Hepatitis C Virus Non-structural Proteins in the Probable Membranous Compartment Function in Viral Genome Replication*

The molecular mechanism of hepatitis C virus(HCV) RNA replication is still unknown. Recently, a cell culture system in which the HCV subgenomic replicon is efficiently replicated and maintained for a long period in Huh-7 cells has been established. Taking advantage of this replicon system, we detected the activity to synthesize the subgenomic RNA in the digitonin-permeabilized replicon cells. To elucidate how and where this viral RNA replicates in the cells, we monitored the activity for HCV RNA synthesis in the permeabilized replicon cells under several conditions. We obtained results suggesting that HCV replication complexes functioning to synthesize the replicon RNA are protected from access of nuclease and proteinase by possible cellular lipid membranes. We also found that a large part of the replicon RNA, including newly synthesized RNA, was present in such a membranous structure but a large part of each NS protein was not. A small part of each NS protein that was resistant to the proteinase action was shown to contribute sufficiently to the synthesis of HCV subgenomic RNA in the permeabilized replicon cells. These results suggested that a major subcellular site of HCV genome replication is probably compartmentalized by lipid membranes and that only a part of each NS protein forms the active replication complex in the replicon cells.

Infection of hepatitis C virus (HCV) 1 is estimated to occur in about 3% of the worlds population. HCV infection frequently causes chronic hepatitis, which often leads to the development of liver cirrhosis and hepatocellular carcinoma after a long period (1)(2)(3). Current combination therapy with interferon-␣ and ribavirin, a nucleotide analogue, is effective in many patients with chronic hepatitis C (4,5). There still are, however, a lot of patients who do not respond to these treatments. Therefore, extensive studies have been performed to develop highly effective anti-HCV drugs. Such a drug, however, has not been produced yet, possibly because of the lack of detailed information about the life cycle of this virus.
HCV is a member of the Flaviviridae family and contains a single-strand RNA genome of positive polarity (6). The RNA genome is ϳ9.6 kb in length and consists of a 5Ј-untranslated region of 341 nucleotides, a large open reading frame encoding a single precursor polyprotein of ϳ3000 amino acids, and a 3Ј-untranslated region of variable length (6 -8). The polyprotein is processed by the host and viral proteinases to generate at least 10 functional viral proteins: core (C), envelope (E) 1, E2, p7, non-structural protein (NS) 2, NS3, NS4A, NS4B, NS5A, and NS5B (from the amino-to the carboxyl-terminal) (9 -12). C, E1, and E2 are believed to form viral particles as structural proteins. p7 was recently reported to form an ion channel-like structure (13,14). NS molecules have been considered to function in the replication of HCV subgenomic RNA (15). Using recombinant proteins produced in either bacterial or insect cells, the proteinase and helicase activities of NS3 and RNA-dependent RNA polymerase activity of NS5B have been biochemically characterized (16 -20). However, it was not clear whether these recombinant proteins function in HCV genomic replication as it has been revealed that HCV genomic sequences are highly variable among all isolates and furthermore, it is unclear whether the genes for these viral enzymes were derived from infectious HCV genomes.
Recently, HCV subgenomic RNA that replicates efficiently and is maintained for a long period in the human hepatoma cell line Huh-7 was developed and called the HCV subgenomic replicon (15). Functional replicons originating from different HCV isolates have been reported (15,(21)(22)(23). The HCV subgenomic RNA was constructed by replacing the structural and part of the non-structural protein-encoding regions (C-NS2) of the HCV genome with the neomycin phosphotransferase gene (neo r ) and an internal ribosome entry site of encephalomyocarditis virus (21). This implies that HCV proteins encoded in this subgenomic RNA (NS3-NS5B) are functional and sufficient for this RNA replication. In this model system, it has been suggested that mutations of particular amino acids in the NS region enhanced the efficiency of the replication (24 -27). It was also demonstrated that the existence of the cis-acting elements in either the 5Ј-or 3Ј-untranslated regions were required for efficient replication (27)(28)(29). Recent observations indicated that the replication of the replicon RNA could be reproduced in vitro using particular cellular fractions from replicon cells (30 -32). It remains, however, to be elucidated how and where the HCV RNA is synthesized in the cells. So, we intended to clarify these points using digitonin-treated replicon cells of which plasma membranes were permeabilized. This permeabilized cell system is often used to monitor several cellular events, such as a nuclear protein transport as well as replication of positive-strand RNA viruses.
