Introduction

Hepatitis C virus (HCV) is a major cause of chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma. With over 170 million people currently infected [2], HCV is a growing public-health burden.

The life cycle of HCV has been difficult to study because cell culture and small animal models of HCV infection are not available. The recent development of HCV replicon systems has permitted the study of HCV translation and RNA replication in human hepatoma-derived Huh-7 cells in vitro [17]. However, these replicon systems cannot produce authentic virions because they lack the infection steps, and analysis of these infection steps is very important for understanding HCV pathogenesis.

Recently, some groups have successfully established in vitro infection systems [16, 21, 26, 28–30]. The strategies of these systems are basically the same as the ones used for transfection of Huh-7 cells or their derivatives with in vitro-generated HCV genome RNA [1]. The non-structural regions used in those studies were from the 2a genotype JFH (Japan Fulminant Hepatitis)-1 clone or the 1a genotype H77 clone. The former is known for its exceptionally vigorous amplification and broad permissiveness in cultured cells other than Huh-7 [3, 12, 13], while the latter shows only poor replication ability. Another group reported a newly established immortalized hepatocyte cell line that is susceptible to HCV infection, but only modest improvement was achieved [10]. There are also reports of a system using a full-genome replicon that has the entire coding region under the control of the internal ribosomal entry site of encephalomyocarditis virus, EMCV-IRES; however, this system also failed to show infectivity in the G418 selection assay [7, 20], and secretion of particles with the putative characteristics of HCV virions could not be confirmed [4].

We now report the establishment of infectious virion-producing replicon cells that utilize an ordinary genotype 1b replicon strain. In order to address the contribution of structural and non-structural gene products to the maturation of HCV particles in vitro, we partitioned these regions in the same cistron of the full genomic sequence, thereby enabling the functions of these structural and non-structural genes to be studied separately. Thus, we termed this construction “divided open reading frame carrying” full genome replicon, or dORF replicon.

Virus particles secreted from cells containing dORF replicon RNA, as confirmed morphologically using electron microscopy, were shown to be able to infect Huh-7 cells. Replication of dORF replicon RNA was so efficient that infected cells could survive and proliferate under G418 selection to form colonies, as seen after transfection with replicon RNA. In addition, a reporter gene was successfully inserted into the construct, and activity of the reporter gene could be transmitted to naive Huh-7 cells by infection.

We believe that the success of this system is due to the difference in the construction of the replicon, namely, having the intact 5′ half extending to NS2 instead of being divided at the beginning of the core region. Although further investigation is required to elucidate whether the encapsidation signal of HCV is located in the region that is divided in the full-genome replicon, this is the first report to describe genome-length replicon-containing cells that can produce virus particles that have the putative characteristics of the HCV virion, in terms of both morphology and biological properties.

Results

dORF replicon RNA can replicate in Huh-7 cells

We began this study with transfection with the dORF replicon RNAs (Fig. 1A). When 30 μg of each RNA was electroporated into 4 × 106 Huh-7 cells, the dORF and dORF bla RNA-transfected cells formed 20 and 5 colonies, respectively, after 3 weeks of G418 selection. No colonies appeared as a result of transfection with polymerase-defective mutants (data not shown). Two colonies were picked, amplified, and designated as dORF replicon cell #1 and #2, and dORF bla replicon cell #1 and #2. Some of these cells were then used for quantification of HCV RNA and northern blot analysis (Fig. 1B). Northern blot analysis showed that these clones contained HCV RNAs of the expected size and that the HCV RNA copy numbers of these clones did not differ substantially from that of the subgenomic replicon, indicating that replication ability had not been hampered by insertion of the structural genes, which is counter to what was expected. Western blot analysis showed that these clones express both structural and non-structural proteins (Fig. 1C). These results confirmed that transfected dORF HCV RNAs can replicate in Huh-7 cells, just as authentic subgenomic replicon RNAs do.

