Hepatitis C Virus Genotype 5a Subgenomic Replicons for Evaluation of Direct-Acting Antiviral Agents

Abstract

Hepatitis C virus (HCV) exists as six major genotypes that differ in geographical distribution, pathogenesis, and response to antiviral therapy. In vitro replication systems for all HCV genotypes except genotype 5 have been reported. In this study, we recovered genotype 5a full-length genomes from four infected voluntary blood donors in South Africa and established a G418-selectable subgenomic replicon system using one of these strains. The replicon derived from the wild-type sequence failed to replicate in Huh-7.5 cells. However, the inclusion of the S2205I amino acid substitution, a cell culture-adaptive change originally described for a genotype 1b replicon, resulted in a small number of G418-resistant cell colonies. HCV RNA replication in these cells was confirmed by quantification of viral RNA and detection of the nonstructural protein NS5A. Sequence analysis of the viral RNAs isolated from multiple independent cell clones revealed the presence of several nonsynonymous mutations, which were localized mainly in the NS3 protein. These mutations, when introduced back into the parental backbone, significantly increased colony formation. To facilitate convenient monitoring of HCV RNA replication levels, the mutant with the highest replication level was further modified to express a fusion protein of firefly luciferase and neomycin phosphotransferase. Using such replicons from genotypes 1a, 1b, 2a, 3a, 4a, and 5a, we compared the effects of various HCV inhibitors on their replication. In conclusion, we have established an in vitro replication system for HCV genotype 5a, which will be useful for the development of pan-genotype anti-HCV compounds.

INTRODUCTION

Hepatitis C virus (HCV) currently infects approximately 185 million people worldwide, increasing their risk of developing liver cirrhosis and hepatocellular carcinoma (1). No vaccine is available against HCV infection, and the standard of care until 2011, consisting of PEGylated interferon (IFN) and ribavirin, cured only 50% of patients. However, the addition of direct-acting antiviral agents (DAAs) over the past few years has revolutionized HCV treatment, with cure rates now approaching 80 to 90% (2–5). The recently published results of phase 3 clinical trials report even better regimens with >95% cure rates (6, 7). These are exciting developments, yet the issues of drug resistance, side effects, and drug-drug interactions will remain a challenge. Therefore, the quest to find safe drugs with a pan-genotype activity and a high barrier to drug resistance will continue. To aid these efforts, it is important to have cell culture replication systems for all HCV genotypes. Of the six major HCV genotypes, replication systems for HCV genotypes 1 and 2 were developed over a decade ago (8–10) and provided a strong foundation for the development of currently approved antivirals. Recently, replication systems for genotypes 3, 4, and 6 have also been reported (11–14). However, a similar system for genotype 5 is still lacking. This genotype is restricted mainly to South Africa (15) and has been associated with sporadic cases of HCV infection in Canada, Brazil, The Netherlands, Spain, and Belgium (16–18). In 2004, a high prevalence of genotype 5a was reported in central France (19). These findings highlight the importance of developing an in vitro replication system for this genotype.

HCV is an enveloped, positive-stranded RNA virus in the family Flaviviridae. The viral genome is ∼9.6 kb long, with 5′ and 3′ untranslated regions (UTRs) flanking an open reading frame (ORF) of approximately 9,000 nucleotides (nt). This ORF encodes a polyprotein of ∼3,000 amino acids that is co- and posttranslationally cleaved by host and viral proteases to yield 3 structural and 7 nonstructural (NS) proteins (20). The observation that the nonstructural proteins were sufficient for replication of a closely related bovine viral diarrhea virus (21) laid the foundation for the development of HCV subgenomic replicons. These replicons comprise three distinct elements: (i) an HCV internal ribosomal entry site (IRES), containing the 5′ UTR and the first 12 to 19 amino acids from the capsid protein, fused with a drug-selectable marker, neomycin phosphotransferase II (NPTII), which upon expression confers resistance to G418, (ii) an IRES from encephalomyocarditis virus (EMCV), which drives expression of the HCV nonstructural proteins, and (iii) the HCV 3′ UTR. When RNA transcripts from these replicons are introduced into Huh-7.5 cells by transfection or electroporation, followed by selection with G418, cell colonies harboring autonomously replicating viral genomes are obtained that can then be used to study various aspects of HCV replication. In this report, we generated the first subgenomic replicon for genotype 5a, using a strain isolated from the plasma of an infected blood donor. We identified novel cell culture-adaptive mutations that significantly increased the viral RNA replication levels. Furthermore, we used this newly generated replicon to measure the responsiveness of genotype 5a to various anti-HCV compounds.

MATERIALS AND METHODS

Cells, antibodies, and chemicals.

Huh-7.5 cells were maintained in Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% fetal bovine serum and 0.1 mM nonessential amino acids (NEAA), at 37°C in a humidified atmosphere containing 5% CO₂. G418 (Sigma, MO) was added to the medium at a concentration of 750 μg/ml to select and maintain cells carrying HCV replicons. Anti-NS5A antibody 9E10 has been described previously (22). Interferon alpha 2a (IFN-α2a), ledipasvir, and sofosbuvir were purchased from PBL Assay Science (NJ, USA), MedChem Express (NJ, USA), and Acme Bioscience (CA, USA), respectively. Danoprevir (RG7227/ITMN-191) and daclatasvir were purchased from Selleck Chemicals (TX, USA).