Cell biological and biochemical analyses have demonstrated that all HCV NS proteins are directly or indirectly associated with inner cellular membranes and colocalize on the rough endoplasmic reticulum (ER) membranes (33)(34)(35)(36). So, we expected the active HCV replication complexes to be retained on the inner cellular membranes in permeabilized replicon cells. In this paper, we report that the functional replication complexes are retained in permeabilized replicon cells and its activity is easily detected by using this system. We also obtained data suggesting that a part of each NS protein in the cells, which is probably located in a membranous compartment, forms the active replication complex and contributes to the synthesis of HCV subgenomic RNA.
Sequencing Analysis-The nucleotide sequence of the HCV subgenomic replicon RNA in MH-14 cells was determined by reverse transcription-PCR-based DNA sequencing as described previously (21).
Plasmid Construction-The plasmid pGEM-NN was constructed by inserting the cDNA fragment of the subgenomic replicon from pNNRZ2 (21) into the TA site of pGEM-T-Easy vector (Promega, Madison, WI). The cDNA fragment was obtained by PCR using the oligonucleotides 5Ј-GCCAGCCCCCGATTGGGGGCGACAC-3Ј and 5Ј-ACATGATCTG-CAGAGAGGCCAG-3Ј and the plasmid pNNRZ2 as primers and a template, respectively.
In Vitro Transcription-pGEM-NN was linearized with PvuI or SpeI for plus or minus strand subgenomic HCV RNA synthesis, respectively, and used as a template for in vitro RNA synthesis with MEGA script Sp6 or T7 kit (Ambion, Austin, TX), respectively.
Preparation of RNA and Northern Blot Analysis-RNA was extracted from cells and a reaction mixture with Sepasol RNA I and II super reagent (Nacalai Tesque, Kyoto, Japan), respectively, according to the manufacturer's protocol. Northern blot analysis was performed as described previously (37). For the preparation of the 32 P-labeled probe, the EcoRI fragment of pNNRZ2 was labeled with a Ready-to-Go DNA labeling beads (-dCTP) kit (Amersham Biosciences) in the presence of 50 Ci of [␣-32 P]dCTP (Amersham Biosciences).
Indirect Immunofluorescence-Indirect immunofluorescence analysis was essentially performed as described previously (38), with minor modifications. 1.0 ϫ 10 5 cells were seeded on poly-L-lysine (Sigma)coated coverslips. Three days post-seeding, the cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 , and 1.4 mM KH 2 PO 4 ) for 30 min at room temperature. The cells were treated with pre-chilled 100% methanol for 10 min at Ϫ20°C after the fixation. In the case of permeabilized cells, permeabilization with digitonin followed by washing with buffer B (see below) was performed before or after fixation with 4% paraformaldehyde and then the cells were permeabilized completely with chilled 100% methanol after fixation. The antibodies against NS5A (␣-NS5A), NS5B (NS5B-12), and protein-disulfide isomerase (StressGen) were used as primary antibodies. NS5B-12 was a gift from Dr. M. Kohara (Tokyo Metropolitan Institute of Medical Science). The fluorescent secondary antibodies were Alexa 568 goat anti-mouse IgG (HϩL) and Alexa 488 goat anti-rabbit IgG (HϩL) conjugates (Molecular Probes, Eugene, OR). The nucleus was visualized by staining with 4Ј,6-diamidino-2-phenylindole. All imaging experiments were performed on a Leica SP2 confocal microscope (Leica Microsystems, Germany).
Cell Permeabilization and Synthesis of HCV Subgenomic RNA-Cells of about 80% confluency in 12-or 6-well plates were precultured in complete Dulbecco's modified Eagle's medium containing 5 g/ml actinomycin D (Nacalai Tesque) for 2 h, then washed with cold buffer B: 20 mM HEPES-KOH (pH 7.7 at 27°C), 110 mM potassium acetate, 2 mM magnesium acetate, 1 mM EGTA, and 2 mM dithiothreitol. The cells were permeabilized by incubation in buffer B containing 50 g/ml digitonin for 5 min at 27°C and the reaction was stopped by washing twice with cold buffer B. The permeabilized cells were, then, incubated for 4 h at 27°C in the labeling reaction mixture: 2 mM manganese(II) chloride, 1 mg/ml acetylated bovine serum albumin (Nacalai Tesque), 5 mM phosphocreatine (Sigma), 20 units/ml creatine phosphokinase (Sigma), 50 g/ml actinomycin D, 500 M ATP, CTP, and GTP (Roche Diagnostics), and 10 Ci of [␣-32 P]UTP (Amersham Biosciences) in buffer B (pH 7.7), for 4 h at 27°C unless otherwise specified. The reaction was terminated by the addition of Sepasol RNA I or II super reagent.