Fig. 1
figure 1

Confirmation of “divided open reading frame carrying” (dORF) replicon cells. (A) Schematic representations of replicon RNAs used in this study. All the replicon constructs contained inserts just after the T7 promotor. UTR, untranslated region; NS, non-structural protein; neo, neomycin phosphotransferase II; EMCV, encephalomyocarditis virus; IRES, internal ribosomal entry site; FMDV, foot-and-mouth disease virus; bla, beta-lactamase. (B) Northern blot analysis. A 10-μg amount of total RNA from each cell sample was loaded. Subgenomic replicon RNA: 108 copies of in vitro-generated subgenomic RNA. Numbers below the lanes are the HCV copy number per microgram of total RNA. Huh-7 cell, subgenomic replicon cell, dORF replicon cell #1, #2, dORF bla replicon cell #1, #2. (C) Western blot analysis. A 10-μg amount of each cell lysate was loaded. Huh-7 cell, Huh-7-JFH1: Huh-7 cell transfected with JFH1 viral RNA, subgenomic replicon cell, full-genome replicon cell, dORF replicon cell #1, #2, dORF bla replicon cell #1, #2

Full size image

dORF replicon cells secrete virus particles

In a previous study, HCV subgenomic replicon cells secreted RNase-resistant subgenomic RNA into the culture supernatant [4, 7, 20]. We also detected a similar amount of RNase-resistant HCV RNA in the culture supernatant of our dORF replicon cells, as well as of the subgenomic and full-genome replicon cells. These supernatants showed no significant differences in terms of distribution of HCV RNA in buoyant density gradient analysis (Figs. 2A, B, open square). In contrast, there was a clear difference between these supernatants after NP-40 treatment. While almost all of the HCV RNA in the supernatant of the subgenomic replicon cells was eliminated by NP-40 treatment (Fig. 2A, filled triangle), there remained a peak of HCV RNA at a density of 1.18 g/mL in the supernatant of the dORF replicon cells (Fig. 2B, filled triangle). These results were confirmed in the same experiment, using concentrated culture supernatant (Figs. 2C, D). We also confirmed the results of previous reports [7, 20], which showed no genomic RNA resistant to NP-40 treatment in the supernatant of full-genome replicon cells (Fig. 2E). Secreted core proteins in the concentrated supernatant showed a different density gradient distribution compared to genomic RNA (Fig. 2F, open circle) in that the core proteins were present at densities of 1.1–1.2 g/mL, while HCV RNA was more broadly distributed in the range of 1.06–1.22 g/mL. Thus, HCV RNA and core proteins were not always associated with each other. However, after NP-40 treatment, core proteins were found only in the same fraction as HCV RNA, at 1.19 g/mL (Fig. 2F, filled triangle). Taken together with the results of the report mentioned above [20], our replicon cells harboring dORF RNA appeared to secrete particles with core proteins that were assembled into nucleocapsids as well as particles without core proteins that were sensitive to NP-40 treatment, like the ones from subgenomic and full-genome replicon cells. We concluded that the broader distribution of the HCV genome RNA in the density gradient than that of the core protein was caused by the overlapping distribution of these two particle types, and that the remaining peaks of genome RNA and core protein after NP-40 treatment were of nucleocapsids that had had their envelopes stripped off by NP-40 [11].

Fig. 2
figure 2

Density gradient analysis of supernatants. Culture supernatants were treated with RNaseA and loaded directly onto a sucrose density gradient without treatment (open square) or after NP-40 treatment (filled triangle). Quantification of HCV RNA in each fraction of supernatant from the subgenomic replicon (A) and dORF replicon (B). Analysis of concentrated culture supernatant from the subgenomic replicon (C) and dORF replicon (D). Concentrated culture supernatant from the full-genome replicon NNC#2 was also analyzed (E). Quantification of HCV core protein in each fraction of supernatant from the dORF replicon (F)

Full size image

According to our hypothesis, the distribution of core proteins in the density gradient represented that of the intact virion, and we therefore tried to observe virions directly by electron microscopy, using the fraction in which the core protein was present. We easily identified numerous round-shaped virus particles approximately 50 nm in diameter by scanning electron microscopy (Fig. 3A). Furthermore, when the immunogold method using anti-E2 RR6 antibody was applied to samples fixed on the mesh, transmission electron microscopy could be used to visualize virus particles labeled with colloidal gold (Fig. 3B). These findings provide evidence of intact virion production from our dORF replicon cells.

