and insect cells or following in vitro translation of the protein (4,5,(13)(14)(15)(16). The protein region essential for NS2/3 cleavage activity has been approximately mapped to amino acids 898 -1207 of the HCV open reading frame (4,13,14). The catalytic mechanism of NS2/3 cleavage is unknown but speculated to be either a metalloprotease (18) or cysteine protease (19). Cleavage activity of in vitro translated NS2/3 is inhibited by EDTA, and activity is restored with metal ion re-addition (13,16).
There are no reported sequence homologies of the NS2 protein with proteins of known structure, and its N terminus is believed to be a transmembrane polypeptide (14). The NS3 protease domain structure is known in its mature form, however. The NS3 N terminus formed by NS2/3 cleavage is markedly affected by association with the cofactor peptide, NS4A. The x-ray crystallographic structure of NS3 bound to a peptide containing NS4A amino acids 21-34 (amino acids 1672-1685 of the HCV polyprotein) reveals that NS3 N-terminal residues 2-9 interact directly with NS4A to compose one of eight strands in an antiparallel ␤-sheet extending through the NS3 protease (20,21). In contrast, without NS4A, the N terminus of NS3 is poorly organized (22).
Considering the paucity of data regarding the biochemistry of the NS2/3 reaction, it is of interest to probe the effects of peptides that represent natural ligands of NS2/3 upon the NS2/3 reaction rate. These peptides include sequences containing the NS3 binding region of NS4A as well as sequences containing the NS2/3 cleavage site. Because the detailed investigation of the NS2/3 reaction has been hampered by the lack of assays using defined components, we have employed here the demonstrated technique of in vitro translation to produce active NS2/3. The unexpected result of this study is that NS4A is a potent inhibitor, whereas the reaction is unaffected by potentially competing cleavage-site peptides.

MATERIALS AND METHODS
DNA Constructions-Two DNA constructs were made for the synthesis of the HCV NS2/3 J strain RNA and its subsequent translation to proteins, which lack the N-terminal membrane binding region of NS2 but contain HCV residues 849 -1240, referred to as 849 -1240J and Mal849 -1240J. Codons 849 -1240 were amplified by polymerase chain reaction from pT7 (a generous gift from Dr. Nicola La Monica (14)). For 849 -1240J the HCV sequence was cloned into pET3c (Novagen), and for Mal849 -1240J the DNA was inserted into pETMalcH (a generous gift of Dr. B. Leiting (23)), to produce an open reading frame encoding the fusion protein Mal849 -1240, which includes Escherichia coli maltose-binding protein His 6 tag-HCV residues 849 -1240 (23). DNA for pCITE 810 -1615BK was a generous gift from Dr. Nicola La Monica. Upon transcription and translation, pCITE 810 -1615BK produces HCV residues 810 -1615 of the BK strain, all of NS2, and most of NS3 (810 -1615BK). The translation of this protein has been described pre-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Site-directed mutagenesis was performed with the Stratagene Quick Change method to generate nonprocessing mutants H952A and C993A in the expression constructs described above.
Peptides-Peptides were obtained by custom synthesis from Midwest Biotech (Fishers, IN) and were greater than 95% pure as judged from reverse-phase high pressure liquid chromatography. Effective molecular weights were obtained by quantitative amino acid analysis. All peptides were dissolved and diluted in Me 2 SO, so that the final concentration of Me 2 SO in every reaction was 5%.
In Vitro Transcription and Translation-Circular DNA plasmids were linearized with BLP1 (Bpu 1102) and purified with a Qiagen QiaEX II kit before transcription. RNA was transcribed with T7 RNA polymerase (Ambion Megascript kit), phenol/CHCl 3 -extracted, and ethanol-precipitated. Translations were with Promega or Ambion in vitro rabbit reticulocyte lysate translation kits at 30°C for 30 -40 min using [ 35 S]methionine as a label (Amersham Pharmacia Biotech). Translation was then inhibited by the addition of cycloheximide (250 M final concentration), and samples were immediately frozen on dry ice.
NS2/3 Autocleavage Reactions-Translated NS2/3 was thawed on ice and cleavage was initiated by incubation at 20°C, either in the original translation mixture or following a 10-fold dilution into a 10,000 molecular weight filtrate of reticulocyte lysate produced with Amicon Microcon-10 filters. Samples taken at the times indicated were combined with SDS-gel sample buffer and frozen on dry ice. NS2/3 cleavage reactions with the 810 -1615BK were performed with 1% Triton X-100 present, as described previously (16). At the completion of an experiment, frozen samples were placed in boiling water for 5 min, and radiolabeled proteins were separated by SDS-PAGE (14%). Prestained Novex molecular weight standards were used in estimation of molecular weights of the products. For peptide inhibition measurements, incubations were initiated by the addition of diluted lysate to a Me 2 SO solution of peptide in a tube held at 20°C.
The distribution of 35 S-labeled proteins on dried gels was determined with a PhosphorImager (Molecular Dynamics). Product bands were quantified and expressed as a proportion of total signal in the gel lane so that variations in gel lane loading were normalized. The product NS2 from 810 -1615BK was used to generate data shown for screening of peptides and IC 50 calculations because of its migration on gels in a region with less background than the higher molecular weight products and its ability to initiate the 810 -1615BK reaction with Triton X-100 (14). The IC 50 values were determined by first expressing the product level found as a fraction of the no-inhibitor control product level and then fitting the following equation, fractional activity ϭ a ϩ [b/(1 ϩ (x/c) d ), to the data, where a is the minimal level of fractional activity (tending to 0), a ϩ b is the maximal level (tending to 1), x is the concentration of inhibitor, c is the IC 50 , and d is a slope coefficient.

