Highly efficient infectious cell culture of three hepatitis C virus genotype 2b strains and sensitivity to lead protease, nonstructural protein 5A, and polymerase inhibitors


  • Santseharay Ramirez,

    1. Copenhagen Hepatitis C Program (CO-HEP), Department of Infectious Diseases and Clinical Research Centre, Copenhagen University Hospital, Hvidovre and Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Yi-Ping Li,

    1. Copenhagen Hepatitis C Program (CO-HEP), Department of Infectious Diseases and Clinical Research Centre, Copenhagen University Hospital, Hvidovre and Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Sanne B. Jensen,

    1. Copenhagen Hepatitis C Program (CO-HEP), Department of Infectious Diseases and Clinical Research Centre, Copenhagen University Hospital, Hvidovre and Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Jannie Pedersen,

    1. Copenhagen Hepatitis C Program (CO-HEP), Department of Infectious Diseases and Clinical Research Centre, Copenhagen University Hospital, Hvidovre and Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Judith M. Gottwein,

    1. Copenhagen Hepatitis C Program (CO-HEP), Department of Infectious Diseases and Clinical Research Centre, Copenhagen University Hospital, Hvidovre and Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Jens Bukh

    Corresponding author
    1. Copenhagen Hepatitis C Program (CO-HEP), Department of Infectious Diseases and Clinical Research Centre, Copenhagen University Hospital, Hvidovre and Department of International Health, Immunology and Microbiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    • Address reprint requests to: Jens Bukh, M.D., Department of Infectious Diseases, Copenhagen University Hospital, Hvidovre, Kettegaard Alle 30, DK-2650 Hvidovre, Denmark. E-mail: jbukh@sund.ku.dk; fax: +4536474979.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • This study was supported by research grants from The Lundbeck Foundation (to S.R., Y.P.L., J.M.G., and J.B.), The Danish Cancer Society (to J.M.G. and J.B.), The Novo Nordisk Foundation (to Y.P.L., J.M.G., and J.B.), The Region H Research Fund (to S.R., J.M.G., and J.B.), The A.P. Møller og Hustru Chastine Mc-Kinney Møllers Fondation (to J.B.), and the Danish Council for Independent Research, Medical Sciences (to S.R., Y.P.L., and J.B.). S.R. is the recipient of an Individual Postdoctoral Stipend from the Danish Council for Independent Research of Medical Sciences. The authors thank C.M. Rice (Rockefeller University, New York, NY) for providing valuable reagents.


Hepatitis C virus (HCV) is a genetically diverse virus with multiple genotypes exhibiting remarkable differences, particularly in drug susceptibility. Drug and vaccine development will benefit from high-titer HCV cultures mimicking the complete viral life cycle, but such systems only exist for genotypes 1a and 2a. We developed efficient culture systems for the epidemiologically important genotype 2b. Full-length molecular clones of patient strains DH8 and DH10 were adapted to efficient growth in Huh7.5 cells by using F1468L/A1676S/D3001G (LSG) mutations. The previously developed J8cc prototype 2b recombinant was further adapted. DH8 and J8 achieved infectivity titers >4.5 log10 Focus-Forming Units/mL. A defined set of DH8 mutations had cross-isolate adapting potential. A chimeric genome with the DH10 polyprotein coding sequence inserted into a vector with J8 untranslated regions was viable. Importantly, we succeeded in generating DH8, J8, and DH10 viruses with authentic sequences in the regions targeted by lead direct-acting antivirals. Nonstructural protein (NS)5B inhibitors sofosbuvir, mericitabine, and BI207127 had activity against 1a (strain TN), 2a (strains JFH1 and J6), and the 2b strains, whereas VX-222 and filibuvir only inhibited 1a. Genotype 2b strains were least sensitive to seven lead protease inhibitors, including MK-5172 with high overall potency. NS5A inhibitor daclatasvir was exceptionally potent, but efficacy was affected by the HCV strain. Conclusion: Highly efficient HCV full-length 2b culture systems can be established by using consensus clones with defined mutations. Lead protease and NS5A inhibitors, as well as polymerase inhibitors sofosbuvir, mericitabine, and BI207127, show cross-activity against full-length 1a, 2a, and 2b viruses, but important sensitivity differences exist at the isolate level. Infectious cultures for different HCV strains will advance studies on viral biology and pathogenesis and promote individualized patient treatment. (Hepatology 2014;59:395–407)


amino acid


direct-acting antivirals


half-maximal effective concentration


Focus-Forming Units


hepatitis C virus




Los Alamos HCV Sequence Database


nucleoside/nucleotide inhibitor


non nucleoside inhibitor


nonstructural protein




open reading frame


protease inhibitor




untranslated region


wild type.

Approximately 150 million people are infected with hepatitis C virus (HCV) worldwide, and over 350,000 are estimated to die from associated chronic liver disease each year. The economical and social burden of hepatitis C is enormous, and efficient therapies and vaccines are needed. Infectious culture systems are important for HCV studies, contributing to drug and vaccine development. However, only a few HCV strains can be studied because of the lack of efficient culture systems.

