Impact of a baseline polymorphism on the emergence of resistance to the hepatitis C virus nonstructural protein 5a replication complex inhibitor, BMS-790052

Authors


  • Potential conflict of interest: Drs. Nower, Fridell, Valera, Wang, Gao, Roberts, and O'Boyle own stocks in Bristol-Myers Squibb. Dr. Nettles owns stock in Bristol-Myers Squibb. He also owns stock in and is an employee of Johnson & Johnson.

  • This study was supported by Bristol-Myers Squibb.

Abstract

The influence of naturally occurring polymorphisms on the potency of the HCV nonstructural protein 5A (NS5A) replication complex inhibitor, BMS-790052, was investigated by evaluating hybrid replicons in which the entire NS5A coding region of genotype (GT) la and 1b laboratory (lab) strains (H77c and Con1) were replaced with the corresponding regions of specimens collected from 10 GT-1a- and 6 GT-1b-infected subjects. For baseline (BL) specimens, with no previously observed resistance variants identified by population sequencing, the median 50% effective concentration (EC50) values for BMS-790052 were similar for the clinically derived and lab strains. A Q30R variant was observed at viral breakthrough (VBT) in one of the GT-1a-infected subjects. Because the lowest plasma exposure of BMS-790052 observed in this subject was 117 nM and the median 50% effective concentration value for a GT-1a H77c replicon containing a Q30R substitution is ∼7 nM, a rigorous investigation was initiated to determine the basis for resistance. Three approaches were used: (1) replacement of the entire H77c NS5A or (2) replacement of the N-terminal region of NS5A, with sequence from BL and day 14, and (3) substitution of specific amino acids. A BL polymorphism (E62D) did not contribute resistance to BMS-790052; however, the linked variant, Q30R-E62D, conferred high-level resistance in vitro and is likely responsible for VBT in vivo. Conclusion: Our data show that a BL polymorphism with minimal effect on the anti-HCV effect of BMS-790052 can affect the emergence of resistance and significantly affect clinical outcome. This work establishes a clear, systematic approach to monitor resistance to NS5A inhibitors in the clinic. (HEPATOLOGY 2012;55:1692–1699)

Chronic hepatitis C virus (HCV) infection is one of the most common causes of liver disease and is estimated to affect 170 million people worldwide.1 Many infected patients progress to liver cirrhosis and hepatocellular carcinoma.2 Currently, the most common treatment for chronic HCV infection consists of pegylated interferon plus ribavirin (Peg-IFN/RBV), and treatment efficacy varies markedly depending on viral genotype (GT).3 There are six major HCV genotypes with multiple subtypes. GT-1 is the most difficult to eradicate with Peg-IFN/RBV, as has been reviewed elsewhere.4, 5 The cure rate or sustained viral response (SVR) for GT-1 is ∼45%.4, 5 Combining one of the recently approved nonstructural protein (NS)3 protease inhibitors (e.g., telaprevir or boceprevir) with Peg-IFN/RBV significantly improves the SVR rate.6, 7

The HCV genome is a single-stranded positive RNA that encodes a single polyprotein of ∼3,000 amino acids. The HCV polyprotein is processed by cellular and viral proteases into at least 10 individual proteins, as has been reviewed elsewhere.8, 9 Based on their functions in the viral life cycle, these proteins can be divided into two groups: structural and nonstructural proteins. Nonstructural proteins NS3, NS4A, NS4B, NS5A, and NS5B are the viral proteins required for HCV RNA replication. The development of direct-acting antivirals (DAAs) to treat HCV has been predominantly focused on inhibitors of NS3 and NS5B. NS3 is a serine protease responsible for processing the viral polyprotein, whereas NS5B is an RNA-dependent RNA polymerase (RdRp) and is responsible for viral RNA synthesis.

