The NS5A Protein of the Hepatitis C Virus Genotype 1a Is Cleaved by Caspases to Produce C-terminal-truncated Forms of the Protein That Reside Mainly in the Cytosol*

The nonstructural 5A (NS5A) protein of the hepatitis C virus (HCV) is a multifunctional protein that is implicated in viral replication and pathogenesis. We report here that NS5A of HCV-1a is cleaved at multiple sites by caspase proteases in transfected cells. Two cleavage sites at positions Asp154 and 248DXXD251 were mapped. Cleavage at Asp154 has been previously recognized as one of the caspase cleavage sites for the NS5A protein of HCV genotype 1b (1, 2) and results in the production of a 17-kDa fragment. The sequence 248DXXD251 is a novel caspase recognition motif for NS5A and is responsible for the production of a 31-kDa fragment. Furthermore, we show that Arg217 is implicated in the production of the previously described 24-kDa product, whose accumulation is affected by both calpain and caspase inhibitors. We also showed that caspase-mediated cleavage occurs in the absence of exogenous proapoptotic stimuli and is not related to the accumulation of the protein in the endoplasmic reticulum. Interestingly, our data indicate that NS5A is targeted by at least two different caspases and suggest that caspase 6 is implicated in the production of the 17-kDa fragment. Most importantly, we report that, all the detectable NS5A fragments following caspase-mediated cleavage are C-terminal-truncated forms of NS5A and are mainly localized in the cytosol. Thus, in sharp contrast to the current view we found no evidence supporting a role for caspase-mediated cleavage in the transport of the NS5A protein to the nucleus, which could lead to transcriptional activation.

The nonstructural 5A (NS5A) protein of the hepatitis C virus (HCV) is a multifunctional protein that is implicated in viral replication and pathogenesis. We report here that NS5A of HCV-1a is cleaved at multiple sites by caspase proteases in transfected cells. Two cleavage sites at positions Asp 154 and 248 DXXD 251 were mapped. Cleavage at Asp 154 has been previously recognized as one of the caspase cleavage sites for the NS5A protein of HCV genotype 1b (1, 2) and results in the production of a 17-kDa fragment. The sequence 248 DXXD 251 is a novel caspase recognition motif for NS5A and is responsible for the production of a 31-kDa fragment. Furthermore, we show that Arg 217 is implicated in the production of the previously described 24-kDa product, whose accumulation is affected by both calpain and caspase inhibitors. We also showed that caspase-mediated cleavage occurs in the absence of exogenous proapoptotic stimuli and is not related to the accumulation of the protein in the endoplasmic reticulum. Interestingly, our data indicate that NS5A is targeted by at least two different caspases and suggest that caspase 6 is implicated in the production of the 17-kDa fragment. Most importantly, we report that, all the detectable NS5A fragments following caspase-mediated cleavage are C-terminaltruncated forms of NS5A and are mainly localized in the cytosol. Thus, in sharp contrast to the current view we found no evidence supporting a role for caspase-mediated cleavage in the transport of the NS5A protein to the nucleus, which could lead to transcriptional activation.
Caspases are cysteine-dependent proteases that constitute the central executioners of apoptosis (3)(4). Caspases are expressed as proenzymes, and they become activated through cleavage at internal aspartic acid residues by other caspases (4 -5). Procaspases with a long N-terminal prodomain, called initiator caspases (caspase 2, 8, 9, and 10), are activated first and then cleave and activate procaspases with a shorter N-terminal prodomain, called executioner caspases (caspase 3, 6, and 7) (4). The activated executioner caspases then cleave a number of target proteins, important for several cellular functions, leading to the ultimate destruction of the cell (6). Up to date, 14 mammalian caspases have been identified that are implicated in different aspects of cell death, but the exact function of each individual caspase is still largely unknown (4).
Surprisingly, however, growing evidence now indicates a participation of caspases and other apoptotic regulators in nonapoptotic cellular processes such as cell cycle control, cell differentiation, and inflamma-tion (7)(8)(9). Most interestingly, caspases are also utilized by a number of viruses for cleavage of their own proteins (10). Several caspases have been shown to target the structural proteins of different viruses, including transmissible gastroenteritis coronavirus (TGEV) (11), human astrovirus (HAstv) (12), influenza A virus (13), and feline calicivirus (FCV) (14), as well as nonstructural viral proteins, such as the NS1 from the Aleutian mink disease parvovirus (ADV) (15), the adenovirus (Ad) E1A (16), or the immediately early protein 22 from herpes simplex virus (HSV-1) (17), with an impact on viral pathogenesis. Interestingly, caspase 3, the executioner caspase that targets the majority of cellular substrates during apoptosis, is also responsible for cleaving many viral proteins, and numerous studies have implicated caspase 6, 7, and 2 in the cleavage of different viral protein substrates (11,14,16). Because most viruses encode inhibitors of caspases to evade cellular antiviral mechanisms that lead cells to apoptosis, (18 -21) it is not immediately apparent why viruses rely on caspases for cleavage of their own proteins. On the other hand, some viruses induce apoptosis, resulting in virus dissemination, whereas certain viruses do both at different stages in their propagation (10,19). Thus, the cumulative data on the novel requirement of caspase activity for virus propagation combined with the growing evidence for the participation of caspases in several nonapoptotic cellular processes, has triggered the hypothesis that caspase activation may represent an important point of control for virus replication that remains to be explored (10).
Hepatitis C virus (HCV), 2 a small hepatotropic RNA virus, affects ϳ3% of the population worldwide, leading to major health problems. HCV infection is associated with high rates of progression to chronic infection, which often lead to liver cirrhosis and hepatocellular carcinoma with fatal outcome (22)(23)(24). Currently, no vaccine against HCV is available. The hepatitis C virus is classified within the Hepacivirus genus of Flaviviridae family (25). The viral genome consists of a 9.6-kb, singlestranded, positive-sense RNA molecule, which encodes a precursor polyprotein of about 3,000 amino acids. Proteolytic processing of the polyprotein by host and viral proteases yields at least 10 mature viral proteins (core, E1, E2, p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B) (26). An additional protein known as F or ARFP or coreϩ1 was recently identified. This protein is encoded by an alternative reading frame within the core coding region, but its function remains unknown (27)(28)(29). The 5Ј-and the 3Ј-untranslated regions of the viral genome are highly conserved and contain control elements for viral replication and translation of the viral polyprotein (26).