HCV is an enveloped, positive-strand RNA virus from the family Flaviviridae. Its genome contains ∼9,600 nucleotides (nts) with a single open reading frame (ORF) flanked by 5′ and 3′ untranslated regions (UTRs). The polyprotein of ∼3,000 amino acids (aa) is processed into structural (Core, E1, and E2) and nonstructural proteins (p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B) with complex roles in the viral life cycle.[1] HCV presents significant genetic diversity with six epidemiologically important genotypes and numerous subtypes. Viruses recovered from infected individuals are referred to as isolates or strains, and they circulate as quasispecies. Genotypes 1, 2, and 3 are the most prevalent and globally distributed. Genotype 2 is highly prevalent in West Africa[2] as well as Asian countries such as China and Japan.[3] In the United States, genotype 2 is the second in prevalence, with subtype 2b accounting for 10% of all infections.[4] Genotypes 1 and 2 respond differently to standard interferon (IFN)/ribavirin (RBV) therapy, with genotype 2–infected patients achieving higher clearance rates.[5] However, limited information is available about the sensitivity of genotype 2 viruses to direct-acting antivirals (DAAs), both in vivo and in vitro.

DAAs are expected to improve HCV clearance rates in patients with chronic hepatitis C. The subgenomic replicon systems[6] have been of major importance for discovery, development, and preclinical testing of these compounds; however, replicons do not recapitulate the complete viral life cycle and are not suitable for testing the effect of drugs beyond viral replication. Therefore, it is essential to develop infectious culture systems for different HCV strains from the major genotypes and subtypes that reproduce all viral functions. Nevertheless, development of such systems has been a major challenge, and they are only available for a few adapted genotype 1 and 2 isolates.[7-12] Among these, only chimeric J6/JFH1(2a), JFH1(2a), J6(2a), and TN(1a) could release high-titer infectious virus particles in cell culture.

In this study, we developed efficient full-length cell-culture systems for three genotype 2b isolates. After long-term culture, viruses reached high infectivity titers by acquiring specific cell-culture-adaptive mutations that were then used to generate highly efficient molecular recombinants. These systems permitted analysis of sensitivity to frontline HCV DAAs, in comparison with previously developed 1a and 2a full-length viruses. We found differential sensitivity to NS3/NS4A protease, NS5A, and NS5B inhibitors, both at the genotype and isolate level. Importantly, we also identified front-line drugs that are efficient against all viruses, thus possibly overcoming HCV genetic diversity.

Materials and Methods

Viral Sequence Analysis and Reverse Genetics

The consensus ORF sequence of two genotype 2b isolates (DH8 and DH10) was determined in sera of chronically infected patients from Denmark (Supporting Materials); the core-NS2 sequence was previously described.[13] Full-length recombinants were assembled by chemical synthesis (GenScript; Piscataway, NJ) or standard molecular cloning; for DH10, the ORF was cloned into a vector with J8 UTRs (Supporting Materials). Nt changes for reverse genetics studies were introduced by standard cloning proceedures. The complete HCV sequence of final plasmid preparations (HiSpeed Plasmid Maxi Kit; Qiagen, Hilden, Germany) was sequence confirmed (Macrogen Inc., Seoul, Korea).

For determination of ORF sequences of cell-culture–derived viruses, we performed direct sequencing of overlapping amplicons (Supporting Materials, Supporting Table 1). Nt and aa positions were numbered according to the corresponding 2b genomes.

Table 1. Characterization of DH8 Full-Length Viruses
   H77 Reference13861672184026022606271929442980364144984539473153555697577561816231708073357591759492779336
 Peak Titer Log10 FFU/mL (Day) Peak Titer Log10 FFU/mL (Day)HCV RNA Log10 IU/mL                       
  1. Changes at indicated nt and aa positions identified in direct ORF sequencing are listed; the corresponding H77 (AF009606) reference number is also given. All dominant coding changes are shown; minor and 50/50 changes are depicted only at positions where dominant changes in other genomes were found; for an overview of all changes, see Supporting Table 2. The associated HCV gene is indicated. The original DH8CF sequence is shown at the top, and engineered mutations are shown in light shadings (LSG in dark shading). Dots indicate identity with DH8CF.

  2. a

    Viruses collected from different time points after passage were mixed to generate stock for antiviral treatment assays.

  3. b

    Recombinant named DH8cc.

Transfection 1, culture A3.1 (46)1st3.7 (8)7.7...C.......CTG/A.C...C.G.
  2nd4.3 (9)7.1...C.....G.CTA.C...C/t.G.
  3rd4.7 a7.5...C.......CTA.C...C.G.
Transfection 1, culture B3.9 (60)1st3.2 (8)7.8.......T/c...CT..C.....GT
Transfection 2, culture A3.1 (41)1st4.2 (8)7.9.....CT....CT..CGG.T/c.G.
Transfection 2, culture B3.9 (55)1st3.1 (8)7.3G...A...C..CT.TC..A.CG.
Transfection 33.1 (42)1st3.1 (13)7.6.CA.......C/tCT..CG...T/cG.
DH8_LSG plus mutations                          
L884P/V1951A/ I2440T/L3021F3.8 (11)2nd3.9 (9)7.4.......C...CT..C....CGT
L758S/A1790T/ V1951A/I2439Tb4.1 (8)2nd4.3 (11)7.2...C.......CTA.C...C.G.
L758S/A1790T/ V1951A/I2439T/ L3021F4.3 (8)2nd3.5 (6)6.3...C.......CTA.C...C.GT
   Recombinant aa349444502758759797872884110413901404146816761790181619511968224723322439244030013021
   H77 Reference349444500754755793868880110013861400146416721786181219471964224723322417281829792999

