Poliovirus/Hepatitis C Virus (Internal Ribosomal Entry Site-Core) Chimeric Viruses: Improved Growth Properties through Modification of a Proteolytic Cleavage Site and Requirement for Core RNA Sequences but Not for Core-Related Polypeptides

Abstract

H.-H. Lu and E. Wimmer (Proc. Natl. Acad. Sci. USA 93:1412–1417, 1996) have demonstrated that the internal ribosomal entry site (IRES) of poliovirus (PV) can be functionally replaced by the related genetic element from hepatitis C virus (HCV). One important finding of this study was that open reading frame sequences 3′ of the initiating AUG, corresponding to the open reading frame of the HCV core polypeptide, are required to create a viable chimeric virus. This made necessary the inclusion of a PV 3C protease (3C^pro) cleavage site for proper polyprotein processing to create the authentic N terminus of the PV capsid precursor. Chimeric PV/HCV (P/H) viruses, however, grew poorly relative to PV. The goal of this study was to determine the molecular basis of impaired replication and enhance the growth properties of this chimeric virus. Genetic modifications leading to a different proteinase (PV 2A^pro) cleavage site between the HCV core sequence and the PV polyprotein (P/H701-2A) proved far superior with respect to viral protein expression, core-PV fusion polyprotein processing, plaque phenotype, and viral titer than the original prototype PV/HCV chimera containing the PV 3C^pro-specific cleavage site (P/H701). We have used this new virus model to answer two questions concerning the role of the HCV core protein in P/H chimeric viral proliferation. First, a derivative of P/H701-2A with frameshifts in the core-encoding sequence was used to demonstrate that production of the core protein was not necessary for the translation and replication of the P/H chimera. Second, a viral construct with a C-terminal truncation of 23 amino acids of the core gene was used to show that a signal sequence for signal peptidase processing, when present in the viral construct, is detrimental to P/H virus growth. The novel P/H chimera described here are suitable models for analyzing the function(s) of the HCV elements by genetic analyses in vivo and for antiviral drug discovery.

Hepatitis C virus (HCV) is associated with 95% of cases of posttransfusion hepatitis and over 50% of sporadic non-A, non-B hepatitis (27). Based on the genotype of HCV cDNA that was isolated in 1989, HCV was classified into Flaviviridae (7, 21, 28, 33), a family comprising many enveloped positive-strand RNA animal viruses. The HCV genomic RNA is about 9.4 kb long, encoding a single open reading frame (ORF). Translation of the ORF produces a polyprotein that is processed by host signal peptidase and two viral proteinases to yield at least 10 different structural and nonstructural proteins. The length of the 5′ nontranslated region (NTR) appears to be approximately 340 bases, depending on specific viral strains, and contains complex secondary structures (15). Tsukiyama-Kohara et al. (43) have shown that a highly ordered sequence within the HCV 5′ NTR (Fig. 1B and reference 5) can function as an internal ribosome entry site (IRES), an observation that has been confirmed by others (25, 38, 39, 45). IRESes, originally discovered in the studies of translational control of picornaviruses (16, 17, 35, 36), promote the initiation of translation without the requirement for a 5′-terminal cap structure (29, 46).

FIG. 1.

Characterization of the P/H701 chimeric viruses containing different proteinase cleavage sites upstream of the PV structural proteins. (A) Schematic diagram of the genomic organization for the P/H701 chimera with the proteinase cleavage sites. The cloverleaf-like RNA structure of PV, an essential cis-acting replication signal, is located at the 5′ end of the genome, and noninitiating AUG codons found in the HCV 5′ NTR are denoted by stars. The structure of the HCV 5′ NTR is based on results from reference 15. The shaded (HCV) and clear (PV) boxes depict the ORF encoded by the viral polyprotein; the position of the HCV core protein gene (ΔC) is marked. Locations of the PV-encoded proteinases 2A and 3C are shown within the polyprotein, and their respective cleavage sites are expanded below. (B) In vitro translation protein products templated by PV and the two P/H701 transcript RNAs. Protein bands corresponding to the PV-encoded polypeptides are indicated, as is the position of the unprocessed Δcore-P1 precursor. Numbers above the lanes represent amounts of T7-derived RNA transcripts used per 12.5 μl of translation reaction. (C) Plaque phenotypes of the PV and P/H701 chimeric viruses on HeLa R19 cell monolayers. Infected cells were stained to visualize plaques after incubation at 37°C for 48 h. The titers of virus obtained after RNA transfection of PV(M), P/H701-2A, and P/H701 were about 10⁸, 10⁷, and 10⁶ PFU/ml, respectively. (D) One-step growth kinetics of PV and P/H701 viruses on HeLa cell monolayers. Aliquots of synchronously infected cells were harvested at the time points shown and processed for measurements of viral titers by plaque assay.

Poliovirus (PV) is the prototype member of the family of Picornaviridae. Its positive-strand RNA genome is characterized by a long (742-nucleotide [nt]) 5′ NTR that contains two important cis-acting elements: the 5′-cloverleaf-like structure and the IRES element. These structures are involved in the initiation of viral RNA replication (3) and cap-independent translation of viral mRNA, respectively (46). The PV IRES element spans approximately from nt 100 to 590, a size that is characteristic of all picornaviral IRES elements. Unlike genomes of the flaviviruses, the PV genome encodes a polyprotein that, except for maturation cleavage, is proteolytically processed only by virus-encoded proteinases (20).