The HCV NS5A protein is a multifunctional serine phosphoprotein of 56 -58 kDa in size (30 -32), that participates in viral replication, translation, and HCV-mediated pathogenesis (33)(34)(35)(36). As with most of  the nonstructural viral proteins, NS5A participates in the formation of  the viral replication complex, on the cytoplasmic side of the ER, where it  can also directly interact with the RNA-dependent RNA polymerase of  the virus (37-39). An amphipathic ␣-helix located in the N-terminal region of NS5A is responsible for its anchorage on the ER membranes despite the presence of a functional NLS in the C-terminal part of the protein (40 -41). Thus, recombinant N-terminal-deleted forms of NS5A, which lack the membrane anchoring signal, are almost exclusively localized to the nucleus and exhibit transactivation properties (42)(43). Furthermore, NS5A interacts with a number of cellular proteins implicated in the interferon-mediated antiviral response (44 -47), transcription (42-43, 48 -51), apoptosis (52)(53)(54)(55)(56)(57), cell growth, and differentiation (58 -59), lipid metabolism (60), and signal transduction (38,45,47,61), denoting the potential of the protein to affect the host environment. NS5A also perturbs Ca 2ϩ homeostasis, induces oxidative stress and activates the Ca 2ϩ -dependent calpain proteases (62)(63)(64). Interestingly, it was recently shown by our laboratory that calpains cleave NS5A to produce shorter N-terminal forms of the protein, suggesting that in addition to phosphorylation, calpain cleavage may modulate the numerous activities of the NS5A protein (65). Additionally, it was recently shown that under certain conditions, caspase-like proteases can cleave NS5A of HCV-1b into a few fragments, producing N-and C-terminal-truncated forms of the protein that can potentially enter the nucleus and activate transcription (1-2). Two cleavage sites were mapped at the aspartic acid residues at positions 154 and 398 (Asp 154 and Asp 398 ). However, no other caspase cleavage sites were determined, and Asp 389 is present only in genotype 1b. Based on those findings, it was hypothesized that cleavage of NS5A by caspase-like proteases may modulate the transport of NS5A in the nucleus, and the ability of the protein to function as transcriptional activator (1)(2).
The present study was undertaken to characterize the caspase-dependent cleavage of the NS5A protein from HCV genotype 1a. We report the presence of two caspase cleavage sites, one at an aspartic acid residue at position 154 (Asp 154 ) similar to that of HCV-1b and a novel one at position 248 -251 ( 248 DXXD 251 ). The latter site represents a consensus sequence for the majority of HCV genotypes and subtypes. We also provide evidence implicating caspase 6 in this process. Additionally, we report that the arginine residue at position 217 (Arg 217 ) is critical for the production of the previously described 24-kDa N-terminal fragment of NS5A, which depends on the proteolytic activity of both caspases and calpains (65). Most importantly we report that, in contrast to the current view (2, 10), caspase-mediated cleavage of NS5A results in the generation of C-terminal-truncated forms of the protein, which are mainly localized in the cytoplasmic-perinuclear space.

MATERIALS AND METHODS
Plasmids-All the plasmids were constructed using standard technology, and each time the sequence of the amplified NS5A fragment was verified by sequencing analysis (MWG-Biotech Co.). pHPI 728, expressing the full-length NS5A protein from HCV 1a strain H77, and pHPI 1403, expressing the N-terminal His 6 -tagged full-length NS5A protein (NS5A/His N ), under control of the human cytomegalovirus (HCMV) immediate-early promoter, have been described before (65). pHPI 1408, carrying a small N-terminal-deleted form of NS5A, lacking the first 32 amino acids, was constructed following amplification of the corresponding sequence from pHPI 691 (65), HindIII digest, Klenow, and cloning into the XbaI-blunt-ended site of pCI. The primers used were: sense, 5Ј-CCAAGCTTGCCATGGGGATTCCCTTTGT-3Ј (HindIII and NcoI restriction sites are underlined, the translation initiation codon is in bold); antisense, 5Ј-CTCGAGAAGCTTAGCAGCA-CACGA-3Ј (XhoI and HindIII restriction sites are underlined; the complementary sequence of a stop codon is in bold). pHPI 1411, encoding an N-terminal-deleted fragment of NS5A lacking the first 129 amino acids, was constructed following amplification of the corresponding sequence from pHPI 611 (66), HindIII digest, Klenow, and cloning into the XbaIblunt-ended site of pCI. The primers used were: sense, 5Ј-CCAAGCT-TGATATCATGGTATCGGGTATGA-3Ј (HindIII and EcoRV restriction sites are underlined; the translation initiation codon is in bold); the antisense primer was as for pHPI 1408. pHPI 1407, encoding an N-terminal-deleted fragment of NS5A lacking the first 235 amino acids, was constructed following amplification of the corresponding sequence from pHPI 611, HindIII digest, Klenow, and cloning into the XbaIblunt-ended site of pCI. The primers used were: sense, 5Ј-CCAAGCT-TGCCATGGCTCCATCTCTC-3Ј (HindIII and NcoI restriction sites are underlined; the translation initiation codon is in bold); the antisense primer was as for the pHPI 1408. pHPI 1409, encoding a C-terminaldeleted form of NS5A lacking the last 94 amino acids, was