Functional and Therapeutic Analysis of Hepatitis C Virus NS3·4A Protease Control of Antiviral Immune Defense*

Chronic hepatitis C virus (HCV) infection is a major global public health problem. HCV infection is supported by viral strategies to evade the innate antiviral response wherein the viral NS3·4A protease complex targets and cleaves the interferon promoter stimulator-1 (IPS-1) adaptor protein to ablate signaling of interferon α/β immune defenses. Here we examined the structural requirements of NS3·4A and the therapeutic potential of NS3·4A inhibitors to control the innate immune response against virus infection. The structural composition of NS3 includes an amino-terminal serine protease domain and a carboxyl-terminal RNA helicase domain. NS3 mutants lacking the helicase domain retained the ability to control virus signaling initiated by retinoic acid-inducible gene-I (RIG-I) or melanoma differentiation antigen 5 and suppressed the downstream activation of interferon regulatory factor-3 (IRF-3) and nuclear factor κB (NF-κB) through the targeted proteolysis of IPS-1. This regulation was abrogated by truncation of the NS3 protease domain or by point mutations that ablated protease activity. NS3·4A protease control of antiviral immune signaling was due to targeted proteolysis of IPS-1 by the NS3 protease domain and minimal NS4A cofactor. Treatment of HCV-infected cells with an NS3 protease inhibitor prevented IPS-1 proteolysis by the HCV protease and restored RIG-I immune defense signaling during infection. Thus, the NS3·4A protease domain can target IPS-1 for cleavage and is essential for blocking RIG-I signaling to IRF-3 and NF-κB, whereas the helicase domain is dispensable for this action. Our results indicate that NS3·4A protease inhibitors have immunomodulatory potential to restore innate immune defenses to HCV infection.

Hepatitis C virus (HCV) 2 establishes persistent infections in the majority of exposed individuals. There are nearly 200 mil-lion people worldwide with chronic HCV infection, and it is a major cause of hepatitis and liver disease. Moreover, HCV is associated with the development of liver cancer and is currently the leading indication for liver transplants (1,2). HCV is a member of the Flaviviridae family of enveloped, single-stranded, positive-sense RNA viruses. The 9.6-kilobase genome of HCV encodes a single polyprotein that is post-translationally processed into at least 10 structural and nonstructural (NS) proteins by a combination of host and viral proteases (3). In addition to their primary role in HCV genome replication and virion maturation, various HCV proteins have been shown to antagonize host immune defenses. HCV control of the innate antiviral response and interferon (IFN) ␣/␤ antiviral defenses may provide a cellular foundation for viral persistence (4).
The innate antiviral response to RNA virus infection is triggered when intermediates of viral replication, including viral RNA or protein products, are recognized by specific cellular pathogen recognition receptors (5). Retinoic acid inducible gene-I (RIG-I) and melanoma differentiation-associated gene 5 (MDA5) are cytosolic pathogen recognition receptors that bind to viral RNA and double-stranded RNA, albeit with variable efficiency (6,7). Both are DEXD/H-box RNA helicases and encode tandem amino-terminal caspase activation and recruitment domains (CARDs) (6). RIG-I is an essential pathogen recognition receptor for various negative-strand viruses and Flaviviridae (8), including HCV (9,10). RNA binding by RIG-I and MDA5 promotes a conformation change that potentiates signaling by the CARDs (9,11) to a CARD-containing adaptor protein termed IFN-␤ promoter stimulator 1 (IPS-1; for review, see Ref. 12), located on the mitochondrial membrane. IPS-1 confers downstream signaling to the latent cytoplasmic transcription factor IFN regulatory factor-3 (IRF-3) and nuclear factor-B (NF-B) (13)(14)(15)(16). IRF-3 activation commences with its virus-induced phosphorylation, dimerization, and nuclear translocation, whereupon it binds to the promoter region of IRF-3-dependent genes, including IFN-␤, IFN-stimulated gene (ISG) 15, and ISG56 (17)(18)(19). Parallel NF-B activation induces a variety of proinflammatory genes and cooperates with IRF-3 to induce IFN-␤ expression (20). ISGs have antiviral and immunomodulatory activity that limit HCV infection (4,(21)(22)(23), and their expression marks the effector stage of the innate antiviral response (24). HCV evades the innate antiviral response in part through the actions of the viral NS3⅐4A protease-helicase complex, which ablates RIG-I and MDA-5 signaling (6,10,25) through proteolytic cleavage and inactivation of IPS-1 (16,26,27). NS3⅐4A cleaves IPS-1 at Cys-508, thereby releasing it from the mitochondrial membrane and preventing its downstream activation of immune effectors (27). However, the NS3⅐4A structural motifs and features that direct IPS-1 cleavage and loss of RIG-I and MDA5 signaling have not been defined.