The biological and pathogenic properties of PV and HCV are vastly different. PV is a highly infectious and cytolytic agent that proliferates rapidly to high titers in suitable cell cultures. In contrast, HCV replicates slowly in the host, initially causing acute disease that is often mild or even asymptomatic. About 50% of acute hepatitis C cases are followed by chronic hepatitis, and 20% of the patients with chronic hepatitis may develop cirrhosis and hepatocellular carcinoma (6, 10, 40). Experiments to proliferate HCV in tissue culture cells have largely failed.

Since the discovery of HCV, great efforts have been made to develop effective treatments for chronically infected HCV carriers. These efforts have been hindered by the poor replication of HCV in experimental animals and in cell culture and by the restricted host range of the virus. In a recently described (18) infectious cDNA clone, infectivity was demonstrated by injection of in vitro-transcribed RNA into the livers of healthy chimpanzees. Although this has provided proof that the existing HCV cDNA contains all information necessary for viral proliferation and for inducing hepatitis, its use as a model for drug discovery is limited.

Previously, Lu and Wimmer (25) have demonstrated that the IRES of PV can be functionally replaced by the related genetic element from HCV. One important finding of this study was that an ORF sequence corresponding to the core gene is required to create a viable PV/HCV chimeric virus (referred to as a P/H virus), a feature unique among other characterized IRES elements. Specifically, addition to the 5′ NTR of a segment of 21 nt immediately following the AUG initiating polyprotein synthesis converted a nonreplicating chimeric genome to a virus with a minute-plaque morphology. This functional requirement for sequences downstream of the initiating AUG has also been observed by Reynolds and colleagues (38) in relation to HCV IRES-directed translation in vitro. However, further addition of core sequences improved the growth properties of the P/H hybrid virus; for example the construct P/H701 encodes the N-terminal 123 amino acids of the core protein. It was suggested that protein fragments encoded by the core coding sequence may be beneficial for expression of the viral polyprotein and, hence, for viral replication (25). Nevertheless, these chimeric viruses still expressed poor growth properties relative to those of PV (25). The construction of the chimeras such as P/H701 made it necessary to insert a PV proteinase 3C (3C^pro) cleavage site at the junction of the HCV core-encoding sequences and the PV structural protein precursor P1. This facilitated the generation of a proper N terminus of the PV polyprotein by 3C^pro processing (Fig. 1). Cleavage at this junction by 3C^pro, however, was found to be slow (25).

The goal of this study was to determine molecular parameters responsible for the poor growth phenotype of the chimeric virus. We reasoned that creating a new P/H chimera with enhanced growth properties would facilitate genetic studies of the HCV IRES. To this end, genetic modifications to the proteinase cleavage site that are necessary for the complete removal of HCV core protein sequences from the PV polyprotein have been introduced. This report describes the construction of P/H chimeras replacing the artificial PV 3C^pro-specific cleavage site (25) with an artificial PV proteinase 2A (2A^pro)-specific cleavage site, creating P/H701-2A. This study demonstrates that the P/H701-2A RNA translates better in vitro and in vivo, processes the core-PV polyprotein fusion more efficiently, produces larger plaques, and displays better growth kinetics compared with the construct containing PV 3C^pro-specific cleavage site. We propose that the P/H chimeric constructs bearing the PV 2A^pro-specific cleavage site are suitable (i) for analyzing functions of these HCV elements in vivo and (ii) as a model for screening potential drugs that target these genetic elements of HCV. To begin addressing the first goal, we carried out an analysis of mutant chimeric viruses to test a hypothesis (25) that the core protein, or deletion versions thereof, aid in the replication of the virus, possibly by stimulating HCV IRES function. Our data provide evidence that expression of a core protein fragment does not enhance proliferation of the P/H chimera. With this chimeric virus model, we have also found that a hydrophobic signal sequence at the C terminus of the core protein is the only hindrance to the expression of full-length core protein in this PV-based system.

MATERIALS AND METHODS

Cells, viruses, and plasmids.

HeLa R19 cell monolayers were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 5% bovine calf serum (BCS). PV type 1 strain Mahoney [PV1(M)] and its derivatives were amplified in HeLa cells as described previously (26). The titer of the virus stocks was determined by standard plaque assay on HeLa R19 monolayers (24). Briefly, HeLa cells were infected with cell lysates derived from transfection with the corresponding RNA. After incubation at 37°C for 48 h, viral plaques were developed by 1% crystal violet staining. Plasmid pT7PVM is a derivative of pT7XL, a full-length cDNA clone of PV1(M) constructed in this laboratory (44).

Constructions of plasmids with 2A^pro-specific cleavage site-encoding sequence.