Infection with HCV results in a highly heterogeneous virus population, a consequence of its rapid replication turnover rate (∼1012 virions/day)10 and the lack of a proofreading function in the NS5B RdRp. Therefore, mutations at every position of the HCV genome are possible, and variants resistant to individual DAAs are predicted to preexist at baseline (BL) in infected subjects. Consistent with this prediction, preexisting variants with resistance to protease inhibitor telaprevir have been observed in virtually all HCV-infected patients.11

NS5A is an essential component of the HCV RNA replication complex and does not possess any known enzymatic activities, as reviewed previously.12 BMS-790052, a NS5A replication complex inhibitor, is a novel class of inhibitor.13In vitro, it exhibited exceptional potency with broad GT coverage. Median 50% effective concentration (EC50) values of BMS-790052 are 50 and 9 pM against GT-1a and GT-1b replicons, respectively. Its exceptional in vitro potency translated to a robust initial anti-HCV effect in clinical studies, achieving a mean reduction in HCV RNA of >3 log10 IU/mL 24 hours after single doses of 10 or 100 mg.13 However, viral breakthrough (VBT) was observed in the majority of subjects before or at the end of 14 days of monotherapy with BMS-790052.14 Genotypic and phenotypic analysis of clinical specimens indicated the following: (1) there is a direct correlation between substitutions found in the subjects with VBT and the resistance substitutions observed in the replicon system; (2) for GT-1a, the frequently observed resistance substitutions identified were at residues 28, 30, 31, and 93 of NS5A, whereas for GT-1b, major resistant substitutions identified were at residues 31 and 93; and (3) in general, variants with single amino-acid substitutions in the GT-1b replicon displayed minimal resistance to BMS-790052 (<30-fold), whereas GT-1b variants with double amino-acid substitutions and GT-1a variants with single amino-acid substitutions conferred much higher levels of resistance (>1,000-fold).13, 15, 16

To examine the influence of naturally occurring polymorphisms on the potency of BMS-790052, the entire NS5A coding region of the GT 1a and 1b laboratory (lab) strains, H77c and Con1, respectively, were replaced with the corresponding regions derived from specimens collected from 10 GT-1a- and 6 GT-1b-infected subjects. EC50 values of the NS5A inhibitor, BMS-790052, on the clinically derived BL specimens were similar to the lab strains. Through these analyses, we have developed a systematic approach to phenotype clinical specimens, especially for those in vitro resistance profiles that do not reflect the anti-HCV effect observed in the clinic.

Abbreviations

BL, baseline; cDNA, complementary DNA; DAA, direct-acting antiviral; EC50, median 50% effective concentration; GT, genotype; HCV, hepatitis C virus; lab, laboratory; NS, nonstructural protein; PCR, polymerase chain reaction; Peg-IFN/RBV, pegylated interferon plus ribavirin; RdRp, RNA-dependent RNA polymerase; SD, standard deviation; SVR, sustained viral response; VBT, viral breakthrough; WT, wild type.

Materials and Methods

Cell Culture and HCV NS5A Inhibitors.

Genotype 1a (H77c) and 1b (Con1) replicon cell lines have been described previously.17 The previously described cured Huh-7 replicon cell line, which is highly permissive for HCV replicon replication, was used for transient replication assays.17 The NS5A inhibitor, BMS-790052, has been described previously.13

Construction of HCV Replicons.

Viral RNA was isolated from the plasma of HCV-infected subjects, at BL and indicated time points, using the QiaAmp Viral RNA extraction kit (Qiagen, Hilden, Germany). First-strand complementary DNA (cDNA) was synthesized from random hexamer primers with a SuperScript III First-Strand Synthesis System for reverse-transcription polymerase chain reaction (PCR) (Invitrogen Corporation, Carlsbad, CA). The NS5A coding region was amplified with GT-specific primers and Platinum Taq high-fidelity DNA polymerase (Invitrogen). To replace the replicon NS5A with subject sequences, H77c was reconstructed to contain MfeI and SpeI restriction enzyme sites and Con1 was modified to contain Afl2 and Sac2 sites, which frames the NS5A coding region. A recombinant PCR18 was performed on NS5A cDNA generated from specimens using subject-specific primers containing the unique restriction enzyme sites mentioned above, along with primers specifically designed to change amino-acid residue 232 from serine to isoleucine (S2204I), a replication-enhancing adaptive mutation, into the subject NS5A protein. Finally, the PCR products were digested with the restriction enzymes, MfeI and SpeI, for GT-1a subjects and were ligated into the similarly digested HCV H77c replicon DNA. For GT-1b subjects, the PCR products were infused onto previously digested Con1 replicon DNA, which was cut with Afl2 and Sac2 restriction enzymes, using the protocol recommended by the manufacture (In-Fusion Cloning Kit; Clontech Laboratories, Mountain View, CA).