For the treatments with microccocal nuclease and/or Nonidet P-40, the permeabilized cells were pre-washed twice with buffer D: 20 mM HEPES-KOH (pH 7.7 at 27°C), 110 mM potassium acetate, 2 mM magnesium acetate, 2 mM dithiothreitol, and 1 mM CaCl 2 . Then the permeabilized cells were incubated in buffer D containing 0.1 unit/ml microccocal nuclease (United States Biochemical Corp.), with or without 0.5% Nonidet P-40 for 15 min at 37°C or 0.5% Nonidet P-40 for 15 min on ice.
Slot Blot Hybridization-RNA products synthesized in the permeabilized cells in the presence of 200 Ci of [␣-32 P]UTP were fractionated by denaturing agarose gel. The 8-kb RNA bands were eluted from the gel using RNade (BioOne, Carlsbad, CA). To increase the hybridization signal, the 32 P-labeled 8-kb RNA eluted from the gel was subjected to alkaline hydrolysis to generate fragments of ϳ250 nucleotides in length and used in hybridization. Newly synthesized replicon RNA in intact replicon cells was metabolically labeled by adding 1200 Ci of [ 32 P]orthophosphate to the culture medium (see below) and handled in the same manner. For detection of plus and minus strand replicon RNA, minus and plus strand replicon RNA were prepared as riboprobes by in vitro transcription, respectively, as described above. Then 2 g of each riboprobe was applied to a nylon membrane filter (Hybond-N, Ambion). Slot blot hybridization was performed as described previously (39), except that ULTRAhyb (Ambion) was used for hybridization buffer.
Proteinase K Treatment-After permeabilization of the replicon cells, the cells were scraped into 400 l of cold buffer B and transferred to a siliconized tube. The cells were treated with various concentrations of proteinase K at 37°C for 5 min. The reaction was terminated by the addition of 1 mM phenylmethylsulfonyl fluoride, and followed by trichloroacetic acid precipitation. The trichloroacetic acid precipitates were solubilized in 1ϫ sample buffer containing 1 mM phenylmethylsulfonyl fluoride and used in Western blotting analysis.

Selection of Replicon Cell Clones in Which Replicon RNA
Efficiently Replicated-We reported the establishment of a cell clone, NN 50-1, in which HCV subgenomic replicon RNA originating from the HCV genome isolated from the cultured human T cell line MT-2C infected with HCV in vitro efficiently replicated (21). Furthermore, we obtained a cell clone, MH-14, in which the HCV subgenomic RNA level is nearly 5-fold higher than that in NN 50-1 (data not shown). Nucleotide sequence analysis revealed that the replicon RNA in MH-14 cells bore two point mutations in the NS4B and NS5A encoding regions. One was a thymine to cytosine transition at nucleotide position 5985 (the number corresponds to the nucleotide number of the HCV genotype 1b genome) in the NS4B-encoding region without an amino acid substitution. Another was a cytosine to adenine transition at nucleotide 6953 in the NS5A-encoding region, resulting in the substitution of arginine for serine at amino acid position 2204. The substitution at amino acid 2204 has already been reported as one of the adaptive mutations enhancing the efficiency of viral replication (25). Then we used MH-14 cells, refereed as the replicon cells, in the following experiments to analyze the mechanism of HCV genomic replication.
