|Home | About | Journals | Submit | Contact Us | Français|
Sequences in the 5′ and 3′ termini of plus-strand RNA viruses harbor cis-acting elements important for efficient translation and replication. In case of the hepatitis C virus (HCV), a plus-strand RNA virus of the family Flaviviridae, a 341-nucleotide-long nontranslated region (NTR) is located at the 5′ end of the genome. This sequence contains an internal ribosome entry site (IRES) that is located downstream of an about 40-nucleotide-long sequence of unknown function. By using our recently developed HCV replicon system, we mapped and characterized the sequences in the 5′ NTR required for RNA replication. We show that deletions introduced into the 5′ terminal 40 nucleotides abolished RNA replication but only moderately affected translation. By generating a series of replicons with HCV-poliovirus (PV) chimeric 5′ NTRs, we could show that the first 125 nucleotides of the HCV genome are essential and sufficient for RNA replication. However, the efficiency could be tremendously increased upon the addition of the complete HCV 5′ NTR. These data show that (i) sequences upstream of the HCV IRES are essential for RNA replication, (ii) the first 125 nucleotides of the HCV 5′ NTR are sufficient for RNA replication, but such replicon molecules are severely impaired for multiplication, and (iii) high-level HCV replication requires sequences located within the IRES. These data provide the first identification of signals in the 5′ NTR of HCV RNA essential for replication of this virus.
Hepatitis C virus (HCV) is a distinct member of the family Flaviviridae, comprising a group of enveloped viruses to which the flaviviruses like Yellow fever virus and the pestiviruses Border disease virus, Classical swine fever virus (CSFV), and Bovine viral diarrhea virus (BVDV) belong (33). In the majority of cases, HCV causes a persistent infection that is frequently asymptomatic or associated with only mild symptoms. However, persistently infected patients are at high risk to develop a chronic liver disease that can lead to liver cirrhosis and eventually hepatocellular carcinoma (39). In the absence of efficient therapies, chronic hepatitis C has become the main indication for liver transplantation.
HCV possesses a single-stranded RNA genome of positive polarity. This plus-strand RNA carries a single long open reading frame that encodes a polyprotein of ~3,010 amino acids. Mature viral proteins are generated by a series of proteolytic cleavages that are mediated by host cell signal peptidases and two viral proteinases (for a review, see reference 4). The structural proteins core, envelope protein 1 (E1), and E2 are located in the amino-terminal one-third of the polyprotein, and the nonstructural proteins NS2, NS3, NS4A, NS4B, NS5A, and NS5B are located in the remainder. Of the latter, NS3 to NS5B are required and sufficient for RNA replication (28). NS3 possesses proteinase as well as nucleoside triphosphatase (NTPase) and helicase activities and NS5B is the RNA dependent RNA polymerase (RdRp) (3, 6, 17, 23, 26, 49). The function of the hydrophobic polypeptide NS4B is unknown. NS5A may be involved in the resistance to the antiviral activity of alpha interferon, and it is probably required for some steps in RNA replication (8, 15, 16, 24). This process most likely is mediated by a replicase complex that is tightly associated with intracellular membranes (38). Replication proceeds via a minus-strand RNA intermediate that serves as the template for the production of excess copies of plus-strand RNA molecules (28).
Translation of the HCV polyprotein is mediated via the 5′ nontranslated region (NTR) that carries an internal ribosome entry site (IRES) (50, 52). Such RNA elements have first been discovered with the poliovirus (PV), the prototype member of the Picornaviridae (36). Analogous to HCV, PV encodes a single polyprotein that is translated in a cap-independent manner via a highly structured 5′ NTR. Genetic studies indicated that the cloverleaf-like structure located at the extreme 5′ end of the PV genome is not required for translation—but has a modulatory effect (48)—and that it acts as an independent RNA element that is sufficient for replication (1, 2, 45). However, several reports suggest that sequences within the PV IRES are required for RNA multiplication, too (9, 21, 47).
Numerous studies have convincingly demonstrated the existence of an IRES in the HCV 5′ NTR (reviewed in reference 44). Unlike the case of PV, this RNA element permits the direct binding of the 40S ribosome subunit in the absence of additional translation factors in a way that the initiator AUG of the long open reading frame is directly positioned in the P-site of the ribosome (37). Moreover, the secondary, and probably also the tertiary, structures of the HCV and PV IRESs as well as their activities in cell extracts of various sources are very different (30, 31). Based on these and several other differences, the IRESs of PV and HCV were classified as type 1 and type 3 elements, respectively (25).