Human subjects.

The samples from volunteer blood donors collected by the South African National Blood Services (SANBS) were anonymized and used for research purposes only after clearance from the ethics committee of the SANBS. Experiments were performed with authorization from the Institutional Review Board at the Rockefeller University.

Amplification of viral genomes.

Plasma samples from four HCV genotype 5a-infected blood donors were used to isolate HCV RNA. Briefly, total RNA was extracted from 140 μl of plasma using a QIAamp viral RNA minikit (Life Technologies, CA) by following the manufacturer's protocol. An aliquot of 5 μl was subjected to cDNA synthesis using Superscript III reverse transcriptase (Life Technologies) and random hexamers. The reaction was carried out at 50°C for 50 min, followed by enzyme inactivation at 80°C for 15 min. PCR primers to amplify the cDNA were designed using sequences of two genotype 5a isolates, SA13 and EUH1480 (GenBank accession no. AF064490 and NC_009826), available in the HCV database (http://hcv.lanl.gov). For the sequences of these primers, see Table S1 in the supplemental material. Using these primers, the complete HCV ORF and partial UTRs were amplified into 8 overlapping fragments (see Table S1, fragments 2 to 9) by nested PCR using Phusion High Fidelity DNA polymerase (New England BioLabs, MA). The same conditions were used for both rounds of PCR: initial denaturation at 98°C for 5 min, 35 cycles of PCR (98°C for 30 s, 55°C for 30 s, 72°C for 2 min), and a final elongation step at 72°C for 10 min. The 5′ rapid amplification of cDNA ends (RACE) to amplify the terminal 5′ end of the viral genome was performed exactly as described by Saeed et al. (11). The terminal 3′-end sequence was determined by 3′ RACE. Briefly, RNA extracted from 250 μl of plasma samples, using TRIzol LS reagent (Life Technologies), was tailed at the 3′ end with GTP using yeast poly(A) polymerase (Affymetrix, CA). The tailed RNA was then reverse transcribed into cDNA using SuperScript III reverse transcriptase and a 27-mer reverse primer containing a stretch of eight cytidines at the 3′ end. The resulting cDNA was immediately amplified by nested PCR using Ex Taq DNA polymerase (Toyobo, Japan) and the primers described in Table S1. The PCR conditions consisted of an initial denaturation step at 95°C for 2 min, 35 cycles of PCR (95°C for 30 s, 58°C for 30 s, 72°C for 2 min), and a final elongation step at 72°C for 10 min.

Determination of HCV consensus sequences.

The PCR amplicons were cloned into the TOPO TA vector (Life Technologies), followed by transformation of DH5α. Plasmid DNA purified from 5 to 10 bacterial colonies for each amplicon was subjected to sequence analysis, and the consensus sequence was determined by adopting the most abundant nucleotide at each position. The strains recovered from four plasma samples were tentatively named SA1, SA2, SA3, and SA4. The subgenomic regions from the isolated strains were aligned using MacVector, and their percent identity was calculated. Furthermore, the polyprotein regions from the newly sequenced genotype 5a isolates and representative isolates from all major HCV genotypes available in the database were subjected to phylogenetic analysis using the neighbor-joining method.

Construction of subgenomic replicons.

Based on the consensus sequence of SA1, we assembled SA1/SG-neo. Briefly, clones covering the NS3-3′ UTR region of the viral genome were pieced together by overlapping PCR or by use of restriction enzymes. Since finding individual clones for each amplicon that were similar to the consensus sequence at all nucleotide positions proved difficult, the clones deviating from the consensus sequence at the nucleotide level, but not at the amino acid level, were sometimes selected to ligate to the neighboring fragment. This resulted in the inclusion of a few silent mutations in the region of the replicon encoding NS3-NS5B proteins. A cassette containing the NPTII gene, followed by an EMCV IRES, was cloned upstream of NS3. This construct was then ligated at its 5′ end with the IRES from SA1, consisting of the 5′ UTR and the first 19 codons from the capsid protein. To facilitate in vitro transcription, a T7 promoter was cloned upstream of the HCV IRES and an XbaI runoff site was introduced downstream of the 3′ UTR. The NPTII gene of SA1/SG-neo was replaced with a chimeric gene encoding firefly luciferase protein fused in-frame with NPTII to synthesize SA1/SG-Feo, as described elsewhere (12, 23).

RNA synthesis and transfection of cultured cells.

Plasmids linearized with XbaI and purified with a MinElute PCR purification kit (Qiagen Sciences, MD) were in vitro transcribed using a T7 RiboMAX Express large-scale RNA production system (Promega, WI). To ensure complete removal of template DNA, an additional on-column treatment with RNase-free DNase was performed at room temperature (RT) for 15 min. The quality of in vitro-transcribed RNA was assessed by agarose gel electrophoresis. Huh-7.5 cells were electroporated with RNA transcripts using a BTX ElectroSquarePorator as described previously (12). To obtain cell colonies carrying autonomously replicating HCV genomes, electroporated cells were exposed to G418 (final concentration of 750 μg/ml) at 48 h postelectroporation, and medium was replaced every third day with a fresh medium containing 750 μg/ml G418. Three weeks postelectroporation, G418-resistant cell colonies were either isolated and expanded for downstream analysis or fixed with 7% formaldehyde and stained with crystal violet.