Transfection and Passage of Recombinant Full-Length HCV in Huh7.5 Cells

Transfections with RNA transcripts were performed as previously reported.[10, 14] For viral passage, cell-free supernatants were incubated with 4 × 105 Huh7.5 cells overnight. Cultures were split every 2-3 days, and progression of infection was monitored by immunostaining on cells seeded in slides and co-stained with primary antibodies anti-core C7-50 (Enzo Life Sciences) and anti-NS5A 9E108 at a 1:250 dilution.[10] Infectivity titers expressed as log10 focus-forming units (FFU)/mL were determined as previously described.[15, 16] To improve staining, we used a mixture of anti-core C7-50 and anti-NS5A 9E10 at dilutions 1:450 and 1:1,000, respectively. HCV RNA titers (IU/mL) on passage supernatants with peak infectivity titers were determined as previously described.[17]

Antiviral Treatments in Huh7.5 Cells

HCV-infected cultures were treated with NS3 protease, NS5A, and NS5B polymerase inhibitors (purchased from Acme Bioscience, Palo Alto, CA) in a high-throughput assay.[15, 16] Immunostaining of fixed 96-well plates was performed using anti-core C7-50 and anti-NS5A 9E10 at dilutions 1:450 and 1:1,000, respectively. Concentration-response curves and half-maximal effective concentration (EC50) values were calculated as previously described.[17] Noncytotoxic dose ranges for all drugs were determined with a cytotoxicity assay (CellTiter 96 AQueous One Solution Cell Proliferation Assay; Promega, Madison, WI).[17]


Development of a Highly Efficient Full-Length HCV Culture System for Genotype 2b, Strain DH8

Based on the consensus sequence of patient strain DH8, we generated a novel full-length clone (DH8CF [Supporting Materials]; GenBank accession no.: KF420335); the genome consisted of 9,663 nts, encoding a polyprotein of 3,033 aa. The DH8CF ORF differed by 8.4% and 5.3% at the nt and aa levels from the prototype 2b, full-length clone J8CF (JQ745651).[10] RNA transcripts of DH8CF were transfected into Huh7.5 cells, in three independent experiments. No HCV-positive cells were detected during 53, 12, and 10 days of follow-up, respectively. Thus, wild-type (WT) DH8 was considered nonviable in Huh7.5 cells.

We recently described mutations F1468L (NS3 helicase), A1676S (NS4A), and D3001G (NS5B), designated LSG, that were instrumental for culture adaptation of full-length clones J6CF,[10] J8CF,[10] and TN.[12] We performed three independent transfections of Huh7.5 cells with DH8CF containing LSG, designated DH8_LSG; HCV-positive cells were observed from day 1. At day 9, two transfection cultures were split into two replicates (A and B) for a total of 5 cultures followed over the long term (Table 1 and Supporting Table 2). Spread of infection (defined as ≥80% HCV-antigen–positive cells) was observed at days 36-53, with HCV infectivity titers of 3.1-3.9 log10 FFU/mL. In the first viral passage to naïve Huh7.5 cells, viruses spread at days 5-8, reaching 3.1-4.2 log10 FFU/mL. The ORF sequences of DH8_LSG recovered from all five first passages were determined. Coding changes were present in multiple genes, but V1951A (NS4B) was a common aa change. Also, four of five cultures had I to T change at either aa position 2,439 or 2,440 in NS5A.

Table 2. Characterization of J8 Full-Length Viruses
   H77 Reference369121515892163260226442919379039513978473147684794535556175697618162317129759490409277
 Peak Titer Log10 FFU/mL (Day) Peak Titer Log10 FFU/mL (Day)HCV RNA Log10 IU/mL                      
  1. Changes at indicated nt and aa positions identified in direct ORF sequencing are listed; the corresponding H77 (AF009606) reference number is also given. All nt changes are shown. The associated HCV gene is indicated. The original J8CF sequence is shown at the top, and engineered mutations are shown in light shadings (LSG in dark shading). Dots indicate identity with J8CF.

  2. a

    Data from J8cc.[10]

  3. b

    Viruses collected from different time points after passage were mixed to generate stock for antiviral treatments.

  4. c

    Recombinant and recovered viruses contain noncoding mutation A5324G.

  5. d

    Viruses harvested at this time point were used in treatment assays.

J8cc3.2a (7)4th4.8 (4)7.8GCC/TA/c.GCCAA/TC..TA/G..GTC/tGG
  5th4.6 b7.3GCT/cA.GCCATC..TG/a..GTCGG
J8cc-HTc4.7 (7)2nd4.3 (9)7.7.C.A.GCCATC..TG..GTCGG
J8_LSG/STAT3.8 (11)2nd4.1 (17d)7.8....C.....CT/CA/GT.AC..C.G
   Recombinant aa10292416612758772864115412081217146814801489167617631790195119682263244029223001
   H77 Reference10292416608754768860115012041213146414761485167217591786194719642263241829002979

To generate an efficient DH8 recombinant (Supporting Table 2), we tested the effect of V1951A (NS4B), which emerged in all transfections, in DH8_LSG. However, the virus did not spread until day 27, with titers of 3.0 log10 FFU/mL. After passage, the virus had acquired multiple changes, including I2439I/S. Because changes of I2439 were observed during DH8_LSG passages, we next tested DH8_LSG/V1951A/I2439T. After transfection, the virus spread at day 33, with titers of 3.3 log10 FFU/mL, and showed additional coding changes. Thus, V1951A alone or the combination V1951A/I2439T was not sufficient for the full adaptation of DH8_LSG.