RESULTS
NS2/3 Processing Reactions-Typical NS2/3 processing reactions are shown in Fig. 1. The reaction occurred on a time scale of minutes, with the rate and final extent of reaction varying somewhat with the sequence expressed. NS2/3 810 -1615BK was cleaved to as much as 60% with a 3-h incubation, and the maltose-binding protein fusion, Mal849 -1240J, to nearly 100%. In all constructs, the mutations H952A or C993A prevented the appearance of products, as reported previously (4,13). Both Mal849 -1240J and 810 -1615BK were used in subsequent characterization of the NS2/3 reaction and inhibition. NS2/3 from 810 -1615BK was produced as a single 90-kDa band that cleaved itself to products of 65 (NS3) and 25 kDa (NS2), close to the expected molecular weights (Fig. 1A). In addition, 810 -1615BK did not begin cleavage until the addition to Triton X-100, as has been reported (16), thereby allowing reactions to be initiated at will without background cleavage products generated during the translation phase of the experiment. The translation products of Mal849 -1240J had the expected precursor molecular weight of 80 kDa, but also a smaller protein of 67 kDa (data not shown), possibly because of internal initiation, thus complicating the use of this version of NS2/3 for quantification of processing rates and inhibitor potencies.
Dilution of the NS2/3 precursor 10-fold into water completely prevented the processing reaction (data not shown). Dilution into a 10,000 molecular weight filtrate of rabbit reticulocyte  Table I.

Inhibition of NS2/3 Cleavage by NS4A 34512
lysate supported the reaction at a rate similar to that observed in undiluted lysate. Greater dilution of precursor (up to 40-fold) did not further change the rate of processing. The necessity of low molecular weight cellular component(s) for NS2/3 reactions was previously noted (16). Thus, for all subsequent measurements, a 10-fold dilution of in vitro synthesized NS2/3 into lysate filtrate was used. The accumulation of products for 810 -1615BK occurred at a rate of 0.04 min Ϫ1 , a rate comparable to the qualitative results previously published using different extensions of the NS2/3 polypeptide (4,13,16).
Peptide Inhibition of NS2/3 Processing-Peptides containing the cleavage site sequence of NS2/3, RLL*API (where * denotes the scissile bond), were tested for their effect upon NS2/3 processing in reactions of 60 min. No significant effect was observed with a variety of substrate or product-like peptides at a concentration of 500 M, as shown in Fig. 1A and Table I. NS4A peptides were inhibitory, as shown for peptide 33 in Fig. 1A. The inhibition appeared to occur immediately, since no pre-incubation of NS2/3 with peptide was performed before initiation of the reaction. Also, changes in inhibitor potency were not observed using 20-or 45-min incubations (data not shown).
The inhibition by NS4A peptides could be titrated; typical results are shown in Fig. 1, B and C. Potencies of 5.7 and 3.4 M were obtained for peptides 33 and 37, respectively. Peptides that represent only a portion of the region known to bind to NS3, such as peptides 66 and 67, did not inhibit (Table I). A peptide with the same amino acid composition as peptide 33 but with a randomized sequence (4ApepB) inhibited with an IC 50 nearly 25-fold greater than peptide 33 (Table I). The inhibition patterns shown in Table I were observed with both 810 -1615BK and Mal849 -1240J. Peptide HCV 33 also inhibits processing of the NS2/3 protein, 849 -1240J; therefore, the effect is not construct-or strain-specific (data not shown). Peptides unrelated to NS4A or the NS2/3 cleavage site were not inhibitory (Table I). DISCUSSION Despite numerous reports of NS2/3 processing reactions in cells and in vitro translation systems, a defined assay system containing purified NS2/3 protein is yet to be described. Existing techniques with in vitro translated NS2/3 have been exploited here to study the effects of potential peptide ligands of NS2/3. In the experiments described, the NS2/3 concentration is extremely low, possibly as low as 1 nM based upon an estimated specific activity of the radiolabeled protein produced (data not shown). The finding that dilution has no effect on the proportion of NS2/3 cleaved within a given time is consistent with an intramolecular mechanism. In addition, the lack of significant inhibition by potentially competing substrates, that is, peptides representing the cleavage site, is also consistent with an intramolecular cleavage. 2 NS4A is identified here as an inhibitor of NS2/3 processing. In judging the potency of the inhibition, it is most reasonable to consider the case of a mechanism of NS2/3 cleavage as depicted in Scheme 1, where the rate constant k determines the rate of product accumulation, and a simple association reaction of NS2/3 with inhibitor (I) is governed by an equilibrium constant, K i .