Computer predictions and structure probing revealed four distinct RNA domains in the HCV 5′ NTR (Fig. (Fig.1)1) (44). The short stem-loop 1 is formed by residues 5 to 20 and it is not required for IRES activity. In fact, its deletion was found to enhance RNA translation, suggesting a regulatory role (20, 22, 42, 43). The importance of stem-loop 2 for IRES function is discussed controversially, but most studies describe an enhanced translation when this structure is present (14, 20, 40, 42). Stem-loop 3 represents the core of the IRES. It has the highest degree of structural conservation and it participates in the formation of an RNA pseudoknot that is important for IRES activity (51). The smaller stem-loop 4 that harbors the initiator AUG codon is not essential for translation of the RNA. In fact, the stability of this stem-loop inversely correlates with translation efficiency (19).
The 5′ border of the HCV IRES was mapped between residues 38 and 46, i.e., just upstream of stem-loop 2 (18, 41, 52). Interestingly, in most studies, the 3′ border was mapped to within the core coding region, although it is not clear whether this RNA sequence participates in IRES structure or serves as a spacer to prevent unfavorable base pairings between the IRES and downstream sequences (discussed in reference 44). Core protein is not required for IRES function (53).
While the role of the 5′ NTR for translation has been studied in detail, its importance for RNA replication so far was not addressed. This was due to the lack of an efficient cell culture system. We have recently developed subgenomic HCV RNAs (replicons) that amplify to high levels in the human hepatoma cell line Huh-7 (28). These RNAs are composed of the following elements: the HCV 5′ NTR up to the 3′ end of the IRES that directs translation of the gene encoding the neomycin phosphotransferase (NPT), the IRES of the encephalomyocarditis virus (EMCV) directing translation of the NS3-5B region, and the authentic HCV 3′ NTR (Fig. (Fig.1).1). Upon transfection of Huh-7 cells with these bicistronic RNAs and selection with G418, cell lines were established that carried high amounts of self-replicating HCV RNAs. Moreover, we and others have recently identified cell culture adaptive mutations that increase RNA replication to a level that is sufficient for detection in transient assays (8, 24, 27). The most efficient replicon we developed contains two mutations in NS3 and one in NS5A that increase RNA replication synergistically. Upon replacement of the neo sequence by the gene encoding the firefly luciferase, replication of this cell culture-adapted RNA can be monitored in a transient assay by measurements of the activity of the reporter gene (24).
In this study, we analyzed the sequences at the 5′ end of the HCV genome required for RNA replication. By using the replicon cell culture system, we show that the 5′ terminal 125 nucleotides of the HCV genome are sufficient for RNA replication, albeit at a low level that is strongly enhanced when the complete 5′ NTR is included. These data suggest that the signals required for RNA replication overlap with those necessary for translation.
Cell monolayers of the human hepatoma cell line Huh-7 (35) were routinely grown in Dulbecco's modified mininal essential medium (Life Technologies, Karlsruhe, Germany) supplemented with 2 mM l-glutamine, nonessential amino acids, 100 U of penicillin, 100 μg of streptomycin, and 10% fetal calf serum.
Standard recombinant DNA technologies were used for all constructions (46). The basic plasmids pFK-I389neo/NS3-3′/5.1 and pFK-I389luc/NS3-3′/5.1, carrying two cell culture adaptive mutations in NS3 and one in NS5A, have been described recently (24). Deletions were introduced into the 5′ NTR by PCR-based mutagenesis using primers S Δ24-40-Hind or S Δ5-20-Hind and A neo-3′Pme (Table (Table1).1). PCR fragments were restricted with HindIII and AscI and inserted together with an AscI-XhoI fragment derived from the parental neo or luc replicon construct into the HindIII- and XhoI-restricted vector carrying the luc or neo replicon sequence. For the construction of RNAs with an HCV-PV chimeric 5′ NTR, plasmid pFKnt1-PI-lucEI3420-9605/5.1 was generated carrying the following elements (in the 5′-to-3′ direction): an SbfI restriction site, the T7 RNA polymerase promoter, a PmeI restriction site, the PV IRES from nucleotide 108 to 746 of the PV genome, the luciferase gene, the EMCV IRES, and the HCV sequence from NS3 up to the 3′ end of the genome. This plasmid was obtained by insertion of the hybridized oligonucleotides S T7 nt1-Hind-Eco and A T7 nt1-Hind-Eco