Because the DH8_LSG transfection 1 culture B virus had only the dominant change, L3021F (NS5B), in addition to V1951A, we tested DH8_LSG/V1951A/L3021F. This recombinant had delayed spread (day 27) with low titers, and recovered viruses had acquired several changes, including quasispecies L884L/P (NS2) and W2429W/R (NS5A). Thus, we added I2440T (NS5A) with or without L884P to DH8_LSG/V1951A/L3021F. Spread of DH8_LSG/V1951A/I2440T/L3021F and DH8_LSG/L884P/V1951A/I2440T/L3021F (KF42 0337) occurred at days 27 (3.5 log10 FFU/mL) and 6 (3.8 log10 FFU/mL; Fig. 1A) post-transfection, respectively. DH8_LSG/L884P/V1951A/I2440T/L3021F reached titers of 3.9 log10 FFU/mL in the second passage. The recovered virus had nt changes, but none resulted in dominant aa substitutions. Thus, we had developed an efficient full-length infectious DH8 recombinant, designated as DH8_LSG/PATF.

Figure 1.

Transfection of Huh7.5 cells with RNA transcripts of full-length genotype 2b recombinants DH8, J8, and DH10. The different 2b recombinants contained F1468L/A1676S/D3001G (LSG)10 and additional adaptive mutations, as indicated. (A and B) Two independent transfection experiments; J65′UTR-NS2/JFH1 was included as control.[14] Graphs show percentage of Core/NS5A antigen-positive cells (left y axis) and HCV infectivity titers in collected supernatants (right y axis: log10 FFU/mL, mean of triplicate infections ± standard error of the mean is shown) at the analyzed days (x axis); supernatants from day 1 were not analyzed. Stars indicate infectivity titers below the assay limit of detection (2.5 and 2.7 log10 FFU/mL in A and B, respectively).

In an alternative approach, we engineered the four dominant coding changes from the third passage of transfection 1 culture A, with infectivity titers of 4.7 log10 FFU/mL (Table 1 and Supporting Table 2). The resulting recombinant DH8_LSG/L758S/A1790T/V1951A/I2439T (KF420336) spread at day 6 after transfection, reaching 4.1 log10 FFU/mL (Fig. 1A,B); it had peak titers of 4.4 and 4.3 log10 FFU/mL after the first and second passage, respectively. The ORF sequence of the second passage virus showed no additional mutations, suggesting that the virus was genetically stable. Because NS5B mutations had an important role in the adaptation of previously developed full-length cell-culture systems,[10, 12] we explored the benefit of adding L3021F; after transfection, DH8_LSG/L758S/A1790T/V1951A/I2439T/L3021F spread at day 6 with titers of 4.3 log10 FFU/mL (Fig. 1A). After the second passage, the recovered virus had three coding changes as quasispecies (Supporting Table 2). Thus, the addition of L3021F did apparently not improve viability.

Taken together, the reverse genetics experiments demonstrated that L758S, A1790T, V1951A, and I2439T mutations most efficiently adapted DH8_LSG, resulting in high HCV infectivity titers. Thus, we named this recombinant DH8cc.

Development of a High-Titer Culture System for the Prototype 2b Strain, J8

Because we had generated a highly efficient DH8 culture system using mutations identified from serial passage of DH8_LSG, we set out to improve the efficiency of our published genotype 2b system, J8cc, by serial passages. J8cc contained LSG/F772C/W864R/A1208T/I1968V/E2263V/H2922R and reached 3.2 log10 FFU/mL after transfection.[10] Infectivity titers increased to >4.6 log10 FFU/mL in the fourth and fifth passages in Huh7.5 cells, and seven additional coding changes were identified (Table 2); one of these changes, I2440T, had also appeared in DH8_LSG. We then generated J8cc/M292L/L612M/G1154A/N1217Y/Q1763R/I2440T (KF420338), which spread at day 3 post-transfection and reached infectivity titers of 4.7 log10 FFU/mL (Fig. 1B); the recovered passaged virus had no nt changes (Table 2). This efficient J8 recombinant was designated J8cc-HT, for “J8 cell culture, high titer.”

Development of a Chimeric Genotype 2b Culture System Expressing the Polyprotein of Strain DH10 Using J8 5′ and 3′ UTR Sequences

Genetic divergence of HCV isolates is concentrated in the ORF, which encodes the proteins that participate in the viral life cycle and are targeted by the most advanced DAAs. In contrast, the UTRs are the most conserved genomic elements of HCV, required for replication and translation. Determination of 3′ UTR sequences from HCV isolates is a technical challenge, often requiring high-titer serum samples or liver tissue. Therefore, we explored the possibility of generating a viable chimeric genome by inserting the ORF of a 2b patient isolate (strain DH10) into a vector containing the 5′ and 3′ UTRs of the prototype 2b (strain J8). The length of the DH10 ORF is 9,099 nt, encoding 3,033 aa. At the nt level, the DH10 ORF differs from J8 and DH8 by 8.8% and 8.5%, respectively, and at the aa level, the difference is 5.9% and 5.5%. To test viability of this chimeric genome in vitro, we introduced the LSG mutations.