NS3⅐4A is a complex, bifunctional molecule ( Fig. 1) that is essential for NS protein processing and viral RNA replication (28,29). The amino-terminal region of NS3 contains a serine protease that non-covalently binds its cofactor, NS4A. NS4A is a small peptide necessary for efficient processing of the HCV polyprotein by NS3 and is thought to be involved in tethering the complex to intracellular membranes (30,31). The carboxyl terminus of NS3 possesses a DEXD/H box helicase with nucleoside triphosphatase activity believed to participate in RNA unwinding during viral RNA replication (32). Its helicase activity was found to be positively modulated by the protease domain and NS4A (33)(34)(35). Conversely, conserved motif VI of the helicase domain has been shown to affect protease activity (33, 36 -38). Moreover, structural studies of full-length NS3 indicate that the carboxyl terminus of the helicase domain may interact with the protease active site (39), suggesting a structural and functional interconnectivity of the two domains. NS3 also encodes an arginine-rich region shown to bind other host proteins (40,41). NS3⅐4A interaction with its cellular targets could, therefore, occur through a variety of processes mediated by its helicase or protease domains.
Several peptidomimetic NS3⅐4A protease active site inhibitors are in preclinical development as HCV antiviral drugs (42). Future use of these compounds could represent an alternative or supplement to the current treatment regimen for HCV infection, IFN-␣ and ribavirin combination therapy, which is only effective in about 50% of patients (43). NS3⅐4A protease inhibitors may also exhibit immunomodulatory properties by removing the protease-dependent blockade to the innate antiviral response during HCV replication (23,25), although the impact of IPS-1 in this process is not known. The current study was undertaken to define the domain structure and function of NS3⅐4A that regulates the host cell innate antiviral response.

EXPERIMENTAL PROCEDURES
Expression Cloning and Site-directed Mutagenesis-pFLAG NS3⅐4A and pFLAG NS3 were generated as described (23). Constructs ( Fig. 1) containing NS3 helicase domain truncations at amino acids (aa) 1206 (protease domain alone, Prot), 1238, 1486, and 1501 as well as a protease domain deletion (NS3 aa 1207-1658, Hel) were made as follows. NS3 mutants were cloned by PCR (with primers containing a 5Ј HindIII site and a 3Ј XbaI site) using BD Advantage-HF 2 PCR kit (BD Biosciences) into pCR2.1 vector (TA cloning kit, Invitrogen). These constructs were then subcloned into the HindIII and XbaI sites of pFLAG-CMV-2 vector (Sigma). A protease active site mutant (NS3⅐4A S1165A) and deletion of an arginine-rich region (aa 1487-1501; NS3⅐4A ⌬1487-1501) were constructed using QuikChange XL site-directed mutagenesis kit (Strat-agene). Because NS4A forms part of the protease active site and is indispensable for full protease activity (31), a scheme was devised allowing expression of NS4A with the NS3 mutants. Expressing NS4A from a separate plasmid was not sufficient to result in efficient protease activity (data not shown). To circumvent this problem we created a primer containing an aminoterminal KpnI site, 12 amino acids of NS4A (aa 21-32; which intercalate into the NS3 protease active site), and a flexible linker (GSGS) region and subcloned this into the HindIII sites of the pFLAG NS3 1026 -1206, creating a fully functional, single-chain (SC) NS3⅐4A protease domain (SC Protease). This strategy was similar to that described by Taremi et al. (44). The flexible linker was also subcloned into pFLAG NS3 (SC NS3), pFLAG NS3 1026 -1238 (SC NS3 1026 -1238), pFLAG NS3 1026 -1486 (SC NS3 1026 -1486), and pFLAG NS3 1026 -1501 (SC NS3 1026 -1501) to generate the SC constructs indicated in parentheses. NS3⅐4A, NS3⅐4A S1165A, and the SC Protease were subcloned into pEF1/Myc-His B (Invitrogen) to allow high efficiency expression of each. NS3⅐4A and NS3⅐4A S1165A were cut with HindIII, blunt-ended with Kleenow fragment, digested with XbaI, and subcloned into the EcoRV and XbaI site of pEF1/ Myc-HisB. The SC Protease was subcloned into the SacI and XbaI sites of pEF1/Myc-His B. Primer sequences for all constructs are available upon request.