To replace the N-terminal of NS5A with clinical specimens, NS5A cDNA generated from patient specimens was further amplified by PCR using Platinum Taq high-fidelity DNA polymerase (Invitrogen) and forward primer 5'-TCCGGTTCCTGGCTAAGGGAC ATCTGG-3', which contains an EcoNI site, and reverse primer 5'-CACTACGTATCGGGTATGACTA CTGACAATC-3', which contains a SnaBI site. These primers were verified by comparison to the coding region of subject-specific NS5A and modified, as necessary, according to the sequence. The PCR product was digested with EcoNI and SnaBI, and the correct-sized fragments were gel purified and ligated into similarly digested HCV H77c replicon DNA. Clones containing the correct sequence were identified by restriction digestion and confirmed by sequencing analysis.

Specific site changes, such as Q30R, E62D, Q30R+E62D, and V75A mutants, were generated in the H77c replicon DNA by using the Agilent Quick-change mutagenesis kit (Agilent Technologies, Inc., Santa Clara, CA), according to the manufacture's instruction.

Transient Replication Assays and Isolation of Replicon Cell Lines.

Transient replication assays and isolation of replicon cell lines have been described previously.13, 15 The EC50 value was calculated as the concentration of inhibitor required for a 50% reduction in luciferase activity.

Results

Antiviral Activity of BMS-790052 on HCV Replicons Derived From Clinical Specimens.

HCV GT 1b (Con1) and 1a (H77c) replicons have been used to evaluate the antiviral profile of NS5A inhibitors, including the clinical lead, BMS-790052. To further validate the replicon system for use in phenotyping specimens from clinical studies of BMS-790052, hybrid replicons were evaluated.

The entire NS5A coding region of a GT-1b Con1 replicon was replaced with cDNA of NS5A derived from six BL specimens of GT-1b-infected subjects14, 16 (Table 1A). HCV NS5A sequences derived from clinical specimens of GT-1b share a high degree of amino-acid identity with the GT-1b Con1 replicon (≥95.2%). As expected, even greater identity (≥98.9%) was observed between multiple clones derived from the same specimen (Table 1A). Special attention was given to the signature polymorphisms of each specimen to ensure no cross-contamination among different specimens and/or replicons (data not shown). The replication-enhancing adaptive mutation, S2204I, in NS5A was introduced into all clones to enhance the ability for replicon replication. To obtain reliable EC50 values, hybrid replicons with a replication window (i.e., signal-to-noise ratio) ≥3 were used in transient replication assays (Table 1A). The Con1 replicon was used as a wild-type (WT) control for EC50 determination and also as a comparator for replication ability. Averaged EC50 and standard deviation (SD) values for multiple clones derived from each specimen are shown in Table 1B. NS3 protease and NS5B polymerase inhibitors were used as assay controls.

Table 1A. Profile of Clinical Specimens Derived From HCV GT-1b-Infected Subjects
Subject No.*Time of Specimens Collectedv-RNA (IU/mL) at Time Specimen AnalyzedTested ClonesReplicated ClonesReplicated WindowIdentity to Subject Consensus (%)
 GT-1b Con1 (control)1115-277100.0
WBL7.50E+06533-3796.0
GBL3.20E+05932-495.5
CBL5.45E+049319-29396.9
4BL2.59E+06746-2695.2
5BL1.17E+06416-2097.2
TBL2.0E+0714122-3695.3
Table 1B. NS5A Inhibitor BMS-790052 Antiviral Activity (EC50) for HCV Replicon-Derived GT-1b Clinical Specimens
Subject No.*Average EC50 (nM)
BMS-790052NS3 Protease InhibitorNS5B Polymerase Inhibitor
  • *

    Specimens W, G, C, and T were derived from the 14-day monotherapy of BMS-79005214, 16; specimens 4 and 5 were derived from other BMS clinical studies.

