Analysis of Hepatitis C Virus Superinfection Exclusion by Using Novel Fluorochrome Gene-Tagged Viral Genomes

Abstract

Studies of the complete hepatitis C virus (HCV) life cycle have become possible with the development of an infectious cell culture system using the genotype 2a isolate JFH-1. Taking advantage of this system in the present study, we investigated whether HCV infection leads to superinfection exclusion, a state in which HCV-infected cells are resistant to secondary HCV infection. To discriminate between viral genomes, we inserted genes encoding fluorescent proteins in frame into the 3′-terminal NS5A coding region. These genomes replicated to wild-type levels and supported the production of infectious virus particles. Upon simultaneous infection of Huh-7 cells, coreplication of both viral genomes in the same cell was detected. However, when infections were performed sequentially, secondary infection was severely impaired. This superinfection exclusion was neither due to a reduction of cell surface expression of CD81 and scavenger receptor BI, two molecules implicated in HCV entry, nor due to a functional block at the level of virus entry. Instead, superinfection exclusion was mediated primarily by interference at the level of HCV RNA translation and, presumably, also replication. In summary, our results describe the construction and characterization of viable monocistronic HCV reporter genomes allowing detection of viral replication in infected living cells. By using these genomes, we found that HCV induces superinfection exclusion, which is primarily due to interference at a postentry step.

Hepatitis C virus (HCV) is an enveloped virus of the genus Hepacivirinae within the family Flaviviridae (48). At least six HCV genotypes are known, which can be further grouped into various subtypes differing in nucleotide sequence by 20 to 25%. Over 170 million people are persistently infected with HCV, and approximately 38,000 new cases are registered annually in the United States alone (43). In 50 to 80% of all cases, the virus establishes a persistent infection, often leading to chronic liver disease, such as fibrosis, steatosis, cirrhosis, and hepatocellular carcinoma (20, 44). Current therapy comprises a combination of polyethylene glycol-conjugated alpha interferon and ribavirin (43), but success rates are limited, and severe side effects as well as high costs restrict the use of this therapy (13).

The HCV genome is an uncapped linear single-stranded RNA molecule of positive polarity and has a size of ∼9.6 kb. The RNA encodes a large polyprotein of about 3,000 amino acids in a single open reading frame which is flanked at the 5′ and 3′ ends by nontranslated regions (NTRs). Both the 5′ and 3′ NTRs are required for RNA translation and replication (reviewed in reference 4). Expression of the polyprotein is initiated at an internal ribosome entry site (IRES) located in the 5′ NTR. The polyprotein is processed co- and posttranslationally by cellular and viral proteases into the structural proteins (core, E1, and E2) and the nonstructural proteins (p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B). In addition, the production of a novel HCV protein generated by internal translation initiation or by ribosomal frameshifting has been reported, but its function remains unclear (reviewed in reference 8). The nonstructural proteins NS3 to NS5B are sufficient for RNA replication (35), and distinct functions in the HCV life cycle have been described for most of them. However, the functional role of the phosphoprotein NS5A is still enigmatic. Based on tryptic digestion patterns and biochemical as well as structural analyses, NS5A is composed of three domains separated by low-complexity regions (46) (Fig. 1). In addition, an amphipathic α-helix resides at the very N terminus of the protein and mediates NS5A membrane attachment (9). Recently, the crystal structure of NS5A domain 1 was solved, and it was suggested that NS5A interacts with viral RNA (21, 41, 47). Interestingly, NS5A domain 3 is only poorly conserved between different genotypes and tolerates deletions as well as insertion of large heterologous sequences, such as green fluorescent protein (GFP), which allows the visualization of replication in living cells (2, 38).

FIG. 1.

Genomic HCV constructs used in this study and their RNA replication capacities. (A) Schematic diagram of the NS5A structure according to reference 46. An amphipathic α-helix (AH) contributing to NS5A membrane attachment resides at the N terminus. The three domains are indicated by shaded boxes and are separated by trypsin-sensitive regions with presumably low structural complexity (low-complexity sequence [LCS]). The site of insertion of the GFP gene between codons 382 and 383 of the NS5A coding region is indicated (shaded bar), as are the flanking amino acid sequences. (B) Structures of HCV constructs used in this study. The chimeric genome Jc1 is composed of the indicated genome fragments of the two genotype 2a isolates J6CF and JFH-1 fused at a distinct position within NS2 (42). With the exception of JFH1ΔE/GFP, which contains only JFH-1 sequences, all other genomic constructs used in this study were derived from Jc1. The regions encoding fluorescent proteins (GFP, RFP, and Venus-GFP), either inserted into the NS5A coding sequence or expressed from a separate cistron (Venus-Jc1), are shown as striped boxes. The mutant JFH1ΔE/GFP comprises a large in-frame deletion encompassing most of the coding region of E1 and E2. EI, EMCV IRES. (C) RNA replication of HCV genomes in transfected Huh7-Lunet cells, as determined by Northern blot analysis (upper panel). Total RNA was prepared from transfected cells at the time point specified above each lane, and HCV RNA was detected by Northern hybridization, whereas the lower part of the blot was probed with a β-actin-specific RNA probe. The given amounts of HCV in vitro transcripts of a subgenomic replicon spiked into total RNA of naive Huh7-Lunet cells were used as a positive control (lanes 1 to 3 and 17 to 19), total RNA of naive cells served as a negative control (lane 4 and 20), and a replication-deficient JFH-1 genome (JFH1/GND) carrying a mutation in the active site of NS5B served as a negative control for RNA replication (lanes 29 to 32). HCV-specific signals were quantitated by phosphorimaging and, after normalization for RNA amounts loaded into the gel, HCV RNA copy numbers were calculated (lower panel). Levels and kinetics of RNA replication of the monocistronic RNAs were comparable to those of wild-type Jc1, whereas the bicistronic genome Venus-Jc1 replicated to much lower levels and with slower kinetics.

Very recently, the first productive HCV cell culture system was described. It is based on the transfection of genomic HCV RNAs of the genotype 2a isolate JFH-1 into Huh-7 cells (49, 53). Virus particles released from transfected cells are infectious in naive Huh-7 cells and can establish productive infection in vivo (49). Besides the JFH-1 wild type, two groups have described the generation of chimeric HCV genomes allowing the production of higher-titer virus stocks (33, 42). Most efficient is Jc1, a chimera in which the JFH-1 region encoding core up to the first transmembrane segment of NS2 is replaced by the analogous region of the genotype 2a isolate J6 (42).

Several reports have shown that a cell infected with a virus often becomes resistant to secondary infection with the same (homologous) virus, whereas infection with nonrelated (heterologous) viruses is frequently unaffected. This phenomenon, called superinfection exclusion, is brought about by several mechanisms acting at various stages of the viral life cycle. These include interference at the level of virus entry, inhibition of translation, and interference with replication of the incoming RNA (1, 12, 18, 22, 23, 32, 51). For HCV, it is not clear whether superinfection exclusion is induced upon primary infection.