Fig. 3
figure 3

Electron microscopy analysis of virus-like particles. The core-protein-rich fraction collected from the density gradient was further concentrated by ultracentrifugation and observed by scanning electron microscopy (A). The same fraction attached to formvar-coated grids was incubated with rabbit anti-E2 RR6 antibody, treated with goat anti-rabbit IgG coupled to 10-nm colloidal gold, negatively stained with uranyl acetate, and then examined by transmission electron microscopy (B)

Full size image

Secreted virus particles can infect naive Huh-7 cells

Next, we examined the infectivity of these virus particles. The culture supernatants of these dORF replicon cells were collected, and 3 kinds of naive Huh-7 cells, one purchased from the J.C.R.B. (Japanese Collection of Research Bioresources) and the other two, designated as the cured cells F2 and K4, generated by IFN-α treatment of 1bneo/delS replicon cells, were infected with these supernatants. After two sequential passages and three weeks of G418 selection as described above, a number of colonies appeared, as shown in Fig. 4A. The largest number of colonies was produced from the cured cells K4, and slightly fewer colonies were produced from the cured cells F2, while no colonies appeared when normal Huh-7 cells were used (data not shown). The same infection experiment carried out with full-genome replicon cells produced no infectivity in the supernatant (data not shown). Under the most efficient conditions, the titer of the supernatant reached as high as 20 cfu (colony-forming units) per milliliter when the putative doubling time of these cells was approximately 24 h. Furthermore, the appearance of colonies was abolished by addition of the antibody JS-81 (BD Pharmingen), an antibody to CD81, a possible co-receptor of HCV [22] (Fig. 4B).

Fig. 4
figure 4

Infectivity of supernatants from various replicon cells. Colonies of cells infected with the indicated supernatant. Numbers shown below the plates are the average of a total of four plates per condition (A). Inhibition of infection by anti-CD81 antibody. Cured cell K4 cells (No.2 in Fig. 4A) were treated with mouse IgG1 as the negative control or anti-CD81 before infection (B)

Full size image

Next, we propagated some of these colonies for further analysis. Northern blot analysis showed that these clones carry HCV RNAs of reasonable size (Fig. 5A), including subgenomic RNA (7994 bases), dORF RNA (10994 bases), and dORF bla RNA (11840 bases). Western blot analysis revealed that the cell clones that were infected with supernatant from Huh-7 cells containing the dORF replicon expressed structural proteins (Fig. 5B), indicating that the colonies were not just the reappearance of subgenomic replicons hidden in the cured cells.

Fig. 5
figure 5

Northern blot analysis of colonies formed after infection. 107, 109: amounts of in vitro-generated subgenomic replicon RNA loaded. Numbers below the lanes are the HCV copy number per μg of total RNA (A). Huh-7 cells, subgenomic replicon cells, dORF replicon cell #2, dORF bla replicon cell #2, subgenomic replicon sup: colony from cells transduced with subgenomic replicon supernatant, colony No.1, 2 of dORF replicon #2 sup: colonies from cells infected with dORF replicon #2 supernatant, colony No.1, 2, and 3 of dORF bla replicon #2 sup: colonies from cells infected with dORF bla replicon #2 supernatant. Western blot analysis of colonies formed after infection (B). The order of the lanes is identical to that for the northern blot, except for the dORF and dORF bla replicons, which represent two clones in this figure

Full size image

Together, our findings indicate that these particles in the supernatant infected the Huh-7 cells through a CD81-associated pathway and that infected cells formed colonies after G418 selection, similar to what was observed with electroporation with subgenomic RNA.

A reporter gene inserted into the dORF replicon RNA can be transmitted through infection

First, we confirmed that the beta-lactamase gene in the dORF bla replicon RNA was active in established replicon cell clones and able to process the green fluorescent substrate into blue fluorescent product (Fig. 6A). Next, we attempted to detect the activity of the beta-lactamase gene in the cloned infected colonies. Three clones grown from cells infected with the dORF bla supernatant were treated using a GeneBLAzer In Vivo Detection Kit. One clone was positive for blue fluorescence (Fig. 6B), demonstrating that a reporter gene inserted into the dORF replicon could be transmitted to naive Huh-7 cells through secreted virus particles in the culture supernatant.

