Genotype differences in susceptibility and resistance development of hepatitis C virus to protease inhibitors telaprevir (VX-950) and danoprevir (ITMN-191)

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Protease inhibitors (PIs) have proven to be effective adjuncts to interferon/ribavirin treatment of hepatitis C virus (HCV) infections. Little clinical or in vitro data exists, however, on their effectiveness for nontype 1 genotypes that predominate in Europe, the Middle East, Africa, and most of Asia. NS3 protease and NS4A genes from genotypes 1-6 were inserted into the JFH clone to generate replication-competent intergenotype chimeras. Susceptibility to PIs was determined by replication and infectivity assays. To study resistance development, chimeras were cultured in subinhibitory concentrations of PIs and mutations phenotypically characterized. Marked differences in susceptibility of different genotypes to danoprevir (ITMN-191) and telaprevir (VX-950) were observed. Genotypes 1, 4, and 6 showed median inhibitory concentration (IC50) values of 2-3 nM, >100-fold lower than genotypes 2/3/5 (250-750 nM). Telaprevir susceptibilities varied over a 4-fold range, with genotypes 1 and 2 being most susceptible and genotypes 4 and 5 most resistant. Culture of genotypes 1-6 in PIs induced numerous mutations in the NS3 protease domain, highly variable between genotypes. Introduction of danoprevir and BILN 2061-induced mutations into the original clones by site-directed mutagenesis (n = 29) all conferred resistant phenotypes, with particularly large increases (1-2 log greater IC50 values) in the initially susceptible genotypes 1/4/6. Most introduced mutations and showed little or no effect on replicative fitness. Conclusion: Major differences were found between genotypes in their susceptibility and resistance development to PIs. However, equal sensitivities of genotypes 1, 4, and 6 to danoprevir and a broader efficacy range of telaprevir between genotypes than initially conceptualized provide strong evidence that PIs might be effectively used beyond their genotype 1 target group. (HEPATOLOGY 2011;)

Infections with hepatitis C virus (HCV) are a major cause of chronic liver diseases, such as cirrhosis and hepatocellular carcinoma.1 Clinical management of HCV and the effectiveness of currently used interferon-α (IFN)/ribavirin (RBV) therapy of infected individuals is greatly influenced by genetic heterogeneity of the virus. There are, for example, major differences between the six genotypes of HCV in response rate to therapy and evidence for some genotype-associated variability in the rate of disease progression and associated liver pathology.2, 3 HCV replication is additionally associated with high mutation rates; this confers on HCV, in common with human immunodeficiency virus 1 (HIV-1), considerable adaptive capacity to escape from immunological or drug-treatment pressure. The effectiveness of newly developed protease and polymerase inhibitors for HCV, at least as monotherapy, is indeed likely to be substantially impaired through the acquisition or selection for preexisting amino acid mutations that confer antiviral resistance.

Genetic heterogeneity between HCV genotypes translates into significant molecular and clinical differences. For example, individuals infected with genotype 1 or 4 show lower response rates to the current standard of care of IFN/RBV combination treatment than those infected with genotype 2 or 3.4-6 Furthermore, substantial differences were also reported in the susceptibility of the individual genotypes towards the different antivirals currently in clinical trials.7 The first widely used protease inhibitor (PI), BILN 2061, was developed based on the structure of the NS3 protease of genotype 1. In early clinical trials it was found to be substantially less effective in individuals infected with genotype 2 or 3.8-11 Similarly, VX-950 (telaprevir), another PI, showed potent activity against HCV genotypes 1 and 2,12 but almost no efficacy against genotypes 3 and 4.13, 14

Genotype 1-infected individuals have been almost exclusively targeted for antiviral therapy in current, ongoing clinical trials, partly because of the lack of information about the true effectiveness of PIs against nontype 1 genotypes and because the response rate of type 1 to conventional IFN/RBV therapy is problematically low (40%-50% clearance) compared to genotypes 2 and 3 (˜80%).4 Genotype 1 is highly prevalent in the USA, Europe, and the Far East15 and therefore represents a treatment priority. This generic focus, although understandable, does, however, ignore growing problems with clinical management and therapy of other genotypes, particularly genotypes 4 and 6, which frequently respond poorly to IFN/RBV and which are extensively distributed and rapidly spreading throughout Southern Europe, the Middle East, and South East Asia.2