In this study, we developed a panel of HCV genomes that allow direct visualization of infected cells, using fluorescence microscopy, and discrimination between viral genomes in infected cells. By utilizing this system to study superinfection exclusion, we found that preceding infection with HCV strongly interferes with secondary HCV infection. This block is exerted already 24 h after primary infection, and it is mainly due to interference at the level of RNA translation/replication, whereas viral entry is largely unaffected.

MATERIALS AND METHODS

Cells and viruses.

Cell monolayers were routinely grown in Dulbecco's modified minimal essential medium (DMEM; Life Technologies, Karlsruhe, Germany) supplemented with 2 mM l-glutamine, nonessential amino acids, 100 U/ml of penicillin, 100 μg/ml of streptomycin, and 10% fetal calf serum. Viruses were produced by transient transfection of Huh7-Lunet cells (17) with in vitro-transcribed RNA, as described recently (29, 49). Infection assays were performed using Huh7.5 cells kindly provided by C. M. Rice, Rockefeller University, NY (7). Concentrated virus stocks for superinfection experiments were prepared by ultrafiltration of supernatants collected at 24, 48, and 72 h posttransfection by using Amicon ultrafiltration devices (Millipore) according to the manufacturer's protocol. The cell line carrying the subgenomic replicon sg/neo/5A-RFP represents a cell pool obtained by transfection of Huh7-Lunet cells and G418 selection for about 3 weeks (V. Lohmann, unpublished data). To obtain the cured sg/neo/5A-RFP cell line, sg/neo/5A-RFP cells were treated for 2 weeks with 2 μg/ml Debio-025 (40) in the absence of G418, and removal of the replicon was monitored by the loss of red fluorescent protein (RFP) autofluorescence.

Plasmid construction.

All nucleotide and amino acid numbers refer to the JFH-1 genome. Construction of the Jc1 chimera was described recently (42). To generate fluorochrome gene-tagged reporter plasmids, we performed two separate PCRs, using antisense primer Alinker_Xba_Pme_NS5A_aa383 (5′-TTAAACCCAGGTCTAGAACCGCTCGAGGGGGGCTGGCCAAAG-3′) and sense primer SRsrII (5′-CCGAATCCGGCGGTCCGACGTCCC-3′) or sense primer Slinker_Xba_Pme_NS5A_aa383 (5′-TCTAGACCTGGGTTTAAACGTGATGCAGGCTCGTCCACGGG-3′) and antisense primer AHpaI (5′-GACTTGATGTGGTTAACGGCCCTC-3′) for insertion of the linker at NS5A amino acid position 383 and antisense primer Alinker_Xba_Pme_NS5A_aa383 (5′-TTAAACCCAGGTCTAGAACCGCTCGAGGGGGGCTGGCCAAAG-3′) and sense primer SRsrII or sense primer Slinker_Xba_Pme_NS5A_aa383 (5′-TCTAGACCTGGGTTTAAACGTGATGCAGGCTCGTCCACGGG-3′) and antisense primer AHpaI for insertion at NS5A amino acid position 378. Amplicons were combined by a second PCR resulting in fragments that comprise the complete NS5A coding region plus the linker, which were HpaI/RsrII digested and transferred into pFKi389LucNS3-3′_dg_JFH. This plasmid contains the T7 promoter sequence fused to nucleotides 1 to 389 of the JFH-1 consensus sequence, followed by the firefly luciferase gene, which is flanked by in-frame AscI and NotI restriction sites; the encephalomyocarditis virus (EMCV) IRES; the NS3-to-NS5B coding sequence; the 3′ NTR of JFH-1; the hepatitis delta virus genomic ribozyme (dg); and the T7 terminator sequence (Lohmann, unpublished data). Venus-GFP (39), kindly provided by C. M. Brown, University of Otago, Dunedin, New Zealand, emerald GFP, and RFP reporter genes were amplified by PCR and inserted, after restriction digestion with XbaI and PmeI, into the linker plasmids to create reporter plasmids pFKi389LucNS3-3′_dg_JFH-1_NS5Aaa2354_Venus, pFKi389LucNS3-3′_dg_JFH-1_NS5Aaa2359_emGFP, pFKi389LucNS3-3′_dg_JFH-1_NS5Aaa2354_RFP, and pFKi389LucNS3-3′_dg_JFH-1_NS5Aaa2359_RFP, respectively. To generate NS5A fluorochrome gene-tagged Jc1 genomes, an NsiI/MluI fragment including the NS5A coding region was transferred from subgenomic replicon plasmids to the pFK-J6/C3 chimera carrying the Jc1 virus sequence (42). The construct Venus-Jc1 is described elsewhere (30) and represents a bicistronic reporter virus genome carrying the coding region of Venus-GFP in the first cistron and the chimeric Jc1 open reading frame in the second cistron, expressed by means of the EMCV IRES. Plasmid pFKi389neoNS3-3′_dg_JFH-1_NS5Aaa2359_RFP was used to generate sg/neo/5A-RFP and constructed by replacing the firefly luciferase gene in pFKi389LucNS3-3′_dg_JFH-1_NS5Aaa2359_RFP with the gene encoding neomycin phosphotransferase, using AscI and NotI restriction sites flanking the luciferase gene. All PCR-amplified DNA fragments were analyzed by automated nucleotide sequencing with an ABI 310 sequencer (Applied Biosystems). Big Dye Terminator, version 1.1 (Applied Biosystems), was used for cycle sequencing according to the manufacturer's protocol. The HCV E1E2 expression construct phCMVΔcE1/E2-Con1 was derived from the phCMV-7a expression vector (5) by replacing the DNA fragment encoding the last 60 residues of HCV core and all of E1 and E2 of the genotype 1a H77 isolate with the corresponding fragment of the HCV genotype 1b isolate Con1 (GenBank accession number AJ238799) (35). The HCV E1E2 expression constructs pcDNA3ΔcE1E2-JFH1 and pcDNA3ΔcE1E2-J6CF were constructed in an analogous way, and they direct the expression of envelope glycoproteins of the HCV isolates JFH-1 (28) and J6CF (52), which both belong to genotype 2a.

In vitro transcription, electroporation of HCV RNAs, and Northern blot analysis.

In vitro transcripts of the individual constructs were generated by linearizing the respective plasmids by digestion with MluI. In vitro transcription and electroporation of HCV RNAs were performed as described previously (29). For generation of capped transcripts encoding Renilla luciferase, the same protocol was applied except that the nucleotide mix contained 1 mM of the cap analog m7G(5′)ppp(5′)G (New England Biolabs), 1.25 mM GTP, a 2.5 mM concentration (each) of ATP, CTP, and UTP, and pRL CMV (Promega) as a template after linearization with BamHI. For Northern hybridization, total RNA was prepared by a single-step isolation method (10). After gel electrophoresis and membrane blotting, the membrane was incubated with a ³²P-labeled negative-sense riboprobe complementary to nucleotides 6273 to 9678 of the JFH-1 genome as described previously (29). Hybridization with a β-actin-specific antisense riboprobe served to correct for the total RNA amount loaded in each lane of the gel. Signals were detected and quantified by autoradiography using Kodak photofilms or phosphorimaging.