DNA cloning was accomplished by following standard procedures (41). For construct P/H701 (25) nt 108 to 742, the PV IRES plus spacer region (46) of the PV cDNA was replaced with nt 9 to 332 of the 5′ NTR of genotype 1b HCV (43) plus the N-terminal 123 codons of the core ORF of HCV, producing a fusion of the 123 amino acids with N-terminal sequence of the PV polyprotein (Fig. 1C). This construct contains a PV 3C^pro-specific cleavage site (amino acids LALFQ*G [underlined letters denote the most important cleavage signals; the asterisk denotes the scissile bond]) between the truncated core protein-encoding sequence (Δcore) of HCV and P1 of PV, ensuring the release of the HCV Δcore from the PV polyprotein. P/H701 was used as the template for PCR-based mutagenesis. Mutagenesis oligonucleotides HCV2A (5′-GTAAGGTCATCGAGCTCAAAGGTCTCACAACATATGGTGCTCAGGTTTC-3′), which contains the PV 2A^pro-specific cleavage site (amino acids KGLTTY*G), and PVBsm I (5′-GGGAACACAAAGGCATTCC-3′) were used in the mutagenesis PCR. The region between SacI (nt 808) and BsmI (nt 1605) of P/H701 was removed to produce vector P/H701dSB. The PCR product was digested with SacI and BsmI and ligated to P/H701dSB to yield P/H701-2A. The construct P/H701-2A, containing the PV 2A^pro-specific cleavage site, was selected by restriction mapping and verified by sequencing.

The region between EcoRI and SacI of P/H701-2A was removed to produce vector P/H701-2AdES, lacking HCV IRES and core sequence. For constructs P/H905 (25) and P/H905ΔS (containing a nearly full length core gene, lacking only the C-terminal 23 hydrophobic amino acids codons [22]), the wild-type (wt) PV IRES was replaced with nt 9 to 332 of the 5′ NTR of genotype 1b HCV (43) plus 191 or 173 codons of HCV core-encoding sequences, respectively (Fig. 2). Each of the above constructs contains a PV 3C^pro-specific cleavage site (amino acids LALFQ*G) between the respective truncated core protein encoding sequence of HCV and P1 of PV. The 5′ NTR plus respective Δcore encoding sequences were released by digestion of the above P/H constructs with EcoRI and SacI and cloned into P/H701-2AdES to create a 2A counterpart that bears the PV 2A^pro-specific cleavage site. All constructs were verified by sequencing and restriction mapping.

FIG. 2.

Characterization of P/H701-2A and its frameshift derivative P/H701SH2-2A. (A) Schematic diagrams of N-terminal portions of ORFs of HCV core sequences corresponding to P/H701-2A and P/H701SH2-2A. The primary sequence of each core gene fragment, written in codon form, is displayed upstream of the SacI cloning linker site, the PV 2A^pro-specific cleavage site, and the start of the PV ORF. The dashed-line arrow with a number indicates the number of codons (Codons^#) within the bracket. For construct P/H701SH2-2A, the identities and positions of inserted and deleted dinucleotides are emphasized with arrows. (B) Protein profiles of translation products produced under the direction of RNAs of P/H701-2A and P/H701SH2-2A in a HeLa cell-derived in vitro translation system. RNA concentrations used in translation reactions (total volume) are shown as in Fig. 1B. (C) Plaque phenotypes of chimeras P/H701-2A and P/H701SH2-2A on HeLa cell monolayers. The dilution factors of the virus-containing supernatants used are 10⁻⁵ for both variants.

Construction of P/H701SH2-2A plasmids.

P/H701-2A was used as the template in PCR-based mutagenesis (14) to yield a frameshift mutation construct which displaced the core-encoding reading frame immediately after the initiating AUG codon by inserting two nucleotides (CC) and returned the ORF to the PV reading frame by deleting two nucleotides next to the 3′ cloning site. Mutagenesis oligonucleotides P/HEcoRI (5′-GCGAATTCGGCGACACTCCACC-3′) and P/H701SH2− (5′-GGATTTGTGCTGGCATGATGCACGG-3′) were used in PCR-A. P/H701SH2+ (5′-CCGTGCATCATGCCAGCACAAATCC-3′ and P/H701SH2 (5′-CCTTTGAGCTCTGACCTTACCC-3′) were used in PCR-B. Gel-isolated PCR fragments from both PCR-A and PCR-B were used in the PCR-C with oligonucleotides P/HEcoRI and P/H701SH2 to produce the PCR fragment with the designed mutation. The mutated PCR fragment was digested with EcoRI and SacI and cloned into P/H701-2AdES to yield P/H701SH2-2A, which was selected by restriction mapping and verified by sequencing.

In vitro transcription and RNA transfection.

For the production of RNA transcripts in vitro, 1.0 μg of full-length cDNA was linearized at a unique PvuI site downstream of the viral genome. RNA transcripts were produced from linear cDNA by T7 RNA polymerase in an in vitro system described previously (44). HeLa R19 monolayers cultured in 35-mm-diameter dishes were transfected by the DEAE-dextran method as described elsewhere (24) and grown in 2 ml of DMEM containing 2% BCS at 37°C for up to 4 days. Virus yield from transfection was titrated by plaque assay (24).

Characterization of viral growth phenotype.

The plaque phenotypes of wild-type (wt) PV and the P/H chimeras were characterized by plaque assay on HeLa R19 cell monolayers seeded on 35-mm-diameter dishes (24). After a 2-day incubation at 37°C, the plaques were developed by staining with 1% crystal violet. To measure one-step growth kinetics, HeLa R19 cells in 35-mm-diameter dishes were infected with virus at a multiplicity of infection (MOI) of 10 per cell. The infected cells were washed and then cultured in 2 ml of DMEM containing 2% BCS at 37°C, and individual dishes were harvested at 0, 2, 4, 7, and 12 h postinfection (p.i.). Virus titer in each cell lysate was determined by plaque assay (24).