  • Transient replication assays. An EC50 value for the NS5A inhibitor, BMS-790052, was determined for each hybrid replicon. Averaged EC50 and SD values for multiple clones derived from each specimen are shown.

  • The specimen derived from subject T had ∼100% NS5A Q54H-Y93H substitutions at BL.16

GT-1b Con1 (control)0.003 ± 0.00090.9 ± 0.452.1 ± 1.1
W0.002 ± 0.00110.6 ± 0.200.9 ± 0.62
G0.001 ± 0.00081.2 ± 0.371.5 ± 1.06
C0.002 ± 0.00070.7 ± 0.282.2 ± 1.08
40.003 ± 0.00051.7 ± 0.353.4 ± 0.32
50.002 ± 0.00102.3 ± 0.864.6 ± 2.4
T0.050 ± 0.0140.5 ± 0.091.9 ± 0.63

Previously characterized resistant substitutions were not identified by population sequencing in the BL specimens, except for subject T.14, 16 The EC50 values for BMS-790052 with clones derived from BL ranged from 0.001 to 0.003 nM, which is similar to WT (Con1) replicon (0.003 nM for BMS-790052; Table 1B). The specimen derived from subject T had ∼100% NS5A Q54H-Y93H substitutions at BL.16 The EC50 value for BMS-790052 on this variant was 0.050 nM, or ∼23-fold resistance to BMS-790052 (Table 1B).15, 16

The entire NS5A coding region of a GT-1a (H77c) replicon was replaced with cDNA of NS5A derived from 12 clinical specimens of 11 GT-1a-infected subjects.14, 16 Ten cDNAs were derived from BL specimens, one from a T4 specimen (4 hours after the first dosing), and one from a day 14 specimen (T312) (Table 2A) of subject P who received 60 mg of BMS-790052 once-daily as monotherapy for 14 days.16 The replication-enhancing adaptive mutation, S2204I, in NS5A was introduced into all clones to enhance replicon replication. A total of 12 clones derived from subject E with a replication window (i.e., signal-to-noise ratio) less than 2 were not used for EC50 determinations. The amino-acid sequence identity between the NS5A consensus of each specimen and the GT-1a replicon, H77c, is ≥92.6%, and the identity between each clone and the consensus of the individual specimen is ≥93.3% (Table 2A). The EC50 values of BMS-790052 were determined with these GT-1a hybrid replicons (Table 2B). No previously characterized resistance substitutions were identified by population sequencing in the BL specimens. The averaged EC50 values ranged from 0.003 to 0.025 nM for multiple clones derived from each specimen and were similar to values with WT H77c replicon (0.012 nM). Overall, this analysis indicated that the NS5A sequence heterogeneity present in GT-1a and GT-1b BL specimens had a minimal effect on the potency of BMS-790052.

Table 2A. Profile of Clinical Specimens Derived From HCV GT-1a-Infected Subjects
Subject No.*Time of Specimens Collectedv-RNA (IU/mL) at Time Specimen AnalyzedTested ClonesReplicated ClonesReplicated WindowIdentity to H77c (%)Identity to Subject Consensus (%)
ABL1.06E+062435-794.298.4
FBL1.96.E+061553-795.398.9
IBL1.40E+071696-28496.298.2
JBL1.26E+062243-595.398.7
PBL2.44.E+071566-1694.999.1
MBL1.66E+072044-892.697.8
NBL1.84.E+062056-2794.698.2
QBL1.05.E+061683-2094.298.7
SBL1.04.E+071663-6295.197.5
UBL1.95.E+071654-3197.198.2
ET4†1.00E+05120<292.993.3
PT312 (Day 14)‡4.40E+062453-3294.698.7
Table 2B. NS5A Inhibitor BMS-790052 Antiviral Activity (EC50) for HCV Replicon-Derived GT-1a Clinical Specimens
Subject No.*Subject (Time)Average EC50 (nM)§
BMS-790052NS3 Protease InhibitorNS5B Polymerase Inhibitor
  • *

    All specimens were derived from the 14-day monotherapy of BMS-790052.14, 16

  • T4: 4 hours after first dosing.