TABLE I
Inhibition of NS2/3 by peptides NS2/3 reactions with 810 -1615BK were performed for 60 min, as described under "Materials and Methods." The NS2/3 cleavage site-derived peptides 17 and 18 correspond to HCV amino acids 1020 -1033, J strain and BK strain, respectively. Other cleavage site-derived peptides are smaller segments of 17 or 18. The asterisk indicates the cleavage point, as determined in the NS2/3 protein (4,12). Peptide 33 represents NS4A residues 21-34 and has lysine residues appended to each end to enhance solubility. Peptide 4ApepB has the same amino acids as 33, but in a random order. Similar results were obtained with NS2/3 Mal849 -1240J. Abu, L-␣-aminobutyric acid.

Ac-AIILSGR
Ϫ12 74 Ac-RIILSGRK Inhibition of NS2/3 Cleavage by NS4A 34513 The rate of product (P) formation is given by dP/dt ϭ k⅐[NS2/ 3] n , where n is the order of the reaction with respect to concentration. We have determined with initial dP/dt measurements as a function of NS2/3 concentration that n ϭ 1. Additionally, according to Scheme 1, K i ϭ [I][NS2/3]/[NS2/3:I]. Thus, inhibitor binding that reduces the value of [NS2/3] 1 and dP/dt by 50% (the IC 50 ) is the K i . It is of interest, therefore, to compare the K i values found here, 3.4 and 5.7 M for peptides 33 and 37, respectively, with the K d values determined for a NS3:NS4A complex. It has been reported that with a NS4A peptide (KKKGSVVIVGRIILSGR-NH 2 ) at 1 mM, the association with NS3 protease is not observable in the absence of glycerol (24). Only with 50% glycerol is the reported K d (5.3 M) as low as the K i found here in the absence of glycerol. The longer NS2/3 may be more predisposed to avid NS4A association. 3 The x-ray crystallographic structures of NS3 protease alone and in an NS4A-bound form allow an explanation for the NS4A inhibition of NS2/3 processing. As depicted schematically in Fig. 2, NS4A binding brings the N terminus of NS3 into a stable ␤-sheet structure that is an integral part of the NS3 domain. The overall structure of NS2/3 may be affected similarly by NS4A binding, with residues critical for cleavage not positioned for the reaction in the NS4-bound state. Specifically, the cleavage site may be rendered inaccessible by NS4A binding, as depicted in Fig. 2.
The NS2/3 cleavage site sequence and the residues essential for cleavage are absolutely conserved in sequences of isolated strains of HCV, so that NS2/3 is presumed to be essential for viral replication. It may seem odd, therefore, that an essential cofactor for NS3 protease activity, the NS4A, would inhibit another essential process. The available data can be reconciled by the hypothesis that NS2/3 cleavage normally precedes NS4A release and NS3 protease processing of the rest of the NS region. In the absence of methods for replication of HCV in cell culture, the definition of the temporal sequence of HCV proteolysis events has not been possible. Our data presented here suggest that NS4A must not be released prematurely, or the NS2/3 cleavage would be inhibited. Although this idea appears inconsistent with in vitro translation studies showing that NS3/NS4A cleavage can occur in polyproteins wherein NS2/3 cleavage has not occurred (25), the rates of NS4A release have not been determined in those studies. The possibility remains that the normal sequence of proteolytic events in HCV replication includes NS2/3 cleavage as preceding NS3 protease action. The proper timing of proteolytic events has been demonstrated to be essential for viral replication in other systems such as HIV (26). The data presented here suggest that premature occupancy of the NS4A binding site is now a conceivable avenue for therapeutic intervention in the hepatitis C virus life cycle. The organizing effect of NS4A peptide upon the NS2/3 cleavage site is illustrated using the published structure of the NS3/NS4A complex (21). The magenta strand is NS3 residues 1-24 (HCV residues 1027-1050). In the upper diagram, the NS3 N terminus is drawn as a random coil because of its lack of definition from crystallographic studies.