Translation in vitro and in vivo.

Translation in vitro was performed in a HeLa cell extract (30) supplemented with either PV1(M) RNA or P/H chimeric RNAs at 30°C for up to 16 h. RNA quantities used are given in Results and figure legends.

To label viral polypeptide synthesis in vivo, 35-mm-diameter dishes of HeLa cells were infected with lysates derived from cells transfected with PV1(M) or P/H RNA. Specifically, cells were infected with PV1(M) at an MOI of 10 and incubated for 3.5 h, while cells infected with P/H chimeras (P/H701-2A and P/H905ΔS-2A) were exposed to an MOI of 1 and incubated for 20 h. The supernatants were removed by aspiration, and the infected cells were incubated at 37°C in 2 ml of DMEM with 2% BCS. Prior to labeling, the infected cells were washed twice with Met-deficient DMEM and starved in Met-deficient DMEM for 1 h. The starved cells were then supplemented with 30 μCi of Tran³⁵S-label (ICN Biomedicals, Costa Mesa, Calif.) for 1 h. The labeling medium was removed, and the cells were washed three times with 2 ml of cold isotonic buffer (10 mM Tris-HCl [pH 7.5], 140 mM NaCl, 1.5 mM MgCl₂). The cells from each dish were lysed in 200 μl of 0.5% Nonidet P-40 lysis buffer (10 mM Tris [pH 7.5], 1 mM EDTA, 0.5% Nonidet P-40, 100 mM NaCl), and the lysates were clarified by a low-speed centrifugation for 3 min.

In vitro- or in vivo-labeled proteins were analyzed by sodium dodecyl sulfate–12.5 or 13.5% polyacrylamide gel electrophoresis (SDS-PAGE).

Western blotting of core protein.

Western blotting was performed after SDS-PAGE of in vivo-radiolabeled protein samples. Briefly, proteins from infected cells, pulse-labeled 4 h p.i., were electrotransferred onto a nitrocellulose membrane (Schleicher & Schuell), which was blocked in 5% nonfat dry milk and incubated with mouse monoclonal anticore antibodies C7-50 and C8-59 (a kind gift from J. Wands, Massachusetts General Hospital) diluted 1:1,000 in Tris-buffered saline–Tween 20 including 5% nonfat dry milk for 1.5 h. Alkaline phosphatase-conjugated anti-mouse immunoglobulin gamma and light chains (Biosource International) were used as secondary antibody. The core protein was visualized with Sigma Fast 5-bromo-4-chloro-3-indolylphosphate–nitroblue tetrazolium tablets as suggested by the manufacturer (Sigma). The Western blot was exposed to autoradiography film to detect all radiolabeled protein bands.

RESULTS

Construction of P/H701-2A.

The construction of a viable PV chimera in which the PV IRES (Fig. 1A) was replaced by the HCV 5′ NTR plus HCV sequences of the HCV ORF provided conclusive evidence that the HCV genome indeed encodes a functional IRES element in a replicating RNA genome in vivo (25). Among all of the P/H chimeras with a PV 3C^pro-specific cleavage site at the junction between the HCV core sequences and the PV polyprotein, P/H701 proliferated most efficiently in comparison to wt PV. P/H701, however, still exhibited a minute- to small-plaque phenotype. Based on published work (4), it was anticipated that cleavage at the PV 3C^pro-specific cleavage site would readily liberate a free N terminus of P1; this, in fact, was not the case. Approximately 50% of the core-P1 polyprotein precursor remained unprocessed in vitro and in vivo (25). Moreover, P/H701 grew very poorly. Cytopathic effect was not observed after P/H transcript RNA transfection in cell culture until 4 days (24 h for wt PV RNA) posttransfection (p.t.). Once virus was collected, the plaques formed were very small even after a 3-day incubation, and in vivo labeling of viral protein was most efficient at 20 (4 to 6 for wt PV-infected cells) h p.i. In addition, the in vitro translation efficiency was low, requiring more P/H701 transcript RNA per reaction than wt PV transcript RNA to achieve optimal protein production. Notably, the plaque size phenotype was stable after several passages (25).

To facilitate the characterization of HCV IRES function in this chimeric virus system, the possibility that slow processing between Δcore and P1 is rate limiting in chimeric viral proliferation was explored. Viral 2A^pro is known to efficiently cleave the viral polyprotein in trans (23, 24). Therefore P/H701-2A was constructed by replacing the PV 3C^pro-specific cleavage site (amino acids LALFQ*G; underlined letters indicate residues important for efficient cleavage) at the junction between the core-encoding sequences of HCV and P1 of PV in P/H701 with the PV 2A^pro-specific cleavage site (amino acids KGLTTY*G [Fig. 1A and 2A]). Chimeric cDNAs of P/H701, P/H701-2A, and wt PV were transcribed in vitro to yield infectious viral RNAs, which were then assayed for the ability to be translated in vitro and to replicate in vivo.

Translation of chimeric viral RNA transcripts in vitro.