Immunofluorescence analysis.

Cells were seeded onto glass coverslips that were placed in 24-well cell culture plates. At different time points after transfection or 72 h after infection, cells were washed three times with phosphate-buffered saline (PBS) and fixed with 500 μl of 4% paraformaldehyde for 10 min at room temperature (RT). Subsequently, cells were washed three times with PBS, permeabilized by a 5-min incubation in 500 μl of 0.5% Triton X-100 in PBS, and washed three times with PBS prior to incubation with the first antibody. The JFH-1 NS5A-specific antibody 9E10 (33), kindly provided by T. Tellinghuisen and C. M. Rice, Rockefeller University, NY, was diluted 1:200 in PBS containing 5% goat serum. After 45 min of incubation at RT, cells were washed three times with PBS for 10 min each and incubated with a 1:1,000 dilution of the Alexa 488-conjugated secondary antibody. After 45 min of incubation in the dark, cells were washed once with PBS, incubated for 1 min with a 1:5,000 diluted 4′,6′-diamidino-2-phenylindole dihydrochloride (DAPI)-PBS solution, and immediately washed three times for 10 min each with PBS. Finally, cells were washed with water and mounted on glass slides with Fluoromount G (Southern Biotechnology Associates, Birmingham, AL). For detection of dengue virus 2 (DEN-2) infection, an NS5-monospecific polyclonal antiserum raised in rabbits was used as described recently (36). Bound antibodies were detected by using a goat anti-rabbit Alexa 488-conjugated secondary antibody (Molecular Probes, Leiden, The Netherlands).

FACS analysis.

Huh7-Lunet cells were transfected with 10 μg in vitro transcript of Jc1/GFP, Venus-Jc1, JFH1ΔE/GFP, or Jc1, seeded into six-well culture dishes, and detached by trypsin treatment at 24, 48, and 72 h posttransfection. Infected cells were grown in 12-well culture dishes and analyzed at 72 h postinfection. Cells were resuspended in complete DMEM and centrifuged for 5 min at 4°C at 110 × g in a tabletop centrifuge. The cell pellet was resuspended in 3% paraformaldehyde and stored at 4°C until use. Immediately before analysis, cells were centrifuged for 5 min at 4°C at 110 × g, the paraformaldehyde solution was aspirated, and cells were resuspended in PBS supplemented with 0.2% bovine serum albumin and 0.02% sodium azide (fluorescence-activated cell scanning [FACS] assay buffer) and analyzed directly by using a FACSscan apparatus and Cell Quest Pro software (Becton Dickinson Biosciences). In case of simultaneous determination of cell surface expression of HCV receptor CD81 or scavenger receptor BI (SR-BI) with GFP-derived autofluorescence, detached cells were stained for 1 h on ice with a CD81-specific (115-136-071; Dianova, Hamburg, Germany) or SR-BI-specific (anti-CLA1; Becton Dickinson Biosciences) monoclonal antibody diluted 1:200 or 1:50, respectively, in FACS assay buffer. Subsequently, cells were washed with PBS, and bound antibodies were detected by a 1-h incubation on ice with mouse-specific secondary antibodies conjugated to allophycocyanin (Dianova, Hamburg, Germany) at a dilution of 1:100 in FACS assay buffer. Cells were analyzed by using a FACSCalibur apparatus and Cell Quest Pro software (both from Becton Dickinson Biosciences).

Luciferase assays.

Transient HCV RNA replication assays were performed as described previously (31). In brief, 4 × 10⁶ cells were transfected with 5 to 10 μg in vitro-transcribed RNA and resuspended in 12 ml or 20 ml culture medium (in cases of subgenomic or genomic RNA, respectively), and 2-ml aliquots were seeded into each well of a six-well culture dish. At 4 to 72 h posttransfection, cells were lysed and luciferase activity was determined. Renilla luciferase activity was measured in the same lysates, using an assay buffer containing 15 mM potassium phosphate (pH 7.8), 25 mM glycyl-glycine (pH 7.8), 15 mM MgSO₄, 4 mM EGTA, and 50 μg/ml coelenterazine (PFK Chemikalien, Kleinblittersdorf, Germany).

Core ELISA.

To quantify the amount of HCV core protein in transfected cells or cell culture supernatant, an Ortho Trak-C enzyme-linked immunosorbent assay (ELISA) kit was used (Ortho Clinical Diagnostics, Neckargemünd, Germany). Four million cells were transfected with 5 to 10 μg in vitro-transcribed RNA and resuspended in 20 ml culture medium, and 2-ml aliquots were seeded into each well of a six-well culture dish. Cell lysates and cell culture supernatants were analyzed at different time points between 4 h and 72 h posttransfection. Culture supernatants were used directly for core ELISA, whereas for determination of intracellular core amounts, cells were lysed by the addition of 500 μl PBS containing 1% Triton X-100, a 1/10,000 volume of aprotinin (1 U/ml), a 1/1,000 volume of leupeptin (4 mg/ml), and a 1/100 volume of phenylmethylsulfonyl fluoride (100 mM). After 5 min of centrifugation at 18,000 × g, supernatants were processed for ELISA according to the manufacturer's protocol. Colorimetric measurements were performed using a Sunrise colorimeter (Tecan Trading AG, Switzerland). To determine the efficiency of core protein release, the ratio of total core protein (the sum of intra- and extracellular core amounts) to extracellular core protein was determined.

TCID₅₀ determination.

Determination of 50% tissue culture infective dose (TCID₅₀) values was performed as described recently, with slight modifications (33). Huh7.5 target cells were seeded at a concentration of 10⁴ cells per well of a 96-well plate in a total volume of 180 μl complete DMEM. Twenty-four hours later, serial dilutions of virus-containing supernatant were added to eight wells per dilution. Virus-containing supernatant was replaced at 4 h postinoculation by fresh medium, and 72 h later, cells were washed with PBS and fixed by the addition of 100 μl ice-cold methanol and overnight incubation at −20°C. Cells were washed three times with PBS, permeabilized, and blocked overnight by the addition of 40 μl blocking solution (1× PBS, 0.5% saponin, 1% bovine serum albumin, 0.2% dried milk, 0.02% NaN₃). Cells were then incubated with PBS-0.3% H₂O₂ (vol/vol) for 5 min at RT. After three washes with PBS and one wash with PBS containing 0.5% saponin, infected cells were detected by using the NS5A antibody 9E10 (1:200 dilution in PBS-0.5% saponin) and a goat anti-mouse secondary antibody coupled to horseradish peroxidase (A4416; Sigma). After a 1-h incubation at RT, the cells were washed three times with PBS and once with PBS-0.5% saponin. Horseradish peroxidase activity was detected by using a Vector Nova Red substrate kit (Linaris, Weinheim, Germany). Reactions were terminated by replacing the substrate solution with water. Plates were analyzed by microscopy, and virus titers were determined by using the calculation method of Spearman and Kärber (26, 45).

