Synergistic Activity of Combined NS5A Inhibitors

Abstract

Daclatasvir (DCV) is a first-in-class hepatitis C virus (HCV) nonstructural 5A replication complex inhibitor (NS5A RCI) that is clinically effective in interferon-free combinations with direct-acting antivirals (DAAs) targeting alternate HCV proteins. Recently, we reported NS5A RCI combinations that enhance HCV inhibitory potential in vitro, defining a new class of HCV inhibitors termed NS5A synergists (J. Sun, D. R. O’Boyle II, R. A. Fridell, D. R. Langley, C. Wang, S. Roberts, P. Nower, B. M. Johnson F. Moulin, M. J. Nophsker, Y. Wang, M. Liu, K. Rigat, Y. Tu, P. Hewawasam, J. Kadow, N. A. Meanwell, M. Cockett, J. A. Lemm, M. Kramer, M. Belema, and M. Gao, Nature 527:245–248, 2015, doi:10.1038/nature15711). To extend the characterization of NS5A synergists, we tested new combinations of DCV and NS5A synergists against genotype (gt) 1 to 6 replicons and gt 1a, 2a, and 3a viruses. The kinetics of inhibition in HCV-infected cells treated with DCV, an NS5A synergist (NS5A-Syn), or a combination of DCV and NS5A-Syn were distinctive. Similar to activity observed clinically, DCV caused a multilog drop in HCV, followed by rebound due to the emergence of resistance. DCV–NS5A-Syn combinations were highly efficient at clearing cells of viruses, in line with the trend seen in replicon studies. The retreatment of resistant viruses that emerged using DCV monotherapy with DCV–NS5A-Syn resulted in a multilog drop and rebound in HCV similar to the initial decline and rebound observed with DCV alone on wild-type (WT) virus. A triple combination of DCV, NS5A-Syn, and a DAA targeting the NS3 or NS5B protein cleared the cells of viruses that are highly resistant to DCV. Our data support the observation that the cooperative interaction of DCV and NS5A-Syn potentiates both the genotype coverage and resistance barrier of DCV, offering an additional DAA option for combination therapy and tools for explorations of NS5A function.

INTRODUCTION

Hepatitis C virus (HCV) infections can be effectively cured using combinations of small-molecule direct-acting antivirals (DAAs) with different mechanisms of action. The nonstructural 5A replication complex inhibitor (NS5A RCI) class of compounds represented by daclatasvir (DCV) (BMS-790052, Daklinza; Bristol-Myers Squibb) has activity against genotypes (gts) 1a, 1b, 2a, 3a, 4a, 5a, and 6a and is the most potent class of HCV inhibitor yet described (1–3). NS3 protease inhibitors and NS5B polymerase inhibitors have thus far been the preferred DAA partners for DCV combination therapies (1). Combinations of small-molecule DAAs are required for effective curative oral therapies, because resistant variants existing in the quasispecies population in untreated HCV-infected patients are enriched during monotherapy. To prevent viral breakthrough or relapse and selection of resistant variants, multiple highly effective pangenotypic DAA combination therapies, with NS5A RCIs as the backbone, have been approved or are currently being developed (1, 4, 5).

The binding site(s) for NS5A RCIs has been modeled based on biochemical and crystal structure data, implicating at least one site that spans a region between NS5A proteins that associate as dimers (6). Available evidence for NS5A inhibitor binding suggests that NS5A RCIs, such as DCV, bind tightly and specifically to NS5A protein dimers present in infected cells (6).

A recent report (7) of DCV binding to Escherichia coli-expressed NS5A protein provided the first direct evidence that DCV binds to NS5A. The authors demonstrated that DCV binding can reduce the affinity of NS5A protein for viral RNA, supporting one potential mechanism of inhibition. An additional report of direct binding to NS5A protein of DCV and the related compound ledipasvir also supports the direct association of NS5A RCIs with the NS5A protein (8). Immunomicroscopy analysis indicates that DCV binding results in the apparent disassociation of NS5A from the replication complex, consistent with other evidence that NS5A RCI binding has multiple effects (9–13).