The ability of the transcript RNAs to direct protein synthesis was tested by in vitro translation in a HeLa cell extract (30). The transcribed RNAs from P/H701 and P/H701-2A were competent to produce the same PV-specific polypeptides as wt PV, an observation demonstrating the integrity of the chimeric constructs (Fig. 1B). P/H701-2A was found to direct protein synthesis significantly more efficiently than P/H701 RNA. At a transcript RNA concentration of only 6.7 mg/ml, P/H701-2A RNA gave better translation profiles than seen with P/H701 RNA at optimal translation conditions (54 mg of transcript RNA per ml). Comparing the apparent amount of the Δcore-PV polyprotein precursor at the top of the gel, we found that an estimated 70% of the precursor was processed in the course of the P/H701-2A translation, whereas about 30% of the precursor was processed in translations with P/H701 RNA. It was therefore predicted that during HeLa cell infection by P/H701-2A virus, more PV structural proteins may be available for packaging of virion particles. The reason for the increased cleavage efficiency of the PV 2A^pro-specific cleavage site is unknown, but the difference may be due to changes in protease to substrate recognition.

Replication of the P/H701 chimeras with different cleavage sites.

To further study the P/H701-2A chimera, RNA transcripts from P/H701-2A and P/H701 were transfected into HeLa R19 cells. It was found that transcript RNA from the P/H701-2A cDNA induced cytopathic effect at about 27 h p.t., significantly faster than P/H701, which, as mentioned above, required more than 4 days. The HeLa R19 cells transfected with P/H701-2A RNA transcripts were harvested 2 days p.t., whereas cells transfected with P/H701 transcripts were harvested 4 days p.t. Lysates from both P/H701-2A and P/H701 were subjected to standard plaque assays along with the wt PV control. The results clearly show that infection with P/H701-2A virus produces plaques larger than those formed from infection with P/H701 virus (Fig. 1C). Analysis of viral growth kinetics (Fig. 1D) revealed that by 10 h p.i., the P/H701-2A virus grew to about 1 order of magnitude higher in titer than the original P/H701 virus. However, the eclipse phase in the one-step growth curve, an indication of the early events of uncoating, translation, and replication, lasted for 4 h p.i. for both P/H701 and P/H701-2A virus infections (wt PV eclipse lasts 2 h). Based on these results and the in vitro translation data, the plaque phenotype of P/H701-2A is proposed to be mainly the result of better processing of the Δcore-PV polyprotein precursor. An example of the processing of the chimeric polyprotein in vivo is shown in Fig. 3A.

FIG. 3.

Pulse-labeling/time course of proteins produced in HeLa cell monolayers during infection with P/H701-2A (A) and its frameshift derivative P/H701SH2-2A (B). Equivalent radioactive counts per minute from each time point sample were loaded on each lane of the gel. “Mock” lanes demonstrate host proteins produced at the 1-h time point. PV, radiolabeled PV protein marker. Virus stocks used in this experiment were from the sixth passage of the original RNA transfection lysates.

Construction of a frameshift mutation in the HCV core gene.

Previous results (25) demonstrated that sequences of the coding region of the core protein downstream of the HCV 5′NTR are necessary for the HCV IRES activity since constructs bearing only nt 9 to 332 of the 5′ NTR of HCV, fused directly to the PV ORF, did not yield chimeric virus upon transfection. When at least eight codons from the HCV ORF were added to the HCV 5′ NTR, a viable virus with a minute-plaque phenotype and very poor growth properties was recovered. Further extension of the core gene fusion significantly increased viral proliferation. Indeed, it was hypothesized that a polypeptide fragment encoded by the core ORF was involved in this effect of enhancement (25). Using our newly constructed cDNAs, we designed an experiment to distinguish the role of the core protein from the role of the RNA sequence of the core gene in HCV IRES function.

We constructed clone P/H701SH2-2A, in which the core-encoding reading frame immediately after the initiating AUG codon was interrupted by inserting two nucleotides (CC), followed by the engineered restoration of the ORF to the PV reading frame through the deletion of two nucleotides (TC), just preceding the 2A cleavage site (Fig. 2A). The choice of the two nucleotides to be inserted was based on the following considerations: first, insertion of only one nucleotide will result in a production of short polypeptide with a stop codon in the middle of the truncated core sequence; second, the two nucleotides inserted will not base pair with any nucleotide sequence in stem-loop V, a stem-loop structure where the initiating AUG is positioned within the single-stranded loop. This is considered advantageous for the initiation of core polypeptide translation (5, 15). We realized, however, that the insertion of the two cytidylic acid residues would disturb the context of the initiating AUG codon from AUCAUGA to the less favorable AUCAUGC (19).

In vitro translation of P/H701SH2-2A RNA transcripts was carried out along with P/H701-2A and wt PV RNA (Fig. 2B). Although the translational efficiency in this particular experiment was low for all constructs tested, all of the PV-specific polyproteins and cleavage products were observed. Indeed, synthesis and processing of the P/H701SH2-2A polyprotein was as efficient as that of P/H701-2A. Transcribed RNAs from constructs P/H701SH2-2A, P/H701-2A, and wt PV were also used to transfect HeLa R19 cells. Plaque assay results (Fig. 2C) showed that the titer and plaque size of P/H701SH2-2A and P/H701-2A were nearly the same. The results of the translation of viral RNAs of P/H701SH2-2A and the plaque assays demonstrate that the core-encoding RNA sequence, rather than the core protein itself, is essential for the in vitro translation and in vivo replication of the P/H chimeras.