Replicon RNA and NS Proteins Were Retained in Digitoninpermeabilized Replicon Cells-Previous papers suggested that all HCV NS proteins are directly or indirectly associated with the inner cellular membranes, especially rough ER membranes, and form replication complexes on the membranes (10,40). This suggested to us that functional HCV replication complexes could be retained in the replicon cells whose plasma membranes had been permeabilized with digitonin. To test this possibility, first, the fate of NS proteins in the replicon cells was investigated after digitonin treatment by Western blotting. In this experiment, ectopically expressed mouse DHFR was used as a cytoplasmic protein marker. After digitonin treatment, DHFR was not detected in the permeabilized cells (Fig.  1A, lanes 2 and 3), indicating that cytoplasmic soluble proteins were washed efficiently out of the cells under these conditions. On the other hand, HCV NS proteins (NS3-NS5B) were detected just like BiP/Grp78, an ER marker, in the permeabilized replicon cells as in the intact cells that were not treated with digitonin, as expected ( Fig. 1A, lanes 2 and 3). Moreover, when we analyzed the RNA by Northern blotting, the retention of replicon RNA in the permeabilized replicon cells was observed ( Fig. 1B, lanes 2 and 3). Treatment of the permeabilized replicon cells with the high salt buffer containing 2 M KCl did not greatly influence the amount of replicon RNA and NS proteins retained in the cells (data not shown). To investigate whether newly synthesized replicon RNA in the intact replicon cells was retained after permeabilization with digitonin, we performed metabolic labeling of the cells with [ 32 P]orthophosphate. As shown in Fig. 1C, newly synthesized replicon RNA was detected after permeabilization (Fig. 1C, lane 3), although the amount was slightly decreased compared with that in the intact cells (Fig. 1C, lane 2). The localization of NS5A and NS5B in the permeabilized replicon cells was also analyzed by indirect immunofluorescence. These proteins were seen to accumulate around the perinuclear region and to be mostly colocalized with protein-disulfide isomerase, an ER marker, in the permeabilized replicon cells (Fig. 1D, panels i-p), just as in the intact replicon cells (Fig. 1D, panels a--h). This indicated that treatment with digitonin did not markedly affect subcellular localization of these proteins. From all of these results, it seemed likely that the replication complexes including replicon RNA were retained in the permeabilized replicon cells just like in the intact cells.
The Replication Complexes in the Permeabilized Replicon Cells Functioned to Synthesize the HCV Subgenomic RNA-To see whether the replication complexes in the permeabilized replicon cells were active in HCV RNA synthesis, permeabilized or intact cells were incubated in reaction mixtures including [␣-32 P]UTP for 4 h. Actinomycin D, which showed no inhibitory effect on the RNA-dependent RNA polymerase activity of HCV NS5B (15), was also added to the reaction mixture to inhibit the cellular activities of DNA-dependent DNA and RNA synthesis. After the reaction, total RNA of the cells and the reaction supernatants were extracted and analyzed by denaturing agarose gel electrophoresis followed by autoradiography. A radiolabeled product ϳ8 kb in length, which was equivalent to the subgenomic replicon RNA in size, was found in RNA from the permeabilized replicon cells but not from Huh-7 cells ( Fig. 2A, lanes 1-4). When overexposed, the 8-kb band became detectable in RNA sample from the intact replicon cells, although the signal was much lower than that from the permeabilized replicon cells (data not shown). This 8-kb product, however, was not detected in all the reaction supernatants ( Fig. 2A, lanes 5-8). The radiolabeled 8-kb product was degra-dated by treatment with RNase A (data not shown), suggesting that it was HCV subgenomic RNA synthesized by the replication complexes in the permeabilized replicon cells. To examine whether the 8-kb RNA synthesized in the permeabilized cells was actually HCV subgenomic RNA, we performed slot blot hybridization analysis using each plus and minus strand HCV RNA synthesized in vitro as probes on the nylon membrane filter. After denaturing agarose gel electrophoresis, the radiolabeled 8-kb RNA derived from the permeabilized replicon cells was eluted from the gel and hybridized with the probes on the filter. Then the radioactivity hybridized with either the plus or minus strand-specific probe on the filter was detected by autoradiography (Fig. 2B, 8 kb RNA), whereas radiolabeled RNA prepared from permeabilized Huh-7 cells in the same way did not show any hybridization signals on the filter (data not shown). By the same procedure, we also confirmed that the metabolically radiolabeled 8-kb RNA in Fig. 1C was actually replicon RNA (data not shown). The ratio of plus to minus strand RNA synthesized in the permeabilized replicon cells was estimated to be 2.7 Ϯ 0.4 (average Ϯ S.D.) by three independent experiments. These results indicated that the radiola-beled 8-kb RNA was actually HCV subgenomic RNA synthesized in the permeabilized replicon cells and contained both plus and minus strands of the HCV RNA, implying that the functional HCV replication complexes were present in the permeabilized replicon cells.