The use of functional NS5A affinity reagents (inhibitory cross-linkers, labeled NS5A proteins) and genetic variants has provided evidence for the presence of NS5A dimers and oligomers in cells, suggesting that NS5A-NS5A dimers may be necessary for HCV replication and NS5A function (7, 14, 15). Experiments using NS5A constructs tagged near domain 1 for binding of fluorescent labels detected a correlation between inhibitor activity and changes in fluorescent signal that is consistent with conformational changes in NS5A upon compound binding (14). In addition, active compound immunoprecipitated more NS5A than an inactive analog, suggesting that binding of the inhibitor potentially stabilizes dimeric and higher-order NS5A multimers while modulating NS5A conformation.

Computer-simulated docking of DCV to NS5A dimers has provided a framework for understanding the potency of DCV on different genotypes and the resistance variants selected by DCV monotherapy (7, 9, 16). A recent report of two additional crystal forms of NS5A domain 1 (gt 1a), together with the previously reported structures (17–19), suggests that NS5A domain 1 has multiple conformations and a flexibility that may be important for an association of NS5A function and dimers. The ability to model the binding of NS5A compounds in a nonsymmetrical manner may also reflect this property (20). The picomolar potency of DCV toward diverse HCV genotypes coupled with the abundance of the NS5A protein in infected cells supports a polymer-based interaction model of NS5A dimers, in which the binding of DCV to one NS5A dimer influences proximal and distal NS5A dimers that are arranged in a polymeric structure. The polymer-based interaction model (18, 21) is consistent with our recently reported observations with NS5A synergists (22). We hypothesize that DCV binds to resistant NS5A dimers (a conformation different from the wild type [WT]) without disrupting NS5A function(s); however, DCV binding causes a conformational change that accommodates the binding of a 2nd inhibitor (synergist [Syn]) on adjacent NS5A dimers to disrupt function(s) of the oligomer. This model is further supported by studies indicating that NS5A compounds can bind to resistant mutants with no detectable consequence (6).

NS5A synergists were discovered during the characterization of NS5A RCI specificity, when pulldown experiments in NS5A replicon cells revealed compounds that competed for binding even though they had little inhibitory activity (22). In addition, some weakly active compounds in combination with DCV were found to enhance inhibitory potency (22). The observed synergistic effect led us to investigate the structure-activity relationship (SAR) of this chemotype on DCV-resistant variants. Two compounds discovered through these SAR studies and that are reported here were used to further investigate the impact of NS5A-Syn on multiple HCV genotypes and the resistance barrier. Replicon elimination studies and a comparison of the 50% effective concentrations (EC₅₀s) on multiple genotypes suggest that synergy inhibition is conserved for gts 1a through 6a. The ability to cure infections by hybrid viruses from the most DCV-recalcitrant genotypes (1a, 2a, and 3a) with DCV–NS5A-Syn and successfully eliminate DCV-resistant virus variants demonstrates the cooperative and powerful nature of the NS5A inhibitor combinations. The use of NS5A-Syn compounds in combination with DCV in this report further confirms the pangenotypic anti-HCV mechanism while providing additional tools for studying NS5A function.

MATERIALS AND METHODS

Replicon cell lines and virus stocks.

All replicons and virus stocks contained a Renilla-based luciferase reporter incorporated in the genome and were described previously (1, 3, 23, 24). The replicon cell lines and viruses used in the study contained sequences derived from gt 1b (Con1 [GenBank accession no. AJ238799.1]), gt 1a (H77c [GenBank accession no. AF011751.1]), gt 2a (JFH and JFH-1 [GenBank accession no. AB047639]), gt 2a (NIH and L31M NS5A [amino acids {aa} 1 to 425]), gt 2a (infectious clone-pJ6CF [GenBank accession no. NC_009823.1]), gt 3a (NS5A [aa 1 to 429], HCV3a1 [GenBank accession no. JX944789]), gt 4a (NS5A, HCV4a-23 [GenBank accession no. JQ347515]), gt 5a (NS5A [aa 1 to 430], gt 5a-7 [GenBank accession no. KJ719453]), and gt 6a-16 [GenBank accession no. KJ719455]).

Cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco) with 10% fetal bovine serum (FBS; Invitrogen) and 0.25 mg/ml to 1.0 mg/ml G418 and penicillin-streptomycin (pen-strep), as previously described (25, 26). Cells were passed twice weekly in T175 flasks. Replicon cells from the same passage were used concurrently for both EC₅₀ evaluations and colony elimination studies.

Virus activity assays.

Assays were conducted as previously described, with minor modifications (22–24). Briefly, virus stocks harvested from transfected cell medium were used to infect fresh Huh-7.5 cells in T175 flasks and were monitored for luciferase activity. The infected cells were split and passaged until the luciferase activity was ∼2E6 units per ml of cell pellet. The infected cells were divided and placed into individual T25 flasks with the appropriate treatment (compound concentrations are listed in the figures) or dimethyl sulfoxide (DMSO). Following incubation, typically for 2 or 3 days, cells were trypsinized and resuspended in 4 ml of total DMEM, and 1 ml was used for monitoring luciferase, RNA isolation for reverse transcription-PCR (RT-PCR), or continued passage by transfer to a new flask.

Replicon elimination assays.

Individual replicon cell lines (∼250,000 cells) were plated in 10 ml of DMEM–10% FBS–1.0 mg/ml G418-pen-strep with compounds (concentrations listed in the Figures) or DMSO (10 μl/plate) in duplicate 100-mm-diameter dishes. The medium was changed to compound-free medium (10 ml) following the indicated days of treatment exposure (range, 1 to 7 days), and following a 10-ml rinse with compound-free medium. The plates were microscopically inspected and medium changed when a large number of cells were visibly floating, indicating cell death. At the end of the colony growth period (typically 3 weeks, with clearly visible colonies) or if the cell monolayer was very confluent with no cell death apparent, the medium was removed, and plates were allowed to air dry. The plates were then stained using a solution of 0.2% crystal violet (Sigma, St. Louis, MO, USA) in 20% methanol. The dried plates were digitally recorded using an InGenius camera system (Syngene, Frederick, MD, USA).

Replicon EC₅₀ evaluations.

Replicon cells (10,000 cells/well) were plated in 96-well plates in DMEM–10% FBS–pen-strep (200 μl/well). Compound (1.0 μl) titrated in DMSO (Sigma) or DMSO (1.0 μl) was added.

Two-way titrations with DCV and NS5A-Syn compounds were prepared in a 96-well format. DCV with or without a synergistic compound was placed into individual wells beginning with 1 μM and diluted 4-fold, for a final concentration of DCV equal to 15 pM and a final concentration of synergist equal to 244 pM. DCV or NS5A-Syn compounds were also titrated singly and used to quantify the magnitude of synergistic effects. Cells were incubated for 3 days. The medium was removed, and a DMEM-EnduRen (6 μM final) (Promega, Madison, WI, USA) mix was added (50 μl/well). The cells were returned to the incubator for 15 min, and luminescence was read in a TopCount reader, as previously described (26). The average value from DMSO-treated cells was used as 100% activity, while background activity was subtracted using the signal obtained from the combination of 1 μM DCV and 1 μM NS5A-Syn. The percent activity remaining was calculated to generate the reported 50% activity level.

RT-PCR and sequencing of virus RNA.

RNA was isolated from 1 ml of resuspended infected cells using the RNeasy protocol (Qiagen, Gaithersburg, MD, USA), as recommended by the manufacturer. Total RNA was used for RT using random hexamer primers and the Platinum Taq polymerase high-fidelity kit (Invitrogen, Grand Island, NY, USA). The primers for the 2a JFH hybrid viruses consisted of 5′-CTTACTATAACCAGCCTACTCAGAAGACTCCAC-3′ and 5′-CTCAAAGGGTTGATTGGCAACTTTTCCTCTTC-3′, and PCR was performed as previously described (26). The PCR products were purified by the QIAquick PCR purification kit (Qiagen), quantified, and sequenced, as described previously (26).