  • Day 14 (T312): 24 hours after last dosing.

  • §

    Transient replication assays. An EC50 value for the NS5A inhibitor, BMS-790052, was determined for each hybrid replicon. Averaged EC50 and SD values for multiple clones derived from each specimen are shown.

GT-1a H77c (control)0.012 ± 0.0081.0 ± 0.52.6 ± 1.6
ABL0.007 ± 0.0031.4 ± 0.41.7 ± 1.0
FBL0.003 ± 0.0011.7 ± 0.83.6 ± 2.4
IBL0.010 ± 0.0070.9 ± 0.51.9 ± 1.5
JBL0.010 ± 0.0062.0 ± 1.71.3 ± 0.7
PBL0.006 ± 0.0011.5 ± 0.51.6 ± 0.4
MBL0.025 ± 0.0131.4 ± 0.62.7 ± 2.2
NBL0.009 ± 0.0061.2 ± 0.73.4 ± 2.0
QBL0.009 ± 0.0041.3 ± 0.52.0 ± 1.1
SBL0.003 ± 0.0020.9 ± 0.41.8 ± 1.2
UBL0.009 ± 0.0051.5 ± 0.91.2 ± 0.5
ET4No transient replication window
PDay 14 (T312)159 ± 201.7 ± 0.61.3 ± 0.5

Because a significant number of GT-1a replicon clones derived from human specimens did not replicate well in transient replication assays (Table 2A), a total of 75 replicon cell lines were established from clones that were replication competent and noncompetent in the transient assay as well as the control H77c clone. The cell lines were used to determine whether compensatory mutations necessary to establish cell lines affected the potency of BMS-790052 (Table 3). Average EC50 values were similar in cell lines isolated from replication-competent (0.003-0.019 nM) and -noncompetent clones (0.004-0.027 nM; Table 3), suggesting that compensatory mutations had minimal effect on the potency of BMS-790052.

Table 3. BMS-790052 Antiviral Activity (EC50) for HCV Replicon Cell Lines Containing NS5A Derived From GT-1a Clinical Specimens
Subject No.*Cell Lines Established From Replicated ClonesAverage EC50 (nM)Cell Lines Established From Nonreplicated ClonesAverage EC50 (nM)
  • *

    All specimens were derived from the 14-day monotherapy of BMS-790052.14, 16

  • Abbreviation: NA, not applicable.

H77c40.012 ± 0.008NANA
A30.003 ± 0.00120.005 ± 0.004
F40.005 ± 0.0040NA
I40.014 ± 0.00830.013 ± 0.011
J40.005 ± 0.00310.004
P60.014 ± 0.00910.006
M10.006 ± 0.00330.004 ± 0.002
N40.014 ± 0.00810.012
Q80.019 ± 0.01130.021 ± 0.011
S60.008 ± 0.00420.008 ± 0.001
U50.009 ± 0.00840.027 ± 0.018
E0NA30.013 ± 0.010
P3151 ± 140NA
Total52 23 

Systematic Approach to the Analysis of HCV NS5A Inhibitor Resistance Derived From Clinical Specimens.

The EC50 value for BMS-790052 on the day 14 specimen from subject P was 159 nM (Table 2B). This variant, with ∼100% Q30R substitution in NS5A (percentage estimated by population sequencing), was the only virus detected on days 7 (T144) and 14 (T312) using two different sets of primers.16 The lowest plasma exposure of BMS-790052 in this subject during the 14-day treatment period was 117 nM or 86.8 ng/mL, whereas the EC50 value for a GT 1a H77c replicon containing the Q30R variant is ∼7 nM or 5.4 ng/mL.13, 15, 16 Because a concentration of 117 nM would be expected to completely inhibit the previously characterized Q30R variant with an EC50 value of 7 nM, a rigorous investigation of the NS5A clones derived from subject P was triggered by the observation.

Amino-acid alignment of NS5A protein from H77c, subject P BL, and day 14 (T312) sequences is shown in Fig. 1. There are 23 amino-acid differences between H77c and BL NS5A, and only one amino-acid difference at residue 30 (Q30R) between BL and day 14 specimens.