Assessment of the efficacy of PIs against different genotypes has been greatly hampered by the lack of a convenient animal model or a method for in vitro culture of HCV other than the type 1/2-based replicons and the infectious genotype 2a clone, JFH1. To address this problem, we have previously constructed a panel of replication competent chimeric Jc1 (pFK JFH1/J6/C-846) clones containing protease and NS4A coding sequences from all six major genotypes16 (Fig. 1; Supporting Information Fig. S1). RNA transcripts expressed from these clones are infectious on transfection and virus from transfections can be used to determine directly the susceptibility of different genotypes to PIs. In the current study, this in vitro phenotypic assay was used to investigate susceptibilities of genotypes 1-6 to the two structurally distinct PIs in advanced clinical trials (telaprevir and danoprevir). We have additionally genetically and phenotypically characterized a large number of mutations that developed on in vitro passage of different genotypes in subinhibitory concentrations of each PI and determined their effect on viral replication fitness. This study provides the first evidence-based assessment of the applicability of PIs to nontype 1 genotypes using the full-length viral replication cycle, and may contribute to the more effective use of antiviral therapy in future HCV management.

Figure 1.

Genome diagram of HCV showing position of inserted NS3 (protease domain) and NS4A (protease cofactor) sequences from heterologous sequences. All positions are numbered relative to the H77 numbering reference sequence.40 Further information on clone construction and restriction site creation is provided elsewhere.16

Abbreviations

HCV, hepatitis C virus; IFN, interferon; PIs, protease inhibitors; RBV, ribavirin.

Materials and Methods

Construction of Intra- and Intergenotypic Recombinants.

Construction and the successful expression of replication-competent virus from the intra- and intergenotypic recombinants J1b1b, J2a2a-T1066S, J3a3a, J4a4a-19, J5a5a-Q1247L, and J6a6a-V1040L with Jc117, 18 has been described.16 For reverse genetic studies, mutations were introduced using mutated primers with the Quick Change Site-Directed Mutagenesis Kit (Stratagene). Modified fragments were verified by sequencing.

Huh7.5 Cell Culture, RNA Synthesis, Transfection, and Immunostaining.

Procedures for cell culturing, RNA synthesis, transfection, and immunostaining were described.16 Briefly, linearized and blunted DNA templates were cleaned by phenol/chloroform extraction, followed by ethanol precipitation and RNA synthesized using T7 RNA Polymerase (Promega). RNA was transfected into Huh7.5 cells by electroporation and viral spread assessed by NS5A immunostaining with polyclonal sheep anti-NS5A serum. RNA transcripts from the replication-deficient JFH1-based genome containing the GND mutation in the NS5B polymerase served as a negative and that of Jc1 as positive control, respectively.

Drug Inhibition Studies.

The HCV-specific PI BILN 2061 (a gift from GlaxoSmithKline) was resuspended at 5 mM in dimethylsulphoxide, telaprevir, and danoprevir (both purchased from Acme Bioscience, Palo Alto, CA) at 20 mM dimethylsulphoxide. The effect of the different PIs on the replication of the intra- and intergenotypic recombinants was assessed as described.16 Briefly, RNA was transfected into Huh7.5 cells and cultures incubated with or without PIs. Alternatively, naïve cells were infected with infectious supernatant, washed, and incubated with or without PIs. Antiviral efficacy was measured as relative inhibition of RNA replication by staining for NS5A and determining the percentage of HCV-positive cells or counting the number of foci forming units/mL. Each concentration was assayed in triplicate.

In Vitro Selection and Susceptibility Testing of PI-Resistant Recombinants.