Cells, Viruses, and Protease Inhibitors-Huh7 human hepatoma cells (47), Huh7.5, and HEK 293 cells were propagated in Dulbecco's modified Eagle's medium (Cellgro) supplemented with 10% fetal bovine serum (Hyclone), 1ϫ nonessential amino acids, 2 mM L-glutamine, antibiotic-antimycotic solution, and 1 mM sodium pyruvate as described previously (25). Huh7.5 cells, a subline of Huh7 cells kindly provided by C. Rice, contain a defect in RIG-I and do not signal through the RIG-I pathway (10). UNS3⅐4A and UNS3 cells (a gift from D. Moradpour) are osteosarcoma cells conditionally expressing HCV 1a NS3⅐4A or HCV 1a NS3, respectively, and were described previously (30,48). HP and K2040 cells are Huh7 cells harboring cell cultureadapted subgenomic HCV 1b replicons (11,22). Cell culture methods for UNS3⅐4A, HP, and K2040 cells have been described elsewhere (11,22,23). For Sendai virus infection (Cantell strain, Charles River Laboratories), 1 ϫ 10 5 or 2 ϫ 10 4 cells/well were seeded into 12-or 48-well dishes respectively, and where indicated, transfected with expression constructs. Twenty-four hours after seeding or transfection, cells were washed in phosphate-buffered saline and mock-infected or infected with 100 hemagglutinin units of Sendai virus per ml of serum-free media for 1 h. Three volumes of culture media were then added to wells, and cells were incubated another 17 or 19 h as described (23). HCV 2a stock was produced as described previously (27). For HCV infection ϳ4 ϫ 10 4 cells seeded in a 12-well plate were infected at a multiplicity of infection of 1 with HCV 2a in culture media for 3 h and washed in phosphatebuffered saline, and culture media was added. Cells were then incubated until collection at the indicated time points. SCH6 (kindly provided by Schering-Plough) (23) and ITMN-C (a gift from Dr. S. Seiwert at Intermune, Brisbane, CA) protease inhibitor treatments were conducted by replacing culture medium with medium containing 20 M SCH6 or the indicated amount of ITMN-C.
Confocal Microscopy-2 ϫ 10 4 cells were plated in 4-chamber slides, 200 ng of the indicated expression plasmids were transfected, and cells were collected 12 or 24 h later. Cells were fixed in 4% paraformaldehyde and blocked in phosphate-buffered saline containing 10% fetal bovine serum and 1% Triton X-100. Antibodies against the FLAG epitope (Sigma), IPS-1 (a gift from Z. Chen), IRF-3 (kindly provided by M. David), and HCV NS3 NCL (Novacastra laboratories) were used in combination with Alexa 488-or Alexa 594-conjugated secondary antibodies (Invitrogen). Slides were mounted with Prolong Gold (Invitrogen) and examined using a Zeiss Pascal laser scanning confocal microscope and Zeiss LSM software. The images shown are Յ7 m optical slices.
Promoter Luciferase Reporter Assays-2 ϫ 10 4 cells/well were seeded into 48-well dishes and cotransfected with 25 ng of pIFN-␤luc or pPRDII-luc and 5 ng of pCMV-Renilla along with the indicated amounts of expression constructs. Cells were then cultured alone or subjected to virus infection and harvested 20 h later for luciferase assay as described (23). Results were calculated relative to Renilla luciferase values.
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assay analysis of NF-B DNA binding activity was conducted on UNS3⅐4A and UNS3 nuclear extracts as described (46). Briefly, cells were grown in the presence (ϪNS3⅐4A) or absence (ϩNS3⅐4A, ϩNS3) of 1 mg/ml tetracycline, and protease inhibitor (20 M SCH6 or 10 M ITMN-C) or Me 2 SO (mock treatment) was added to the media for 24 h. Cells were then mock-infected or infected with 100 hemagglutinin units/ml of Sendai virus, and infection was allowed to proceed 18 h. As a positive control for NF-B activation, some wells were washed twice in phosphate-buffered saline and treated for 30 min with 20 ng/ml tumor necrosis factor-␣ in serum-free media. Antibody supershifts were performed using antibodies against NF-B p50 or p65 subunits (Santa Cruz Biotechnology). NF-B consensus and mutant oligonucleotides were obtained from Santa Cruz Biotechnology. Non-radiolabeled, cold competitor was used as a negative control.
Statistical Analysis-Differences among treatments were examined using unpaired t-tests and were considered statistically significant when p Ͻ 0.02 and where marked with an asterisk. Statistical differences in which p Յ 0.001 were marked with two asterisks.