Figure 1.

Alignment of GT-1a NS5A proteins for H77c (top),17 BL (middle), and day 14 (bottom) specimens from subject P. Amino-acid identities are indicated with dots. The adaptive mutation at the residual 2204 (polyprotein) or 232 (NS5A) are boxed with gray color. In all in vitro assays, the serine at this position was changed to isoleucine (S2204I). Two primers (B1 and A1) were used for baseline and day 14 specimens. Because the results were identical, only one set of data (primer B1) is shown.

Three different approaches were used to investigate why BMS-790052, at a plasma concentration of ∼117 nM, did not suppress replication of GT-1a Q30R variant in subject P during 14 days of monotherapy. First, the entire NS5A coding region of the H77c replicon was replaced with NS5A derived from subject P specimens (BL and day 14). Second, the N-terminal region of H77c NS5A (5-129 amino acids) was replaced with subject P–derived two sequences (BL and day 14). Finally, the effects of specific amino-acid substitutions present in subject P, but not in the H77c replicon clone, were examined.

When the entire NS5A coding region of the GT-1a H77c replicon was replaced with cDNAs derived from specimens of subject P, the average EC50 value for six BL clones was 0.006 nM (Table 2B), similar to the control GT-1a H77c replicon value of 0.012 nM, but the average EC50 value derived from five clones from the day 14 specimen was 159 nM (Table 2B). Genotypic analysis revealed that clones derived from the BL specimen matched the population-consensus sequence. The clones contained E62D, V75A, K107T, and R123Q substitutions in the first 129 amino acids of NS5A (compared to GT-1a replicon H77c; Fig. 1). Similarly, these four substitutions were present in the majority of clones derived from the day 14 specimen, which contained an additional Q30R substitution (Fig. 1).

When sequences encoding the first 129 amino acids of NS5A from the GT-1a H77c replicon were replaced with cDNAs derived from BL and day 14 specimens of subject P, reliable data were not obtained because of low replication ability of the replicons (<2-fold above background after multiple attempts) in transient replication assays. Therefore, replicon cell lines were selected. Population-sequencing analysis of cDNAs derived from these replicon cell lines confirmed four amino-acid changes in the first 129 amino acids of NS5A (E62D, V75A, K107T, and R123Q) from the BL specimen and an additional Q30R substitution from the day 14 specimen. EC50 values of BMS-790052 in replicon cells with the first 129 amino-acid coding region of NS5A derived from the BL specimen was 0.043 nM (Table 4), similar to the value in H77c replicon cells (0.014 nM) and the value of 0.038-0.050 nM previously reported.13, 15 The EC50 value derived from the day 14 specimen was 149 nM, similar to the EC50 value of 159 nM derived from the replicon with replacement of the entire NS5A coding region (compare values in Table 2B). These results demonstrated that the five amino-acid changes in the first 129 amino acids of NS5A from the day 14 specimen are sufficient to dramatically decrease the susceptibility to BMS-790052.

Table 4. BMS-790052 Antiviral Activity (EC50) for HCV Replicon Cell Lines Containing the First 129 Amino Acid of NS5A Derived From BL and Day 14 of Subject P
Cell LineEC50 (nM)Amino-Acid Changes in NS5A
BMS-790052NS3 Protease InhibitorNS5B Polymerase Inhibitor
H77c0.014 ± 0.0111.9 ± 1.12.0 ± 0.3None
Subject P (BL)0.043 ± 0.0162.5 ± 0.82.4 ± 1.2E62D-V75A-K107T-R123Q
Subject P (day 14)149 ± 705.1 ± 2.72.3 ± 0.9Q30R-E62D-V75A-K107T-R123Q