into pFK-I389luc/NS3-3′/5.1 after restriction with EcoRI and HindIII. The resulting construct was linearized with PmeI and SfiI and ligated with a PmeI-SfiI fragment that carried the PV IRES fused to the luciferase gene (V. Lohmann and R. Bartenschlager, unpublished data) and an SfiI fragment corresponding to nucleotides 3622 to 8499 of the HCV genome. Constructs carrying 5, 12, or 24 nucleotides of the HCV 5′ NTR downstream of the T7 promoter were obtained by insertion of the complementary sense and antisense oligonucleotides listed in Table Table11 via the SbfI and PmeI restriction sites. To obtain plasmids that were used for synthesis of replicons carrying longer HCV sequences at their 5′ ends, PCR fragments were generated by using the sense primer S T7 nt1-Sbf and either of the following antisense oligonucleotides: A23-43-Pme, A61-84-Pme, A102-125-Pme, A271-296-Pme, and A310-341-Pme (Table (Table1).1). After restriction with SbfI and PmeI, the fragments were inserted into pFKnt1-PI-lucEI3420-9605/5.1, resulting in plasmids that allowed synthesis of replicons with 43, 84, 125, 296, or 341 nucleotides of the HCV 5′ NTR upstream of the PV IRES. The 55-nucleotide-long spacer element was obtained by hybridization of the oligonucleotides S60-sp-Pme and A60-sp-Pme and treatment with DNA polymerase in the presence of high NTP concentrations. The double-stranded DNA fragment was restricted with PmeI and inserted into the various luc replicon constructs described above. This construction resulted in the insertion of a total of 63 nucleotides between the 5′ NTR sequences of HCV and the PV IRES. Plasmids carrying the neo replicons and the chimeric 5′ NTRs were obtained by insertion of a PmeI-ApaI fragment isolated from pFKnt341-sp-PI-lucEI3420-9605/5.1 and an ApaI/NotI fragment isolated from pBSK-PI-neo (Lohmann and Bartenschlager, unpublished) into each of the luc replicon constructs described above. Constructs with deletions in stem-loop 2 of HCV were generated by PCR using primers S T7nt1-SbfI and A125Δ61-104-Pme or A125Δ75-91-Pme. After restriction of the purified fragments with SbfI and PmeI, they were inserted into pFKnt341-sp-PI-luc-EI3420-9605/5.1 or pFKnt341-sp-PI-neo-EI3420-9605/5.1.
Sequences were verified using a Thermo Sequenase Fluorescent Labeled Primer Cycle Sequencing Kit with 7-deaza-dGTP (Amersham-Pharmacia Biotech, Freiburg, Germany) and IRD-41 labeled primers (MWG-Biotech, Ebersberg, Germany) by following the instructions of the manufacturer. Reaction mixtures were analyzed on a model 4000 Licor DNA sequencer (MWG-Biotech).
These methods have been described in detail elsewhere (27). In brief, plasmid DNA was restricted with AseI and ScaI (New England Biolabs, Bad Schwalbach/Taunus, Germany) and after purification by phenol extraction and ethanol precipitation was dissolved in RNase-free water. For an in vitro transcription reaction, 5 μg of restricted plasmid DNA was added to a buffer containing 80 mM HEPES, pH 7.5, 12 mM MgCl2, 2 mM spermidine, 40 mM dithiothreitol (DTT), 3.125 mM of each NTP, 50 U of RNasin (Promega, Mannheim, Germany), and 30 U of T7 RNA polymerase (Promega) in a total volume of 50 μl. After 2 h at 37°C, 15 U of T7 RNA polymerase was added, and the reaction mixture was incubated for another 2 h. Transcription was terminated by the addition of 6 U of RNase-free DNase (Promega) and 30 min of incubation at 37°C. DNA was extracted with acidic phenol and chloroform, precipitated with isopropanol, and dissolved in RNase-free water. The RNA concentration was determined by measurement of the optical density at 260 nm, and the integrity was checked by denaturing agarose gel electrophoresis. For the electroporation of selectable replicons, 0.3 to 100 ng of in vitro transcript adjusted with total RNA from naïve Huh-7 cells to a final amount of about 9 μg was mixed with 400 μl of a suspension of 107 Huh-7 cells per ml in a cuvette with a gap width of 0.4 cm (Bio-Rad, Munich, Germany). After one pulse at 960 μF and 270 V with a Gene pulser system (Bio-Rad), cells were immediately transferred to 8 ml of complete Dulbecco minimal essential medium (DMEM) and seeded in a 10-cm-diameter culture dish. After 24 h, medium was replaced by complete DMEM supplemented with 500 μg of G418 (Geneticin; Life Technologies)/ml. Three to 4 weeks later, colonies were stained with Coomassie brilliant blue (0.6 g/liter in 50% methanol, 10% acetic acid). To determine the efficiency of colony formation of a given construct, serial dilutions of in vitro transcripts were transfected in parallel. Representative results of multiple independent transfections are shown.