In vivo protein pulse-labeling of P/H701-2A and P/H701SH2-2A.

To follow viral protein production directed by the P/H701-2A and P/H701SH2-2A viruses in infected HeLa cells over time, we performed pulse-labeling experiments using radiolabeled amino acids. HeLa R19 cell monolayers were infected with sixth-passage viruses (not plaque purified) of P/H701-2A and P/H701SH2-2A at an MOI of 10. At each time point indicated in Fig. 3, the infected cells were washed, starved of methionine, and pulsed with radiolabeled amino acid to visualize the newly synthesized proteins after separation by SDS-PAGE.

Both P/H701-2A (Fig. 3A) and P/H701SH2-2A (Fig. 3B) induced the shutoff of host cell translation by 3 h p.i. Viral protein synthesis reached its maximum level at about 5 h p.i. for both P/H701SH2-2A and P/H701-2A. The Δcore-PV polyprotein precursors were fully processed in vivo for both P/H701-2A and P/H701SH2-2A, demonstrating the effectiveness of the newly introduced PV 2A^pro cleavage site at the Δcore/P1 junction in vivo. These data support the contention that P/H701SH2-2A carries all signals necessary for efficient viral replication and host protein synthesis shutdown.

Analysis of in vivo-synthesized core protein by Western blotting.

RNA viruses display a remarkable level of genetic flexibility, related to the quasispecies nature of such genomes. This is due to the inherent infidelity of RNA-dependent RNA polymerase, as well as to recombination (46). These considerations also apply to the chimeric P/H genomes and to mutant derivatives like P/H701SH2-2A. If the HCV core protein or a fragment thereof is essential to HCV IRES function, be it via interaction with the viral RNA or an undetermined host cell protein(s), then selective pressures may lead to the selection of genomes that have restored, in frame, the core ORF.

To address the existence of the core-related polypeptides and demonstrate the genetic stability of the P/H701SH2-2A mutations, HeLa R19 cells were infected with sixth-passage viruses of P/H701-2A and P/H701SH2-2A at an MOI of 10, and the newly synthesized viral polypeptides were pulse-labeled 4 h p.i. Radiolabeled infected cell lysates were processed for Western blotting, and the nitrocellulose membrane was probed with a cocktail of mouse monoclonal anticore antibodies (31). A truncated core protein signal was detected only in the P/H701-2A lysate (Fig. 4A, lane 3), which indicated that even after six blind passages of P/H701SH2-2A virus (passage of bulk virus), there was no selection for variants that restored the production of a protein immunologically related to the HCV core protein. To confirm the genetic stability of the P/H701SH2-2A, a number of plaque-purified viruses from six consecutive passages were analyzed by reverse transcription-PCR and sequencing. No changes in the primary sequence or in the length of the core gene were detected (47). Upon completion of the Western blot processing shown in Fig. 4A, the nitrocellulose membrane was exposed to autoradiography film to detect all radiolabeled proteins. The complete set of PV-specific proteins was found in all samples tested (Fig. 4B, lanes 1 to 3), an observation demonstrating equivalent protein expression directed by the different viral genomes: PV, P/H701-2A, and the frameshift derivative virus.

FIG. 4.

Detection of the truncated core protein produced by P/H701-2A during an in vivo pulse-labeling experiment. HeLa R19 cells were infected with P/H701-2A and P/H701SH2-2A viruses (as for Fig. 3), and newly synthesized viral polypeptides were pulse-labeled with Tran³⁵S-label at 4 h p.i. The gel was processed for Western blot analysis using HCV core-specific monoclonal antibodies as the probe. After visualization of the core antigen (A), the nitrocellulose was exposed to autoradiography film to detect all radiolabeled proteins (B). Positions of PV-encoded polypeptides are marked on the right; those of protein molecular weight markers used for the Western blot are indicated on the left. Under the gel conditions used, the truncated core protein migrates at about 20,000 (the predicted molecular weight is about 16,000).

Taken together, these data show that this P/H chimeric virus, containing a frameshift mutation, is not under a selection pressure to restore the integrity of the core ORF. Together with experiments in which the core sequences were deleted to eight codons without loss of viability (47), we conclude that the core protein is not absolutely required for translation or chimeric virus growth.

Effects of a signal sequence on P/H virus translation and replication.

To increase the size of the HCV-derived ORF downstream of the HCV 5′ NTR, we inserted the entire core ORF into the chimeric virus. We had previously observed that a P/H virus containing the complete 191 codons of the core gene (P/H905) was not viable (25). The core ORF encodes a signal sequence at its C terminus that is necessary for signal peptidase-mediated processing of the HCV structural proteins (42). Based on previous data (24), we considered it possible that the signal sequence was responsible for the null phenotype, a hypothesis that we tested by deleting 23 amino acids from the C terminus of construct P/H905-2A, thereby creating construct P/H905ΔS-2A (Fig. 5A). In vitro translations of the transcribed RNAs from P/H905-2A and P/H905ΔS-2A were performed with a HeLa cell extract (30). Transcript RNAs of both constructs were efficiently translated, producing the expected PV-specific polypeptides (Fig. 5B). In these translation reactions of P/H905-2A and P/H905ΔS-2A RNAs, we observed an additional protein migrating slightly faster than the PV VP3 protein. The novel protein in the P/H905ΔS translation is slightly smaller than the one seen in the P/H905 sample, reflecting the fact that P/H905ΔS contains a truncated core gene. Indeed, reverse genetics and Western blotting have both been used to identify this protein as the core protein (47). It should be noted that the inability to radiolabel the core fragment produced by P/H701-2A in such in vitro translation reactions is due to the lack of methionine or cysteine codons found in that segment of the core gene (43).