Generation of HCV pseudoparticles (HCVpp) and infection studies.

Human immunodeficiency virus (HIV)-based pseudotypes bearing vesicular stomatitis virus glycoproteins (VSV-G) or HCV Con1-, JFH-1-, or J6CF-derived E1 and E2 proteins were generated by cotransfection of 293T cells (14). Briefly, 2.5 × 10⁶ 293T cells were seeded into 10-cm-diameter plates 1 day before transfection with 2.7 μg envelope protein expression construct pczVSV-G (24), phCMVΔcE1-E2-Con1, pcDNA3ΔcE1E2-JFH1, or pcDNA3ΔcE1E2-J6C, 8.1 μg HIV Gag-Pol expression construct pCMVΔR8.74 (15), and 8.1 μg of a firefly luciferase transducing retroviral vector (29, 55) by using Lipofectamine 2000 (Invitrogen, Karslruhe, Germany). The medium was replaced 8 h after transfection, and supernatants containing the pseudoparticles were harvested 48 h later. Transfected cells were supplemented with fresh culture medium, and virus-containing cell culture fluid was harvested again 24 h later (at 72 h posttransfection). Both harvests were pooled, cleared by passage through 0.45-μm-pore-size filters, and concentrated by using Amicon ultrafiltration devices (Millipore). Concentrated (20-fold) pseudotype virus stocks were utilized to infect naïve Huh7.5 cells or Huh7.5 cells that had been infected with Jc1 virus 4 days earlier, using a multiplicity of infection (MOI) of 0.8. The efficiency of pseudotype virus infection was evaluated by luciferase assays at 72 h postinfection. HIV-based retroviral constructs were kindly provided by D. Trono, Swiss Institute of Technology, Lausanne, Switzerland.

DEN-2 and HCV coinfection analysis.

Huh7.5 cells (1 × 10⁵) were seeded onto glass coverslips and infected 24 h later with Jc1 virus at an MOI of 10. Cells were grown for 24 h and infected for 4 h with DEN-2 (New Guinea C strain) at an MOI of 2 as described recently (36). After 4 h, the inoculum was replaced by DMEM, and cells were incubated for another 24 h. Finally, cells were analyzed for HCV and DEN-2 infection by NS5A- and NS5-specific immunofluorescence, respectively.

RESULTS

Replication competence of fluorochrome gene-tagged HCV genomes.

Detailed studies of simultaneous or successive infections by two different HCV variants require a method that allows discrimination between the two viral genomes. Based on previous studies by us and others showing that heterologous sequences can be inserted into the NS5A coding region without gross effects on RNA replication of subgenomic replicons (2, 38), in this study we constructed HCV genomes that carry a GFP or RFP reporter gene inserted in frame into the NS5A coding sequence. These genomes should allow easy detection of infected cells by fluorescence microscopy of living and fixed cells and discrimination between two viral genomes in the same cell. To this end, GFP or RFP was inserted into domain 3 of the NS5A coding sequence at two sites corresponding to positions described by Moradpour and colleagues for the Con1 isolate (amino acid positions 378 and 383 of NS5A) (38). RNA replication was first analyzed in the context of a subgenomic luciferase reporter replicon and found to be unaffected by either of the insertions (data not shown). We then arbitrarily selected the insertion at amino acid position 383 of NS5A and introduced the modified NS5A protein into the chimeric full-length HCV genome Jc1 (Fig. 1B), which was recently shown to produce a high titer of infectious virus (42). Since efficient RNA replication is a prerequisite for efficient particle production, we first performed a comparative replication analysis of Jc1/GFP and Jc1/RFP with the wild-type Jc1 genome. JFH1ΔE/GFP, lacking most of the envelope protein coding region, served as a negative control for particle production, and JFH1/GND, carrying an inactivated NS5B polymerase, served as a negative control for RNA replication. In addition, we used Venus-Jc1, a bicistronic full-length genome expressing the Venus variant of GFP (39) from the first cistron and the chimeric Jc1 open reading frame with unmodified NS5A from the second cistron (Fig. 1B) (30). This bicistronic construct design has been used in the past both for replicons and for a JFH-1 reporter virus (49) and therefore was used for a comparative analysis with the monocistronic constructs in order to determine the advantages and disadvantages of both systems.

In a first experiment, RNA replication in transfected cells was monitored by Northern hybridization (Fig. 1C). Replication of the monocistronic genomes Jc1/GFP and Jc1/RFP and of the JFH1ΔE/GFP control was comparable to that of the Jc1 wild type, arguing that the insertion into NS5A had no influence on RNA replication capacity and kinetics. In contrast, we observed a clear delay in replication kinetics of the bicistronic genome Venus-Jc1. While replication of all monocistronic genomes had already increased significantly within the first 24 h posttransfection, the RNA amounts for the bicistronic genome Venus-Jc1 dropped below the input level during this time. Comparable results were obtained when transfected cells were examined by FACS analysis (data not shown). Moreover, we tested the analogous series of GFP constructs in the context of the authentic JFH-1 genome and found that replication of the monocistronic reporter genome JFH1/GFP was comparable to that of the JFH-1 wild type, while replication of the bicistronic Venus-JFH1 construct was lower and kinetics were delayed, similar to those of the Jc1 constructs (data not shown). Thus, the in-frame insertion into NS5A did not interfere with efficient RNA replication.

Production of infectious HCV particles from cells carrying fluorochrome gene-tagged HCV genomes.

To analyze particle production from cells transfected with the different reporter genomes, we measured the amounts of core protein released into cell culture supernatants in relation to total core protein amounts expressed in these cells. Supernatants and cells were harvested at 4, 24, 48, 72, and 96 h posttransfection and used for core-specific ELISA. As shown in Fig. 2A, comparable amounts of core protein were detected in cells 4 h after transfection, reflecting very similar transfection efficiencies. While core protein accumulated to comparable levels in cells transfected with the monocistronic constructs Jc1/GFP, Jc1/RFP, and JFH1ΔE/GFP, core amounts obtained with the bicistronic Venus-Jc1 genome were reduced nearly 10-fold at 24 h posttransfection, most likely due to the delayed RNA replication kinetics. In contrast, the reduced intracellular core amounts in Jc1-transfected cells were probably the result of a very efficient core release already at 24 h posttransfection (Fig. 2B). Surprisingly, despite efficient RNA replication and intracellular core protein production, the relative amount of extracellular core protein released from cells that had been transfected with the monocistronic Jc1/GFP or Jc1/RFP genome was about 10-fold lower than that for the wild type (Fig. 2A, lower panel). In essence, in the case of Jc1, about 10% of total core protein was released into the supernatant 24 h after transfection (Fig. 2B). Core release achieved with the bicistronic Venus-Jc1 genome was up to 5% of the total core protein, while for the monocistronic genome <1% of the core protein was released from transfected cells. These results show that both the bicistronic and the monocistronic reporter genomes are impaired in virus production, but the defect is much more pronounced in the latter.

FIG. 2.