constructed following amplification of the corresponding sequence from pHPI 611, EcoRV-HindIII digest, Klenow, and cloning into the XbaI-blunt-ended site of pCI. The primers used were: sense, 5Ј-AGATATCATGAGCTC-CGGTTCCTG-3Ј (EcoRV and SacI restriction sites are underlined; the translation initiation codon is in bold); antisense, 5Ј-ACCCTC-GAGAAGCTTACGGAGGTACCGG-3Ј (XhoI and HindIII restriction sites are underlined; the complementary sequence of a stop codon is in bold). pHPI 1405, encoding an N-terminal NS5A fragment (1-233 amino acids), was constructed following ligation of the BamHI-PvuII blunt-ended fragment from pHPI 611 into the XbaI-blunt-ended site of pCI. pHPI 1570 encodes an N-terminal NS5A form (amino acids 1-248), that contains the mutations Asp 154 3 Glu 154 and Arg 217 3 Gly 217 , and corresponds to the 31-kDa NS5A product. For the construction of pHPI 1570, firstly the two point mutations Asp 154 3 Glu 154 and Arg 217 3 Gly 217 were introduced sequentially in the NS5A coding sequence of pHPI 728, by using the primers described below, to yield plasmid pHPI 1439. Then, the nucleotide sequence encoding the mutated 31-kDa NS5A product was amplified by PCR, and inserted into the HincII site of pUC19, yielding the plasmid pHPI 1569. The following primers were used: sense, 5Ј-AGATATCATGAGCTCCGGTTC-CTG-3Ј (EcoRV and SacI restriction sites are underlined; the translation initiation codon is in bold); antisense, 5Ј-CTCGAGAAGCT-TAGTCATGGTTGGC-3Ј (XhoI and HindIII restriction sites are underlined; the complementary sequence of a stop codon is in bold). Finally, the (EcoRI-HindIII)-blunt-ended fragment encoding the 31-kDa mutated NS5A product from pHPI 1569 was cloned into the XbaI-blunt-ended site of pCI, generating plasmid pHPI 1570. pHPI 1406, a pCI-based plasmid encoding an N-terminal-deleted (Ϫ162 amino acids) fragment of NS5A, has been described (65). pHPI 1316 is a pA-EUA2-based plasmid 3 that carries two expression cassettes transcribed in opposite directions. The first comprises the HCMV immediate early promoter that controls the expression of the full-length NS5A protein, and the second comprises the herpes simplex virus type 1 immediate early (a22/a47) promoter that controls the expression of the green fluorescent protein (GFP). GFP protein expressed from this plasmid permits the estimation of the number of cells that also express the NS5A protein. For the construction of pHPI 1316, the HindIII fragment from pHPI 691 (65), encoding the full-length NS5A, was blunt-ended and ligated into the blunt-ended XbaI site of the pA-EUA2 vector.
Finally, plasmid pHPI 1602 encodes the NS3 3 NS5B polyprotein under the control of the HCMV promoter. The construction of this plasmid was as follows: (a) The StuI-HindIII NS3 3 NS5B fragment (lacking the first 35 nt from NS3 and the last 1502 nt from NS5B) from pFL90 was cloned into the SmaI-HindIII site of pUC19, giving rise to pHPI 1573. (b) For the reconstruction of the NS3 sequences, the N-terminal part of NS3 (nt 1-610) encoding the first 204 amino acids was amplified by PCR, using the following primers: sense, 5Ј-GGACATG-CATGCATCTAGAAGGATGGGGCCCATCACGGCGTACG-3Ј (SphI and XbaI restriction sites are underlined; the translation initiation codon is in bold); antisense, 5Ј-CCGGTGGGAGCATGCAGGTGGG-CCACCTGGAAGC-3Ј, and ligated into the HincII site of pUC19, giving rise to pHPI 1574. (c) Next, the SphI fragment containing the N-terminal part of NS3 (nt 1-610) from pHPI 1574 was inserted into the SphI site of pHPI 1573, giving rise to pHPI 1669. (d) For the reconstruction of NS5B sequences, the C-terminal part of NS5B (nt273-1775) encoding the last 500 amino acids was amplified by PCR, using the following primers: sense, 5Ј-GCTATCCGTAGAGGAAGC-TTGCAGCCTGGCG-3Ј (nt 263-1775) and antisense, 5Ј-GGCCAA-GCTTGGCTAGTCTAGACTAGTTACGGTTGGGGAGGAGGT-AG-3Ј (HindIII and XbaI restriction sites are underlined; the complementary sequence of a stop codon is in bold) and ligated into the HincII site of pUC19, giving rise to pHPI 1668. (e) Following this, the HindIII fragment of this NS5B (nt 263-1775) from pHPI 1668 was inserted into the HindIII site of pHPI 1669, giving rise to pHPI 1670. (f) Finally, for expression in mammalian cells, the XbaI-blunt-ended fragment of the NS3 3 NS5B coding sequence from pHPI 1670 was inserted into the XbaI-blunt-ended site of pA-EUA2, generating the pHPI 1602 plasmid. pHPI 1663 encodes the full-length NS5A 1b and was constructed following PCR amplification of the corresponding sequence from pTM/NS5A 1b (kindly provided by Dr. Bartenschlager), digestion with XbaI, Klenow, and ligation into the XbaI-blunt-ended site of pCI. The primers used were: sense, 5Ј-CTAGTCTAGACTAGATACCAT-GGCCTCCGGCTCGTGGC (XbaI and NcoI restriction sites are underlined; the translation initiation codon is in bold); antisense, 5Ј-CT-AGTCTAGACTAGCTAGTTCAGCAGCAGACGACGT-3Ј (XbaI restriction site is underlined; the complementary sequence of a stop codon is in bold).
Cells, Transfections, and Apoptosis Assays-Vero (green monkey kidney fibroblasts), HuH-7 (human hepatoma), and HeLa (human cervix adenocarcinoma) cells were obtained from the American Type Culture Collection. WRL 68 (human liver embryonic hepatoma) cells were kindly provided by A. Budkowska (Institute Pasteur Paris, France). All the cells were maintained as previously described (65).