Replicon RNA in the Permeabilized Replicon Cells Was Resistant to Nuclease-The production of radiolabeled HCV subgenomic RNA in the permeabilized cells seemed to continue until 4 h and reached a maximum until ϳ5 to 6 h (data not shown). Then the amount of radiolabeled HCV RNA was stably maintained even after 8 h. One possible explanation for this stability was the removal of RNase from the cell by permeabilization. Therefore, we examined the sensitivity of the newly synthesized HCV RNA in the permeabilized replicon cells to exogenously added nuclease. After the reaction for RNA synthesis, the permeabilized replicon cells were treated with micrococcal nuclease. As shown in Fig. 3A, the radiolabeled HCV RNA remained almost intact even after nuclease treatment (lanes 1 and 2), although 28 S rRNA was efficiently degradated under the same condition (lanes 2 and 3). This lower sensitivity of the replicon RNA to nuclease suggested that one of the reasons for the stability of the RNA in the permeabilized cells was its resistance to RNase activity. On the other hand, the radiolabeled HCV RNA was sensitive to nuclease in the presence of a nonionic detergent, Nonidet P-40 (Fig. 3A, lane 3), although a small portion of the HCV RNA was likely to remain resistant to nuclease. When the permeabilized replicon cells were treated with only Nonidet P-40, the radiolabeled HCV RNA in the cells was still detectable but apparently diminished. However, the radiolabeled molecules became detectable in the reaction supernatant, possibly because of leakage from the cells because of Nonidet P-40 (Fig. 3A, lanes 4 and 8). From these results, HCV subgenomic RNA newly synthesized in the permeabilized cells seemed resistant to nuclease in a cellular lipid membrane-dependent manner. Moreover, we also observed that the replicon RNA synthesis was equally carried out when the permeabilized cells were pretreated with nuclease before the reaction of the RNA synthesis (data not shown). This suggested that the HCV RNA used as a template in the viral RNA synthesis was also resistant to nuclease and present in the same environment as the newly synthesized RNA products mentioned above. Furthermore, we investigated the replicon RNA that was newly synthesized in the intact replicon cells. After metabolic labeling of the replicon cells with [ 32 P] orthophosphate, the cells were permeabilized with digitonin and treated with micrococcal nuclease as above. As shown in Fig. 3B, about 30% of radiolabeled replicon RNA in the intact replicon cells was found to be lost after permeabilization ( lanes  1 and 2), probably implying that a part of the newly synthesized RNA flowed out with the cytoplasm from the permeabilized cells (see discussion). A large part of newly synthesized replicon RNA retained in the permeabilized cells showed resistance to the nuclease action in the absence of Nonidet P-40 (Fig. 3B, lane 3) but was tuned to be sensitive to nuclease in the presence of Nonidet P-40 (Fig. 3B, lane 4). This suggested that the fate of the newly synthesized replicon RNA in the permeabilized cells was similar to that retained in intact cells after permeabilization. Then we analyzed the nuclease sensitivity of total replicon RNA in replicon cells by Northern blotting. After permeabilization, replicon cells were treated with nuclease in the presence or absence of Nonidet P-40 as described above. As shown in Fig. 3C, we found that most of the replicon RNA was intact even after treatment (Fig. 3C, lanes 2 and 3). As in the case of the newly synthesized replicon RNA, the replicon RNA pre-existing in the replicon cells was sensitive to nuclease in the presence of Nonidet P-40 (Fig. 3C, lane 4). From these  3, 4, 7, and 8) digitonin. Total RNA extracted from the cell fraction (Cell), as well as the reaction supernatant (Sup), of these cells after incubation in the reaction mixtures for 4 h was analyzed by denaturing agarose gel electrophoresis followed by autoradiography. The 8-kb RNA specifically found in the cell fraction of replicon cells is indicated by an arrowhead. B, characterization of the radiolabeled 8-kb RNA synthesized in the permeabilized replicon cells. Equal amounts of the unlabeled plus and minus strand replicon RNA synthesized in vitro were blotted on nylon membrane filter and used as specific probes for detection of the minus and plus strand replicon RNA synthesized and radiolabeled in the replicon cells, respectively, in the slot blot hybridization experiment (8 kb RNA). To show the strand specificity of this experiment, the results of the experiment performed by using the plus and minus HCV RNA 32 P radiolabeled in vitro instead of those in the replicon cells are presented (in vitro transcript, plus and minus). results, it was suggested that replicon RNA existed in a subcellular compartment that was formed with cellular lipid membranes.