Two independent transfections of DH10_LSG were performed (Supporting Table 3). In transfection 1, positive cells were observed at day 10 and the virus spread at day 47, reaching 2.6 log10 FFU/mL. However, in transfection 2, positive cells were not observed consistently until day 64, and spread occurred at day 83, reaching 2.3 log10 FFU/mL. We performed serial passages of transfection 1 virus and reached infectivity titers of 3.7 log10 FFU/mL in the fourth passage. The ORF sequence presented various coding changes, including NS4B mutations V1951A (observed also in DH8_LSG) and N1931S (observed in J6 and TN full-length viruses[10, 12]). Second-passage viruses from transfection 2 had titers of 3.6 log10 FFU/mL, and, interestingly, changes V1951A and N1931T (S in transfection 1) were also present.

Table 3. Median EC50 Values Obtained in Treatment Assays of HCV Genotype 1a, 2a, and 2b Full-Length Viruses for Different Protease, NS5A, and Polymerase DAAs
   EC50 (nM)
TargetClinical DevelopmentDrug NamesJFH1 (2a)aJ6 (2a)DH8 (2b)J8 (2b)DH10 (2b)TN (1a)
  1. EC50 values were obtained from the sigmoidal dose-response curves shown in Figures 2 and 3.

  2. a

    J6core-NS2/JFH1 recombinant.8

  3. Abbreviation: Ni, not inhibited.

NS3LicensedTelaprevir (VX-950)493458694179993109
 Boceprevir (Sch503034)5891,1239811462,069128
Phase IIIVaniprevir (MK7009)8812921912277816
 Simeprevir (TMC435)912667195333
 Faldaprevir (BI-201335)8912319017679619
 Asunaprevir (BMS650032)1598991,4744443,28464
Phase IIMK-51721.27188241.3
NS5APhase IIIDaclastavir (BMS-790052)0.0950.0923120.03
NS5BPhase IIISofosbuvir (PSI-7977; GS-7977)558256658615607774
Phase IIMericitabine (PSI-6130;RG7128)6,9312,6115,4839,2557,98711,808
 Filibuvir (PF-00868554)NiNiNiNiNi138
Figure 2.

Concentration-response curves for genotype 1a, 2a, and 2b full-length viruses treated with NS5B polymerase inhibitors and NS5A inhibitor daclatasvir. Values represent the means of triplicates and standard error of the mean (bars). *J6core-NS2/JFH1 recombinant.[8] In the different graphs, each curve corresponds to a single virus, following the color code as shown at the bottom of the figure. (A) Nucleoside/tide analogs (NI) sofosbuvir (nucleotide) and mericitabine (nucleoside). (B) Non-nucleotide analogs (NNI) VX-222 and filibuvir (thumb II inhibitors). Note that for isolates JFH1, J6, DH8, J8, and DH10, a concentration-response curve could not be generated because the viruses were not inhibited at noncytotoxic concentrations of filibuvir and VX-222. (C) NNI BI207127 (thumb I inhibitor). (D) NS5A inhibitor daclatasvir. For this drug, data were generated in three independent experiments (one including J6 and JFH1, another for TN and previously reported,[12] and a third J8, DH8, and DH10).

Figure 3.

Concentration-response curves for genotype 1a, 2a, and 2b full-length viruses treated with lead PIs. Values represent the means of triplicates and standard error of the mean (bars). *J6core-NS2/JFH1 recombinant.[8] In the different graphs, each curve corresponds to a single virus, following the color code indicated. (A) Licensed PIs telaprevir and boceprevir. (B) Phase III PIs vaniprevir, simeprevir (for TN and J6, the experimental data from this inhibitor were previously reported[10, 12]), asunaprevir, and faldaprevir. (C) Phase II PI MK-5172.

To generate an efficient DH10_LSG recombinant, we initially tested the adaptive potential of V1951A (NS4B). After transfection, DH10_LSG/V1951A spread at day 49, confirming that V1951A was not sufficient for adaptation. We then combined seven aa changes observed in the first passage of DH10_LSG, generating two recombinants, DH10_LSG/G351S/Y792N/A992V/I1824V/N1931X/V1951A/D2434N (X corresponds to T or S, respectively). After transfection, both viruses behaved similarly, spreading at day 5 and reaching 3.6 and 3.3 log10 FFU/mL, respectively (Fig. 1B; Supporting Table 3). The N1931T virus had several dominant coding changes, whereas the N1931S virus only showed one dominant coding change, N534T (E2); this latter recombinant (KF420340) was therefore named “DH10cc.” Overall, we had demonstrated that it was possible to generate a 2b culture system by using the ORF from a patient isolate inserted into a vector with J8 5′ and 3′ UTRs.