Fig. 6
figure 6

Detection of beta-lactamase activity in dORF replicon cells. Parental dORF bla replicon #2 cell (A) and colony no. 3 cloned from cells infected with dORF bla replicon #2 cell supernatant (B). Blue fluorescence shows high beta-lactamase activity, indicating that the reporter gene functioned normally after infection

Full size image

Discussion

There have been several previous reports of full-genome HCV replicons that can replicate well in Huh-7 cells and express sufficient amounts of structural proteins [1, 4, 7, 14, 20]. Pietschmann et al. (2002) observed the secretion of an RNase-resistant HCV genome into the supernatant from both full-genome and subgenomic replicon cells and nonspecific uptake of these genomes by naive Huh-7 cells. Ikeda et al. (2002) were also unable to detect any infectivity in the supernatant of their full-genome replicon cells. They assumed that the reason for this failure was the inability of Huh-7 cells to release intact virions or to be infected by the virus, although this was later demonstrated not to be the case by a series of reports on infection using the JFH-1 clone [16, 26, 30].

First, we attempted to improve the efficiency of the full-genome replicon in two ways, namely, by modifying the construct and reducing the genome size. Numerous studies have examined the encapsidation signal in the genomic RNA of positive-sense single-stranded viruses [5, 8, 9]. Frolova et al. [5] showed that the encapsidation signal of Sindbis virus lies in the nsP1 gene and is 132 nucleotides long. Johansen et al. [9] found that the IRES of poliovirus had the ability to enhance the efficiency of packaging of the polio subgenomic replicon. We think that these findings indicate that the construction of the genome could affect the efficacy of encapsidation, and we therefore decided to change the site of genome division from the beginning of the core region to the middle of the NS2 region. Regarding the size of the genome, there have been reports that the insertion of a foreign gene of significant size can result in the deletion of a portion of the chimeric genome during replication [18, 19]. We therefore removed the second half of the NS2 region, because this region appears to be unnecessary for both replication and packaging in Huh-7 cells, and this deletion was found to have no influence on the efficacy of encapsidation, as there were no apparent differences between the NS2-deleted construct and the one containing the entire NS2 region (data not shown).

Our established dORF replicon was able to replicate well in Huh-7 cells and express sufficient amounts of structural proteins, similar to the previously reported full-genome replicon. Although both the dORF replicon cells and the previously reported full-genome replicons secreted RNase-resistant genomes, there was a striking difference between these two full-genome replicons when NP-40 treatment was carried out on their supernatants. There was no RNase-resistant genome left in the NP-40-treated supernatant of full-genome replicons, although density gradient analysis of the NP-40-treated supernatant of dORF replicon cells clearly showed the coexistence of the HCV genome and core proteins at a peak of 1.18 g/mL. This peak may represent NP-40-resistant nucleocapsids. The distribution of core proteins in the density gradient analysis of the concentrated supernatant of the dORF replicons did not match that of the HCV genome. A reasonable explanation for this mismatch is that the lighter side of the broad peak of the HCV genome was not representative of intact virions and is instead an indication of secretion by a pathway used in subgenomic replicon cells, which differs from the natural process. The fact that the peak of the HCV genome of full-genome replicons was located in a narrow range on the lighter side compared to that of the dORF replicons supports this hypothesis. We observed round particles in the concentrated core protein fraction using electron microscopy, and those particles also seemed to contain core proteins. These findings indicate that our dORF replicon cells produced both intact virions and artificial membranous particles, with the former having the morphological and biophysical characteristics of putative virions.

The colony-forming assay clearly demonstrated the ability of the supernatants of our dORF replicon cells to infect Huh-7 cells efficiently. The reason for the difference in efficacy between the two cured cells is uncertain but may involve the ability to support replication or the level of receptor expression. This needs to be clarified in order to improve the efficiency of HCV infection in vitro. Differences in the efficiency of infection were also noted between clones of the same dORF replicon cells, which may have been due to the accumulation of different mutations in the structural region, although we have not yet confirmed this hypothesis. We also observed colonies being formed by cells that were treated with supernatant containing subgenomic replicons, and these colonies most likely represent the so-called “non-specific transduction” of the subgenomic replicon. Although this dORF supernatant infection could be blocked by the anti-CD81 antibody reported previously [30], we cannot exclude the possibility that the infection we observed was due to highly efficient “non-specific transduction,” as we could not determine whether “non-specific transduction” also could be affected by the anti-CD81 antibody because of the low colony-forming ability of the supernatant of subgenomic replicons.