Intra- and intergenotypic recombinants were passaged for 2 to 3 weeks under subinhibitory concentrations of PIs as described.16 To establish an infection, recombinants were first passaged for 3 to 4 days without PIs. Through cell splitting, recombinants were further passaged with PIs and 10 clones of each culture sequenced to identify variants with low and high copy numbers. To determine whether identified substitutions conferred resistance to the corresponding PI, they were introduced into the original recombinant by site-directed mutagenesis. Recombinants containing single mutations were then assessed for their susceptibility towards PIs as described above.

Role of the Funding Source.

No contribution to the study design, collection or analysis of data, writing, or publication decision was made by the funding source (BBSRC).

Results

Susceptibility of HCV Genotypes to Danoprevir and Telaprevir.

The intra- and intergenotypic chimeric recombinant viruses J1b1b, J2a2a-T1066S, J3a3a, J4a4a-19, J5a5a-Q1247L, and J6a6a-V1040L were chosen to represent each major genotype. Naïve Huh7.5 cells were infected with infectious virus or alternatively electroporated with synthetic RNA and the reduction in frequency of NS5A-positive cells upon PI treatment assessed (Fig. 2A; individual median inhibitory concentration [IC50] values and a comparison with those from BILN 2061 are listed in Table 1). Danoprevir showed a similar efficacy pattern of genotype susceptibility we determined for BILN 2061, a structurally similar cyclic PI.16 The chimeras J1b1b, J4a4a-19, and J6a6a-V1040L showed a 100- to 400-fold greater susceptibility postelectroporation than J2a2a-T1066S, J3a3a, and J5a5a-Q1247L (Fig. 2A). The susceptibility of the infectious clones J2a2a-T1066S, J5a5a-Q1247L, and J6a6a-V1040L to danoprevir as determined by measurement of infectivity reductions closely matched results from the replication assays. In particular, genotype 6a showed a similar greater susceptibility to danoprevir treatment postinfection than genotypes 2a, 3a, and 5a (Fig. 3A).

Figure 2.

Antiviral Inhibition of Jxx's in transfected cells. One to 10 μg RNA was electroporated into Huh7.5 cells and incubated for 24 hours. Cells were then washed and incubated in media containing 0.1% DMSO, as a carrier control, with or without the indicated doses of (A) danoprevir or (B) telaprevir for a further 72 hours. The percent inhibition of replication was determined at 96 hours postelectroporation (mean ± standard error of the mean [SEM]; n = 3), and calculated as the ratio of NS5A-positive cells in PI-treated cells to those of the no-antiviral control.

Figure 3.

Antiviral inhibition of Jxx's in supernatant infected cells. After 8 hours inoculation with Jxx (MOI 0.015), Huh7.5 cells were washed and incubated in media containing 0.1% DMSO, as a carrier control, with or without the indicated doses of (A) danoprevir or (B) telaprevir. Inhibition was calculated at 72 hours postinfection as reduction in foci forming units/mL (mean ± SEM; n = 3) after antiviral addition compared to foci forming units/mL of the no-antiviral control.

Table 1. IC50's for Chimeras of Genotypes 1-6 Against Three PIs*
CloneGenotypeBILN 2061DanoprevirTelaprevir
  • *

    IC50 values for BILN-2061 have been reported16 and are included here for comparative purposes.

  • Dose-response curves (Figs. 2 and 3) were fitted to a four parameter logistic function to obtain half maximal inhibitory concentrations (IC50s) using SigmaPlot.

(A) Postelectroporation
 J1b1b1b3nM3nM830nM
 J2a2a-T1066S2a720nM750nM1100nM
 J3a3a3a105nM280nM1410nM
 J4a4a-194a1nM2nM2300nM
 J5a5a-Q1247L5a480nM670nM2700nM
 J6a6a-V1040L6a1nM2nM650nM
CloneGenotypeBILN 2061DanoprevirTelaprevir
(B) Postinfection
 J2a2a-T1066S 2a210nM150nM520nM 
 J3a3a3a80nM130nM600nM
 J5a5a-Q1247L5a110nM60nM1360nM
 J6a6a-V1040L6a2nM0.05nM150nM

Genotype-associated differences in susceptibility to telaprevir were also observed. This linear PI showed the greatest efficacy posttransfection in genotype 1b and 6a-infected cells (Fig. 2B, Table 1; IC50 values of 840 and 650 nM). Genotypes 2a and 3a showed intermediate susceptibility to telaprevir posttransfection (1,100 and 1,410 nM), whereas genotypes 4a and 5a were resistant (2,300 and 2,700 nM). Similarly, genotype 6a infectivity reductions were over 5 times greater than those of genotypes 2a, 3a, and 5a (Fig. 3B).