For the transfection procedure, all the cells were seeded in 12-well culture plates (Nunc) the day before transfection in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. After 18 h, the cells (60 -70% confluence) were transfected with 0.5 g/well of highly quality purified DNA, by using the Lipofectamine Plus reagent (Invitrogen), as specified by the supplier. For drug utilization at selected times p.t., cells were treated with the appropriate amounts of the cell permeable protease inhibitors for 12 h.
To measure the mortality in cells expressing the NS5A protein, cells were transfected with pHPI 1316, which expresses NS5A and GFP under different promoters, as described above. GFP protein expressed from this plasmid permits the estimation of the number of cells that also express the NS5A protein. Forty-eight hours p.t., cells were incubated with 0.1% trypan blue, and the percent mortality of NS5A-expressing cells was estimated as follows: ((number of green cells stained with trypan blue in the fields analyzed) : (number of total green cells in the same fields)) ϫ 100. To induce apoptosis in GFP-expressing cells, 43 h p.t. cells were treated with TNF␣ (0.01 ng/l) and CHX (50 g/ml) for 5 h.
For the analysis of chromatin condensation in the presence of the NS5A protein, WRL68 cells were transfected with the pHPI 1316, as before, and at 48 h p.t. they were fixed with methanol for 10 min at Ϫ20°C, followed by incubation with Hoechst 33258 (10 g/ml) for 15 min at room temperature. Images were taken in Zeiss microscope using an IM-50 Leica camera. Cells treated with TNF␣ and CHX as above, serve as positive controls.
For the PARP cleavage assay WRL68 cells transfected either with pHPI 728 or with the empty vector pCI were harvested at 48 h p.t. Approximately 80 g of total proteins were analyzed on a denaturing 12% polyacrylamide gel and immunoblotted with the PARP rabbit polyclonal antibody, dilution 1:500 (Santa Cruz Biotechnology). Cells treated with TNF␣ and CHX, as before, serve as positive control.
Immunoblot Analysis, Antibodies, and Immunofluorescence Analysis-Cell monolayers were harvested at the indicated times p.t., rinsed with ice-cold PBS-A and lysed (10 min on ice) in triple detergent buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.1% SDS, 100 g ml Ϫ1 of phenylmethylsulfonyl fluoride, 1% Nonidet P-40, 0.5% sodium deoxycholate), in the presence of protease inhibitor mixture (Sigma), as specified by the manufacturer. Approximately 30 -40 g of total protein were analyzed each time, unless indicated otherwise. Following this, SDS-polyacrylamide gel electrophoresis loading buffer was added to each sample, the samples were boiled for 3 min, separated on a dena-turing 12% polyacrylamide gel, and transferred onto nitrocellulose sheets. After blocking in PBS supplemented with 0.02% (v/v) Tween 20 (PBST), and 5% (w/v) dried milk for 40 min at room temperature, the membranes were incubated overnight at 4°C with an NS5A rabbit polyclonal antiserum (65) diluted 1:100 in PBST-1% dried milk. After several washes with PBST-1% dried milk, the membranes were incubated for 1-2 h at room temperature with the anti-rabbit horseradish peroxidaseconjugated antibody, diluted in the same buffer. Finally, following extensive washes, first with PBST-1% dried milk and then with PBS, the membranes were soaked in Pierce enhanced chemiluminescence reagent and exposed to film (Kodak).
For immunofluorescence analysis of the NS5A-expressing cells, we used the NS5A polyclonal antibody, which was purified by a slightly modified affinity chromatography method based on CNBr-activated Sepharose 4B beads, as previously described (65). Forty-eight hours p.t. cells were fixed with 3.7% paraformaldehyde and analyzed as before (65). Briefly, after fixation, cells were neutralized with 100 mM of glycine in PBS-A, permeabilized with 0.1% Triton-X in PBS-A supplemented with 2 mg/ml bovine serum albumin (permeabilization buffer) and incubated with the NS5A polyclonal antibody diluted in the same buffer. After washes with the permeabilization buffer, cells were incubated with the anti-rabbit Alexa-fluor 488-conjugated antibody, diluted in the above buffer. Finally, the samples were washed, mounted and analyzed by laser scanning confocal microscopy (Leica TCS-SP microscope equipped with Leica confocal software).
In Vitro Cleavage Assays-The 35 S-labeled NS5A protein and its mutated forms were produced using the TNT transcription-translation kit (Promega) under standard conditions, as specified by the suppliers. Samples were cooled to stop the reactions, and the recombinant human caspases 2 and 6 (Biomol) were separately applied to the in vitro translated NS5A proteins. The cleavage assays were carried out at 37°C for 6 h in a final volume of 20 l, containing 15 l of the TNT reaction and 250 units of the respective recombinant caspase in cleavage buffer (50 mM HEPES, pH 7.4, 100 mM NaCl, 0.1% CHAPS, 1 mM EDTA, 5% sucrose, and 10 mM dithiothreitol). The reactions were stopped by adding SDS-PAGE loading buffer, and the proteins were analyzed on a denaturing 12% polyacrylamide gel. The products were visualized by autoradiography.

RESULTS
Caspase-mediated Cleavage of NS5A-Earlier studies from this laboratory have shown that NS5A from genotype 1a was cleaved in two C-terminally truncated fragments (40 and 24 kDa), when expressed in cells infected with an HSV-based vector expressing the protein (65). Furthermore, it was shown that calpain proteases were responsible for the generation of both fragments, whereas the accumulation of the 24-kDa fragment was also dependent on caspase activity (65). However, when NS5A was expressed in transiently transfected Vero cells, two additional bands of 31 and 17 kDa in size were detected. Interestingly, the accumulation of the latter fragments was unaffected by the use of calpain inhibitors, whereas the use of the pancaspase inhibitor Z-VAD-FMK completely blocked their production (65), suggesting that both the 31-and 17-kDa proteins may represent caspase-dependent NS5A cleav- age products. Thus, it is likely that transiently expressed NS5A induces caspase-dependent activity which is blocked in the environment of HSV-1-infected cells.