Colony titration assay.

Huh-7.5 cells, seeded in 6-well plates at a density of 400,000 cells/well, were grown overnight. Cells were transfected with 2 μg of in vitro-transcribed RNA using a TransIT-mRNA transfection kit (Mirus Bio, WI), by following the manufacturer's recommendations. Six hours later, cells from each well were harvested and plated in new 6-well plates at multiple densities (between 2 × 10² and 2 × 10⁵ cells/well). The total numbers of cells in each well were maintained at equal levels by using cells transfected with RNA transcripts from replication-defective HCV genomes. G418 selection was performed as described above. Three weeks posttransfection, colonies were stained with crystal violet and quantified manually. The colony formation efficiency was determined by calculating the percentage of transfected cells that survived after selection.

Analysis of HCV RNA replication in replicon cells.

The isolation of colonies, RNA quantification, and flow cytometry were performed exactly as described previously (12).

Feo replicons from HCV genotypes 1a, 1b, 2a, 3a, and 4a.

The genotype 2a replicon, JFH1/SG-Feo, and the genotype 3a replicon, S52/SG-Feo(AII), have been previously described (12). The genotype 4a replicon, ED43/SG-Feo(VYG), was made by introducing a newly identified mutation, leading to an M1205V substitution in the NS3 protein of ED43/SG-Feo(Y), as described by Saeed et al. (12). It contains three cell culture-adaptive substitutions: M1205V (NS3), D1431Y (NS3), and R2882G (NS5B). This replicon replicates to higher levels than ED43/SG-Feo(Y) (unpublished results). The genotype 1a replicon, H77/SG-Feo(L+8), was generated from H77/SG-neo(L+8) (24). It contains four adaptive substitutions: P1496L (NS3), V1880A (NS4B), A1940V (NS4B), and C1968R (NS4B). The genotype 1b replicon, Con1/SG-neo(I) (8), was modified to generate Con1/SG-Feo(I). This replicon contains an S2204I substitution in the NS5A protein.

HCV inhibitor assay.

HCV Feo replicons were used to measure the effect of various inhibitors on HCV RNA replication. Huh-7.5 cells carrying these replicons were seeded in 96-well plates with clear bottoms (PerkinElmer, MA) at a density of 1 × 10⁴ cells/well. The following day, the cells were exposed to various concentrations of compounds and incubated for 72 h. Cell lysates were prepared by using 1× cell culture lysis reagent (CCLR; Promega), and firefly luciferase expression was measured with a luciferase assay system (Promega) and a Synergy NEO HTS multi-mode microplate reader (BioTek). The 50% effective concentrations (EC₅₀) relative to that for the untreated cells were determined using GraphPad Prism software (San Diego, CA). Replicon cell viability in the presence of inhibitors was tested in parallel using CellTiter-Glo assay (Promega), according to the manufacturer's recommendations.

Nucleotide sequence accession numbers.

The consensus nucleotide sequences of SA1, SA2, SA3, SA4, and SA1/SG-neo (the subgenomic replicon derived from SA1) have been deposited in the GenBank database under accession numbers KJ925146 to KJ925150.

RESULTS

Isolation of genotype 5a sequences from patients.

Plasma from four voluntary blood donors who were found to be infected with HCV genotype 5a served as a source of HCV genomes described in this report. Quantitative PCR (qPCR) showed that these plasma samples contained high levels of HCV RNA, ranging from 9.8 × 10⁵ to 1.4 × 10⁶ copies/ml (see Fig. S1 in the supplemental material). Complete viral genomes were recovered by amplifying 10 overlapping cDNA fragments, and the consensus sequence was determined by sequence analysis of 5 to 10 clones for each fragment. The terminal 5′ and 3′ ends of HCV genomes were amplified by 5′ RACE and 3′ RACE, respectively, as described in Materials and Methods. Strains isolated from the four patients, named SA1, SA2, SA3, and SA4, were composed of 9,562, 9,566, 9,566, and 9,584 nucleotides, respectively, and contained three structural elements: the 5′ UTR (nt 1 to 339), ORF (nt 340 to 9381), and 3′ UTR (nt 9382 to the 3′ end). The difference in lengths of these strains was due to variable lengths of the poly (U/UC) region in the 3′ UTR. In all strains, the ORF was 9,042 nt long, encoding a 3,014-amino-acid-long polyprotein. By phylogenetic analysis, all four strains clustered with previously sequenced genotype 5a isolates (Fig. 1).

FIG 1.