In Vitro Selection of PI-Resistant Recombinant Viruses.

To identify mutations conferring resistance to the PIs, Huh7.5 cells, transfected with the different recombinants, were passaged by cell splitting under subinhibitory concentrations of the individual PI as described.16 The addition of increasing concentrations of PI did not result in any visual cytopathic effects. Substitutions indicative of resistance development were those that occurred in both replicas of the experiment but not in the control, many of which corresponded to those previously observed in treated patients.19-24 Although in our previous study,16 passaging in BILN 2061 drove the emergence of mutations only at positions 156 and 168 in all six genotypes, passaging under danoprevir resulted in a diverse pattern of substitutions (Fig. 4; Table 2). Besides substitutions at known resistance loci (NS3 amino acid positions 36, 41, 43, 138, 168), potential resistance mutations were also identified at positions 28, 77, 80, 85, 98, 133, 160, and 174, of which those at 77, 80, and 174 °Ccurred in both replicates of the same genotype and absent in the control passage, observations clearly indicative of resistance-induction. The mutation Q41R was found either as a double mutant with E168A or as a triple mutant with E28G and E168A, but not as a single mutant. P85L was only observed in combination with T98R (Supporting Information Fig. S3).

Table 2. Acquisition of Mutations in NS3 on Passage with Danoprevir
DanoprevirProportion Remaining Wildtype - Mutations
CloneGenotypeConc. / nM*DayPositionWTReplicate 1Replicate 2Control
  • *

    Final passage concentration of danoprevir.

  • Numbers of clones retaining original amino acid in two replicate passage experiments (10 clones analysed in each) and in a control passaged without danoprevir (10 clones analyzed). Mutations are indicated if they occurred in at least two clones or were previously reported.

J1b1b1b3009138S100%90% - 1P100%
    168D30% - 3A/V, 1G30% - 2A/Y, 1E/G/V100%
    174S70% - 3P60% - 4P100%
J2a2a-T1066S2a10,0002185P100%60% - 4L100%
    98T100%60% - 4R100%
    160S%0 - 10C100%100%
    168D100%50% - 5A100%
J3a3a3a10,0002143F90% - 1S100%100%
    77N90% - 1S90% - 1S100%
    80Q80% - 2R80% - 2R100%
J4a4a-194a3009168D30% - 4A, 3G10% - 5V, 2A/G100%
J5a5a-Q124L5a7,0002136L100%90% - 1P100%
    133S70% - 3G100%100%
    168E50% - 3A, 2K60% - 4A100%
J6a6a-V1040L6a3002128E80% - 2G100%100%
    41Q100%70% - 3R100%
    168D0% - 5H, 4A, 1E0% - 7A, 2Y, 1T100%
Figure 4.

Position of resistance-induced mutations in the protease gene on passaging genotypes 1-6 through danoprevir and telaprevir. Alignment of the NS3 protease gene for genotypes 1-6, showing its main secondary structural features and catalytic site (based on the structure annotations in41). The positions and relative frequency of amino acid changes occurring on passage in danoprevir and telaprevir are shown above and below each sequence (red and blue filled circles), with the substituted amino acid identified (x indicates more than one different amino acids were substituted at that site; full information on the frequencies and genetic linkage is provided in Tables 2, 3, and Figs. S3-S4 in the Supporting Material).

Table 3. Acquisition of Mutations in NS3 on Passage with Telaprevir
TelaprevirProportion Remaining Wildtype - Mutations
CloneGenotypeConc. / nM*DayPositionWTReplicate 1Replicate 2Control
  • *

    Final passage concentration of telaprevir.