As caspase activity may depend on the cellular context (4) 1 and 2). 48 h p.t., cells were harvested, and the proteins were analyzed on a denaturing 12% polyacrylamide gel, transferred to a nitrocellulose sheet and detected with the NS5A polyclonal antibody. As shown in Fig. 1A, both the 31-and 17-kDa fragments (bands b and d, respectively) are visualized in the absence of Z-VAD-FMK (lane 2) but not in the presence of the inhibitor (lane 3) in all cell lines tested. As expected, the production of the 24-kDa product (band c) was also reduced in the presence of the inhibitor but to a lesser extent compared with the 17-and 31-kDa fragments. These results demonstrate that NS5A of HCV-1a is cleaved at multiple sites by caspases, in transfected cells, in the absence of external apoptotic stimuli.
To investigate the nature of the 31-and 17-kDa products, the NS5A protein was tagged with a His 6 epitope at its N terminus (NS5A/His N ) and the expression of the His-tagged and untagged NS5A forms was analyzed in Vero-transfected cells by Western blot analysis. As shown in Fig. 1B (lane 3), all cleavage products of the NS5A/His N protein exhibited an increase in size, which was consistent with the predicted size increase of 3 kDa because of the tag, compared with the untagged NS5A protein (lane 2). In contrast, Vero cells that express the NS5A/His C protein show no apparent differences in the mobility of the NS5A cleavage products (data not shown). These data demonstrate that all the detectable NS5A cleavage products originate from the N-terminal part of the NS5A protein.
Finally, we compare the subcellular localization of the NS5A protein in the presence or absence of the Z-VAD-FMK inhibitor by immuno-fluorescence analysis. For this purpose, Vero or HuH-7 cells were either transfected with the plasmid pHPI 728 (Fig. 1C, panels b, c, e, f) or with the empty vector pCI (Fig. 1C, panels a and d), and at 36 h, p.t. cells were either mock-treated or treated with Z-VAD-FMK for 12 h, fixed, and stained with the affinity-purified NS5A rabbit polyclonal antibody. As shown in Fig. 1C, both treated (panels c and f) and untreated (panels b and e) cells exhibited cytoplasmic, ER-like, perinuclear staining for NS5A with no apparent different characteristics. Thus, in contrast to the current view (2, 10), our results do not support translocation of the NS5A protein to the nucleus after caspase-mediated cleavage. Instead, our data indicate that caspases cleave the NS5A of HCV-1a to generate ER-associated N-terminal forms of the protein.
Mapping of the NS5A Cleavage Sites-Examination of the amino acid sequence of the NS5A protein of HCV-1a revealed a number of putative caspase recognition sites that could produce N-terminal NS5A fragments similar in size to those observed in transfected cells. To identify the bona fide caspase recognition sites, a site-directed mutagenesis analysis was performed to convert candidate aspartic acid residues at positions 154 (Asp 154 ), 205 (Asp 205 ), 248 (Asp 248 ), or 251 (Asp 251 ) to glutamic acid residues ( Table 1). The proteolytic processing of the NS5A protein was analyzed in Vero, HeLa, and WRL 68 cells transfected separately with plasmid pHPI 728 expressing the wild-type NS5A protein (Fig. 2, A and B, lane 2), or plasmids expressing the mutated forms of NS5A. These included plasmids pHPI 1426 expressing the NS5A/ Asp 154 3 Glu 154 form (Fig. 2A, lane 3), pHPI 1565 expressing the NS5A/ Asp 248 3 Glu 248 form (Fig. 2B, lane 3), pHPI 1423 expressing the NS5A/ Asp 251 3 Glu 251 form (Fig. 2B, lane 4), or the empty vector pCI (Fig. 2,  A and B, lane 1). Cells were harvested at 48 h p.t., and the proteins were visualized by Western blot analysis with the NS5A polyclonal antibody. As shown in Fig. 2A (Fig. 2B, lanes 3 or 4, respectively). On the other hand, the mutation Asp 205 3 Glu 205 has no effect on the cleavage pattern of the NS5A protein (data not shown). Collectively, these results indicate that Asp 154 as well as the 248 DXXD 251 sequence belong to bona fide caspase recog- nition sites, which are responsible for the production of the 17-kDa and 31-kDa products, respectively. Notably, Asp 154 has been previously identified as a caspase recognition site for the NS5A of HCV-1b (2), whereas the 248 DXXD 251 is a novel caspase recognition site that represents a consensus sequence among the majority of HCV isolates. On the other hand, the Asp 205 does not appear to belong to a caspase recognition motif, as its conversion to Glu 205 had no effect on the NS5A proteolytic processing. For the production of the 24-kDa NS5A fragment, it is predicted that a putative cleavage within the amino acid stretch between residues 200 and 235 of the NS5A sequence could produce an N-terminal product of ϳ24 kDa in size. However, provided that apart from the Asp 205 , no other predicted caspase recognition motifs were identified within this sequence, and that the generation of the 24-kDa fragment is sensitive to both calpain and caspase inhibitors, we assumed that the 24-kDa fragment is the result of calpain-mediated cleavage, whereas caspases may modulate this cleavage (67)(68)(69). As calpain recognition sites are not well defined, our first attempt to identify the cleavage site for the 24-kDa fragment was to examine the cleavage pattern of a number of NS5A mutated forms carrying multiple changes within the corresponding region of NS5A protein. 4 The first construct contained three mutations at positions 224, 225, and 227 (Pro 224 3 Ser 224 , Ser 225 3 Pro 225 , and Ala 227 3 Asp 227 ), the second carried four mutations, targeting simultaneously the above three amino acids as well as the residue at position 205 (Asp 205 3 Gly 205 ), the third contained five mutations targeting simultaneously the above four amino acids as well as the residue at position 232 (Ser 232 3 Ile 232 ), and the fourth NS5A mutant combined the five previously described changes with an alteration at position 217 (Arg 217 3 Gly 217 ) ( Table 1). Whereas none of the first three NS5A mutants had an effect on NS5A cleavage, the latter suppressed the production of the 24-kDa product, suggesting that Arg 217 may be a critical amino acid in the generation of this product (data not shown). To test this possibility further, two additional mutated NS5A forms were constructed, with a conversion of arginine at position 217 to either glycine (Arg 217 3 Gly 217 ) or to lysine (Arg 217 3 Lys 217 ). Vero, HeLa, or WRL68 cells were transfected separately either with the plasmid pHPI 1438 expressing the mutated NS5A/Arg 217 3, band c), whereas no apparent effect on the NS5A cleavage was observed when Arg 217 was converted to its synonymous amino acid Lys 217 (lane 4). These data confirmed that Arg 217 is critical for the production of the 24-kDa fragment and suggest that Arg 217 may play a role in the maintenance of the proper conformation of the cleavage site. This is consistent with the finding that calpain mediated cleavage is affected by the three-dimensional structure of its substrates (70). Collectively, our data demonstrate that Asp 154 , 248 DXXD 251 and Arg 217 residues are critical for the production of the 17-, 31-, and 24-kDa fragments, respectively.