We also demonstrated that the reporter gene that was inserted in addition to the neomycin resistance gene could be transmitted to the new generation of viruses. This finding raises the possibility of producing sufficient amounts of reporter virus constitutively.

In summary, we established an infectious-particle-producing HCV replicon system. This achievement should yield more precise information about the encapsidation signal of HCV, which was kept intact despite the partitioning of the genome. This system also allows analysis of the pathway of HCV infection, including adsorption of virions to cell-surface receptors, penetration, uncoating, virus particle assembly, and HCV release. Moreover, the dORF replicon system may be used as a convenient tool to investigate the utility of the newly established siRNA system [14, 27] and evaluation of compounds that are effective against subgenomic replicons.

Although we believe that the reason for our success is our new construct, further examination is necessary to verify our findings.

Materials and methods

Construction and RNA transcription

To construct dORF replicon RNA, the second half of the NS2 region of the HCV-R6 strain [25] was replaced in frame with the foot and mouth disease virus (FMDV) 2A protease gene, the neomycin resistance gene, and the encephalomyocarditis virus (EMCV) internal ribosomal entry site (IRES). In addition, the region from NS3 to the beginning of NS5B was replaced with the 1bneo/delS replicon sequence from the N strain of genotype 1b [6](kindly provided by Dr. Seeger). This construct was designated as the “divided open reading frame carrying full genome” (dORF) replicon. The subgenomic replicon construct was also prepared from the R6 strain and also contained the 1bneo/delS replacement. For the reporter assay, the FMDV 2A protease gene and beta-lactamase gene (bla; Invitrogen) were inserted after the remaining NS2 gene to produce the dORF bla replicon construct. Replication-deficient versions of these three replicons were also prepared by deleting 27 nucleotides, including the GDD motif of NS5B polymerase.

In vitro transcription of these replicon RNAs was performed using the MEGAscript kit (Ambion).

Cell culture and electroporation

Huh-7 cells were cultured in DMEM (SIGMA) with 10% fetal bovine serum. Replicon cells were maintained in the same medium supplemented with 300 μg/mL G418 (Invitrogen). These cells were passaged 3 times a week at a 4:1 splitting ratio. Electroporation of replicon RNA was performed as described previously [17]. The subgenomic replicon (1bneo/delS replicon) cells were treated with 1000 IU of IFN-α for 2 months and cloned by the limited dilution method. Two of these clones were designated as HCV replicon-cured Huh-7 cells F2 and K4. The cell line containing the full-genome replicon of genotype1b, namely the NNC#2 clone [15], was a kind gift from Dr. Shimotohno of Keio University.

Northern blot analysis and quantification of HCV RNA

Total RNA was purified from cells using ISOGEN (Nippon Gene) for northern blot analysis or ABI prizm6100 (Applied Biosystems) for real-time RT-PCR. Purified RNAs were quantified by absorbance at 260 nm. For northern blot analysis, 30 μg of each total RNA was used with a Northern Max Kit (Ambion) according to the manufacturer’s instructions. The probe for detection of HCV RNA was a PCR fragment of the NS5B region (nucleotide numbers 7629–7963) that had been biotin-labeled using a BrightStar Psoralen-Biotin Kit (Ambion) according to the manufacturer’s instructions. Following hybridization of the membranes, the probe was detected using a BrightStar BioDetect Kit (Ambion) according to the manufacturer’s instructions, and luminescence was detected using the LAS1000 detection system (Fujifilm). Measurement of the HCV RNA copy number by real-time RT-PCR was performed using an ABI PRISM 7900 system (Applied Biosystems) as described previously [24].

Western blot analysis

Western blot analysis was carried out using the conventional semi-dry blot method. Cells were lysed with buffer containing 100 mM Tris-HCl (pH 7.4) and 4% sodium dodecyl sulfate. A 10-μg amount of protein from each sample was separated by SDS-PAGE through a 4–20% gradient gel (Invitrogen) and transferred to the membrane according to the gel manufacturer’s protocol. The antibodies used in this study were anti-core mouse monoclonal antibody (MAb), anti-E1 MAb, anti-E2 MAb (reported previously; [25]), anti-NS3 antiserum (reported previously; [25]), anti-NS5B antiserum (Upstate), and anti-beta-actin MAb (Abcam). Horseradish peroxidase-labeled anti-mouse and anti-rabbit IgG goat antibodies (Santa Cruz Biotechnology and DAKO, respectively) were used as the secondary antibody. The membranes were treated using an ECL Plus kit (Amersham) according to the manufacturer’s instructions, and luminescence was detected using an LAS1000 system (Fujifilm).