Huh-7 cells were transfected by electroporation as described above using 5 μg of a replicon carrying the firefly luciferase gene. After addition of 9 ml of complete DMEM, aliquots of the cell suspension were seeded in 3-cm-diameter culture dishes and harvested at various time points. To determine the luciferase activity, cells were washed three times with phosphate-buffered saline, scraped off the plate into 350 μl of ice-cold lysis buffer (1% Triton X-100, 25 mM glycylglycine, 15 mM MgSO4, 4 mM EGTA, 1 mM DTT). Then, 100 μl of lysate was mixed with 360 μl of assay buffer (25 mM glycylglycine, 15 mM MgSO4, 4 mM EGTA, 1 mM DTT, 2 mM ATP, 15 mM K2PO4, pH 7.8) and, after addition of 200 μl of a 200 μM luciferin stock solution, measured in a luminometer (Lumat LB9507 from Berthold, Freiburg, Germany) for 20 s. Values obtained with cells harvested 4 h after electroporation were used to correct for the transfection efficiency.
Extracts from HeLa cells were prepared as described elsewhere (32). The preparation of Huh-7 cell extracts was performed in the same way, with minor modifications. Prior to harvest, cells were washed twice with a buffer containing 140 mM NaCl, 5 mM KCl, 10 mM HEPES (pH 7.4), and 1 g of glucose/liter, detached from the culture dish by treatment with trypsin, and washed three times. All further steps were as described for HeLa cell extracts (32). Rabbit reticulocyte lysates were purchased from Promega. For in vitro translation in HeLa cell extracts, 4.5 μl in vitro-transcribed RNA (corresponding to 0.5 μg) was mixed with 8 μl of master mix (75 μl of HeLa cell extract, 11 μl of 2 M KAc, 1.5 μl of 50 mM MgAc, 6 μl of 30 mM MgCl2, 15 μl of 35S-protein labeling mixture [NEN Life Science, Köln, Germany], and 25 μl of translation buffer that was obtained by mixing the following components: 40 μl of 100 mM ATP, 6 μl of 40 mM GTP, 40 μl of 1 M creatine phosphate (Sigma), 10 μl of creatine phosphokinase (10 mg/ml; Sigma), 76 μl of 1 M HEPES buffer (pH 7.6), 80 μl of 100 mM DTT, 20 μl of calf liver tRNA (5 mg/ml; Roche Molecular Biochemicals, Mannheim, Germany), 50 μl of 1 mM amino acid mixture without methionine (Promega), 10 μl of 100 mM spermidine (Sigma), and 168 μl of RNase-free water. After incubation for 14 to 16 h at 30°C, reactions were terminated by the addition of protein sample buffer (200 mM Tris-HCl, pH 8.0, 5 mM EDTA, 3.3% sodium dodecyl sulfate, 2% 2-mercaptoethanol, 10% sucrose, and 0.1% bromophenol blue). For in vitro translations in cell extracts of Huh-7 cells, 0.5 μl of RNA (corresponding to 0.5 μg) was mixed with 16.5 μl of extract, 3.5 μl of translation buffer, 1 μl of 35S-protein labeling mixture, 0.5 μl of RNasin (40 U/μl), and 3 μl of salt mix (900 mM KAc, 7.3 mM MgCl2, 3.3 mM Mg-acetate). After 2 h at 30°C, translations were terminated by the addition of protein sample buffer. In vitro translation reaction mixtures in rabbit reticulocyte lysates were composed of 8.75 μl of lysate, 0.25 μl of RNasin, 0.25 μl of amino acid mixture without methionine, 1.5 μl of 35S-protein labeling mixture, and 1.75 μl of RNA (corresponding to 0.5 μg). Reactions were terminated after 1 h at 30°C as described above. Proteins in all translation reaction mixtures were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and HCV-specific bands were quantified by phosphorimaging.
Five micrograms of in vitro-transcribed luciferase replicon RNA was mixed with 15 μg of an EMCV-IRES-lacZ in vitro transcript and used for electroporation of Huh-7 cells as described above. Cells were seeded in 3-cm-diameter culture dishes, and duplicates were harvested 4 h after transfection. Lysates were used for the determination of luciferase activities as described above. To correct for different transfection efficiencies, beta-galactosidase (β-Gal) activities were determined in the same lysates as follows: 30 μl of cell lysate was mixed with 3 μl of magnesium solution (100 mM MgCl2, 4.5 M 2-mercaptoethanol), 66 μl of ONPG solution (4 mg of o-nitrophenyl-β-d-galactopyranoside [Sigma]/ml dissolved in Na-phosphate buffer, pH 7.5), and 201 μl of Na-phosphate, pH 7.5. After a 5-h incubation at 37°C, reaction mixtures were centrifuged for 10 min at 13,000 × g and optical densities of cleared supernatants were measured at 420 nm.