Caspase-mediated Cleavage of NS5A Does Not Require an Intact NS5A Protein-As previous studies have shown that the NS5A protein is anchored on the ER membrane through an N-terminal amphipathic ␣-helix (40), we sought to examine whether caspase activation was the result of ER stress because of overexpression of NS5A in transfected cells. To assess this, we examined the cleavage pattern of several NS5A deletion mutants, which lack the N-terminal part of the protein that contains the ER anchoring signal (Fig. 4A). For this purpose, Vero cells were separately transfected with plasmids expressing the NS5A lacking either the first 129 (pHPI 1411), 162 (pHPI 1406), or 235 (pHPI 1407) amino acid residues (Fig. 2B), or with a plasmid expressing an NS5A deletion lacking only the first 32 amino acids (pHPI 1408) (Fig. 4C). Thirty-six hours p.t., cells were either treated with the Z-VAD-FMK inhibitor (Fig. 4B, lanes 3, 5, 7, and 9) or remained untreated (Fig. 4B,  lanes 1, 2, 4, 6, and 8), and cell lysates were analyzed by Western blot, as above. As shown in Fig. 4, the production of all N-terminal-truncated forms of NS5A was followed by the appearance of shorter forms of the protein. Interestingly, as shown in B, the accumulation of a 22-kDa fragment (marked by arrows) was inhibited by the presence of the pancaspase inhibitor (lanes 5, 7, and 9), suggesting the involvement of a caspase-mediated cleavage. In contrast, the accumulation of a 33-kDa fragment (marked by stars) remained unaffected in the presence of Z-VAD-FMK, but was affected by calpain inhibitors (data not shown). We assume that both the 22-and 33-kDa cleavage products may represent C-terminal parts of the protein generated by caspase-mediated cleavage at position Asp 251 , or by calpain-mediated cleavage near Arg 217 , respectively, which are produced by an as yet unknown mechanism in the selected N-terminal-truncated forms of NS5A. Notably, the NS5A form lacking only the first 32 amino acids was also cleaved in transfected cells, generating the expected N-terminal forms of NS5A (shown by arrow), shorter by 33 amino acids (Fig. 4C, lane 3). To confirm the intracellular localization of the N-terminal forms of NS5A, we performed immunofluorescence studies (Fig. 4E). As expected all N-terminally deleted forms of NS5A exhibit a diffuse cytoplasmic-nuclear staining pattern (Fig. 4E, panels c, d, e, and f), compared with the full-length NS5A (panel b). Collectively, these data suggest that the induction of caspase cleavage of the NS5A protein in transfected cells is independent of the accumulation of the protein on ER membranes.
Finally, we investigated whether the NS5A protein can be proteolytically processed in the absence of its C-terminal part. For this purpose, two C-terminally truncated forms of NS5A were constructed, encoding either the first 233 (pHPI 1405) or the first 354 amino acids of the protein (pHPI 1409), respectively (Fig. 4A), and their expression was analyzed in transiently transfected Vero cells. As shown in Fig. 4D, both the C-terminal-truncated NS5A forms are proteolytically processed to produce the expected N-terminal fragments, namely the 17-, 24-, 31-, and 40-kDa fragments for pHPI 1409 (lane 3), and the two shorter fragments for pHPI 1405 (lane 4). In contrast, when amino acid substitutions Asp 154 3 Glu 154 and Arg 217 3 Gly 217 were introduced into a plasmid that encodes for the 31-kDa form of NS5A, the production of the 17-and 24-kDa fragments was abolished (lane 5). Collectively, these data demonstrate that neither the N-nor the C-terminal part of the NS5A protein are required for its proteolytic processing by caspases and calpains.
Identification of the Caspases Involved in NS5A Cleavage-To determine which caspase(s) are responsible for the NS5A cleavage, firstly we examined the effect of various specific caspase inhibitors on the accumulation of the 17-and the 31-kDa NS5A products (Fig. 5).  Arrow and a-d denote the full-length NS5A protein and its cleavage products. Molecular mass markers are shown on the right.
To validate the involvement of caspases 2 and 6 in the cleavage of the NS5A protein, we performed an in vitro cleavage assay. For this purpose, 35 S-labeled NS5A protein, which was produced from rabbit reticulocyte lysates was used as a substrate for the recombinant caspases 6 and 2, respectively. Mutated forms of the NS5A protein where the previously described putative cleavage sites for the caspases 6 and 2 have been modified, serve as negative controls. As it is illustrated in Fig. 5E caspase 6 efficiently cleaves the wt NS5A protein in vitro, to produce the 17-kDa fragment. As expected, the production of the 17-kDa fragment by caspase 6 is abolished when the Asp 154 is converted to Glu 154 (lane 12) and remains unaffected when the Asp 248 or the Asp 251 are converted to Glu 248 or Glu 251 , respectively (lanes 10 and 11). On the other hand, treatment by caspase 2 failed to produce the 31-kDa fragment. Instead a smaller by 2 kDa NS5A form is efficiently produced suggesting cleavage close to the N or C terminus of the protein. Notably, there are several putative caspase recognition sites at the C-terminal part of NS5A that could be potentially used by caspase 2 under our experimental conditions. Collectively, these data suggest that caspase 6 is directly involved in the production of the NS5A 17-kDa fragment. However, the possible involvement of caspase 2 on the production of the 31-kDa fragment needs further investigation.