Density gradient analysis and core ELISA

Culture supernatants from replicon cells were loaded onto 10–60% sucrose density gradient tubes with or without 10-fold concentration in an Amicon-100 (Millipore). The tubes were then ultracentrifuged at 100,000 g for 16 h and fractionated. NP-40 was added to the culture supernatants to a final concentration of 0.5%, and they were then incubated at 4°C for 30 min. For electron microscopy, the culture supernatant was concentrated, loaded onto a 60% sucrose cushion, and ultracentrifuged at 100,000 g for 4 h. The interface between the concentrated medium and the sucrose cushion was collected and separated by the density gradient method described above. A 2-mL fraction from 5 ml to 7 mL from the bottom, with a density of 1.1–1.2 g/mL, was examined by electron microscopy after further concentration by the sucrose cushion ultracentrifugation method described above. The amount of core protein in the fractions was quantified using an Ohso ELISA kit in accordance with the manufacturer’s instructions.

Electron microscopy

The concentrated fraction of core protein was observed by scanning and transmission electron microscopy. For scanning electron microscopy, the sample was allowed to settle on the surface of a poly-l-lysine-coated glass cover slip for 30 min, and the attached sample was then fixed with 0.1% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 10 min, washed three times with 0.1 M phosphate buffer, and post-fixed with 1% osmium tetroxide in the same buffer for 10 min. After dehydration through a graded series of ethanol, the samples were dried in a freeze dryer (Hitachi ES-2020, Hitachi) using t-butyl alcohol, coated with osmium tetroxide, approximately 2 nm thick, using an osmium plasma coater (NL-OPC80; Nippon Laser and Electronics Laboratory), and then examined using a Hitachi S-4800 field emission scanning electron microscope at an accelerating voltage of 10 kV [23]. For transmission electron microscopy, the sample was allowed to settle on a formvar-coated nickel grid for 10 min, dried in air, incubated with rabbit anti-E2RR6 antibody (prepared as described in the supplementary information), washed with PBS, and then incubated with goat anti-rabbit IgG coupled to 10-nm colloidal gold (British BioCell). After negative staining with 2% uranyl acetate, the sample was examined using a JEM 1200EX transmission electron microscope (JEOL) at an accelerating voltage of 80 kV.

Rabbit anti-E2 RR6 antibody to the HCV-E2 protein was prepared as follows: The E2 gene of HCV type 1b [25] was cloned under the control of the ATI-P7.5 hybrid promoter of vaccinia virus vector pSFB4 and allowed to recombine with the Lister strain of the vaccinia virus to give vector RVV. Rabbits were infected intradermally with 108 p.f.u. of RVV, and 2 months later, they received two booster injections with the purified E2 protein. HCV-E2 protein was expressed from the RVV vector and purified by lentil lectin column chromatography and affinity chromatography using an anti-E2 monoclonal antibody [25].

Infection

A 2.5-ml aliquot of cleared culture supernatants from replicon cells was added to approximately 70% confluent of Huh-7 cells in 25-cm2 flasks, and the same amount of complete DMEM was added 2 h later. Infected cells were transferred to 75-cm2 flasks the next day and to four 10-cm dishes 2 days later. G418 at a concentration of 300 μg/mL was added to the medium immediately after the second passage. The three types of Huh-7 cells used in this study included the one purchased from J.C.R.B. and the 2 IFN-cured replicon cell lines F2 and K4 described above. The medium was changed every other day. For the blocking experiment, cells were treated with the anti-CD81antibody as described previously [30]. Cells were fixed with 10% formalin/PBS(-) for 10 min after washing with PBS(-) and staining with 1% crystal violet/PBS(-) for 1 h before washing with water.

Beta-lactamase detection assay

Beta-lactamase activity was detected using a GeneBLAzer In Vivo Detection Kit (Invitrogen) according to the manufacturer’s instructions and observed using a fluorescence microscope (Nikon) with UV light excitation.