Phylogenetic tree drawn by aligning polyproteins from the HCV strains isolated in this study and representative strains from various genotypes reported in the database. The strains used in the phylogenetic analysis include HC-J8 (GenBank accession no. D10988), HCV-BEBE1 (accession no. D50409), JFH-1 (accession no. AB047639), HCV-6a63 (accession no. DQ480514), TrKj (accession no. D49374), S310 (accession no. AB691595), HCV K3a/650 (accession no. D28917), NZL1 (accession no. NC_009824), S52 (accession no. GU814263), SA13 (accession no. AF064490), EUH1480 (accession no. Y13184), HCV-02C (accession no. DQ418784), H77 (accession no. AF009606), HCV-JK1 (accession no. X61596), and HCV-JT (accession no. D11168). The length of the horizontal bar corresponds to the number of nucleotide substitutions per site.

To determine the degree of variation among the isolated strains, the genetic heterogeneity in each subgenomic region was calculated (Table 1). At the nucleotide level, the mean genetic heterogeneity of the entire ORF was 8.54%. Among the subgenomic regions, the largest variation was seen in the envelope proteins (E1, 11.11%; E2, 12.59%) and NS2 (11.44%). The 5′ UTR was the most conserved region, with a mean diversity of 0.79%. At the amino acid level, genetic heterogeneity of the entire polyprotein was 4.34%. Among the subgenomic regions, E2, NS2, and E1 displayed the greatest diversity (10.53%, 7.68%, and 6.60%, respectively).

TABLE 1.

Mean genetic heterogeneity between isolated genotype 5a strains^a

Genomic region	Heterogeneity
	Nucleotide level		Amino acid level
	Nucleotidesb	% heterogeneity (mean ± SD)	Amino acidsb	% heterogeneity (mean ± SD)
5′ UTR	1–339	0.79 ± 0.48
Core	340–912	3.93 ± 1.03	1–191	1.83 ± 0.86
E1	913–1488	11.1 ± 0.86	192–383	6.60 ± 1.67
E2	1489–2580	12.6 ± 1.36	384–747	10.5 ± 1.51
P7	2581–2769	9.79 ± 1.60	748–810	1.59 ± 1.74
NS2	2770–3420	11.4 ± 0.63	811–1027	7.68 ± 2.17
NS3	3421–5313	7.16 ± 0.60	1028–1658	1.88 ± 0.53
NS4A	5314–5475	6.28 ± 2.85	1659–1712	0.31 ± 0.76
NS4B	5476–6258	8.17 ± 3.32	1713–1973	1.72 ± 0.79
NS5A	6259–7608	8.98 ± 0.94	1974–2423	6.04 ± 0.99
NS5B	7609–9381	6.91 ± 0.30	2424–3014	2.57 ± 0.39
ORF	340–9381	8.54 ± 0.80	1–3014	4.34 ± 0.41

^aUTR, untranslated region; E, envelope; NS, nonstructural.

^bThe nucleotide and amino acid positions refer to a previously sequenced genotype 5a strain, SA13 (GenBank accession no. AF064490).

Generation of subgenomic replicons.

We generated a G418-selectable subgenomic replicon (Fig. 2A), SA1/SG-neo, from the consensus cDNA clone of SA1, as described in Materials and Methods. Huh-7.5 cells were electroporated with RNA transcripts generated by in vitro transcription of SA1/SG-neo, and G418 selection was imposed 48 h later. However, all cells died after 3 weeks of selection, suggesting a failure of SA1/SG-neo to establish replication (Fig. 2B). We have previously shown that some of the cell culture-adaptive changes, originally described for genotypes 1a and 1b replicons, can support replication of other genotypes (11, 12). Based on this observation, we generated three mutants containing single-amino-acid substitutions: (i) SA1/SG-neo(i), containing a T1281I substitution in NS3 (corresponding to T1280I in genotype 1b strain Con1; GenBank accession no. AJ238799) (25), (ii) SA1/SG-neo(I), containing a S2205I substitution in NS5A (corresponding to S2204I in genotype 1a strain H77; GenBank accession no. AF009606) (8), and (iii) SA1/SG-neo(G), containing an R2888G substitution in NS5B (corresponding to R2884G in Con1) (26). Since combining cell culture-adaptive substitutions in NS3 with those in other nonstructural proteins increases the efficiency of G418-resistant colony formation (12, 25, 27), we generated double mutants containing the T1281I substitution together with S2205I or R2888G. All mutants failed to replicate in Huh-7.5 cells, except SA1/SG-neo (I), for which a small number of colonies were obtained (Fig. 2B). Eight cell colonies were isolated, and HCV RNA replication was confirmed by quantification of HCV RNA and detection of NS5A protein. HCV RNA copies ranged from 2.3 × 10⁶ to 1.8 × 10⁷ copies/μg of cellular RNA (Fig. 2C). NS5A protein could be detected by flow cytometry in 61.5 to 91.6% of cells (Fig. 2D). When naive Huh-7.5 cells were electroporated with RNA extracted from replicon cell clones and subjected to selection with G418, a large number of cell colonies were obtained, suggesting transmissibility of G418 resistance by the replicating viral RNAs (see Fig. S2 in the supplemental material). Taken together, these results indicated that the G418-resistant cell colonies carried autonomously replicating viral RNA.

FIG 2.