Finally, as previous studies have shown that NS5A protein does not induce apoptosis in transfected cells, it was of interest to examine the apoptotic conditions in our experimental settings (52-54, 56 -57, 71). For this purpose, WRL 68 or Vero (data not shown) cells transiently expressing NS5A were assayed for apoptosis by a trypan blue exclusion assay. To monitor the cells that express NS5A, we modified our expression vector to contain, in addition to the full-length NS5A sequence, the GFP gene under a different promoter (pHPI 1316). Cells were transfected with the above plasmid, and at 48 h p.t., they were treated with trypan blue. Subsequently, microscopy analysis was performed, and the percentage of the green cells that had been stained with trypan blue in comparison to the total number of green cells measured in the analyzed fields was estimated. As a negative control, cells transfected with the plasmid vector that expressed only GFP were used. As a positive control cells transfected with the same plasmid vector were treated with TNF␣ (0.01 ng/l) and CHX (50 g/ml) for 5 h to induce apoptosis. As illustrated in Fig. 6A, only 6.7% of the cells that expressed the NS5A protein were stained with trypan blue. This result was comparable to that of the negative control, whereas the percentage of green cells stained with trypan blue after treatment with TNF␣ and CHX was ϳ72.4%. To verify these results, the condensation of chromatin in the nucleus of NS5Aexpressing WRL68 cells was also investigated by Hoechst 33258 staining, under the above experimental conditions. Following the analysis of several fields, representative data shown in Fig. 6B indicate no chromatin condensation in NS5A-expressing cells (panel c) or empty vector- proteins were analyzed by immunoblotting, as before. As illustrated in Fig. 7A (lane 3) NS5A is proteolytically processed in the presence of the other viral proteins, generating the 31-, 24-, and 17-kDa cleavage products, suggesting that caspase-mediated processing of NS5A is compatible to the expression of most HCV nonstructural proteins. On the other hand, all attempts to detect NS5A cleavage when the con.1 replicon was used were unsuccessful, most likely because of the low level of NS5A protein expression (data not shown). However, as the con.1 replicon system is derived from the HCV-1b genotype, we also sought to examine the proteolytic cleavage of NS5A of HCV-1b under our experimental conditions. For this purpose Vero cells were either mock-transfected Fig. 7B (lane 1), transfected with pHPI 728, which expresses the fulllength NS5A 1a (lane 2), or with pHPI 1663, which expresses the fulllength NS5A 1b (lane 3), and analyzed by Western blotting, as before. As illustrated in Fig. 7B, the NS5A of HCV-1b is cleaved to produce a 17-kDa (band d), 24-kDa (band c), and 31-kDa (band b) fragment, implying cleavage at sites Asp 154 , Arg 217 , and 248 DXXD 251 , as for the NS5A of HCV-1a. Interestingly, however, the calpain-dependent N-terminal 40-kDa fragment (band a) is not produced from the NS5A of HCV-1b (65). Instead, a new product of 48 kDa is detected, as reported in previous studies (2). Thus, although our data support cleavage of NS5A in the presence of the viral nonstructural proteins constituting the replication complex, the relationship between NS5A cleavage and virus replication remains to be elucidated.

DISCUSSION
We report here that the NS5A protein from HCV genotype 1a is cleaved by caspases to produce shorter, N-terminal forms of the protein, in the absence of exogenous apoptotic stimuli. Two cleavage sites at the aspartic residues Asp 154 and 248 DXXD 251 were mapped. Cleavage at Asp 154 has been previously recognized as one of the caspase cleavage sites for the NS5A protein of HCV genotype 1b (1)(2) and results in the production of a 17-kDa N-terminal NS5A fragment. The sequence 248 DXXD 251 is a novel caspase recognition motif for NS5A that represents a consensus sequence among the majority of HCV genotypes and is responsible for the production of a 31-kDa N-terminal NS5A fragment. Interestingly, the use of caspase-specific inhibitors combined with an in vitro cleavage assay has implicated caspase 6 in the cleavage of NS5A at position Asp 154 . On the other hand, although the production of the 31-kDa fragment was blocked in transfected cells by the addition of a caspase 2-specific inhibitor, the purified caspase 2 failed to generate this fragment in vitro. Thus, the identity of the caspase responsible for cleavage at 248 DXXD 251 position remains ambiguous. It is possible that changes in the conformation of the in vitro synthesized NS5A substrate, or an indirect role of caspase 2 to be responsible for the failure to produce the 31 kDa in vitro. Alternatively, other caspases that could be blocked by the Ac-VDVAD-CHO inhibitor may be responsible for the production of the 31-kDa fragment. Nevertheless, our data show that at least two different caspases target NS5A, and it appears that there is no FIGURE 6. Transient expression of NS5A protein does not lead to apoptosis. A, WRL 68 cells seeded in 12-well plates were transfected either with pHPI 1316 (bar B) or with the empty GFP expression vector (bars A and C). At 43 h p.t., cells transfected with the GFP expression vector were either treated for 5 h with TNF␣ (0.01 ng/l) and CHX (50 g/ml) to induce apoptosis (bar C) or remained untreated (bars A and B). Following this, all the samples were staining sequentially for 5 min with 0.1% trypan blue at room temperature, and microscopy analysis was performed, where the percentage of green cells stained with the trypan blue was calculated in comparison to the total green cells, in the number of fields analyzed. interplay between those cleavage events in our studies. Notably, the 248 DXXD 251 motif represents a recognition site for other caspases and thus may be recognized under conditions, which permit the activation of those caspases (72). Furthermore, we showed that the arginine residue at position 217 (Arg 217 ) is implicated in the production of the previously described 24-kDa N-terminal NS5A fragment, whose accumulation is affected by both calpain and caspase inhibitors. Because no known recognition motif for caspases is present in the respective region of NS5A, it could be predicted that Arg 217 contributes to a calpain recognition motif. Interestingly, we also found that N-terminally truncated forms of the protein are still processed by caspases suggesting that caspase activation does not correlate with the accumulation of NS5A on the ER membranes. Finally, cleavage of the NS5A protein was also observed in the presence of most HCV nonstructural proteins (NS3-NS4A-NS4B-NS5A-NS5B) (Fig. 7A).