Compounds.

All compounds were synthesized at Bristol-Myers Squibb Co., (BMS, Wallingford, CT, USA) and were >99% pure. The synthesis of these types of compounds and methods used to identify NS5A synergists are available in previous works (27, 28).

RESULTS

Discovery of NS5A synergistic compounds.

It has been reported that the NS5A synergistic compound Syn-395, in a triple combination containing DCV, is as effective against gt 1a as either the NS3 or the NS5B inhibitor that it replaced (22). The initial observations reported for Syn-395 are confirmed and extended in this study of gts 1a, 1b, 2a, 3a, 4a, 5a, and 6a using Syn-690 and Syn-535 (Fig. 1 and Table 1). These compounds were selected from multiple NS5A synergists that were discovered and characterized at BMS because of their inherent inhibitory activity and their ability, in combination with DCV, to restore the picomolar potency of DCV against DCV-resistant variants. Since the potency of DCV on wild-type (WT) cells can mask the inhibitory effects of a DCV–NS5A-Syn combination, replicon cells and viruses that are less sensitive to DCV were used for the characterization. DCV was typically fixed at a concentration of ∼250 nM, the level observed in the plasma samples from patients receiving an effective dose (60 mg) of DCV in clinical trials (see figure legends for details).

FIG 1.

Structures of DCV, Syn-690, and Syn-535. The NS5A synergists exhibit some structural similarity to DCV, with a central biphenyl and bis-imidazole core. Substitutions in the terminal moieties distinguish the inhibitor classes.

TABLE 1.

EC₅₀s of study compounds^a

Genotype	EC50 by compound (mean ±SD) (nM)
	DCV	Syn-535	Syn-690	Syn-395b
1a	0.040 ± 0.01	16 ± 1.6	79 ± 6.2	133 ± 23
1b	0.004 ± 0.002	12 ± 2.6	30 ± 8.6	67 ± 23
2a NIH	54 ± 4	21 ± 1.9	39 ± 5.3	187 ± 70
3a	0.670 ± 0.23	0.950 ± 0.1	4 ± 0.4	8 ± 4
4a	0.020 ± 0.004	>1,000	>1,000	762 ± 680
5a	<0.015	0.920 ± 0.1	20 ± 1.5	39 ± 16
6a	0.170 ± 0.1	0.270 ± 0.02	60 ± 5	NDc

^aReplicons containing gts 1 to 6 had compounds titrated to determine the EC₅₀ of compounds used.

^bData for comparison to a previously reported compound (Syn-395 [22]).

^cND, not done.

Comparing the effectiveness of triple combinations against gt 1.

Colony elimination studies against gt 1a and 1b replicon cells were used to test the effectiveness of Syn-690 as a replacement for either an NS3 or NS5B inhibitor (Fig. 2). Replicon cells were exposed for 7 days to a triple combination of DCV, an NS3 inhibitor (ASV [asunaprevir]), and an NS5B inhibitor (BCV [beclabuvir] BMS-791325), which are DAAs that are currently approved for the treatment of HCV infections (DCV and ASV) or are in advanced clinical trials (BCV) (29). This combination was compared to the effectiveness of triple combinations containing Syn-690 substituted for either ASV or BCV. The compound concentrations are multiples of the respective EC₅₀s for DCV, ASV, and BCV (Fig. 2). The 1× concentration of Syn-690 (5 nM) closely mimics the EC₅₀s of ASV and BCV, even though the EC₅₀s for Syn-690 on gt 1a and 1b are 79 and 30 nM, respectively.

FIG 2.