  • Proportion of clones retaining original amino acid in two replicate passage experiments (10 clones analyzed in each) and in a control passaged without telaprevir (10 clones analyzed). Mutations are indicated if they occurred in at least two clones or were previously reported.

J1b1b1b1,8001336V100%90% - 1A100%
    174S60% - 4P70% - 3P100%
J2a2a-T1066S2a10,0002126K100%30% - 7R100%
J3a3a3a5,0002377N10% - 9S0% - 10S100%
J4a4a-194a5,0001336L100%90% - 1P100%
    54T100%90% - 1A100%
    156A100%90%- 1V100%
J5a5a-Q1240L5a15,00016155A10% - 7S, 2T0% - 7S, 3T100%
J6a6a-V1040L6a6,0002136V90% - 1A100%100%
    54T100%90% - 1A100%
    77N100%0% - 10S100%
    156A0% - 8S, 2V100%100%

Previously described resistance loci found at position 36, 54, 155, 156 in genotype 1-infected patients treated with telaprevir were represented in one or several clones of virus passaged in vitro under PI selection, along with the additional V36A+A156S and T54A+N77S double mutations (Fig. 4; Supporting Information Fig. S4). Replacement of the wildtype codon was furthermore observed at position 174 (genotype 1b) and 77 (genotypes 3a and 6a). These latter mutations have also been found during danoprevir therapy, indicating mutations potentially conferring cross-resistance. As previously shown for BILN 2061,16 the pattern of resistance-associated mutations under danoprevir and telaprevir was highly diverse among the different genotypes, indicating major mechanistic differences in invivo resistance development.

Influence of Amino Acid Substitutions on Viral Replication and PI Susceptibility.

To investigate the influence of the acquired mutations on the viral replication fitness of the individual intra- and intergenotypic recombinants, substitutions were reintroduced into the original recombinants. Mutants were passaged in Huh7.5 cells without antivirals by cell splitting and the spread within the cell culture compared to that of the wildtype (Supporting Information Fig. S2; Supporting Information Material). Most substitutions did not have an obvious effect on the spread of the intra- and intergenotypic recombinants within the cell culture. Interestingly, the same substitutions showed different effects on different genotypes; for example, D168G and A156V reduced replication of genotype 1b but had no effect on 4a. Similarly, D168V dramatically reduced the replicative fitness of genotypes 2a and 6a recombinant viruses, but showed no obvious phenotype with genotypes 1b and 4a.

In order to determine the precise phenotypic effect on PI susceptibility of mutations developing under antiviral pressure, recombinants with the individual substitutions inserted were assessed for their susceptibility towards BILN 2061 and danoprevir. As expected, those at previously described resistance loci conferred an increase in resistance towards the PIs (Fig. 5), as did each of the newly documented substitutions that developed under treatment pressure investigated in the current study. Mutations within genotypes 1b, 4a, and 6a induced the highest fold change in susceptibility towards BILN 2061 and danoprevir, whereas only moderate increases in resistance were induced in genotypes 2a, 3a, and 5a. Generally, mutations from polar, negatively charged amino acids to nonpolar, neutral or positively charged amino acids conferred the highest fold change in susceptibility. Overall, the increase in resistance for the same mutation was 3- to 6-fold higher against BILN 2061 than danoprevir and 1- to 2-fold higher within genotype 1b compared to genotype 4a. Decreased PI susceptibility did not correlate with impaired replicative fitness; for example, the mutant recombinant conferring the greatest resistance of those analyzed (J6a6a-V1040L-D168H), replicated with similar efficiency in the cell culture as the wildtype.

Figure 5.

Influence of PI-induced mutations on antiviral susceptibility. Amino acid substitutions within the NS3 protease region identified during passaging under PIs were reintroduced into the original recombinants and their influence on the susceptibility towards (A) BILN 2061 and (B) danoprevir assessed. Mutant names indicate the genotype of the original recombinant and the amino acid substitution. Note the difference in the y-axis scale between (A) and (B).