One of the most intriguing findings of our work was that only N-terminal fragments of the NS5A protein, which carry the ER associating signal, were detectable following caspase cleavage, whereas detection of the corresponding C-terminal fragments remained elusive. Consistent with these findings, nuclear localization of the full-length NS5A protein under conditions that allow caspase-mediated cleavage was not detected. Therefore, in sharp contrast to the current view, our data failed to provide evidence supporting a role for caspases in NS5A nuclear translocation following proteolysis as a mechanism of tran-scriptional activation (1)(2)(42)(43). Instead, it seems that the NS5A cleavage favors NS5A functions located in the N-terminal part of the protein and are mainly linked with properties of the cell membranes ( Fig. 8) (65). Notably, several recent studies have shown that NS5A can activate transcription, while localized to the cytoplasm. Siddiqui and co-workers (62,63) have shown that NS5A induces activation of STAT3 and NF-B via disruption of intracellular calcium levels. Similarly, Park et al. (56,73) have shown that NS5A modulates the TRAF2-mediated JNK activity, thus affecting c-Jun-mediated transcription, whereas the NS5A-TRAF2 protein interaction was also shown to inhibit TNF␣induced NF-B activation. Furthermore, Qadri et al. (51) reported that NS5A interaction with p53 and TBP affects transcription, and Ghosh et al. (49) showed that NS5A through its direct interaction with the transcriptional factor SRCAP may exerts its negative effect on the p21 promoter. Notwithstanding, Yeh et al. (74) who have recently characterized a mutant form of NS5A isolated from a patient with hepatocellular carcinoma, have shown that although the protein was localized both in the cytoplasm and the nucleus, its transactivation properties were unrelated to nuclear localization.
As NS5A is known to be the classical antiapoptotic protein of HCV virus (52-56, 61, 71), the biological relevance of our findings remains unclear. However, a growing number of studies reveal novel nonapoptotic activities of caspases, indicating that caspases are much more versatile enzymes than originally expected (7)(8)(9)(10)(75)(76). For example, caspase 1 and caspase 11 mediate IL-1 production (8,(77)(78). Caspase 3 is involved in the terminal differentiation of erythrocytes, keratinocytes, monocytes, and epithelial, sperm, skeletal, muscle, osteoblast, and trophoblast cells as well as in B cell proliferation (8 -9, 79 -84). Caspase 8 mediates T cell proliferation, placental, and trophoblast differentiation as well as cell cycle control (7,9,(85)(86). Caspase 12 attenuates inflammatory and innate immune responses (87). Caspase 14 mediates terminal differentiation of keratinocytes (85,88). Finally, caspase 2 has been shown to participate in the activation of the DNA repair machinery when it is activated as a part of a multiprotein complex known as the PIDDsome complex (89). Additionally, proteins from different families seem to serve as substrates for caspase 2, such as golgin-160, PKC␦, and aII-spectrin, implying that caspase 2 can act both as a signaling and an executioner caspase (90 -92). Furthermore, many examples of molecules cleaved by caspases in nonapoptotic cells have been described. Recently, it was shown that the ␤-subunit in the IKK␤ complex can be inactivated by caspase 3 (93), and the p50 and p65 subunits of NF-B are substrates for caspase 3, resulting in the loss of their transcriptional activity (94 -95). Molecules that are implicated in cell proliferation pathways, such as MEK, STAT1, CREB, PKC, and vav1 (96 -100) or molecules involved in cell cycle regulation, such as the cyclin inhibitors p21 Cip1/Waf1 and p27 Kip1 , as well as the protein kinase Wee1, are also caspase substrates. Notably, cleavage of Wee1 by caspases mediates progression through the cell cycle (101). Thus, it appears that caspases have a dual function in life and death (8), and the selectivity of substrate cleavage in nonapoptotic cells is likely to be achieved either by a compartmentalized activation of caspases, activation of antiapoptotic factors or through limited activity of caspases (102).
The impact of caspase-mediated cleavage of NS5A on the virus life cycle is presently unknown. It can be argued that NS5A is cleaved by caspases because it happens to contain caspase cleavage sites in its sequence. On the other hand, however, it is now well recognized that caspase-mediated cleavage of many viral proteins is a direct requirement for propagation of those viruses (10). Specifically, caspase 2 has been implicated in the cleavage of FCV capsid (14), whereas caspase 6 was found to be involved in the processing of human astrovirus HAstV capsid precursor (12), in TGEV nucleocapsid cleavage (11), and in HIV-induced apoptosis of T lymphocytes (103). Inasmuch as limited cleavage of proteins can either lead to the functional inhibition or activation of these proteins (104 -105) we could hypothesize that caspase-mediated cleavage of the NS5A protein may modulate the multiple functions of the protein. On the other hand, we found no evidence supporting a correlation of caspase cleavage and shuttling of the protein to the nucleus.
To date, there are more than 280 known proteins targeted by caspases for cleavage (104). However, for most proteins, the consequences of their cleavage are poorly understood. In many cases, the role of caspase cleavage has been experimentally assessed by expressing substrate proteins that have mutated caspase cleavage sites, or by expressing protein fragments corresponding to caspase-cleaved products. Thus, the identification of caspase cleavage sites for the NS5A protein and the caspases that mediate NS5A cleavage represent an important step toward understanding the functional role of caspase-mediated NS5A cleavage.