RESULTS
Characterization of NS3⅐4A Constructs-To determine the minimal region of NS3⅐4A that could modulate the innate antiviral response, a series of FLAG epitope-tagged NS3 truncation and deletion mutants were constructed (Fig. 1). To assess indi- NS3⅐4A mutants (lower) were constructed as described under "Materials and Methods." The star indicates the serine 1165 (of the catalytic triad) mutation to alanine, which ablates protease activity. NS4A is a 54-aa cofactor for the NS3 serine protease domain. Single-chain constructs contain the 12 aa (aa 21-32) of NS4A essential for intercalation into the NS3 protease active site connected to various NS3 truncation mutants via a flexible linker, enabling direct interaction of the NS4A residues with the NS3 protease without an intramolecular cleavage event. The curved semicircular line represents the flexible linker region. V-shaped lines in NS3⅐4A ⌬1487-1501 represent the area deleted in this construct. vidual domains of NS3 and NS4A in innate immune regulation, we produced constructs encoding NS4A alone (23), the NS3 protease domain (Protease), helicase domain (Helicase), or fulllength NS3 lacking the arginine-rich region (⌬Arg) as well as full-length NS3⅐4A harboring a protease activate site mutation (S1165A) as a negative control. These were placed under the control of the CMV-Immediate-Early promoter. Previous studies have identified a 12-amino acid region of NS4A (aa 21-32) that intercalates into the NS3 protease active site and is indispensable for full protease activity (31). To express full NS3⅐4A protease activity from a single plasmid, we attached these 12 aa of NS4A to each NS3 mutant via a flexible linker, creating a fully functional, SC NS3⅐4A protease encoding full-length NS3 (SC NS3), progressive deletions of the NS3 helicase domain (SC 1501, SC 1486, SC 1238), or the protease domain alone (SC Protease). Use of similar SC constructs as a tool for examining the NS3⅐4A protease structure and function has been well established in crystallization studies (39) and studies identifying small molecule protease inhibitors (49). Moreover, Taremi et al. (44) found that such SC Protease constructs displayed protease activity comparable with wild type (WT) NS3⅐4A. As shown in Fig. 2A, when coexpressed in Huh7 cells, a NS5AB polyprotein substrate was efficiently cleaved by WT NS3⅐4A, SC Protease, SC NS3, ⌬Arg, SC 1238, SC 1486, and SC 1501 constructs (middle panel, lanes 4, 7, 8, 14, 16 -18, respectively). The NS5AB polyprotein was not cleaved by NS4A, NS3⅐4A S1165A, Helicase, SC NS3 S1165A, or protease (lanes 3, 5, 6, 9, 15, respectively) and was only partially cleaved by NS3 alone (Fig. 2A, lane  2). When expressed from the elongation factor-1 promoter expression plasmid (pEF), pEF WT NS3⅐4A, and pEF SC Protease but not pEF NS3⅐4A S1165A also mediated NS5AB cleavage (compare lanes [11][12][13]. These results validate the expression and cleavage activity of the NS3 constructs and show that efficient NS3⅐4A proteolytic activity requires the protease domain and its NS4A cofactor but does not require the NS3 helicase domain. The Helicase Domain Is Dispensable for Inhibition of Innate Defense Signaling-To determine the NS3 structural requirements for inhibition of host antiviral signaling, we evaluated each construct for its ability to regulate IFN-␤ promoter induction in response to Sendai virus infection. It should be noted that Sendai virus and HCV trigger the innate antiviral response through similar mechanisms that are blocked by NS3⅐4A (10,23,25). WT NS3⅐4A significantly suppressed Sendai virus induction of IFN-␤ promoter activity comparable with a dominant-negative IRF-3 control construct (Fig. 2B, IRF3-⌬N). However, constructs lacking NS4A and/or the intact protease domain of NS3 failed to block signaling to the IFN-␤ promoter. Of note, the SC NS3 construct, similar to that used in crystallization studies (39), mediated a weaker albeit significant suppression of virus signaling to the IFN-␤ promoter. Further studies revealed that the SC NS3 protein did not localize to membrane-bound compartments (data not shown), suggesting its decreased ability to block viral activation of the IFN-␤ promoter could be due to aberrant localization. We also assessed Sendai virus-induced ISG expression in cells transfected with constructs encoding WT NS3⅐4A, NS3⅐4A S1165A, or the SC Protease. As shown in Fig. 2C, WT NS3⅐4A and the SC Protease suppressed ISG production, but ISGs were induced in cells expressing the protease-deficient NS3⅐4A S1165A mutant. Together these results indicate that the fully functional NS3 protease domain with its minimal NS4A cofactor is necessary for suppressing the host cell signaling of IFN-␤ promoter expression.