Discussion

The genetic variability of HCV proteins of different genotypes influences the structure of protease and polymerase enzymatic sites and potentially limits the effectiveness of antiviral therapy targeting viral replication proteins.25 Because antiviral drugs were and continue to be developed using genotype 1-based enzymatic assays, they have not been optimized for and frequently show lower efficacies against other genotypes. These differences translate into marked variability in clinical response rates between genotypes for several antivirals, among them the two PIs BILN 2061 and telaprevir8, 10, 12, 13 investigated in the current study. Reduced effectiveness of antiviral drugs for certain genotypes not only unnecessarily exposes patients to severe side effects, but also potentially facilitates the development of resistance mutations. As nongenotype 1 infections become more widespread around the world, it is clearly important to expand the evaluation and potential clinical use of antiviral therapy.

BILN 2061 and its derivate danoprevir are both macrocyclic protease inhibitors, with similar mechanisms of inhibition.26, 27 Using our in vitro assay, we previously reported genotypes 2a, 3a, and 5a to have 100- to 700-fold greater IC50s to BILN 2061 than 1b, 4a, and 6a.16 This concurs with the available clinical data that documents BILN 2061 to be less effective in reducing viral replication in individuals infected with genotype 2 and 3 compared to type 18, 10, 28 and lends support to the value of the in vitro system to predict PI efficacies. Genotype profiles obtained in the current study may therefore be predictive for clinical response.

Our observations that danoprevir was equivalently effective against genotypes 4a and 6a as it was against 1b (each 100- to 350-fold more susceptible than genotypes 2a, 3a, and 5a) provides the clearest indication that 4a and 6a may indeed be equally effectively treated as type 1 in clinical practice. These data are consistent with purely enzymatic data showing genotypes 1b, 4, and 6 are more readily inhibited by danoprevir than genotype 3a,29 despite the 10-100 fold lower IC50 values for each of the genotype constructs assayed that characterize enzymatic assays more generally. With genotypes 4, 5, and 6 representing a substantial proportion of all HCV infections worldwide and with 4a and 6a also being poorly responsive to conventional therapy,2 it is crucial that these genotypes are considered in the future development of PIs and other antiviral therapy.

Although not as potent as BILN 2061, telaprevir has been shown to be clinically effective in genotype 1-infected patients.12 In a phase IIa clinical trial, telaprevir also demonstrated substantial activity in genotype 2-infected patients but only limited efficacy in genotypes 3- and 4-infected individuals, for whom as a result treatment was stopped.13, 14 As demonstrated by our in vitro studies, telaprevir also shows considerable differences in potency against different genotypes, observations that highlight again the potential value of evaluating PIs on all genotypes before clinical assessment. We found that genotypes 1b and 6a were the most susceptible to telaprevir, followed by 2a, then 3a, and genotypes 4a and 5a being the most resistant. Based on the in vitro findings, genotype 6a- but not 4a- and 5a-infected patients might therefore be effectively treated by this antiviral in the future. However, in this specific case where relatively smaller genotype-associated differences in IC50 values for telaprevir have been found, it is necessary to investigate the extent of within-subtype variability in susceptibility and whether this might have a significant impact on clinical effectiveness. For example, in previous studies, catalytic efficiencies within a subtype were shown to vary widely (by up to 7-fold), especially within genotypes 1a and 1b.30, 31 We have shown that different isolates of genotype 3a exhibited at least 3-fold variability in IC50 values (130 nM to 310 nM) against BILN 2061, attributable to naturally occurring amino acid changes in the protease domain of NS3.16 These strain-associated differences may indeed account for the discrepancies between genotype 3 and 4 susceptibilities in the in vitro system with clinical susceptibility data.13, 14 However, the chimera model correctly reports their much poorer response compared to genotype 1 and 2.