The design of the bicistronic replicon RNAs used in this study is shown in Fig. Fig.1.1. The 5′ NTR of HCV was fused either to the neo gene or to the gene encoding the firefly luciferase (luc) that were expressed as fusion proteins carrying 16 amino acids of the core protein at their amino terminus. The EMCV IRES was used to direct translation of the HCV NS3-5B region into which three cell culture-adaptive mutations were introduced. These are located in NS3 (E1202G and T1280I) and in NS5A (S2197P), and they enhance RNA replication synergistically (24). Upon transfection of Huh-7 cells with the neo replicon, RNA replication could be measured by determining the number of G418-resistant colonies obtained after selection with Geneticin. In the case of the luc replicon, RNA replication was analyzed by measuring the luciferase activity at various time points after transfection in comparison to cells transfected with a replication-defective RNA. We have shown that both the number of G418-resistant colonies obtained with a given replicon and the luciferase activity determined 24 to 96 h posttransfection are a direct correlate of the efficiency with which a replicon multiplies in cells (24). The advantage of the neo replicons was the higher sensitivity of detection because 1 μg of the parental replicon rep 5.1 yielded ~5 × 105 colonies and, therefore, mutants with a low capability of replication still could be detected. The advantage of the luc replicons was the determination of RNA replication in a transient assay and, therefore, a rapid test of various mutants. It should be noted that both assays yield congruent results (24).
When studying the importance of sequences in the 5′ NTR of HCV for RNA replication, we had to consider that mutations introduced into this region might influence IRES activity. Since this would affect the expression levels of the selection marker (neo) or the reporter gene (luc), we first analyzed all mutants for RNA translation by using both in vitro translations in cell extracts and after transient expression in Huh-7 cells. Since these assays were always performed with the replicon, we were able to analyze translation in the context of a replication-competent HCV RNA molecule.
A number of studies have shown that the first ~40 nucleotides of the 5′ NTR are not required for IRES function. Based on this observation and the fact that sequences at the 5′ end of a viral RNA genome are very important determinants for replication, we anticipated that the same might be true for HCV. The region upstream of the IRES contains stem-loop 1 that is formed by nucleotides 5 to 20 and that is separated from stem-loop 2 by a 23-nucleotide-long spacer (Fig. (Fig.1).1). To analyze the importance of these elements for RNA replication, two constructs were generated in which either the spacer or stem-loop 1 were deleted (Fig (Fig2A).2A). We did not attempt to produce a replicon that lacks the very first nucleotides because this mutation was expected to be lethal (11). To exclude the possibility that these deletions affected IRES function, the neo replicon RNAs were first used for in vitro translations in cell extracts from different sources: rabbit reticulocytes, HeLa cells, and Huh-7 cells, which are the only ones that support replication of these RNAs. Owing to the bicistronic design of the replicons, translation efficiency of the 5′ (neo) cistron that was under control of the HCV IRES could be determined independently from the translation efficiency of the downstream cistron (NS3-5B) that was under control of the EMCV IRES. Translation from the latter was assumed to be unaffected by mutations in the 5′ region of the molecule, and therefore, NS3 served as an internal control for the quality of the translation reaction. As shown by the representative result in Fig. Fig.2B,2B, within a given cell extract, the ratios of the amounts of the core-NPT fusion protein and NS3 were similar in all cases, irrespective of the overall expression levels that were highest in rabbit reticulocyte lysates. The analogous result was found when HeLa cell extracts were programmed with the corresponding luc replicons or with replicon RNAs in which all heterologous sequences were deleted (not shown). The latter were obtained by fusing the core coding sequence via a ubiquitin cleavage site in frame to NS3. These results suggested that the translation efficiency was not affected by the deletions and independent of the reporter gene. Moreover, the findings were in agreement with previous observations showing that the HCV IRES is active in a multitude of host cells (10). It should be noted that apart from some unprocessed precursors, only NS3 was clearly visible in the in vitro translations. This was probably due to the absence of microsomal membranes that are required for efficient polyprotein cleavage in vitro or for stabilization of the cleavage products (3). However, the appearance of NS3 was sufficient to determine the translation efficiency from the EMCV IRES.
Next, we wanted to know whether the deletions in the 5′ NTR had an effect on RNA translation within a cell. Luc replicons were transfected into Huh-7 cells, and luciferase activity was determined 4 h after transfection. At this time point, only translation from the input but not from progeny RNA was measured because the amount of newly synthesized replicon was too low to exceed the amount of RNA transfected into the cell (24). Variations in transfection efficiencies were determined by cotransfection of an in vitro transcript that encoded the β-Gal gene under control of the EMCV IRES. Luciferase and β-Gal activities were measured in the same cell lysates. As shown in Fig. Fig.2C,2C, luciferase activities of the mutants were reduced two- to threefold compared to the parental luc replicon. This result was surprising since it was not found with the in vitro systems, indicating that the conditions of RNA translation in the cell lysates did not exactly mimic those in transfected cells. In summary, these data suggested that the deletions in the 5′ NTR only moderately affected intracellular HCV IRES activity.