FIG. 5.

Characterization of P/H viruses carrying larger core ORF inserts. (A) Schematic diagrams of the HCV core ORFs for P/H905-2A and P/H905ΔS-2A in comparison to P/H701-2A, arranged as in Fig. 2A. (B) Protein profiles of translation products directed by P/H701-2A, P/H905-2A, and P/H905ΔS-2A RNAs in a HeLa cell-derived in vitro translation system. Protein bands corresponding to PV-encoded polypeptides are indicated on the right. Positions of unprocessed Δcore-P1 precursors are shown at the top of the list of position markers. Locations of core protein-related bands produced during the translations of P/H905-2A (lane 2) and P/H905ΔS (lane 3) RNAs are also indicated; for an explanation of the bands labeled “core,” see text. (C) Translation of P/H701-2A, P/H905ΔS-2A, and wt PV RNAs in vivo. Positions of PV-encoded polypeptides are denoted on the right. Proteins produced during infection by a given virus are visualized by pulse-labeling infected HeLa cell monolayers with Tran³⁵S-label. (D) Plaque phenotypes of P/H701-2A and P/H905ΔS-2A viruses on HeLa cell monolayers. Dilution factors of the virus-containing supernatants used are indicated in parentheses. Note that because no P/H905-2A virus was ever recovered, that construct was not included in experiments represented in panels C and D.

To study the replication ability of the P/H chimeras bearing the full-length core gene and the 2A^pro cleavage site, we transfected transcribed RNAs onto HeLa cells and then analyzed cell lysates for virus production by plaque assays. It was found that P/H905-2A RNA did not produce virus (47), just like the corresponding construct P/H905 carrying the 3C^pro cleavage site (25). However, P/H905ΔS-2A RNA lacking the signal is replication competent (Fig. 5D). Pulse-labeling of viral proteins in HeLa cells infected with the same transfection lysates as above shows that the P/H905ΔS-2A produced wt PV-specific polyproteins; moreover, the Δcore-PV polyprotein precursors were fully processed (Fig. 5C). Plaque assays showed that P/H905ΔS-2A produced small to medium plaques, comparable to those produced by P/H701-2A (Fig. 5D). These results demonstrate that the C-terminal signal sequence of the core protein is detrimental to P/H chimeric virus growth and, based on the plaque phenotypes of P/H701-2A and P/H905ΔS, that RNA sequences necessary for stimulating chimeric virus growth may be localized only within the domain of the P/H701-2A genotype.

DISCUSSION

A remarkable property of viral IRES elements of picornaviruses and certain flaviviruses is their common function in controlling internal initiation of translation despite differences in their apparent structures. This was most dramatically demonstrated through the construction of a viable dicistronic PV whose ORF was divided into two independent ORFs by the heterologous IRES element of encephalomyocarditis virus (genus Cardiovirus) (29) or by the construction of hybrid PVs whose cognate IRES element was exchanged to that of encephalomyocarditis virus (1), human rhinovirus 2 (genus Rhinovirus) (11), or HCV (25). Picornavirus IRESes can be roughly divided into two types, depending on the genus: those belonging to entero- and rhinoviruses (type 1) and those belonging to cardio-, aphtho-, and hepatoviruses (type 2) (46). There is little, if any, sequence homology between the different picornavirus IRES types, as there is no apparent structural homology between the IRESes of picornaviruses and that of HCV.

Apart from its overall structure, the IRES of HCV, a member of Flaviviridae, is further distinct from picornavirus IRESes because it has incorporated into its functional unit nucleotide sequences downstream of the AUG codon initiating the viral polyprotein (25, 37). This was shown by in vitro studies using dicistronic mRNAs (37), a strategy developed by Jang et al. (17). It was independently demonstrated when PV/HCV hybrid viruses carrying the HCV IRES instead of the PV IRES were constructed (25). Although viral mRNA (T7 transcripts) lacking HCV core sequences downstream of the HCV 5′ NTR, could be translated with low efficiency in vitro, no replication of RNA carrying such minimal-IRES constructs was detected (25). However, when 24 nt (eight codons) of the HCV core were added to the HCV 5′ NTR, viral replication was observed. Viral proliferation of this construct, however, was very poor. Extension of the core sequence downstream of the initiating AUG significantly improved replication. The most efficiently replicating P/H hybrid (P/H701) contained 123 codons of the core protein. These observations led to the conclusion that core protein-related peptides may interact with the cognate IRES, thereby increasing its efficiency (25).