Virus release from transfected Huh7-Lunet cells, as determined by core-specific ELISA. (A) After transfection of in vitro transcripts of the HCV genomes specified at the top, cells were lysed at the time points given, and core protein amounts accumulated in the cells were determined (upper panel). Cell culture supernatants harvested from these cells were analyzed by core ELISA in parallel (lower panel). (B) Efficiency of core protein release from cells transfected with the monocistronic genome Jc1/GFP, the bicistronic genome Venus-Jc1, the parental wild type (Jc1), or the envelope gene deletion mutant (JFH1ΔE/GFP). The percentage of released core protein in relation to total core protein (the sum of intra- and extracellular core protein) was calculated for each time point. Core release from cells transfected with monocistronic genomes carrying the insertion in NS5A was strongly reduced. Note that core release obtained with the ΔE mutant was below the detection limit and therefore marked with an asterisk. Data from a representative experiment of three independent repetitions are shown.

In the next step, infectivity of the released particles and infectivity titers were determined. To this end, Huh7.5 cells were infected with culture supernatants of cells harvested at different time points after transfection. Figure 3A shows that infectious virus titers, as determined by TCID₅₀ assay, were reduced about 50-fold in the case of the monocistronic reporter genome Jc1/GFP compared to the parental Jc1 genome. Jc1/RFP reached the same titers as Jc1/GFP (data not shown). In the case of the bicistronic Venus-Jc1 genome, we observed an increase of virus titer between 24 h and 48 h, correlating both with the increase of the amount of core protein released into the cell culture supernatant (Fig. 2A) and with RNA replication (Fig. 1C).

FIG. 3.

Infectivity of virus particles generated with chimeric genomes that express GFP. (A) Titers of infectious virus contained in supernatants of Huh7-Lunet cells at 24 h and 48 h posttransfection. Infectivity titers were determined by TCID₅₀ assay. Mean values and error ranges for duplicates are shown. (B) Detection of GFP by autofluorescence in cells 48 h after transfection with viral genomes (upper row) or 72 h after infection (lower row) with the viruses specified at the top. In the case of Jc1, NS5A was detected by immunofluorescence (IF). Note the endoplasmic reticulum-like staining in the case of Jc1 and the monocistronic Jc1/GFP genome, in contrast to the largely nuclear Venus-GFP staining in the case of the bicistronic genome Venus-Jc1. Also note the divergent infectivity titers of the virus chimeras, as evidenced by the number of cells displaying green fluorescence (autofluorescence or NS5A-specific staining; lower row). (C) Detection of intracellular GFP expression by FACS analysis of infected cells. Naive Huh7.5 cells were inoculated with culture supernatant harvested from transfected Huh7-Lunet cells at 24 h (upper row) or 48 h (lower row) posttransfection, and 72 h after infection of Huh7.5 cells, they were fixed and GFP expression was determined by FACS analysis. HCV constructs used for transfection of Huh7-Lunet cells are specified at the top. The number in the lower right corner of each diagram refers to the percentage of cells expressing GFP. Note that Jc1 does not express GFP and therefore determines the background of the FACS assay.

Cells infected with either reporter virus were also readily detected 72 h after infection by fluorescence microscopy (Fig. 3B). The differences in the subcellular localization of GFP are due to the different expression modes. While in the case of Jc1/GFP the fluorescent protein is fused to NS5A and therefore localizes in cytoplasmic dot-like structures supposed to be the membranous web (19), in the case of Venus-Jc1 the fluorescent protein is fused to the 16 N-terminal residues of HCV core and localizes predominantly to the nucleus. Finally, infected cells were also detectable by FACS analysis, with the number of GFP-autofluorescing cells correlating well with viral titers (Fig. 3C). A steady increase in infected cells corresponding to an increase of virus titer between 24 h and 48 h posttransfection was observed with the bicistronic Venus-Jc1 genome (Fig. 3C, second column), whereas the number of GFP-positive cells obtained after infection with the corresponding supernatants from Jc1/GFP-transfected cells remained at the same level (left column). These results are in line with the data shown in Fig. 2 demonstrating that monocistronic reporter genomes are impaired in core release and show no increase in core release over time. In summary, we were able to generate fluorochrome gene-tagged HCV genomes that replicate to wild-type levels and allow production of infectious progeny virus. However, insertion of the reporter gene into the NS5A coding sequence reduced viral titers about 50-fold, arguing for an as yet unknown role for NS5A in virus particle production or release.

Induction of superinfection exclusion by primary HCV infection.

Having characterized the reporter gene-tagged viral genomes, we utilized them to find out whether primary HCV infection induces superinfection exclusion as described for several other viruses, including the closely related pestiviruses (32, 37). Huh7.5 cells were inoculated simultaneously with the monocistronic Jc1/GFP and Jc1/RFP viruses, with each at a multiplicity of infection (MOI) of 10 TCID₅₀/cell. In parallel, we first infected cells only with Jc1/GFP and then, after 24 h or 48 h, performed secondary infection with Jc1/RFP (Fig. 4A). Seventy-two hours after the last infection, cells were fixed, and the number of GFP- or RFP-positive cells was determined by fluorescence microscopy. As shown in Fig. 4A and B, when cells were infected with Jc1/GFP and Jc1/RFP simultaneously, approximately 18% of all infected cells expressed both GFP and RFP, indicating that they had been infected by both viruses. Interestingly, in doubly infected cells, one virus species appeared to dominate over the other, as inferred from the inverse correlation of the intensities of GFP and RFP fluorescence. Very rarely, cells were detected in which GFP and RFP were expressed to the same level. In contrast, when Jc1/RFP infection was performed 24 h after primary infection with Jc1/GFP, only 2% of all infected cells displayed dual fluorescence, although >80% of the total cell population became infected, with 57% being infected with only Jc1/GFP and 41% being infected with only Jc1/RFP. After a 48-h primary infection with Jc1/GFP, the number of GFP-positive cells increased further, concomitant with a further reduction of RFP-positive cells and a constantly very small number of double-positive cells. These results suggest that primary HCV infection with Jc1/GFP induces a stringent level of superinfection exclusion within 24 h. The finding that even 48 h after primary infection a small number of cells can still be superinfected suggests that some HCV-infected cells may not be entirely refractory to a secondary infection or that these cells had been infected only poorly with the primary virus. Another possibility may be that double-positive cells are the result of a delayed (secondary) infection caused by progeny virus released from primary infected cells. In this case, secondary infection by Jc1/RFP occurred prior to the establishment of superinfection exclusion. However, this possibility is rather unlikely due to the high MOI chosen for primary infection with Jc1/GFP.

FIG. 4.