The Functional NS3⅐4A Protease Domain Inhibits Signaling by RIG-I and MDA5-NS3⅐4A suppresses RIG-I signaling of the innate antiviral response (25). We verified that our WT NS3⅐4A construct could suppress RIG-I-enhanced virus signaling as well as constitutive signaling by ectopic N-RIG (encoding the CARD domains alone) (Fig. 3A). We then assessed regulation of RIG-I signaling to the IFN-␤ promoter by a subset of our NS3 constructs. As shown in Fig. 3B, the SC Protease and ⌬Arg constructs suppressed N-RIG signaling of the IFN-␤ promoter to an extent similar to WT NS3⅐4A. Expression of NS3 alone, NS4A, NS3⅐4A S1165A, or the NS3 helicase domain had no significant effect on promoter activation. Moreover, when expressed in Huh7 cells, WT NS3⅐4A and the SC Protease suppressed the expression of the IRF-3 target genes, ISG56 and ISG15, but NS3⅐4A S1165A and the NS3 protease domain failed to suppress IRF-3 target gene expression (Fig. 3C).
We also examined NS3 construct regulation of MDA5 signaling. To determine the effect of MDA5 regulation in the absence of RIG-I, we utilized Huh7.5 cells lacking RIG-I function (10). As shown in Fig. 4A, Huh7.5 cells failed to activate the IFN-␤ promoter in response to Sendai virus infection, but ectopic MDA5 complemented this defect. Moreover, we verified that our WT NS3⅐4A construct could suppress WT MDA5 or N-MDA5 (encoding the CARD domains) signaling to the IFN-␤ promoter in Huh7.5 cells. To determine the minimal domain of NS3⅐4A necessary to block MDA5 signaling, we evaluated the ability of constructs encoding WT NS3⅐4A (positive control), NS3⅐4A S1165A (negative control), the SC Protease, and Protease to regulate signaling to the IFN-␤ promoter by N-MDA5 (Fig. 4B). The SC Protease inhibited promoter activation to the same extent as WT NS3⅐4A and the IRF3-⌬N control, but neither the NS3 protease domain alone nor NS3⅐4A S1165A blocked promoter signaling. Both WT NS3⅐4A and the SC Protease suppressed ISG15 expression in response to ectopic N-MDA5 (Fig. 4C). Taken together, these results demonstrate that the SC Protease is sufficient to block signaling by both RIG-I (Fig. 3) and MDA5 (Fig. 4) in a process that requires the NS3 protease domain and its NS4A cofactor. The NS3 helicase domain, however, is dispensable for innate immune response regulation.
The SC Protease Blocks Activation of IRF-3 and Exhibits a Subcellular Localization Pattern Similar to WT NS3⅐4A-To verify that the SC Protease was sufficient for inhibition of IRF-3 activation, we examined IRF-3 and SC Protease distribution in transfected Huh7 cells. Sendai virus infection of cells caused the redistribution of IRF-3 from the cytoplasm to the nucleus (23), consistent with its virus-induced activation, but this response was blocked in control cells expressing WT NS3⅐4A (Fig. 5). Expression of the SC Protease also blocked IRF-3 activation, maintaining it in a cytoplasmic context during Sendai virus infection. Notably, despite the lack of the putative amino-terminal membrane-anchoring motif of NS4A (30), the SC Prote-ase localized to a subcellular compartment similar to WT NS3⅐4A. Thus, the NS3 protease domain and a minimal NS4A cofactor are sufficient to confer proper localization and inhibition of IRF-3 activation.