To analyze whether these mutations had an effect on RNA replication, the two mutants were transfected into Huh-7 cells in parallel with the parental replicon and an RNA that, due to an amino acid substitution in the active site of the NS5B RdRp (the GDD motif), could not replicate (389/GND). Luciferase activities were measured 4, 24, 48, and 72 h posttransfection and were corrected for transfection efficiencies as determined with the 4-h value. The results in Fig. Fig.3A3A show that 24 h after transfection, the parental RNA with the full-length 5′ NTR yielded high luciferase activities demonstrating transient replication in the transfected cells. In contrast, only background levels were found with the inactive replicon, and the same was true with the two mutants Δ5-20 and Δ24-40, suggesting that the deletions in the 5′ NTR impaired RNA replication. This conclusion was confirmed by the results obtained with the analogous neo replicons. After transfection of Huh-7 cells and G418 selection, no colonies were obtained with the two mutants, whereas the parental replicon (389) yielded ~500,000 CFU per μg of in vitro transcript. These data show that the sequences between nucleotides 5 and 40 are essential for HCV RNA replication.
To map the minimal sequence in the 5′ NTR required for RNA replication, we constructed a series of chimeric replicons in which various portions of the HCV 5′ NTR were fused to a heterologous IRES (Fig. (Fig.4A).4A). This approach allowed the analysis of HCV replication signals without affecting RNA translation. We chose the IRES of PV for several reasons: (i) it is a well-characterized element that does not require coding sequences for activity, (ii) we found that this IRES was highly active in Huh-7 cells (data not shown), and (iii) both in structural and in functional terms, the HCV IRES is very different from the PV IRES and, therefore, it should not contain sequences or structures that could compensate for deleted sequences of the HCV 5′ NTR. Because of the similarities of the structures of the HCV and the BVDV IRES, we did not attempt to use the latter as a functional substitute of the HCV translation signal. Moreover, full activity of the BVDV IRES requires part of the Npro coding region, presumably to prevent the formation of stable RNA structures downstream of the initiator AUG (7, 12, 34). The EMCV IRES was not chosen because it was already present in the replicon RNA and we wanted to avoid a duplication of such a long sequence. In the first set of experiments, various portions of the 5′ end of the HCV genome were directly fused to the PV IRES in the context of the neo and luc replicons (see Materials and Methods). Translation of the resulting RNAs carrying the first 43, 125, 296, or 341 nucleotides of the HCV genome at the 5′ end was analyzed both by in vitro translation and in transfected cells. We found that expression of the luciferase gene was severely impaired in transfected cells when the full-length HCV 5′ NTR was directly fused to the PV IRES, whereas no reduction of luc translation was observed with the other 5′ variants (data not shown). This result indicated an interference between the HCV and the PV RNA elements similar to what was recently found with a recombinant PV that carries an HCV IRES to direct translation of the PV polyprotein (56).
Given the possibility of an inhibitory interaction between the HCV and the PV IRES, we reasoned that the introduction of a spacer element of random sequence might separate these two elements sufficiently to avoid an interference. Therefore, a series of replicons with chimeric 5′ NTRs was generated, in which the HCV sequences were separated from the PV IRES by a 63-nucleotide-long spacer (Fig. (Fig.4;4; see Materials and Methods). Translation of the resulting replicons was first analyzed in HeLa cell extracts that supported high activity of the PV IRES and yielded a low background (32). We found that translation of both cistrons (neo and NS3-5B) was comparable between the various replicons (Fig. (Fig.5A),5A), and the analogous observations were made with the corresponding replicons carrying the luciferase gene instead of neo (data not shown). When these luc replicons were transfected into Huh-7 cells and luciferase activities were determined 4 h after transfection, an about twofold reduction was found with the constructs carrying a truncated HCV 5′ NTR (Fig. (Fig.5B).5B). As described above, at this time point posttransfection, the measurements were not influenced by RNA replication. This can best be seen by the comparable luciferase activities in cells transfected with either the parental replicon (341-sp/wt) or the replicon with the defective NS5B RdRp (341-sp/GND). Consistent with the previous findings, luciferase activity of the analogous replicon in which the HCV and PV IRESs were directly fused was ~15-fold lower (341/GND). Thus, the insertion of the spacer between both RNA elements restored translation activity of the PV IRES.