Evidence for homologous but not heterologous superinfection exclusion. (A) Schematic of the experimental procedure. Cells were either doubly infected with Jc1/GFP and Jc1/RFP (I) or first infected with Jc1/GFP and 24 h or 48 h later challenged with Jc1/RFP (II and III, respectively). The lower panel shows a quantification of cells infected with only Jc1/GFP (green portion of each column), only Jc1/RFP (red portion of each column), or both viruses (yellow portion of each column). Values written in the columns refer to the percentages of all infected cells displaying either red, green, or both fluorescent signals. The total number of infected cells was set to 100%. The proportion of the total cell population that was HCV infected is given by the height of the respective column. Values refer to one of three independent experiments. In each experiment, 250 cells were counted for each time point. (B) Representative views of infected cells examined by fluorescence microscopy. Cells were analyzed for GFP or RFP fluorescence; nuclei were stained with DAPI (blue). White arrows refer to cells infected with both viruses. Note that after 24 h, doubly infected cells were observed only sporadically. (C) Superinfection exclusion is also observed after primary infection with the Jc1 wild type. The schematic displayed in the upper panel is analogous to that in panel A, except that Jc1 wild-type virus, which produces much higher titers of infectious virus than the reporter viruses and spreads rapidly in cell culture, was used for primary infection. The quantification of cells infected with Jc1/GFP after simultaneous infection with Jc1 (I) or 24 h or 48 h after primary infection with Jc1 (II and III, respectively) is shown in the lower panel. Values were obtained after counting 800 cells for each time point. (D) Analysis of cells by immunofluorescence (in the case of Jc1, which does not contain a reporter gene) or by autofluorescence (in the case of Jc1/GFP). The top row shows cells infected only with Jc1/GFP (left panel) or only with Jc1 (middle panel). Mock-infected cells are shown in the right panel. Nuclei were stained with DAPI. The panels below show representative sections of images in which Jc1/GFP (left lane) or Jc1 (middle lane) was detected. White arrows indicate cells infected with both viruses. Note the decreasing number of Jc1/GFP-infected cells already 24 h after primary infection with Jc1 (II). (E) No block of DEN-2 by prior HCV infection. Huh7.5 cells were either infected with HCV (Jc1) or mock treated, and 24 h later, cells were either inoculated with DEN-2 or mock treated. Twenty-four hours after DEN-2 infection, cells were analyzed for HCV and DEN-2 antigens by NS5A (red) and NS5 (green) immunofluorescence, respectively. The upper panels (from left to right) show only DEN-2 infection, only HCV infection, HCV and DEN-2 coinfection, and mock-infected cells. Note that the number of DEN-2-infected cells between the first and the third panel is unchanged irrespective of the HCV infection. The lower panels represent DAPI staining of the nuclei (blue).

To rigorously assess if primary infected cells are partially or completely resistant to a secondary round of infection by HCV, we performed simultaneous or sequential infections with the Jc1 chimera, our most aggressive virus that produces the highest infectivity titers in tissue culture, and the Jc1/GFP reporter virus (Fig. 4C). In order to establish the most stringent conditions for secondary infection, we utilized a Jc1 preparation with a titer approximately 10-fold higher than that of the Jc1/GFP reporter virus. As expected, when cells were infected simultaneously with both viruses, a large number of infected cells (25%) were dually infected. The higher proportion of Jc1-infected cells than Jc1/GFP reporter virus-infected cells most likely reflects the higher titer of the Jc1 inoculum. Importantly, in line with the previous experiment, the number of doubly infected cells decreased dramatically when the Jc1/GFP reporter virus was administered 24 h or 48 h after primary infection with the Jc1 chimera (Fig. 4D). Nevertheless, a small proportion of Jc1-infected cells were still susceptible to secondary infection, indicating that superinfection exclusion is not absolute.

Having shown that secondary infection of Huh7 cells with a homologous virus (HCV) was blocked, we wanted to exclude that this phenomenon was due to the induction of a general antiviral state as a result of primary infection. Therefore, Jc1-infected cells were subsequently infected with DEN-2. As shown in Fig. 4E, HCV-infected cells remained fully susceptible to infection with DEN-2, arguing for homologous interference being responsible for HCV superinfection exclusion rather than a general antiviral state.

No evidence for down regulation of CD81 and SR-BI on the surfaces of HCV-infected cells.

Superinfection exclusion can be due to interference at the level of virus entry or RNA translation/replication. Superinfection exclusion can also occur at the level of virus assembly or release under conditions of nonsaturating primary infection where virus spread is still possible. To analyze whether HCV infection results in a down regulation of known molecules involved in HCV entry, we performed FACS analysis of Huh7.5 cells that had been infected with Venus-Jc1 for 72 h. We focused our analysis on CD81 and SR-BI, two molecules that are implicated in HCV entry. As presented in Fig. 5A, Venus-Jc1 productively infected about 26 to 30% of the total cell population. Importantly, the CD81- and SR-BI-specific mean fluorescence intensities in Venus-Jc1-infected cells did not significantly deviate from those in uninfected cells, indicating that for up to 72 h postinoculation, HCV infection did not affect CD81 and SR-BI densities on the cell surface. The same results were found with infection times of up to 120 h (not shown) and with higher MOIs (30).

FIG. 5.

HCV-infected cells express normal levels of CD81 and SR-B1 on the cell surface and are fully permissive for HCVpp superinfection. (A) Huh7.5 cells were infected with Venus-Jc1 (lower row) or were left untreated (upper row) and then were used for cell surface staining of CD81 (middle column) or SR-BI (right column). Cells incubated with only the secondary antibody (anti-mouse-allophycocyanin; left column) were used to set the background. Venus-GFP expression in infected cells and cell surface staining were detected by FACS analysis. Note that the amounts of cell surface-expressed CD81 and SR-BI were unchanged in HCV-infected cells. (B) Naive Huh7.5 cells and cells previously infected with Jc1 (Jc1 inf) were inoculated with pseudoparticles carrying an HCV envelope from Con-1, JFH-1, or J6CF (HCVpp) or the VSV-G envelope (VSV-Gpp) or with lentiviral particles produced in the absence of a viral envelope glycoprotein (no env). Cells were harvested 72 h after secondary infection and analyzed by assessing luciferase activity expressed from the transducing lentiviral vector. Mean values for duplicate wells, with standard deviations, are shown. The panel on the right displays Jc1-infected and naive Huh7.5 target cells that were fixed and analyzed for NS5A expression by indirect immunofluorescence.

Primary HCV infection does not interfere with entry during secondary infection.

Although primary HCV infection did not affect the level of CD81 and SR-BI expression on the cell surface, we could not exclude a block of virus entry at some other step. Moreover, neither CD81 nor SR-BI is sufficient for HCV entry, arguing that some other as yet unknown factors are required (6). We therefore performed entry studies by using HCVpp. They are based on lentiviral vectors that carry functional HCV glycoproteins on their surfaces and also carry a luciferase reporter gene. Successful cell entry can therefore be monitored by luciferase expression in the target cell. For comparative purposes, we used HCVpp carrying envelope glycoproteins of the HCV isolate Con1 (genotype 1b) and the HCV genotype 2a isolates JFH-1 and J6CF. It should be noted that the last isolate was also used for the construction of the Jc1 chimera (Fig. 1B). Huh7.5 cells were infected with wild-type Jc1 virus at an MOI of 0.8 TCID₅₀/cell and challenged with HCVpp after 120 h. At this time point, >95% of the Huh7.5 cell population inoculated with Jc1 was infected, as evidenced by NS5A-specific immunofluorescence (Fig. 5B, right panels). For comparison, noninfected Huh7 cells were used for HCVpp challenge in parallel. As shown in Fig. 5B (left panel), luciferase activities were comparable between noninfected cells and Huh-7.5 cells previously infected with Jc1. Moreover, both cell populations were readily infected with pseudoparticles carrying VSV-G, used as a control. For reasons we do not know, in all cases a prior infection with Jc1 rendered Huh7.5 cells even more susceptible to infection by all the different pseudoparticles. In summary, these results show that HCV-induced superinfection exclusion is not due to an entry block.