Replication of SA1/SG-neo in Huh-7.5 cells. (A) Schematic diagram of SA1/SG-neo. NPTII, neomycin phosphotransferase II. (B) Huh-7.5 cells were selected with G418 for 4 weeks following electroporation with RNA transcripts from the indicated replicons. The resulting cell colonies were stained with crystal violet. (C) HCV RNA in the isolated cell colonies was measured by qPCR. Results are expressed as means ± standard errors of the means (SEM) of two independent measurements. (D) Replicon cells were stained with anti-NS5A antibody and analyzed by flow cytometry. FSC, forward scatter.

Effect of candidate adaptive mutations on replication.

The appearance of a few colonies after electroporation with SA1/SG-neo(I), combined with the fact that significantly more colonies were obtained when cellular RNA from these colonies was introduced into naive Huh-7.5 cells, suggested that additional cell culture-adaptive mutations might have been acquired. To test this possibility, RNA was extracted from the isolated cell clones, and the NS3-NS5B coding region of the viral genome was sequenced. While the S2205I substitution was conserved, at least one additional nonsynonymous mutation was present in each clone (Table 2). We engineered these mutations into SA1/SG-neo(I) and measured their effect on HCV RNA replication by titrating the colony formation in Huh-7.5 cells. Since three nonsynonymous mutations were identified in clone 1, various mutants containing different combinations of these substitutions were generated for this clone. A broad range of cell culture adaptation was observed across the mutants; the mutants showed a 55- to 1,700-fold increase in colony formation efficiency compared to that for SA1/SG-neo(I) (Fig. 3) (see Fig. S3A and S3B in the supplemental material). A mutant containing all three substitutions identified in clone 1, henceforth called SA1/SG-neo(SKIP), yielded the highest number of colonies. However, the number of colonies was much lower than that obtained for Con1/SG-neo(I), a highly cell culture-adapted subgenomic replicon from genotype 1b. To examine if SA1/SG-neo(SKIP) acquired an additional mutation(s) which might further increase the colony formation efficiency of this replicon, we isolated three colonies harboring this construct and sequenced the region encoding NS3-NS5B. No additional mutation was found, indicating that this replicon does not need further adaptation. Next, we asked if the identified mutations were able to support colony formation in the absence of S2205I. Elimination of the S2205I substitution from mutants ablated their ability to yield G418-resistant cell colonies (data not shown), suggesting that isoleucine at this position was essential for the adaptive potential of the de novo mutations.

TABLE 2.

Nonsynonymous mutations found in eight SA1/SG-neo(I) replicon clones

Replicon clone	Nucleotide mutationa	Amino acid substitutionb	NSc protein
1	3946 (2301) G→A	1203 (176) E→K	NS3
	4556 (2911) A→G	1406 (379) K→S	NS3
	4557 (2912) G→C
	7435 (5790) C→T	2366 (393) S→P	NS5A
2	4021 (2376) C→T	1228 (201) H→Y	NS3
3	4165 (2520) G→A	1276 (249) D→N	NS3
4	5453 (3808) A→G	1705 (47) Q→R	NS4A
5	4261 (2616) G→A	1308 (281) G→S	NS3
6	4951 (3306) G→C	1538 (511) V→L	NS3
7	4633 (2988) G→A	1432 (405) D→N	NS3
8	5453 (3808) A→G	1705 (47) Q→R	NS4A

^aPosition within full-length SA1 genome and SA1 subgenomic replicon (in parentheses).

^bPosition within full-length SA1 polyprotein and within individual proteins (in parentheses).

^cNS, nonstructural.

FIG 3.

Colony titration assay of SA1/SG-neo(I) containing additional amino acid substitutions described in Table 2. Huh-7.5 cells, seeded in 6-well plates at a density of 4 × 10⁵ cells/well, were transfected the next day with 2 μg RNA from the indicated mutants. Six hours after transfection, cells were harvested and seeded in new 6-well plates at densities ranging from 2 × 10⁵ to 2 × 10² cells/well. The total numbers of cells in each well were brought to 2 × 10⁵ by adding feeder cells transfected with RNA from a replication-defective HCV replicon (the three residues in the active site of NS5B polymerase were changed from GDD to AAG). The cell colonies resulting from 4 weeks of selection with G418 were counted for at least two cell densities, and the percentage of colony formation was calculated (number of colonies in a well/number of cells initially plated in that well × 100). Shown is the fold increase in colony formation relative to the result for SA1/SG-neo(I), the value for which was arbitrarily set at 1. Data are means ± SEM from two independent experiments. *, this replicon had the highest replication levels and was named SA1/SG-neo(SKIP).

Synthesis of the Feo replicon for genotype 5a.

To generate a genotype 5a replicon that can be used to facilitate screening of antiviral compounds, we replaced the NPTII gene of SA1/SG-neo(SKIP) with a fusion protein of firefly luciferase and NPTII. When RNA transcribed from this replicon, named SA1/SG-Feo(SKIP), was introduced into Huh-7.5 cells, followed by selection with G418 (750 μg/ml), a large number of cell colonies were obtained; however, the number was ∼3-fold lower than that obtained for the “-neo” version of this replicon. We pooled cell colonies and compared the levels of HCV RNA replication with Huh-7.5 cells carrying JFH1/SG-Feo, a previously reported highly efficient replicon derived from the genotype 2a isolate JFH-1 (12). The luciferase activity of SA1/SG-Feo(SKIP) was comparable to that of JFH1/SG-Feo, suggesting high levels of HCV RNA replication (Fig. 4A). Furthermore, NS5A protein could be readily detected in most of the cells carrying SA1/SG-Feo(SKIP) (Fig. 4B).