HCV Genetic Divergence Influences the Sensitivity to Lead DAAs

Our developed full-length culture systems permit the study of front-line HCV DAAs in the context of the complete HCV life cycle. Thus, we determined the efficacy of DAAs targetting different viral proteins against genotypes 1a (strain TN[12]), 2a (strain J610 and recombinant J6core-NS2/JFH18, referred to as JFH1 because it has the JFH1 protease, NS5A and NS5B), and 2b (strains DH8, J8, and DH10). For 2b viruses, we used third-passage DH8_LSG, fifth-passage J8cc, and third-passage DH10_LSG (Tables 1 and 2 and Supporting Tables 2 and 3).

Analysis of NS5B Inhibitors

We analyzed the efficacy of the most advanced NS5B nucleoside/tide (NIs) and non-nucleoside inhibitors (NNIs), currently in phase II or III clinical trials (Fig. 2A-C; Table 3). Polymerase inhibitors had not previously been tested on genotype 2b viruses, and in addition, the NNIs had not been tested on genotype 1a viruses.

Sofosbuvir (nucleotide analog) and mericitabine (nucleoside analog) both inhibited 1a, 2a, and 2b viruses in a dose-dependent manner (Fig. 2A). However, sofosbuvir had lower EC50 values, indicating higher potency. For both drugs, the most sensitive virus was J6(2a); the remaining viruses had similar EC50 values.

In contrast to NIs, NNIs apparently lack pan-genotypic activity. In this study, we tested two thumb II inhibitors (VX-222 and Filibuvir) and a thumb I inhibitor (BI207127). VX-222 and filibuvir inhibited TN(1a) in a dose-dependent manner (Fig. 2B), with VX-222 being most potent. However, these two NNI drugs had no or very limited inhibition of the 2a and 2b virus strains at noncytotoxic drug concentrations (Fig. 2B). In contrast, thumb I inhibitor BI207127 was able to inhibit all viruses in a dose-dependent manner (Fig. 2C), being as potent against most genotype 2 strains as sofosbuvir. TN(1a) appeared to be the most sensitive virus to this drug, and moreover, BI207127 was the polymerase inhibitor that most efficiently suppressed infectivity of the 1a virus.

Analysis of NS5A Inhibitor

We assessed the activity of lead NS5A inhibitor daclatasvir, which has not been previously tested on genotype 2b viruses. Overall, this drug inhibited all viruses and showed high potency (Fig. 2D). Daclatasvir is the inhibitor that shows the lowest EC50 values of all antivirals tested in this study (Table 3); however, it is also the drug that shows the higher differences in activity among strains of the same genotype and subtype. Viruses could be divided into two groups: those inhibited at sub-nanomolar drug concentration, such as JFH1(2a), DH8(2b), and TN(1a), and those inhibited at nanomolar concentration, such as J6(2a), J8(2b), and DH10(2b).

Analysis of NS3/4A Protease Inhibitors

The protease inhibitors (PIs) selected in this study have not been previously tested against genotype 2b viruses, and in addition, the data on the MK-5172 are the first reported on full-length 1a viruses. Tested PIs inhibited all viruses in a dose-dependent manner (Fig. 3A-C; Table 3). For the two licensed PIs, boceprevir and telaprevir, J8(2b) and TN(1a) were the most sensitive. TN(1a) was the most sensitive to other PIs in phase III, except for simeprevir, where J8(2b) had a lower EC50 value. Phase II MK-5172 showed higher potency than any other PIs, with TN(1a) and JFH1(2a) being the most sensitive viruses. Overall, licensed and phase III PIs had differential efficacy against 1a, 2a, and 2b full-length viruses, with some marked differences between isolates of genotype 2b. Phase II MK-5172 showed remarkably higher potency than other PIs tested, but also displayed differential efficacy against strains from the same genotype or subtype.

Development of an Efficient J8 Full-Length Culture Virus With Authentic Protease Sequence Reveals That Adaptive Mutations Influenced Sensitivity to PIs

Because of the different pattern of susceptibility to PIs observed for J8(2b), in comparison to DH8(2b) and DH10(2b), we hypothesized that adaptive mutations present in the protease domain of NS3 in the J8cc fifth passage virus (Table 2) might have been responsible for the general increase in sensitivity of this virus; DH8 and DH10 viruses did not have protease mutations. Therefore, we set out to generate an efficient J8 virus that was not dependent on adaptive protease mutations.

Because combination of LSG with mutations in p7 (L758S), NS4B (A1790T and V1951A), and in NS5A (either I2439T or I2440T, which had also emerged in J8cc passages) adapted DH8, and the resulting viruses did not have any protease mutations, we tested their cross-isolate adaptation potential in J8. We generated the recombinant J8_LSG/L758S/A1790T/V1951A/I2440T (KF420339), which spread at day 5 after transfection, reaching titers of 3.8 log10 FFU/mL (Fig. 1B). The second-passage virus ORF sequence had only two changes, both in the NS3 helicase domain (Table 2). Thus, we had generated an efficient J8 virus with authentic NS3 protease (designated J8_LSG/STAT).