Regulation of NF-B by NS3⅐4A-RIG-I and MDA5 signal the parallel activation of IRF-3 and NF-B (6), and we sought to  examine the effect of NS3⅐4A and its requirement for protease activity on regulation of NF-B function. We performed an electrophoretic mobility shift assay to measure NF-B binding to a target probe in nuclear extracts of cells treated with tumor necrosis factor-␣ (TNF-␣; control) or infected with Sendai virus. Tumor necrosis factor-␣ or Sendai virus infection triggered NF-B binding (Fig. 6A, lanes 1 and 5, respectively). Expression of NS3⅐4A (lane 11) but not NS3 alone (lane 17) blocked Sendai virus-induced NF-B binding. The NS3⅐4A blockade of NF-B binding was relieved when cells were treated with SCH6 or ITMN-C peptidomimetic active site inhibitors of the NS3⅐4A protease (lanes 13 and 15, respectively). These results were verified by luciferase reporter assay, in which WT NS3⅐4A suppressed Sendai virus induction of an NF-B-dependent PRDII-luciferase promoter and its enhancement by ectopic RIG-I or MDA5 (Fig. 6B). To determine the domain of NS3⅐4A necessary for this effect, HEK 293 cells were transiently transfected with the indicated NS3 construct followed by Sendai virus infection (Fig. 6C, left panel) or were cotransfected with a construct encoding constitutively active N-RIG to trigger PRDII promoter signaling (Fig. 6C, right panel). The SC Protease inhibited activation of the PRDII promoter element in response to both Sendai virus infection and ectopic N-RIG. Similar results were observed when the PRDII promoter element was stimulated by ectopic N-MDA5 (data not shown). We also examined the influence of NS3⅐4A upon virus-induced phosphorylation of the NF-B inhibitor, IB␣. As shown in Fig. 6D, Sendai virus infection of HEK 293 cells induced the phosphorylation of IB␣ within 20 h post-infection, but expression of either WT NS3⅐4A or the SC Protease but not NS3⅐4A S1165A prevented virus-induced IB␣ phosphorylation (compare lanes 2, 4, 6, and 8). These results link HCV control of NF-B function with NS3⅐4A protease domain regulation of RIG-I and MDA5.
The Protease Domain of NS3⅐4A Is Sufficient to Target and Cleave IPS-1-RIG-I and MDA5 signal IRF-3 and NF-B activation through the IPS-1 adaptor protein (13), which is targeted and cleaved by NS3⅐4A to inactivate signaling (16,27). To determine whether the functional NS3⅐4A protease domain was sufficient to physically target and cleave IPS-1, we examined complex formation between IPS-1 and the SC Protease or a protease-defective derivative (Fig. 7A, wt and SA, respectively) in transfected Huh7 cells. Immunoprecipitation studies revealed that the functional SC Protease and an SC Protease S1165A mutant each bound to IPS-1 (Fig. 7A, lanes 2 and 3). When compared with the SC Protease, we observed an apparent reduction in the amount of SC Protease S1165A mutant bound to IPS-1 in these experiments. Although this could be due to protein expression or stability differences between the SC Protease constructs, it could also be explained by potential changes in the active site conformation caused by the serine to alanine switch, possibly resulting in decreased affinity for cleavage substrates. Importantly, the interaction between IPS-1 and the SC Protease was not disrupted by ITMN-C protease inhibitor treatment, and the SC Protease maintained an interaction with the cleaved form of IPS-1 (Fig. 7A, lanes 3 and 4). Ectopic IPS-1 is a potent trigger of the innate antiviral response (13), and we found that it triggered endogenous ISG15 expression in Huh7 cells (Fig. 7B). This reaction was blocked by both the SC Protease and WT NS3⅐4A concomitant with IPS-1 cleavage. Thus, the SC Protease is sufficient to target and cleave IPS-1. We also found that the SC Protease suppressed IFN-␤ promoter induction by ectopic IPS-1 as well as WT NS3⅐4A (Fig.  7C). Furthermore, the SC Protease blocked NF-B-dependent induction of the PRDII promoter element by ectopic IPS-1, but neither NS3 nor the protease or helicase domains affected promoter induction (Fig. 7D). Thus, the targeted cleavage of IPS-1 by the NS3⅐4A protease domain inhibits downstream signaling to IRF-3 and NF-B.
NS3⅐4A Protease Inhibitor Prevents Cleavage and Restores Localization of IPS-1-To determine the influence of the NS3⅐4A protease and protease inhibitor treatment on IPS-1 localization and function, we examined the subcellular distribution of IPS-1 in the presence of NS3⅐4A alone and in the context of HCV RNA replication when cells were treated with ITMN-C. As shown in Fig. 8A, IPS-1 showed a characteristic thread-like appearance in the cell cytoplasm, consistent with its mitochondrial membrane localization (27). However, in the presence of NS3⅐4A, this localization was disrupted, and IPS-1 exhibited the diffuse cytoplasmic staining pattern previously described (26,27). When the cells were exposed to the ITMN-C protease inhibitor for 24 h, IPS-1 was restored to its proper localization in the presence of NS3⅐4A (Fig. 8A). We further examined this regulation in the context of HCV RNA replication in two cell lines harboring genetically distinct HCV replicons that are differentially resistant (HP replicon) and sensitive (K2040 replicon) to the antiviral actions of IFN therapy (11,22). In the presence of NS3, IPS-1 membrane localization was disrupted, whereas K2040 cells with no visible NS3⅐4A staining typically retained a normal pattern of IPS-1 localization (Fig.  8B, mock-treated). When the cells were treated with ITMN-C, IPS-1 returned to a native distribution within a 24-h time  course even in the presence of NS3⅐4A (Fig. 8B, ϩITMN-C). Moreover, during ITMN-C treatment the colocalization of NS3⅐4A with IPS-1 became apparent, suggesting they are initially present in the same subcellular milieu and that NS3⅐4A cleavage of IPS-1 disperses it from this environment (Fig. 8, A and B).