To determine which amino-acid change(s) were responsible for the clinically relevant resistance phenotype of the day 14 specimen, variants with specific amino-acid substitutions were analyzed (Table 5). To date, all substitutions resistant to BMS-790052 have been mapped to the first 100 amino acids of NS5A; therefore, E62D and V75A substitutions were the first candidates selected for variant construction. In transient replication assays, the EC50 value of Q30R was ∼10 nM, similar to the value reported previously,13, 15 whereas the E62D and V75A variants alone did not confer resistance to BMS-790052 (Table 5). However, when E62D, but not V75A, was combined with Q30R, the EC50 value of the linked variant (Q30R-E62D) was 153 nM, similar to the results obtained from (1) the replicon containing the entire NS5A coding region from the day 14 specimen (Table 2B) and (2) the replicon cells containing the first 129 amino acids of NS5A (Table 4). These results demonstrate that the linked variant, Q30R-E62D, is sufficient to confer a high level of resistance in vitro and suggest that the linked Q30R and E62D substitutions are most likely responsible for the VBT in subject P.

Table 5. The Q30R and E62D Substitutions Are Responsible for the Resistant Phenotype of Subject P
Substitution site(s)EC50 (nM)Replication Level (%)
BMS-790052NS3 Protease InhibitorNS5B Polymerase Inhibitor
  1. *Transient replication assays.

Wild type (H77c)0.014 ± 0.0081.9 ± 0.91.9 ± 1.3100
E62D0.012 ± 0.0062.0 ± 0.93.0 ± 2.489 ± 39
V75A0.012 ± 0.0062.3 ± 1.41.2 ± 0.244 ± 19
Q30R10.2 ± 3.01.7 ± 0.82.0 ± 1.343 ± 26
Q30R-E62D153 ± 182.3 ± 0.92.6 ± 1.144 ± 29
Q30R-V75A11.9 ± 2.23.1 ± 1.81.3 ± 0.230 ± 19

Effect of BL Polymorphism on the Emergence of Resistance.

To further investigate the influence of BL polymorphisms on resistance development, a hybrid replicon cell line containing the entire NS5A derived from the BL specimen of subject P was isolated and was used to select resistance to BMS-790052. The lab strain GT-1a (H77c) replicon cell line was used as a control. Genotypic analysis revealed that multiple resistant substitutions (Q30E, Q30K, Y93H, and Y93N) were selected in the H77c cell line, whereas the only substitution, Q30R (∼100%), was selected in the hybrid cell line (Fig. 2). These results, combined with the BL specimen analysis, strongly suggest that NS5A BL polymorphisms that have minimal effect on the potency of BMS-790052 can significantly influence the emergence of resistant variants and affect the clinical outcome.

Figure 2.

Resistance selection and genotypic analysis. The hybrid replicon cell plated in 100-mm tissue-culture dishes were maintained in the complete Dulbecco's modified Eagle's medium, supplemented with 0.25 mg/mL of G418 and 120 nM of BMS-790052. Cells were fed with fresh medium, containing inhibitor, weekly. After approximately 4 weeks, pooled cells were expanded for genotypic analysis, as described by Fridell et al.15, 16

Discussion

Results from this study are consistent with previous observations that the phenotypes of NS5A resistance variants characterized in the in vitro replicon system correlate well with resistance variants observed in the clinic.13, 15, 16 Subjects without detectable BL NS5A substitutions frequently observed to confer resistance to BMS-790052 in vitro (residues 28, 30, 31, and 93 for GT-1a and 31 and 93 for GT-1b) experienced robust initial HCV RNA decline.13, 14 Both GT-1a and 1b replicons containing NS5A sequences from these BL specimens exhibited similar inhibitory responses, compared to parental replicons (H77c for GT-1a and Con1 for GT-1b). A variant with ∼100% Q54H and Y93H substitutions was identified at BL for subject T infected with GT-1b. The Q54H-Y93H variant displayed minimal resistance to BMS-790052 with an EC50 value of 0.050 nM, similar to the Y93H substitution by itself (Table 1B).15, 16 Consistent with this in vitro resistance profile, subject T experienced >4 log10 viral RNA decline at day 4 (T72).14, 16

Although the replication ability of hybrid GT-1a and GT-1b replicons constructed from clinical specimens varied significantly (Tables 1A and 2A), EC50 values for BMS-790052 determined on hybrid replicons were similar (Tables 1B and 2B). The varying ability of the hybrid replicons to replicate may be related to the fitness of NS5A in the replicon replication complex; however, the consistent EC50 values suggest that the BMS-790052-binding pocket on NS5A derived from different sources is relatively conserved. Consistent with this observation, all resistant substitutions induced by BMS-790052 have been mapped to the first 100 amino acids of NS5A, mainly at residues 28, 30, 31, and 93.13, 15, 19