Superinfection exclusion is induced primarily at the level of RNA translation/replication.

To study whether superinfection exclusion was due to a block at the level of RNA translation/replication, we transfected comparable amounts of a subgenomic reporter replicon containing the firefly luciferase gene in the first cistron and the GFP-tagged NS5A replicase gene in the second cistron into naive Huh7-Lunet cells or cells containing a subgenomic neo/5A-RFP replicon (Fig. 6A). The latter represents a cell pool obtained after transfection of Huh7-Lunet cells and G418 selection. In addition, cured subgenomic neo/5A-RFP replicon cells lacking detectable amounts of HCV (Fig. 6C) were used for comparison. They were generated by passaging of the cells in the presence of a selective inhibitor and served as a control to exclude nonspecific interference with the cells carrying the replicon. Transfected cells were harvested for a luciferase assay to monitor RNA replication (Fig. 6B) or fixed for fluorescence microscopy to study protein expression at the single-cell level (Fig. 6C). Luciferase activity detected 4 h after transfection was set to 1 relative light unit (Fig. 6B). At 24 h posttransfection and all time points thereafter, a 10-fold reduction of luciferase activity was observed with replicon cells compared to that for naive or cured cells, indicating that prior presence of the replicon interfered with replication of the incoming RNA.

FIG. 6.

Superinfection exclusion is primarily due to interference at the level of RNA translation/replication. (A) Schematic diagrams of the bicistronic replicons used in this experiment. The insertions of the RFP and GFP genes into NS5A are indicated by colored boxes, the neo and firefly luciferase (Luc) genes are indicated by open boxes, and the EMCV IRES is indicated by a small dark gray box. (B) Naive Huh7-Lunet cells, a Huh7-Lunet cell pool carrying a stably replicating sg/neo/5A-RFP replicon, or cured sg/neo/5A-RFP replicon cells that no longer express detectable amounts of HCV proteins were transfected with the luciferase replicon, and luciferase activity was measured at the given time points posttransfection. The luciferase activity determined 4 h after transfection was set to 1. Error bars represent standard errors of the means for eight measurements from four independent wells in two independent experiments. RLU, relative light units. (C) Huh7-Lunet cells carrying the sg/neo/5A-RFP replicon, cured sg/neo/5A-RFP cells, and Huh7-Lunet cells were transfected with the sg/luc/5A-GFP replicon and analyzed by fluorescence microscopy at 48 h posttransfection. Results obtained with subgenomic replicon cells are shown in the upper row, with naive Huh7-Lunet cells shown in the bottom row and cured replicon cells shown in the middle row. Note that the number of GFP-expressing cells was much higher in the case of naive Huh7-Lunet cells and cured replicon cells than in neo replicon-containing cells (60% versus 5% in the experiment shown).

Analogous results were obtained when GFP and RFP expression was monitored by fluorescence microscopy. About 60% of GFP-expressing cells were detected 72 h after transfection of naive and cured Huh7-Lunet cells, in contrast to only about 5% in the case of the replicon cells (Fig. 6C). Moreover, for most cells, GFP signals were stronger in naive and cured Huh7-Lunet cells than in replicon cells. Interestingly, after transfection of the latter, GFP-positive cells often showed rather weak RFP signals compared to those in nontransfected replicon cells, arguing for a competition between both RNA species at the level of RNA translation or replication.

To discriminate whether interference occurred at the level of RNA translation or RNA replication, we cotransfected a replication-deficient subgenomic RNA, designated sg/luc/ΔGDD, and a capped Renilla luciferase RNA into subgenomic neo/5A-RFP replicon cells and the corresponding cured cells (Fig. 7). Cells were harvested 2, 4, 6, and 9.5 h after transfection, and luciferase activity was determined. Renilla luciferase activities achieved by cap-dependent translation were similar for both cell pools. In contrast, HCV IRES-dependent translation measured by firefly luciferase activity was reduced nearly twofold in replicon-containing cells compared to that in cured cells. These results raise evidence that part of the superinfection exclusion occurs at the level of RNA translation. In summary, these results suggest that superinfection exclusion is due, at least in part, to interference at the level of RNA translation, but given the moderate reduction, it is likely that exclusion also occurs at the level of RNA replication.

FIG. 7.

Evidence for interference at the level of RNA translation. (A) Structures of transfected HCV RNA sg/luc/JFH1/ΔGDD and the capped Renilla luciferase-encoding RNA. (B) A pool of Huh7-Lunet cells carrying autonomously replicating sg/neo/5A-RFP replicons (rep) or a cured pool of these cells (cured) were cotransfected with the replication-deficient subgenomic firefly luciferase replicon (Sg/luc) and the capped Renilla luciferase RNA (Rluc). Cells were harvested at the indicated time points, and firefly luciferase and Renilla luciferase activities were measured in cell lysates. Error bars represent standard errors of the means for eight measurements from four independent wells in two independent experiments.

DISCUSSION

For this study, we developed monocistronic reporter virus genomes that permit the characterization of HCV RNA replication upon infection of living cells. Moreover, by using different fluorescent proteins, we could establish a way to easily discriminate between infecting viruses. This method allowed us to study coreplication of HCV genomes upon simultaneous infection, as well as superinfection exclusion.

In agreement with an earlier report (29), we found that the RNA replication kinetics of the bicistronic reporter virus genome (Venus-Jc1) was delayed 24 h compared to that of the wild type, concomitant with a delayed release of infectious virus particles. Interestingly, RNA replication of the monocistronic reporter genomes was unaffected, but release of infectious virus particles was significantly reduced. In fact, we observed an about 50-fold lower infectivity titer than that of the Jc1 wild type, which most likely was due to an impairment of virus particle production rather than an impairment of virus infectivity. Currently, it is not clear whether particle assembly or particle release is affected. Alternatively, the insertion into domain 3 of NS5A may indirectly affect the functions of other viral or cellular proteins responsible for virus particle production. Further studies are required to identify a possible role for NS5A in virus production.