A Large Amount of Each NS Protein Was Not Required for Nuclease Resistance of Replicon RNA-To investigate the contribution of HCV NS proteins to the resistance of replicon RNA against nuclease, the sensitivity of the replicon RNA to nuclease was examined in permeabilized replicon cells after treatment with proteinase K at various concentrations. The condition of the replicon RNA and NS proteins in the cells after treatment was analyzed by Northern and Western blotting, respectively. An endogenous Calnexin, a type I transmembrane protein of ER, was detected by using an antibody specific to its NH 2 -terminal region present in the ER lumen (Calnexin-NT) as a control for monitoring the efficiency of proteinase K digestion. At the dose of proteinase K that affected the amount of full-length Calnexin, a fragment, the size of which matched that of the NH 2 -terminal region of Calnexin was detected (Fig.  4, lanes 4 -11), indicating that proteins or segments on the cytoplasmic side of ER membranes were efficiently digested and those in the lumen were protected from the digestion by proteinase K as expected. Under these conditions, a large part of each NS protein was digested by proteinase K at higher concentrations, although the sensitivity of each NS protein to treatment varied (Fig. 4, lanes 4 -7 and 8 -11). On the other hand, we also observed that a small part of each NS protein remained resistant to even prolonged treatment of the permeabilized replicon cells with proteinase. Under these conditions, we observed that nuclease treatment did not affect replicon RNA despite efficient degradation of 28 S rRNA (Fig. 4, lanes   8 -11), although amounts of 28 S rRNA and replicon RNA were not influenced by only the proteinase treatment (Fig. 4, lanes  4 -7). These results indicated that the stability replicon RNA was not dependent on the large parts of the NS protein. This raised the possibility that the precise subcellular localization of replicon RNA was different from that of the majority of each NS protein.
The Active Replication Complex Was in a Similar Environment to the Replicon RNA-As shown above, the majority of each NS protein in digitonin-permeabilized replicon cells were sensitive to proteinase treatment. Furthermore, all NS proteins (NS3-NS5B) in replicon cells were detectable by indirect immunofluorescent analysis after permeabilization only with digitonin following fixation with 4% paraformaldehyde, although the proteins such as protein-disulfide isomerase and BiP/Grp78 that are present in the ER lumen were not detected under this condition (data not shown). These results indicated that a large part of each NS protein was exposed on the cytoplasmic side of the inner cellular membranes. To see whether these NS proteins function in the replication of HCV, the synthesis of the HCV subgenomic RNA was investigated in the permeabilized replicon cells pretreated with proteinase K. After treatment with proteinase at higher concentrations, each NS protein was almost degradated by the treatment as shown in Fig. 4 (Fig. 5A, lanes 2-5). Under these conditions, however, the HCV RNA was synthesized in a quite similar manner to that in the cells not treated with proteinase (Fig. 5A, lanes 1-5). These results suggested that the majority of active replication complex existed in the proteinase-resistant environment and a large part of each NS protein did not participate in the synthe-  , lanes 2, 3, 6, and 7) or without microccocal nuclease (nuclease Ϫ, lanes 1, 4, 5, and  8). The reactions were performed in the presence (Nonidet P-40 ϩ, lanes 3, 4, 7, and 8) or absence of 0.5% Nonidet P-40 (Nonidet P-40 Ϫ, lanes 1, 2, 5, and 6). Total RNA extracted from the cells (Cell, lanes 1-4) and the reaction supernatant (Sup, lanes 5-8) was similarly analyzed as described in the legend to Fig. 2. B, resistance of the newly synthesized replicon RNA in the intact replicon cells to nuclease action. The intact replicon cells were metabolically labeled with [ 32 P]orthophosphate. After incubation, the cells were permeabilized and treated with (lanes 3, 4, 7, and 8) or without micrococcal nuclease (lanes 2, 5, 6, and 9), in the presence (lanes 4, 5, 8, and 9) or absence of Nonidet P-40 (lanes 2, 3, 6, and 7), as described above. C, resistance of the replicon RNA in replicon cells against nuclease action. After permeabilization of the replicon cells with digitonin, the cells were treated with (lanes 3, 4, 7, and 8) or without micrococcal nuclease (lanes 2, 5, 6, and 9), in the presence (lanes 4, 5, 8, and 9) or absence of (lanes 2, 3, 6, and 7) Nonidet P-40 as above. Total RNA from the cell fraction or the reaction supernatant after treatment was analyzed by Northern blotting using a HCV genome-specific probe to detect the replicon RNA (indicated by the arrowhead with replicon, upper panel). 28 S rRNA (28S) was detected by staining with ethidium bromide (A-C, lower panels). In A, the amount of 28 S rRNA seemed to be low probably because of degradation during incubation in the reaction mixture for 4 h. sis of HCV RNA in the cells. This implied that a small part of each NS protein resistant to proteinase should be present in a similar situation to the active replication complexes in the cells. To clarify the existence of such proteins, we examined carefully the proteinase resistance of each NS protein in the permeabilized replicon cells. After treatment of the permeabilized replicon cells with proteinase K at various concentrations, total protein in the cells was analyzed by Western blotting. Under the conditions that the COOH terminus of Calnexin was efficiently digested by treatment as shown in Figs. 4 and 5A, we found that a small part of each NS protein was resistant to the treatment with proteinase K at even higher concentrations (Fig. 5B, lanes 1-6). These results strongly suggested that such proteinase-resistant NS proteins formed the functional replication complex in the cells. Furthermore, we did not detect any activity for the synthesis of the replicon RNA in the permeabilized replicon cells after treatment with nonionic detergents (data not shown). Summing up the above results, it was suggested that pre-existing replicon RNA and a small portion of each NS protein, which were resistant to nuclease and proteinase, respectively, form the replication complexes in lipid membrane structures to participate in the synthesis of replicon RNA in replicon cells.