Having confirmed the functionality of the translation signals in the various chimeras, we next analyzed the replication of these RNAs by using the transient assay described above. The results in Fig. Fig.6A6A demonstrate that the luc replicons carrying 84 or fewer nucleotides of the HCV genome at the 5′ end did not replicate (data for replicons with fewer than 43 nucleotides are not shown). Luciferase activities found with these RNAs were as low as the ones determined with the defective replicon carrying an inactive NS5B RdRp (341-sp/GND). However, a low level of replication was found with the RNA carrying the 125 5′-terminal nucleotides of the HCV genome at its 5′ end. A further increase of the length of the HCV portion led to an increase of replication which was at maximum when the full-length HCV 5′ NTR was placed at the 5′ end of the replicon. The analogous result was found with the neo replicons. After transfection of Huh-7 cells and G418 selection, no colonies were obtained with replicons carrying 84 or fewer nucleotides of the HCV 5′ NTR at their 5′ end. Some colonies were obtained with the 125-sp-PVI replicon, but the efficiency of colony formation of this RNA was only 9,000 ± 2,000 (mean ± standard deviation) CFU/μg in vitro transcript. This number was ~20-fold-higher with the 296-sp-PVI replicon, and the highest efficiency was found with the RNA carrying the full-length HCV 5′ NTR at its 5′ end (~106 CFU/μg). These data demonstrate that the first 125 nucleotides comprising stem-loops 1 and 2 are sufficient for HCV RNA replication, but efficiency is significantly enhanced in the presence of the full-length 5′ NTR.
To further narrow down the replication signals, two deletions were generated. Replicon Δ72-96-sp-PVI lacked the two apical loops of stem-loop 2 whereas RNA Δ61-104-sp-PVI lacked the three upper loops (Fig. (Fig.4).4). These RNAs were translated both in HeLa cell extracts and in transfected Huh-7 cells as efficiently as all other replicons with the chimeric HCV-PV 5′ NTR (Fig. (Fig.55 and data not shown). However, when analyzed for replication both in the transient assay and by selection for stable replicon-harboring cell lines, the replicon RNAs turned out to be defective (Fig. (Fig.6).6). These results suggest that an intact stem-loop 2 is required for HCV replication.
Signals required for replication of plus-strand RNA viruses usually are located in the 5′ terminal regions of the template strands, and they act as promoter elements for initiation of minus- and plus-strand RNA synthesis. In order to map the sequences required for HCV RNA replication, we took advantage of the replicon system that allows high-level self-replication of subgenomic replicons in the human hepatoma cell line Huh-7 (28). By using both selectable replicons and a novel transient-replication system that we developed recently (24), we showed that the 5′-terminal 125 nucleotides of the HCV genome are sufficient for RNA replication, and we mapped the 3′ border of the minimal replication signal between nucleotides 84 and 125. However, replication was significantly improved upon the addition of further sequences, and maximum replication levels in both assays were obtained with the full-length 5′ NTR positioned at the 5′ end of the HCV RNAs. This result is different from what has been described for some other plus-strand RNA viruses. For instance, for PV, it was shown that the 5′ NTR is composed of two elements: a 5′ terminal cloverleaf-like structure spanning nucleotides 1 to 88 that is essential for RNA replication and the IRES spanning at least residues 134 to 556 of the PV genome (for review, see reference 54). Replacement of the PV IRES by the HCV IRES results in viable chimeras in which RNA replication is controlled by the PV 5′-terminal element, whereas translation is mediated by the HCV IRES (29, 57, 58). Thus, sequences of the PV IRES appear to be dispensable for PV RNA replication. However, in agreement with what we describe here for HCV, several recent studies indicate that sequences within the IRES of PV are also involved in RNA replication, suggesting that the signals for translation and replication overlap (9, 21, 47).