Because NS5A does not possess known enzymatic activities and the correlation between the antiviral effect and binding of BMS-790052 to purified NS5A has not been established, the determination of BMS-790052 potency and the analysis of inhibitor resistance phenotype are solely dependent on the cell-based replicon system. A discrepancy between the antiviral effect of BMS-790052 in subject P with a resistant substitution at Q30R in vivo and the Q30R-resistant profile observed in vitro replicon system provided an opportunity to establish a specific, systematic process for investigating NS5A resistance in clinical specimens. Results from this investigation confirmed a direct correlation between the in vitro and in vivo resistance phenotypes.

A systematic in vitro investigation of the NS5A sequence of subject P identified substitutions that delivered the antiviral response observed. Hybrid replicons containing the entire NS5A sequence or the N-terminal 129 amino acids of NS5A- or NS5A-specific amino-acid substitutions from subject P were analyzed. The studies revealed that the BL NS5A variant, E62D, did not confer any detectable resistance to BMS790052, but when combined with Q30R, the double substitution variant (Q30R-E62D) conferred a very high level of resistance. Although Q30 is not present in either of the published crystal structures,20, 21 these structures show that residue E62 is located adjacent to Zn++ coordinate residues C57 and C59. In fact, we predict that the E62D substitution may affect Q30R resistance by influencing Zn++ binding. Studies to determine if this substitution, when combined with Q30R, would affect Zn++ binding are in progress.

Hybrid replicons with the entire H77c NS5A that were replaced with NS5A derived from either BL or day 14 specimens of subject P had decreased, but measurable, replication ability (replication window 3-32; Table 2A). However, little or no replication signal was detected when the first 129 amino acids of the NS5A replicon were replaced with sequence from clinical specimens, suggesting that the mixing of domains from different NS5A proteins (convenient for cloning) may impair the formation of a proper replication complex. Our ability to isolate replication-competent cell lines containing the first 129 amino acids of NS5A from clinical specimens indicated that compensatory mutation(s) must have been selected to enhance replication ability. We did not attempt to identify these mutations, because the primary aim was to resolve the discrepancy between the in vitro and in vivo resistance profiles. When two specific amino acids (Q30R-E62D) were substituted, the replication ability of the replicon was preserved (Table 5) and a reliable EC50 value was determined.

Unlike the HCV-infected population, both lab-strain replicons (H77c and Con1) were constructed from consensus sequences.22, 23 Isolated replicon cell lines have adaptive mutations that enhance HCV RNA replication ability and differ under different selective pressures. In general, the genomic sequences of these cell lines are more homogenous at both the RNA and amino-acid sequence level than clinical isolates. This study indicates that the heterogeneity of HCV sequences in infected specimens with no detectable level of previously identified resistant substitutions has a minimal effect on the potency of BMS-790052, a conclusion similar to that made from observations of compensatory mutations. However, the effect of the polymorphism, E62D, present in the heterogeneous BL sequence of subject P significantly affected the emergence of resistance.16 Our ability to replicate the clinical phenotype by linking the Q30R substitution with the preexisting E62D further demonstrates the utility of the replicon system in clinical studies. These results strongly suggest that BL polymorphisms in NS5A may significantly affect the emergence of resistance, providing additional challenges for the evaluation of variants associated with clinical failures.

We have gained extensive experience selecting and analyzing HCV resistance to BMS-790052 and have developed a comprehensive path to study clinical resistance to NS5A inhibitors. However, regardless of the methods required, the aim of monitoring resistance is to understand the emergence of resistance to guide optimal dose selection and recommend combination treatment strategies.

Acknowledgements

The authors thank Mark Cockett and Nicholas Meanwell for their continuous support. Additionally, the authors thank Aaron Monikowski, Xin Huang, Xingtie Nie, and Xiaoyan Yang for their technique supports.

Ancillary