The rapid selection of drug-resistant genetic variants is a major problem substantially limiting the effectiveness of antiviral therapy for HCV.21 Mathematical modeling has suggested that all possible single- and double-mutant viruses already preexist before treatment32 and can be rapidly selected at the start of antiviral therapy. Identification of potential resistance mutations within the individual genotypes towards the different PIs is crucial to preidentify individuals with preexisting resistant variants and adjust treatment options accordingly. We induced resistance mutations in vitro through passaging the intra- and intergenotypic recombinants under subinhibitory concentrations of PIs. Several new potential resistance loci were identified (Fig. 4; Tables 2, 3).

The development of resistance differed substantially between the different genotypes. Against danoprevir, most genotypes developed substitutions at position 168, confirming the importance of this locus in the resistance mechanism against this class of macrocyclic inhibitors.26, 29 However, genotype 3a showed instead substitutions at position 43, 77, and 80. Position 80 was described as a resistance locus against TMC435, another macrocyclic PI,33 findings that demonstrate that resistance towards macrocyclic inhibitors can not only be induced by a variety of viable changes but also that some confer resistance to all inhibitors of this class. Interestingly, substitutions at position 77 and 174 were both found in passaging with danoprevir and telaprevir, providing further evidence for the potential emergence of cross-resistance between structurally dissimilar PIs. This is further supported by the occurrence of resistance mutations against danoprevir at position 36, 41, and 43 that have been previously identified in linear PIs such as boceprevir and telaprevir.34-36 Resistance loci 43, 41, and 138 observed against danoprevir,37 and at positions 36, 54, 155, 156 against telaprevir21 were reproduced in the in vitro assay, demonstrating its utility for exploring the range of reported resistance-associated mutations in each genotype as well detecting a number of further possible loci.

To confirm that the observed substitutions confer increased resistance to PIs, they were assessed individually for PI susceptibility. Of the 29 tested, all conferred increased resistance towards BILN 2061 or danoprevir. Generally, increases in resistance were higher with mutations in genotypes 1b, 4a, and 6a than genotypes 2a, 3a, or 5a, likely the result of the already intrinsically high resistance barrier of the latter genotypes. Mutations generally conferred smaller increases in resistance against danoprevir than BILN 2061, an observation in agreement with the closer stereochemical fit of BILN 2061 to the NS3 protease compared with danoprevir.27 Our results are in agreement with previous studies showing the highest fold increase in resistance for D168A/V and A156V/T mutations.19, 22

Antiviral drug-resistant mutants vary in their replicative fitness relative to wildtype virus in the absence of drugs.19, 38, 39 Of major clinical concern are mutants that outgrow wildtype during treatment and still replicate to high levels and transmit further following the end of treatment. Assessment of the influence of a mutation on the viral replication kinetic is therefore necessary in the in vitro evaluation of PI resistance. In the current study we observed marked genotype-associated variability in the fitness cost of a range of different mutations. For example, the D168V mutation showed no effect on the replicative ability of the genotypes 1b and 4a recombinants (consistent with previous data obtained from genotype 1b22). However, it dramatically reduced replication of genotype 2a and 6a constructs below the detection limit. Data in the current study on the phenotypic (or fold-resistance) of individual amino acid changes introduced into the genotype chimeras provide the starting point for a system of genotypic assessment of resistance, as widely used for HIV-1 therapy (such as the database http://hivdb.stanford.edu/) and which may be applied for treatment evaluation and appropriate drug selection.

Our in vitro findings demonstrate that complex patterns of susceptibility and resistance development differences exist between genotypes. The simple paradigm of genotype 1-susceptible, nontype 1 genotypes-nonsusceptible that underlies, in part, the current clinical focus on genotype 1 for antiviral therapy is demonstrably incorrect. The macrocyclic inhibitor danoprevir (and BILN 2061) show equivalent effectiveness against genotypes 4 and 6, genotypes that show intermediate response rates to IFN/RBV therapy,2 are highly prevalent on a worldwide basis, and present the greatest problems in clinical management throughout the Middle East and South East Asia. We believe that the in vitro modeling of antiviral susceptibilities and resistance development that we have developed will play an important role in the preclinical evaluation of antivirals and their future clinical targeting.

Ancillary