We performed new PI treatment assays using J8_LSG/STAT (second passage) and the original J8 (with protease adaptive mutations) for direct comparisons (Fig. 4). For all drugs tested, J8_LSG/STAT showed decreased susceptibility, when compared to the original J8, with protease adaptive mutations, and the EC50 values for J8_LSG/STAT were similar to those for DH8 and DH10 (Fig. 4; Table 3), suggesting a role of adaptive aa changes G1154A and/or A1208T in increasing susceptibility to PIs. G1154A (G124A in NS3 protein numbering) is located in a beta strand, and nearby residues are known to confer resistance to PIs (Supporting Fig. 1).[15] Glycine is the most frequent residue at this position (Los Alamos HCV Sequence Database; LAL), and alanine is present at low frequency. Regarding A1208T (A178T in NS3), although alanine is present in other deposited sequences, threonine is the most common aa at this position and is located in an alpha helix stretch.

Figure 4.

Concentration-response curves for J8 full-length viruses with and without adaptive mutations in the NS3/protease domain treated with lead PIs. Values represent the means of triplicates and standard error of the mean (bars). Solid lines refer to the original J8 polyclonal virus with adaptive mutations in NS3 (J8cc fifth passage), and dashed lines represent J8 without protease adaptive mutations (J8_LSG/STAT second passage). EC50 values (nM) of the corresponding drug for the two viruses are indicated in the individual graphs. (A) Licensed PIs telaprevir and boceprevir. (B) Phase III PIs vaniprevir, simeprevir, and asunaprevir. (C) Phase II PI MK-5172.


Full-length HCV culture systems that represent all viral genotypes and important subtypes will benefit drug and vaccine development for this important human pathogen. However, thus far, only adapted recombinants of JFH1(2a), J6(2a), and TN(1a) strains yielded high-titer cultures. In the present study, we developed efficient full-length cell-culture systems for genotype 2b strains DH8, J8, and DH10, of which DH8 and J8 yielded high infectivity titers. The main characteristics of the most efficient recombinants developed for isolates DH8, J8, and DH10 are summarized in Table 4. Recombinants were adapted to growth in Huh7.5 cells, and a defined set of mutations could adapt two strains, proving a cross-isolate effect. Efficient cell culture of DH10 was achieved by determining only the patient-derived ORF sequence, which was inserted into a cassette containing the 5′ and 3′ UTR from J8. These findings might facilitate the development of HCV culture systems for other genotype 2b isolates, as well as for other genotypes and subtypes.

Table 4. Major Characteristics of Developed Genotype 2b Recombinants
Recombinant NameAdaptive aa Changes (Protein, Substitution, and Number)
(All Based on LSG Mutations)TransfectionSecond PassageCore (1-191)E1 (192-383)E2 (384-750)p7 (751-813)NS2 (814-1030)NS3 (1031-1661)NS4A (1662-1715)NS4B (1716-1976)NS5A (1977-2442)NS5B (2443-3033)
  1. Brief description of the most relevant recombinants developed in this study. Peak infectivity titers are indicated, both for transfection and second-passage viruses. Engineered adaptive mutations (aa substitution and number, according to 2b genotype reference sequence, J8CF [JQ745651]) are localized in the corresponding gene/protein. Dash indicates no mutation in the corresponding gene or protein. LSG in bold corresponds to aa substitutions L(F1468L), S(A1676S), and G(D3001G).

  2. a

    Mutations G1154A and A1208T are located in the protease domain of NS3.

DH8_LSG/PATF3.83.9L884PLSV1951AI2440TG, L3021F
DH8cc4.14.3L758SLSA1790T, V1951AI2439TG
J8cc-HT4.74.3M292LL612MF772CW864RG1154Aa, A1208Ta, N1217Y, LSQ1763R, I1968VE2263V, I2440TH2922R, G
J8_LSG/STAT3.84.1L758SLSA1790T, V1951AI2440TG
DH10cc3.63.0G351SY792NA992VLSI1824V, N1931S, V1951AD2434NG

We tested, for the first time, lead HCV inhibitors against full-length genotype 2b viruses. Our results reveal a differential activity of these drugs toward full-length genotype 1a, 2a, and 2b viruses. Genotype 2b is less sensitive to most PIs, and the efficacy of daclatasvir against genotypes 2a and 2b is largely influenced by HCV variability at the isolate level. We have demonstrated that not only NIs (sofosbuvir and mericitabine) are active against non–genotype 1 viruses, but also the NNI, BI201127, possesses activity against full-length 1a, 2a, and 2b viruses, suggesting unique pangenotypic properties among NNIs.

Development of infectious full-length cell-culture systems for HCV has been a major challenge, because molecular clones of HCV generated from patient sequences do not spontaneously replicate and spread in vitro. Approaches that used subgenomic replicon-derived mutations to adapt full-length clones have only led to culture systems with relatively low infectivity,[9, 18] possibly because replicon mutations introduce constraints in viral production.[18, 19] We recently identified three mutations in the NS3 helicase, NS4A, and NS5B, named LSG, that promoted adaptation of HCV genotype 1 and 2 full-length clones,[10, 12] and used them to adapt novel genotype 2b isolates in the present study. Similarly to J8,[10] LSG permitted in vitro growth of DH8 and DH10. Besides LSG, additional mutations were required to produce viruses with high infectivity titers. These mutations represent unique aa that are rarely present in natural patient-derived sequences. A1951 and V1968, in NS4B, are found in less than 1% of genotype 2b sequences (LAL), and T2439 or T2440 are not present in any of the 73 deposited NS5A 2b sequences. In addition, we had previously demonstrated that changes at aa 1,931 (NS4B), 1,968 (NS4B), and 2,439 (NS5A) constitute key adaptive mutations in cell culture systems.[10, 12]