DISCUSSION
By monitoring the synthesis of HCV RNA in permeabilized HCV replicon cells, we obtained results suggesting that the active HCV replication complexes function to synthesize the replicon RNA in subcellular compartments, which are probably formed by cellular lipid membranes. Recently, electron microscopic analysis showed that the expression of HCV proteins in Huh-7 cells induced the formation of a distinct membrane structure, designated a "membranous web," and all HCV proteins were found in the structure (40). Moreover, a similar web-like structure in livers of HCV-infected chimpanzees and HCV subgenomic replicon cells was reported (41,42). Therefore, it seems likely that this membrane structure is a candidate for the site where the HCV RNA genome is mainly located and replicated. Until now the genomes of many positive-strand RNA viruses, such as brome mosaic virus, murine hepatitis virus, and Kunjin virus have been reported to replicate on inner cellular membranes in association with vesicles or other membrane structures (43)(44)(45). These membrane structures seemed to be constructed by viral proteins. For example, it was reported that brome mosaic virus 1a, the multifunctional RNA replication protein, selectively recruits brome mosaic virus 2a polymerase and viral RNA and forms membrane-bound spherules (43,46). A subcellular site for the genome replication of these viruses including HCV has been suggested by the localization of brome uridine-incorporated genomic RNA, which is a FIG. 4. A large part of each NS protein was not required for the nuclease resistance of the replicon RNA. The digitonin-permeabilized replicon cells were treated with proteinase K at various concentrations (0 g/ml for lanes 2 and 3, 0.01 g/ml for lanes 4 and 8, 0.1 g/ml for lanes 5 and 9, 1 g/ml for lanes 6 and 10, 10 g/ml for lanes 7 and 11), followed by treatment with (lanes 3 and 8 -11) or without (lanes 2 and 4 -7) microccocal nuclease. The samples from intact replicon cells without any treatments were similarly analyzed as shown in lane 1. After these treatments, total RNA and protein in the cells were analyzed by Northern and Western blotting, respectively. Antibodies used in the Western blotting were anti-NS4B (NS4B), anti-NS5A (NS5A), anti-NS5B (NS5B), anti-Calnexin (Calnexin-NT), and anti-KDEL (BiP/Grp78) antibodies. Each protein with original size is indicated by an arrowhead. An asterisk denotes the position of the Calnexin NH 2 -terminal segment that is located in the lumen of the ER. BiP/ Grp78, which located on the lumenal side of the ER membrane, was used as a negative control for proteinase digestion. product of the replication reaction, in the cells (43,(47)(48)(49). It has not been biochemically analyzed, however, whether viral RNA is actually synthesized in or around such a membrane structure. In this paper, we showed that a quite similar activity for the synthesis of HCV subgenomic RNA was present in the permeabilized replicon cells even after proteinase treatment, which digested almost all the NS proteins including NS5B, to that in the cells not treated with proteinase. Furthermore, we found that a small part of each NS protein is actually present in the proteinase-resistant environment in the replicon cells. This suggested, therefore, that a small part of each NS protein forms the replication complex and functions in the membranous compartment.
A fairly large amount of each HCV protein accumulated in the perinuclear fraction in the replicon cells, which is exposed to the cytoplasmic environment. In this protein complex, no association of the HCV subgenomic RNA was observed and there was no activity to synthesize the viral RNA. Thus, the significance of the HCV protein complex regarding multiplication of the virus genome remains to be clarified. The replicon cells originating from Huh-7 may produce large numbers of HCV proteins in the perinuclear fraction as a consequence of overproduction and these proteins may play less important roles in the replication of the HCV genome than the active replication complex that we have noted in this paper. Conversely, the protein complex may play important roles in the regulation of not only the multiplication of the HCV genome but also further processes during the maturation of the virus. In this regard, it is important to analyze the presence as well as the function of the HCV protein complexes in cells where HCV proliferates with different degrees of multiplication. To date, several cellular proteins that interact with particular NS proteins have been reported. For example, double strand RNA-dependent protein kinase, soluble NSF attachment protein receptors-like protein, and karyopherin ␤3 are reported to interact with NS5A (50 -52), although the physiological importance of these interactions has been obscure. This larger part of each NS protein might, therefore, participate in several cellular events and/or modulate the replication of the HCV genome through interactions with these cellular proteins.