One possibility for achieving higher virus titers while retaining the reporter gene is to use bicistronic genomes. However, there are several disadvantages inherent to this construct design. First, presumably due to their increased length or the presence of an internal highly structured RNA that may slow down RNA synthesis, replication of these genomes is significantly delayed compared to that of monocistronic genomes. Second, during prolonged passage we observed that expression of the GFP reporter gene was rapidly lost, most likely due to deletion of all or most of the heterologous sequences. Since the mutants have much faster replication kinetics, they rapidly become the predominant virus species. In contrast, replication of the monocistronic reporter viruses is similar to that of the Jc1 wild type. Moreover, deletion mutants in which most or all of the heterologous sequences have been removed are only viable in case of an in-frame deletion, which is less likely in monocistronic reporter viruses than in the bicistronic virus genome, where any deletion removing the heterologous sequence should be viable. Third, in the case of the bicistronic genome, translation of the HCV polyprotein is under the control of the EMCV IRES, which may be undesirable in certain cases because any interference with the activity of this IRES impairs RNA replication and thereby mimics an antiviral effect.

One obvious application of the monocistronic GFP or RFP genomes is in the study of the dynamics of viral replication in living cells. Although subgenomic replicons carrying in-frame GFP insertions in analogous positions of NS5A have been described recently (2, 38), the advantage of this novel system is the possibility to characterize the dynamics of viral replication, and eventually the biogenesis of the HCV replication complex, upon (synchronized) infection. Moreover, upon coinfection of cells with viruses directing the expression of different fluorochrome reporter-NS5A fusion proteins, it is possible to study whether replication complexes derived from individual genomes remain separated from each other or intermingle. In fact, preliminary data obtained with this novel system suggest that mixed replication complexes do form, at least to some extent, but further studies are required to fully understand their biogenesis.

Taking advantage of the novel reporter virus system, we showed that cells previously infected with HCV are largely resistant to secondary HCV infection but not to infection with a heterologous virus, as exemplified for DEN-2. This superinfection exclusion is not due to a block at the level of virus entry, at least up to 120 h post-primary infection, which is different from the superinfection exclusion observed for several other viruses. For instance, in the case of HIV, the immature envelope glycoprotein gp160 forms a complex with CD4 (12, 23). Presumably due to poor intracellular transport of gp160, this complex is retained in the endoplasmic reticulum and contributes to down regulation of the primary CD4 receptor on the surfaces of infected T cells. In addition, the viral factors Vpu and Nef contribute substantially to CD4 receptor down regulation, although Nef mediates superinfection exclusion primarily by some other CD4-independent manner (see reference 51 and references cited therein). In the case of the pestivirus bovine viral diarrhea virus, superinfection exclusion appears to be mediated at two levels, namely, virus entry and RNA replication (32). While the entry block requires expression of the envelope glycoprotein E2, inhibition of replication of the incoming RNA is independent from the structural proteins. The underlying mechanisms are not yet understood.

While we did not observe a block of virus entry for up to 5 days after primary HCV infection, we cannot exclude the possibility that an entry block develops at later time points. We note that Zhong and coworkers found that during prolonged passage of cells infected with a cell culture-adapted JFH-1 variant, cells become largely resistant to HCV infection (54). This resistance is due to selection for cells that have small amounts of CD81 on their surfaces or that support HCV RNA replication with only a low efficiency. Since JFH-1 exerts a cytopathic effect (53), prolonged culture of infected cells most likely results in selection for those cells that are largely resistant to JFH-1 infection, either by down regulation of important entry molecules or by providing a hostile environment for RNA replication. These cells have a growth advantage over infected cells and therefore become predominant in the culture. In contrast, in our study, we characterized cells only early after primary infection and observed no such negative effects at the level of virus entry. In accordance with our FACS and infection analyses (Fig. 5), we therefore conclude that JFH-1 does not induce a block at the level of virus entry, or does so only to a minor extent, but rather forces selection of poorly infectible cells due to its cytopathogenicity.

By using supertransfection experiments with cells carrying stably replicating subgenomic HCV replicons, we found that translation and, presumably, replication of the transfected (secondary) RNA are impaired. This result is in full agreement with previous observations that the replication efficiency of an HCV RNA decreases in the presence of increasing amounts of a viral RNA cotransfected into the same cell (34). Moreover, upon transfection of increasing amounts of replication-competent viral RNA, the replication efficiency per transfected molecule decreases (34). It is therefore assumed that one or more cellular factors required for efficient HCV RNA replication are limiting in a cell. One would therefore expect that cells can support replication of HCV RNA only to a certain level and that the extent of superinfection exclusion directly correlates with the replication fitness of the primary infecting RNA. In fact, it was found that replication fitness of one replicon reduces the capacity of a second replicon to stably replicate in the same culture (16), arguing for a limiting host cell factor(s) and condition(s) that are required for efficient HCV RNA replication. Although similar findings have been obtained with bovine viral diarrhea virus (32), arguing for a common mechanism of superinfection exclusion at the level of RNA replication, alternative explanations are possible. For instance, large amounts of HCV protein(s) may exert a direct negative effect on replication of the incoming RNA. Such a scenario is assumed for Borna disease virus, where resistance to homologous secondary infection has been ascribed to unbalanced levels of viral nucleocapsid components (18). It is hypothesized that the incorrect stoichiometry of nucleocapsid components leads to an inhibition of the polymerase activity of incoming viruses. Alternatively, large amounts of HCV protein(s) may alter correct trafficking of the incoming superinfecting virus, as suggested in the case of the L surface antigen of the duck hepatitis B virus (50). In the case of Sindbis virus, expression of the nonstructural proteins is sufficient to block secondary infection at the level of RNA replication without affecting RNA translation (1, 22). It is assumed that large amounts of the viral protease accumulated in previously infected cells lead to rapid proteolytic cleavage of the replicase expressed from the incoming RNA, resulting in an inhibition of the negative-strand copy from the incoming RNA (27).

Thus far, the JFH-1 isolate is the only one that replicates to very high levels, whereas all other functional HCV clones described thus far require cell culture-adaptive mutations and even then replicate less efficiently (3). Given the likely correlation between superinfection exclusion and the replication fitness of the primary infecting genome, it is not clear how well the observations made with JFH-1 can be extrapolated to other HCV isolates. Even though interference at the level of RNA replication has also been described for replicons derived from the genotype 1b isolate Con1, these RNAs always carried replication-enhancing cell culture-adaptive mutations (16, 34). Thus, the questions of whether superinfection exclusion also occurs with isolates that are less fit in replication and how these isolates codevelop in long-term passages require the establishment of further functional HCV clones that replicate in the absence of adaptive mutations and that, ideally, support infectious virus production, too. For the same reasons, it is unclear to what extent superinfection exclusion occurs in vivo. Nevertheless, one obvious consequence would be that superinfection exclusion reduces the likelihood of recombination between two viral genomes coreplicating in the same cell. In fact, RNA recombination has been observed only sporadically, arguing that it is a rare event (11, 25). While on one hand superinfection exclusion reduces the likelihood of interference between two viral genomes in the same cell, on the other hand it reduces genetic plasticity by lowering the chance for the formation of novel virus recombinants. It remains to be determined how the complex interplay between replication fitness, superinfection exclusion, and RNA recombination is regulated.