Construction of the P/H hybrid viruses required that the core-encoding sequence be fused to the PV ORF. This, in turn, demanded that the peptides fused to the PV ORF be cleaved precisely so that the 5′ end of the PV capsid precursor (P1) could be myristoylated (8, 9, 34). Following the strategy of Andino et al. (3, 4), a 3C^pro-specific cleavage site that would facilitate separation of the fused HCV-specific peptide was created in the hybrid virus (25). Unexpectedly, cleavage of the N-terminal peptide was very inefficient (25). Indeed, we have recently observed that, depending on the size of the foreign ORF fused to the PV ORF, ORF fusion polypeptides are exceedingly debilitating to viral replication (32). Accordingly, the virus carrying such fusion polyproteins is trying to escape the extra genetic burden of a foreign ORF by rapid deletion, sometime during first passage (32). The poor replication of P/H701 could therefore be due to either a toxic effect of the truncated core polypeptide or the sluggish processing of the fusion polypeptide. Genetic analyses of W1-P/H701 favored the second explanation (25) (see also below).

PV proteinase 2A^pro is a highly active proteolytic enzyme that can efficiently cleave polypeptides in trans (12, 13, 23, 24). Therefore, we tested the possibility that the separation of the fused polypeptide from the PV polypeptide could be facilitated more efficiently by 2A^pro. The results presented here show that the exchange of the cleavage site from the 3C^pro-specific cleavage site to that of 2A^pro greatly improved proteolytic processing at the N terminus of the PV P1 precursor in vitro and in vivo. Relative to PV1(M), the kinetics of replication of P/H701-2A approach wt levels, although a delay in uncoating of P/H701-2A, and other early events, is apparent. Indeed, whereas some uncleaved Δcore-P1 polypeptide was detected after translation of P/H701-2A in vitro, this fusion protein was undetectable in P/H701-2A-infected cells. In all experiments, P/H701-2A replicated to higher titers than P/H701, usually by at least an order of magnitude. These results suggest that the N-terminal core protein fusion inhibits processing of the P1 structural precursor, lowering the amount of PV capsid protein available for encapsidation.

The improved growth of the P/H701-2A virus due to the exchange of the protease cleavage site strengthened its use as a model system for genetic studies of these functional HCV sequence elements under conditions of RNA replication. For example, the introduction of a frameshift immediately downstream of the codon AUG initiating the core ORF, followed by a two-base deletion upstream of the 2A^pro-specific cleavage site (in P/H701SH2-2A), did not reduce the plaque size of the P/H701SH2-2A virus compared with the plaque size of P/H701-2A. Together, these results do not support our earlier hypothesis (25) of an involvement of core polypeptide in HCV IRES function within the P/H hybrid virus. It is possible that the poor replication of P/H701, carrying a 3C^pro-specific cleavage site, was further decreased by the generation of fragmented core polypeptides or out-of-frame polypeptides (25).

The efficient replication of P/H701SH2-2A relative to P/H701-2A came as surprise because the construction of the frameshift by insertion of two cytidine residues removed the favorable purine at +1 from the initiating AUG codon. Since the selection of this AUG codon does not occur by scanning but rather by exact positioning of the 40S ribosomal subunit onto the HCV IRES, the context of the initiation codon, initially defined by Kozak (19), may not be very relevant. Indeed, Jackson and colleagues demonstrated that the HCV IRES is tolerant of changes to the initiating AUG itself (38), suggesting that selection of the initiating AUG codon occurs by a very precise process, not necessarily requiring an AUG context that is favorable in cap dependent scanning.

Passage of P/H701-2A or P/H701SH2-2A did not reveal deletion of the sequence of the PV polyprotein (47), in contrast to observations of constructs in which foreign ORFs were fused to the PV polyprotein downstream of the PV IRES (32). It appears, therefore, that the core-specific RNA sequence enhances replication of the chimera, perhaps by providing RNA structures favorable for HCV IRES function. Decreasing the length of the core ORF to 24 nt (eight codons; construct P/H356-2A) dramatically reduced translational efficiency and viral replication of the corresponding RNA, regardless of whether the construct carried a 2A^pro-specific cleavage site (47) or 3C^pro-specific cleavage site (25). Finally, removing all core sequences abolished viral replication (construct P/H335) (25). At least within the context of the P/H hybrid, these results imply that RNA sequences beyond nt 24 (counting from the AUG codon) may contribute to HCV IRES function.

Lu and Wimmer (25) reported that inclusion of the entire core ORF into the P/H chimera abolished replication of the RNA, apparently due to the hydrophobic sequences mapping to the C terminus of the core polypeptide that may function as signal sequence for the E1 glycoprotein of HCV (42). A signal sequence engineered into the PV polyprotein has been found previously to be lethal (24). Indeed, removal of a hydrophobic region at the C terminus of the core ORF (construct P/H905ΔS-2A) leads to the recovery of viral replication. Significantly, the increase of the core ORF from 123 amino acids (P/H701-2A) to 168 amino acids (P/H905ΔS-2A) did not significantly affect gene expression or plaque phenotype, an observation suggesting that even the presence of a prolonged core sequence was of little consequence to translation and viral proliferation.

The P/H hybrid viruses are unique entities in which the replicating genome is dependent on the function of the HCV IRES controlling translation. Beyond the use of these constructs for basic genetic studies of HCV IRES function, as shown in this report, the P/H viruses are practical tools for development of drugs targeting these HCV sequences. Currently, no vaccines against HCV infection are available, and effective anti-HCV chemotherapy has not been developed. It has been estimated that 1.4% of the U.S. population has been infected with HCV (2). Considering that up to 30% of the infected individuals may develop serious, if not fatal, liver diseases, it is important to develop tissue culture assays to test novel antiviral agents. The new and rapidly replicating P/H hybrid viruses described here may be used in such assays.