We also demonstrated that it is possible to develop functional chimeric genomes by inserting the ORF of a 2b isolate into a cassette vector with 5′/3′ UTRs of another 2b isolate, for which the UTRs were known to be functional in vitro. This finding is of major relevance for the culture of clinical isolates, because the sequence of the UTRs is technically difficult to obtain. Similar chimeric genomes have been proven functional in vivo.[20]

The full-length cell-culture systems permit us to explore the evolutionary potential of HCV. Viruses can adapt to cell culture by acquiring different combinations of mutations and maintaining WT sequences in specific genes or domains. We generated genotype 2b viruses that did not have any changes in the NS3 protease, NS5A domain I, and NS5B finger, palm, and thumb domains (except the c-terminal portion), making them optimal tools for the study of most DAAs. Contrarily, replicon-based systems often accumulate mutations in the protease domain of NS3,[6] potentially affecting the natural isolate sensitivity toward PIs. The importance of using viruses without cell-culture adaptive mutations in the NS3 protease domain was demonstrated for the J8 isolate, because viruses with mutations G1154A and A1208T had increased drug sensitivity, when compared to J8 viruses without these changes. Similarly, we succeeded in developing cell-culture–adapted viruses for DH8, which did not contain mutations in p7, and which could be of importance for functional or drug studies targeting this important viral protein. However, we cannot exclude that adaptive changes present in other proteins, outside the drug targets, might affect drug sensitivities of the cell-culture–adapted viruses.

We tested sensitivity of full-length viruses to selected front-line DAAs targeting NS3/4A protease, NS5A, and NS5B polymerase.[21] Currently, there are very limited data on the activity of PIs in genotype 2 patients; small studies have suggested a benefit of telaprevir and boceprevir when added to therapy with IFN/RBV.[5] In our in vitro systems, most PIs had higher activity against TN(1a) than against genotype 2 isolates. Among genotype 2, isolates from subtype 2b were generally less sensitive to tested PIs, in comparison with 2a isolates. These findings stress the importance of subtype determination in the clinical setting, which may be of more relevance in the era of DAA-based therapy.

Our data on PI MK-5172, highlighting its exceptional higher antiviral potency, when compared with other PIs, represent the first reported testing of this drug in HCV full-length cell-culture systems of various genotypes. Similarly to MK-5172, NS5A inhibitor daclatasvir was shown to be a highly potent HCV inhibitor, but its activity was most influenced by HCV genetic divergence at the isolate level for genotype 2b, as previously indicated for genotypes 1a and 2a.[16, 17]

NS5B inhibitors currently in phase II and III clinical trials are among the most promising anti-HCV drugs. However, they have not been extensively studied in cell-culture viruses for different genotypes and subtypes because of the lack of culture systems with genotype-specific polymerases. In the present work, we report on the effect of front-line polymerase inhibitors on full-length viruses of genotypes 1a, 2a, and 2b. As expected, NIs mericitabine and sofosbuvir were active not only against 1a, but also against 2a and 2b, which is in agreement with their apparent pan-genotypic activity in patients infected with genotypes 1, 2, and 3, in combination with IFN/RBV.[22, 23]

We investigated the activity of front-line NNIs currently in phase II (filibuvir, VX-222, and BI207127). Filibuvir has been reported to significantly reduce HCV titers in clinical studies, when used in monotherapy, in genotype 1–infected patients.[24] Our data support the efficacy of filibuvir and VX-222 against genotype 1, but reveal limited or no activity against genotype 2 viruses. Contrarily, the NNI, BI207127, was active against all viruses. Efficacy against genotype 2 was similar to that of sofosbuvir, whereas efficacy against TN(1a) was higher for BI207127 than for sofosbuvir. This difference might be explained by the fact that NNIs target different pocket (or allosteric) sites of the RNA polymerase, which present higher genetic variability among HCV genotypes. In addition, they could be specific for genotype 1 because they have been mostly developed using Con1 replicons. Therefore, significant differences in antiviral activity against genotype 1 and non-1 viruses are expected. Among the different allosteric sites that are targeted by current advanced thumb NNIs, our results with BI207127 support that thumb pocket I is the preferable target of future NNIs with potential pangenotypic activity.

In conclusion, we have established efficient cell-culture systems for three HCV genotype 2b isolates. Approaches applied in viral adaptation should have relevance for advancing culture development for other 2b isolates and, perhaps, other genotype strains. These systems represent authentic patient-derived sequences in all the corresponding targets of the most relevant DAAs, making them optimal models for the study of antivirals. These full-length systems will, for the first time, permit the study of combination treatment with drugs targeting all structural and nonstructural proteins in genotype 2b. They will allow future studies promoting escape to current DAAs, in the context of the whole genetic background of HCV. Finally, because the viruses generated mimics all the steps of the HCV life cycle, they will permit genotype-specific functional studies of all viral proteins, thus promoting drug and vaccine development.


The authors thank J.O. Nielsen and O. Andersen for providing valuable support and L. Mikkelsen, A-L. Sørensen, and L. Ghanem for general laboratory assistance (all from Copenhagen University Hospital, Hvidovre).