The ratio of plus to minus strand RNA synthesized in the permeabilized replicon cells was estimated to be 2.7 Ϯ 0.4 by slot blot hybridization analysis as shown in Fig. 2. On the other hand, that in the intact replicon cells was estimated to be 11.9 Ϯ 2.0 as reported previously (data not shown and Ref. 15), when intact replicon cells were metabolically labeled with [ 32 P]orthophosphate and the newly synthesized and radiolabeled replicon RNA in the cells was analyzed by slot blot hybridization in a similar manner. The discrepancy in the ratio between the permeabilized and intact cells may be explained by the release of the replicon RNA synthesized in the permeabilized cells from the membranous compartment and degradated in the reaction mixture, although the possibility that the regulation of the plus to minus ratio of newly synthesized replicon RNA may be lost in the permeabilized replicon cannot be completely ruled out. Approximately 50% of the replicon RNA newly synthesized in the intact cells was actually lost by nuclease treatment following permeabilization (Fig. 3B, lanes 1  and 3), suggesting that some parts of replicon RNA newly synthesized in intact replicon cells is present in the cytoplasm. Slight degradation of the replicon RNA was also seen in permeabilized cells in the presence of Nonidet P-40 (Fig. 3). We also observed that in vitro synthesized replicon RNA added exogenously to the permeabilized cells was unstable in the reaction mixture (data not shown), as was observed in a recent report in which the replication activity of the HCV replicon was detected using the cell lysate fraction of replicon cells (30). These observations seem to support the former possibility indirectly. As the radiolabeled nucleotide substrate was incorporated in the newly synthesized replicon RNA in the permeabilized replicon cells, a channel-like structure should be present in the membranous compartment including the replication complex. In the case of spherules of brome mosaic virus, the channel-like structure connecting to the cytoplasm with the inside of the spherule was actually detected by the electron microscopic technique (43). This supported the idea that the replicon RNA that was present in the cytoplasm was probably exported from the membranous structure through the channellike structure. A similar phenomenon was already reported for reovirus in that relatively large reoviral mRNA was exported from the viral core particles to the cytoplasm through the channel formed by a viral protein (53). The pore size of the putative channel of HCV, however, seemed to be limited, because nuclease and proteinase did not pass through the channel. The replicon RNA may be specifically recognized by some viral proteins and exported through the channel post-or coreplicationally. These mechanisms, however, have remained to be elucidated.
During the preparation of this article, Shi et al. (54) reported that almost all of the NS5A and part of the NS5B proteins were present in the membrane fraction that was resistant to treatment with 1% Nonidet P-40. Because the replicational activity in that fraction from the cell lysate was not demonstrated, we do not know whether that kind of membrane fraction from the cell lysate would include the replication activity observed here. Furthermore, a part of replicon RNA was reported to localize in the non-ionic detergent-insoluble membrane fractions (54), whereas we showed that the RNase resistance of both total and newly synthesized replicon RNA was reduced in the presence of Nonidet P-40. This discrepancy may be explained by the different methods used for the detection of the replicon RNA, reverse transcriptase-PCR and Northern blotting in that paper and ours, respectively. We also showed data indicating that a negligible amount of replicon RNA was released from the permeabilized cells into the reaction mixture by the detergent as shown in Fig. 3. It may also be relevant to this discrepancy that we and others have observed instability of the free replicon RNA in the reaction mixture (30). We also observed the existence of nuclease-resistant replicon RNA, even in the presence of nonionic detergent (Fig. 3). Such RNA may represent the replicon RNA in the non-ionic detergent-insoluble fractions.
Here we showed the stable nature of the active HCV RNA replication complex in the replicon cells by permeabilization of the cells. Our data suggests that only a small part of each NS protein contributes to RNA synthesis in the replicon cells. This implied that careful investigation would be required for identification of the precise subcellular sites of replication in the replicon cells. Further investigation to reveal how and where this complex is formed in the cells and what is essential for its activity is required for understanding the mechanism of viral replication and the life cycle of HCV.