FIG 4.

Replication levels of Feo replicons from various HCV genotypes. Huh-7.5 cells were electroporated with Feo replicons from genotype 1a (H77), 1b (Con1), 2a (JFH-1), 3a (S52), 4a (ED43), and 5a (SA1) and selected with G418 (750 μg/ml). After 3 weeks, the surviving cells were pooled and passaged twice in G418-containing medium. LU, luciferase units. (A) Replicon cells from each genotype were seeded in 8 wells of a 96-well plate at a density of 2 × 10⁴ cells/well. Twenty-four hours later, cells were lysed and firefly luciferase activity was measured. The results are plotted as means ± SEM from eight replicates. (B) HCV NS5A protein in the cells was detected by flow cytometry.

Effect of HCV inhibitors on replication of various genotypes.

The development of a subgenomic replicon for genotype 5a provided us an opportunity to compare its response to various antiviral compounds with those of other genotypes. To this end, we used a panel of Feo replicons from genotypes 1a [H77/SG-Feo(L+8)], 1b [Con1/SG-Feo(I)], 2a (JFH1/SG-Feo), 3a [S52/SG-Feo(AII)], 4a [ED43/SG-Feo(VYG)], and 5a [SA1/SG-Feo(SKIP)]. JFH1/SG-Feo and S52/SG-Feo(AII) have been previously described (12), while H77/SG-Feo(L+8), Con1/SG-Feo(I), and ED43/SG-Feo(VYG) were generated as described in Materials and Methods. For convenience, these replicons are referred to as H77, Con1, JFH-1, S52, ED43 and SA1 in this report. The G418-resistant Huh-7.5 cells, selected after electroporation with synthetic RNA from each replicon, were pooled, and a firefly luciferase assay was done to compare their replication levels. The firefly luciferase values were within 3-fold of each other, indicating similar levels of HCV RNA replication (Fig. 4A). By flow cytometry, NS5A protein could be detected in 56% to 96% of cells (Fig. 4B). The lower number of NS5A-positive cells for H77 may be due to a lower sensitivity of the antibody for this genotype. Next, we exposed the replicon cells to various concentrations of IFN-α2a and four DAAs: danoprevir (an NS3 protease inhibitor), daclatasvir and ledipasvir (NS5A inhibitors), and sofosbuvir (an NS5B nucleotide polymerase inhibitor) (Fig. 5). IFN-α2a showed comparable inhibitory effects across all genotypes (Table 3), although a cytotoxic effect was observed at higher concentrations, especially for cells carrying genotype 2a replicons (see Fig. S4A in the supplemental material). Among the DAAs, danoprevir was highly effective against genotypes 1a, 1b, 4a, and 5a, with 50% effective concentrations (EC₅₀) of 1.89 nM, 0.67 nM, 3.67 nM, and 12.79 nM, respectively. Intermediate efficacy was seen against genotype 2a (EC₅₀, 36.31 nM), while genotype 3a was least susceptible (EC₅₀, 276.7 nM). Daclatasvir, with EC₅₀ of 38.48 pM, 20.82 pM, 63.79 pM, 18.21 pM, and 57.06 pM, was highly effective against genotypes 1a, 1b, 2a, 4a, and 5a, respectively, whereas the EC₅₀ for genotype 3a was ∼2 nM, indicating a lower inhibitory effect against this genotype. Ledipasvir displayed a broad range of efficacy depending on the genotype: a very strong inhibition of genotypes 1a and 1b (EC₅₀ of 29.94 pM and 8.12 pM, respectively), yet weak activity against genotype 3a (EC₅₀, 455 nM). Sofosbuvir inhibited replication of all genotypes with comparable efficiencies; EC₅₀ ranged from 107.9 nM to 415.2 nM. Treatment of cells with the DAAs at the concentrations used in this study had no effect on cell viability (see Fig. S4 in the supplemental material), indicating that the decrease in luciferase signal was due to the specific inhibition of HCV RNA replication. Taken together, these results demonstrate differential effects of anti-HCV compounds on various genotypes.

FIG 5.

Inhibitory effect of anti-HCV compounds on the replication of various genotypes. Cells harboring persistently replicating Feo replicons from genotypes 1a (H77), 1b (Con1), 2a (JFH-1), 3a (S52), 4a (ED43) and 5a (SA1) were seeded in duplicate in tissue culture-treated, 96-well, white plates at a density of 1 × 10⁴ cells/well using G418-free cell culture medium. The following day, cells were exposed to the indicated concentrations of IFN-α2a (A), danoprevir (B), daclatasvir (C), ledipasvir (D), and sofosbuvir (E). Seventy-two hours later, cells were lysed in 50 μl of 1× CCLR (Promega), and firefly luciferase activity was measured. Data are plotted as means ± SEM from three independent experiments.