ITMN-C Restores the Endogenous Innate Immune Response Triggered by HCV RNA Replication and HCV Infection-To examine the effect of protease inhibitor treatment on IPS-1 and the innate antiviral response against HCV, we examined ITMN-C treatment of Huh7 cells harboring a HCV 2a replicon or infected with the JFH1 strain of HCV 2a. Treatment of HCV 2a replicon cells resulted in alteration of endogenous IPS-1 from a predominantly cleaved form to its native, full-length form concomitant with induction of ISG56 expression over a 48-h time course (Fig. 9, A and B). To verify these results in the context of an actual HCV infection, Huh7 cells were infected with JFH1 (50) at a multiplicity of infection of 1 for 48 h followed by 24, 36, or 48 h of ITMN-C treatment (Fig. 9C). After 48 h of HCV infection, endogenous IPS-1 levels were reduced with a large portion of the remaining protein present as the cleaved form, consistent with previously published data (27). Within 24 h of ITMN-C treatment and throughout the time course, we observed a recovery of full-length IPS-1 levels with corresponding induction of ISG56 expression and reduction in viral protein abundance (Fig. 9C).

DISCUSSION
Our results show that the NS3 protease domain and minimal NS4A cofactor are sufficient to inhibit antiviral signaling through RIG-I and MDA5 and block activation of the downstream transcription factors IRF-3 and NF-B. This is accomplished by protease domain targeting and cleavage of IPS-1.
Our data indicate that the NS3 helicase domain does not play a role in innate immune regulation and inactivation of IPS-1. Moreover, infection studies demonstrated for the first time that NS3⅐4A protease inhibitor therapy of HCV infection can effectively remove the NS3⅐4A blockade to the innate immune response, restoring RIG-I signaling to IPS-1 and activation of ISG expression in infected cells.
The protease domain of NS3 has been shown to mediate internal interactions with the carboxyl-terminal helicase domain, in which the latter has been suggested to potentiate protease activity (37)(38)(39). We found that the helicase domain was dispensable for NS3⅐4A control of the innate antiviral response and signaling by RIG-I and MDA5. That the helicase domain is not involved in innate immune control is further supported by the observations that the SC Protease, lacking the complete helicase domain of NS3, mediated binding and cleavage of IPS-1 similar to WT NS3⅐4A. Taken together, these results imply that the NS3 helicase domain does not play a role in directing NS3⅐4A interactions with IPS-1. Importantly, our results show that neither full-length NS3 nor the NS3 protease domain alone could block signaling by RIG-I or MDA5 when expressed in the absence of NS4A; however, NS4A alone does not regulate the innate immune response. In contrast, each NS3 construct efficiently suppressed signaling when expressed as a SC construct with aa 21-32 of NS4A, encoding the NS4A central core region integral to NS3 protease activity (31). Previous work has associated NS4A aa 2-19 with localizing NS3 to intracellular membranes (30). Despite lacking this domain of NS4A, our SC constructs effectively cleaved NS5AB and blocked innate immune signaling. Moreover, when expressed in Huh7 cells, the SC Protease construct demonstrated a subcellular distribution pattern equivalent to WT NS3⅐4A that similarly supported IPS-1 targeting. We conclude that whereas NS4A is required for full proteolytic action of NS3, localization and IPS-1 targeting by NS3⅐4A are controlled by residues located within the NS3 protease domain and/or the NS4A central core.

Structure-Function Analysis of HCV NS3⅐4A
The amino-terminal 21 aa of NS3 consist of several highly hydrophobic residues, and crystal structure analysis showed that this region extends away from the rest of the protease (51), suggesting it may be involved in membrane tethering. Additionally, a recent study demonstrated a mitochondrial membrane localization of NS4A in the context of HCV RNA replication (52), implicating NS4A in directing NS3 to the mitochondria and site of interaction with IPS-1. These possibilities are presently being examined.