In spite of the lack of substantial primary sequence homology, the 5′ NTRs of HCV and pestiviruses like BVDV share several structural and functional features. Both contain an IRES that is located at the 3′ end of the 5′ NTR and that is composed of three stem-loop structures. The initiator AUG of the polyprotein coding region resides in the loop of stem-loop 4 and a pseudoknot structure is formed by sequences at the 3′ end of stem-loop 3. One important difference is the presence of two stem-loops (Ia and Ib) upstream of the pestiviral IRES, whereas there is only one such structure in case of HCV. However, for neither BVDV nor HCV are these stem-loops required for IRES activity (12, 20, 43). By constructing a series of BVD viruses with chimeric BVDV/HCV or BVDV/EMCV 5′ NTRs in which the BVDV IRES was replaced by the one of HCV or EMCV, Frolov and coworkers (13) made the surprising observation that the addition of the tetranucleotide sequence GUAU to the HCV or EMCV IRES was sufficient for efficient BVDV replication in cell culture. Interestingly, chimeric BVD viruses with Ia and Ib upstream of the HCV IRES replicated almost as efficient as the parental BVDV, whereas chimeras with only Ia were unstable. Upon passage of such viruses, pseudorevertants arose in which most of Ia was deleted in a way that the tetranucleotide sequence GUAU was restored (13). Assuming that the sequence at the 3′ end of minus-strand RNA acts as the promoter for initiation of plus-strand RNA synthesis, this observation implies that the BVDV replicase complex can recognize the minus-strand complement of the HCV or EMCV IRES to initiate plus-strand RNA synthesis and/or that only the first 4 nucleotides direct the specificity of this process. Alternatively, sequences within the BVDV genome, i.e., downstream of the inserted heterologous IRES sequence may play an important role for RNA replication. As described in this report, the requirements for RNA sequences or structures within the 5′ NTR of HCV appear to be more complex because much longer sequences or particular structures within the IRES were necessary for efficient RNA replication. Chimeras carrying only the first 5 nucleotides of HCV (or even the first 84) fused to the PV IRES did not replicate to a level detectable in our assays. This effect was not due to the particular PV IRES or the reporter genes (neo and luciferase) we utilized, because the same results were found with replicons that lacked these sequences. In this case, replication was measured in transient assays with monocistronic replicons in which various portions of the HCV 5′ NTR were fused via the spacer sequence to the EMCV IRES that directed translation of the NS3-5B replicase (P. Friebe and R. Bartenschlager, unpublished data). These results imply that the signals required for efficient BVDV replication are simpler than those required by HCV. However, differences in the experimental approaches used in this study and by Frolov and colleagues (13) might also account for the discrepant results. For instance, we had to utilize subgenomic replicons, whereas the work with BVDV was performed with full-length infectious genomes. Moreover, we cannot exclude that the fusion of the HCV 5′ NTR sequences to the PV IRES (or the EMCV IRES in case of the monocistronic RNAs) affected the structure or function of the HCV replication signal. In this case, the HCV sequences we identified to be essential for RNA replication would primarily be required for the preservation of the structural and functional integrity of the minimal replication signal rather than being a direct part of itself. Further studies will be required to clarify these issues.
Our results that deletions in the 5′ NTR upstream of the HCV IRES reduce RNA translation appear to contradict the observations made with HCV or pestiviral reporter gene assay-based systems (reviewed in reference 44). For instance, in the case of BVDV, it was found that the 5′ terminal stem-loops Ia and Ib are completely dispensable for IRES activity (12). In contrast, when analyzed in the context of a self-replicating subgenomic BVDV RNA, mutations in the 5′ terminal stem-loop Ia affected both RNA translation and replication (55). For instance, translation in BHK-21 cell extracts was reduced by more than 10-fold when Ia was deleted or replaced by the 5′ terminal stem-loop 1 of HCV and these RNAs no longer replicated. However, when a deletion of Ia either alone or together with a deletion of part of Ib was introduced into an infectious full-length BVDV genome, these RNAs still produced infectious virus, although the specific infectivities of the corresponding in vitro transcripts were low (5). Similar to the results obtained with the subgenomic BVDV RNAs, we also observed a reduction of RNA translation with replicons carrying deletions of stem loop 1 or the spacer sequence, although the effect was less drastic (only two- to threefold reduction). Our results therefore suggest that analysis of the HCV IRES in the context of a self-replicating molecule is important to identify possible effects of HCV proteins or sequences outside of the IRES on RNA translation.
The reduction of translation observed with replicons Δ5-20 and Δ24-40 that lack either stem-loop 1 or the spacer sequence does not explain the block of replication. First, owing to the bicistronic design of the replicons, translation of the HCV NS3-5B replicase was directed by the EMCV IRES that was not affected by the manipulations at the 5′ end of the replicons. Second, the reduction of RNA translation was at maximum threefold (Δ24-40). Although even modest effects on translation activity may result in a significantly greater cumulative reduction of RNA replication, some of the replicons with a twofold-lower translation activity still replicated efficiently (for instance, 296-sp-PVI). Thus, the deletions introduced into the first 40 nucleotides of the HCV 5′ NTR primarily affected RNA replication and not translation.
In summary, we have performed the first characterization of sequences in the 5′ NTR of HCV that are required for RNA replication. We found that stem-loops 1 and 2 play an important role suggesting that this region constitutes or is part of the minimal promoter for initiation of minus-strand RNA synthesis. By generating chimeric 5′ NTRs, we were able to genetically uncouple the signals in the HCV genome required for RNA replication and translation. These results provide the first map of HCV replication signals, and they will guide our further studies in this important area.
We gratefully acknowledge the gift of the PV cDNA and HeLa S10 cell extracts from Aniko Paul and Eckard Wimmer and the provision of Huh-7 cell extracts from Mathias Beck and Michael Nassal. We also thank Ulrike Herian for excellent technical assistance and Thomas Pietschmann for critical reading of the manuscript.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB490, Teilprojekt A2) and the European Community (QLK2-1999-00356).