TABLE 3.

Inhibitory effects of antiviral compounds on HCV replication^a

Compound	Genotype (isolate)	EC50	95% CI
IFN-α2a (IU/ml)	1a (H77)	0.77	0.66–0.88
	1b (Con1)	0.64	0.36–1.15
	2a (JFH-1)	3.77	2.98–4.76
	3a (S52)	2.46	1.51–4.02
	4a (ED43)	1.63	1.11–2.40
	5a (SA1)	1.01	0.76–1.33
Danoprevir (nM)	1a (H77)	1.89	1.33–2.70
	1b (Con1)	0.67	0.47–0.95
	2a (JFH-1)	36.31	24.81–53.16
	3a (S52)	276.7	189.30–404.50
	4a (ED43)	3.67	3.10–4.34
	5a (SA1)	12.79	11.10–14.73
Daclatasvir (pM)	1a (H77)	38.48	31.29–47.33
	1b (Con1)	20.82	14.84–29.21
	2a (JFH-1)	63.79	11.90–341.80
	3a (S52)	2008	972.10–4147
	4a (ED43)	18.21	13.15–25.23
	5a (SA1)	57.06	31.07–104.80
Ledipasvir (pM)	1a (H77)	29.94	18.46–48.55
	1b (Con1)	8.12	7.77–8.48
	2a (JFH-1)	12,062	8,175–17,796
	3a (S52)	4.55 × 105	Very wide
	4a (ED43)	498.8	230.2–1081
	5a (SA1)	576.5	463.2–717.5
Sofosbuvir (nM)	1a (H77)	170.1	112.2–258.0
	1b (Con1)	187.1	91.89–380.8
	2a (JFH-1)	263.9	151.9–458.5
	3a (S52)	198.2	92.55–424.5
	4a (ED43)	415.2	293.8–586.8
	5a (SA1)	107.9	47.95–242.2

^aEC₅₀, 50% effective concentration; 95% CI, 95% confidence interval obtained from one experiment. The results were confirmed by two independent experiments, with each experiment performed in triplicate. “Very wide” 95% CI indicates that the standard errors were too high to unambiguously calculate the confidence interval.

DISCUSSION

Since its development in 1999, the HCV subgenomic replicon system has been instrumental in studying various aspects of HCV RNA replication in cultured cells. However, for many years, this system was restricted only to genotypes 1 and 2, partly because these are the most predominant genotypes in the United States, Europe, and Japan. This restriction was recently overcome with the development of subgenomic replicons for genotypes 3a, 4a, and 6a (12, 13). In this study, we report the first genotype 5a replicon that successfully replicated in Huh-7.5 cells, producing high levels of HCV RNA and proteins (Fig. 2C and D). These high levels of replication, however, were obtained only after inclusion of cell culture-adaptive changes identified in various regions of the viral genome (Fig. 3) (see Fig. S3 in the supplemental material). The replicon with the highest replication capacity was further modified to express a convenient reporter, facilitating its use for high-throughput screening and evaluation of antiviral compounds. Taking advantage of this newly developed replicon, we compared the responsiveness of genotype 5a to various anti-HCV compounds with that of other genotypes (Fig. 5).

In the present study, we used plasma from HCV-infected blood donors in South Africa to recover the viral genomes. Phylogenetic analysis showed clustering of these genomes with genotype 5a isolates. The genomic sequences from three different strains of genotype 5a, namely, SA13 (GenBank accession no. AF064490), EUH1480 (GenBank accession no. NC_009826), and ZS631 (GenBank accession no. KC844046.1), are available in the HCV database (http://hcv.lanl.gov); however, none of them contains the complete 3′ UTR sequence. The 5′ and 3′ UTR of HCV, like UTRs from many positive-strand RNA viruses, contain important sequences and structural motifs critical for translation and RNA replication (reviewed in reference 28). Although these regions are highly conserved across HCV genotypes, intergenotypic differences exist which may affect the level of RNA replication. We therefore performed 5′ and 3′ RACE to determine the authentic termini of the genotype 5a strains. As expected, a high degree of conservation was seen among the isolated strains on both termini: >99% identity in the 5′ UTR and 100% identity in the X-tail region of the 3′ UTR. Within the coding region, envelope proteins exhibited the highest diversity (Table 1). This was not surprising, given that these are the primary targets of neutralizing antibodies and that a continuous tug-of-war between the host's immune system and the virus leads to a high genetic heterogeneity in these glycoproteins.

As the NS3-NS5B region was highly conserved among the isolated genotype 5a strains (97% identity), we selected one representative strain, SA1, for which to develop a subgenomic replicon. Replication and G418 selection of SA1/SG-neo required introduction of a previously reported, broadly active, cell culture-adaptive substitution, S2205I (Ser-to-Ile substitution at NS5A position 232). This, however, was not sufficient for efficient replication, as all replicons isolated from randomly chosen cell colonies had acquired additional mutations. Most of these mutations resulted in coding changes in the NS3 protein, although a few in NS4A and NS5A were also seen (Table 2). Interestingly, one of the identified changes, D1432N, corresponds to the D1431N substitution described previously for genotype 4a (12). All of the de novo mutations, when combined with S2205I, led to various degrees of increased colony formation efficiency (Fig. 3) (see Fig. S3 in the supplemental material). These, however, failed to support HCV RNA replication in the absence of S2205I (data not shown). These results concur with most of the existing data showing that adaptive mutations in different parts of the viral genome enhance each other (12, 25, 27). Exactly how these adaptive changes cooperate and their mechanism of action are still not known.