We found that the SC Protease could bind IPS-1 and that this interaction was maintained in the presence of the peptidomimetic protease active site inhibitor ITMN-C, which abrogated IPS-1 proteolysis (Fig. 7A). Under these conditions ITMN-C is expected to occupy the protease active site and catalytic residues (53). Thus, the SC Protease may bind IPS-1 independently of substrate occupancy in the protease active site. This suggests a model in which NS3⅐4A may first interact with IPS-1 through protease domain residues outside the catalytic pocket, thereby positioning it for subsequent proteolysis. NS3⅐4A exhibits a shallow substrate binding cleft that might accommodate other protein substrates (39,54,55). Indeed, the TIR domain-containing adaptor-inducing interferon (TRIF) adaptor protein has been identified as an in vitro substrate of NS3⅐4A proteolysis (56). In this case residues within the 3/10 helix of NS3 located adjacent to the protease active site bind to a region of TRIF near the cleavage site (57). This shared relationship between protein interaction and cleavage could indicate that NS3⅐4A may target and interact with TRIF and IPS-1 through similar processes. The mechanics of the NS3⅐4A-IPS-1 interaction are currently under investigation.
Our study demonstrates that NS3⅐4A proteolysis of IPS-1 blocks signaling by both RIG-I and MDA5 and ablates virus activation of downstream IRF-3 and NF-B, thus confirming that IPS-1 serves as the essential adaptor for RIG-I and MDA5 signaling. In the case of NF-B, we found that the NS3⅐4A or SC Protease blockade of NF-B activation was concomitant with suppression of IB-␣ phosphorylation and that NF-B activation could be fully restored when cells were treated with a NS3⅐4A protease inhibitor. Thus, the RIG-I/ MDA5 axis may signal NF-B activation through a canonical process of kinase phosphorylation of IB and unmasking of NF-B DNA binding activity (58). Similar to the control of IRF-3 signaling (23,25,27), these processes of NF-B signaling are disrupted by IPS-1 proteolysis. As a signaling adaptor protein, IPS-1 may serve to recruit factors of IRF-3 and/or NF-B activation. This idea is supported by studies that have demonstrated IPS-1 interactions with various signaling components, including the IKK⑀ protein kinase (59), TRAF3 (60), and FADD (13) into a complex with IPS-1. In the context of HCV infection, an IPS-1 signaling complex would Cells were then fixed and stained for IPS-1 (green), NS3 production (red), and nuclei (blue), and images from Ͻ0.7 m sections were collected using a Zeiss laser scanning confocal microscope. B, effect of protease inhibitor on subgenomic HCV replicon dispersion of IPS-1. Replicon cells resistant (HP) or sensitive (K2040) to IFN-␣ treatment were cultured in the presence or absence of 1 M ITMN-C protease inhibitor for 12 or 24 h and processed for confocal microscopy as in A. Areas of IPS-1 and NS3⅐4A colocalization are shown by yellow staining. NS3 was undetectable in ϳ25% of K2040 cells, and these cells exhibited a normal IPS-1 mitochondrial distribution pattern. These data are representative of two independent experiments. possibly be disrupted through the targeted proteolysis of IPS-1 by NS3⅐4A to ablate a wide range of immune signaling action.
NS3⅐4A protease inhibitors are being developed for use as future HCV therapeutics (42), and our studies define a novel role for these compounds as immunomodulatory agents. NS3⅐4A cleavage of IPS-1 during HCV RNA replication and infection results in IPS-1 release from the mitochondrial membrane and diffuse redistribution throughout the cytoplasm with associated loss of host immune signaling (27). We found that ITMN-C treatment of cells harboring distinct HCV 1b replicons mediated a potent inhibition of NS3⅐4A, restoring the native distribution of endogenous IPS-1 as early as 12 h after treatment regardless of the differential sensitivities of each HCV replicon to IFN-␣ therapy (11,22). Moreover, protease inhibitor treatment of Huh7 cells harboring an HCV 2a replicon or infected with the HCV 2a JFH1 clone caused the accumulation of IPS-1 from a NS3⅐4A-cleaved form to a full-length form and related induction of ISG56 expression. These results indicate that 1) ITMN-C treatment and inhibition of NS3⅐4A protease function permits a rapid restoration of IPS-1 function, innate immune response signaling, and immune action directly within HCV-infected cells, 2) endogenous HCV RNA is a potent trigger of RIG-I signaling in infected cells (9), and 3) RIG-I-responsive genes, including ISG56 (22), may directly suppress HCV replication. We conclude that NS3⅐4A protease inhibitors possess immunomodulatory activity through prevention of IPS-1 proteolysis and restoration of RIG-I signaling.