Genotype 5a is localized mainly to South Africa, where it accounts for approximately 40% of HCV infections (15, 29). Information about disease severity and treatment outcome of this genotype is very limited. Although preliminary studies suggested that genotype 5 responds well to combination therapy of interferon and ribavirin (30), no large-scale clinical trials were performed to solidify these results. Furthermore, little is known about its response to DAAs, and none of the approved DAAs are currently recommended for the treatment of genotype 5a infection. Chimeric systems, where one or multiple proteins from genotype 5a were substituted in the backbone of JFH-1 genotype 2a replicase, have been used for cell culture-based inhibitor studies (31–33). However, the results obtained with these systems can be affected by suboptimal intergenotypic interactions between replicase components and diminished replicative fitness. Therefore, replicons with the entire replicase from the same genotype offer more desirable systems for testing antivirals. With the first in vitro replication system in hand, we examined the response of genotype 5a to various HCV inhibitors. The results suggested a high sensitivity of this genotype to sofosbuvir, an NS5B polymerase inhibitor currently approved for the treatment of genotypes 1, 2, 3, and 4 (Fig. 5, Table 3). In fact, the EC₅₀ for genotype 5a was lowest among those for all of the genotypes, although the difference was not significant. Recently, a high response of genotype 6a to sofosbuvir was also documented (13), suggesting a pan-genotype profile of this nucleotide inhibitor. Unlike sofosbuvir, other DAAs displayed differential activity between genotypes. For example, danoprevir, a macrocyclic NS3 protease inhibitor, was highly effective against genotypes 1a, 1b, and 4a, followed by genotypes 5a and 2a, for which EC₅₀ were ∼12- and 36-fold higher, respectively, than that for genotype 1b. In contrast, genotype 3a, in accordance with previously published reports (12, 32, 33), was highly resistant to danoprevir, with an EC₅₀ more than 350-fold higher than that of genotype 1b. Conflicting data exist on the sensitivity of genotype 5a to danoprevir. Using a cell culture infection system, where the NS3 protease domain (NS3P) and NS4A from various genotypes were expressed in the backbone of genotype 2a J6/JFH1 virus, Gottwein et al. showed that the EC₅₀ of danoprevir for genotype 5a was approximately 30-fold lower than that for genotype 3a (32). In contrast, Imhof and Simmonds, using a similar system, reported that genotype 5a was less susceptible than genotype 3a (33). This discrepancy might be explained by two different genotype 5a strains used in these studies: while Gottwein et al. obtained NS3P and NS4A from SA13, Imhof and Simmonds used another strain, EUH1480. To test if these differences in genetic background could account for discrepant results, we compared NS3Ps from SA13 and EUH1480. Although 96% of amino acids were conserved, sequence variation was seen at positions 168 and 170, two of the several loci associated with resistance of genotype 1 to danoprevir (34, 35). At both positions, the amino acids present in SA13 (Asp and Ile, respectively) were the same as those found in susceptible genotype 1 isolates, whereas EUH1480 had Glu and Val at these positions. Even though Glu168 and Val170 are not reported to be directly associated with resistance, several other residues at these positions mediated high levels of resistance to macrocyclic protease inhibitors in various HCV genotypes (33, 36). Most notably, among genotype 5a strains isolated in this study, SA1 had Asp at position 168, while the remaining 3 strains had Glu. Similarly, at position 170, SA1 and SA2 had Ile, while SA3 and SA4 had Val. This may explain, at least in part, why SA1 and SA13 are highly sensitive to danoprevir. In addition, these observations clearly highlight the importance of testing multiple strains from various genotypes before drawing general conclusions. Among the NS5A inhibitors, daclatasvir was highly effective against genotype 5a, which concurs with previously reported results using recombinant HCV genomes expressing the genotype 5a NS5A protein in the J6/JFH1 backbone (37). In fact, with picomolar EC₅₀ in the cell culture replication systems, daclatasvir is thus far the most potent drug against various HCV genotypes. As these potencies have translated into robust activity in the clinical studies with genotype 1 patients (38), a favorable clinical response for genotype 5a can be expected.

In conclusion, we have developed a subgenomic replicon for genotype 5a that will be a useful research tool to study various aspects of this poorly studied genotype and to evaluate its response to antiviral compounds. With this report, the replication systems for all major HCV genotypes are now established; however, this repertoire should be further expanded to include hitherto uncharacterized subtypes. In addition, efforts should be made to propagate HCV in human hepatoma cells without the need for cell culture-adaptive mutations, as these are typically not found in natural isolates and may affect virus-host interactions and responses to antiviral compounds.