Replacing an NS3 inhibitor (ASV) or an NS5B inhibitor (BCV) with Syn-690 in gt 1a and 1b replicon elimination assays. Syn-690 was used in triple combinations with DCV to replace either ASV or BCV (underlined) and was compared to the triple combination of DAAs that inhibits 3 different targets (NS5A, NS5B, and NS3). 1× refers to the lowest concentrations used: DCV, 50 pM; ASV, 6 nM; BCV, 5 nM; and Syn-690, 5 nM. Multiples (3, 10, and 30 times) of the 1× concentration were also compared.

The concentration of Syn-690 (50 nM) required to eliminate gt 1a replicons and replace BCV or ASV is the same concentration required for Syn-395 (22), reflecting a similar EC₅₀ on gt 1a (Table 1). The DCV-ASV-BCV combination effectively eliminates gt 1a and 1b at approximately 10 times the respective EC₅₀s (500 pM DCV, 60 nM ASV, and 50 nM BCV; Fig. 2). Inspection of the colonies remaining when Syn-690 was substituted for either ASV or BCV (Fig. 2) shows that Syn-690 is similarly effective. The substitution of Syn-690 for ASV at 1× and 3× on gt 1b appears to be slightly less effective than the substitution for BCV, possibly reflecting the potency of ASV singly (∼2 nM) on gt 1b. Figure 2 indicates that multiples of ≥10× of Syn-690 in place of either ASV or BCV increase the barrier to the emergence of resistance in gts 1a and 1b to levels similar to the clinically effective combination (DCV-ASV-BCV).

NS5A-syn activity against DCV-resistant variants of gt 1.

The EC₅₀ profiles of DCV, Syn-690, and Syn-535 differ dramatically in gts 1a and 1b WT versus clinically significant DCV-resistant variants (Table 2) (3). NS5A-Syn activity in combination with DCV on WT gt 1a and 1b replicons is masked by the picomolar potency of DCV; however, the effects of the combination on DCV-resistant variants (gt 1a Q30E and gt 1b L31V+Y93H) are dramatic (30). The EC₅₀ on gt 1a Q30E of Syn-690 in combination with DCV at a concentration that approximates the EC₅₀ (200 nM) is enhanced >9,000-fold compared to that for Syn-690 alone (EC₅₀, 0.006 versus 57 nM, respectively), and the EC₅₀ for Syn-535 in combination with DCV (200 nM) is enhanced >40,000-fold compared to that with Syn-535 alone (0.67 pM versus 28 nM, respectively). These comparisons demonstrate the ability of Syn-690 and Syn-535 to enhance the inhibitory potential of DCV on DCV-resistant variants (Table 2).

TABLE 2.

EC₅₀s of DCV, Syn-690, and Syn-535 and DCV-Syn combinations^a

Replicon	EC50 (nM)			Fold enhancement for Syn-690	EC50 (nM)		Fold enhancement for Syn-535
	DCV	Syn-690	DCV–Syn-690		Syn-535	DCV–Syn-535
gt 1a	0.040	66	ND	ND	20	ND	ND
gt 1b	0.004	30	ND	ND	12	ND	ND
gt 1a Q30E	183	57	0.006	∼9,500	28	0.001	∼40,000
gt 1b L31V+Y93H	215	262	0.290	∼900	54	0.013	∼4,000

^aEC₅₀s of DCV, Syn-690, and Syn-535 were determined for WT replicons and for clinically significant DCV-resistant variants of gts 1a and 1b. Values for the DCV-Syn combinations on WT replicons were not determined (ND), since the potency of DCV on WT cells masks the potential inhibitory effects of a combination. The enhancement (fold shift in potency) of the EC₅₀s of Syn-690 and Syn-535 in combination with DCV (200 nM) were calculated. No enhancement of cytotoxicity was observed with combination treatment (data not shown): 50% cytotoxic concentration (CC₅₀) of Syn-535, ∼40 μM; CC₅₀ of Syn-690, ∼12 μM; CC₅₀ of DCV, ∼7 μM.

NS5A-Syn activity against WT gt 2a NIH and 3a replicons.

A colony elimination assay was used to assess the impact of NS5A synergy against 2 WT replicons (gts 2a NIH and 3a) that are less sensitive to DCV than gts 1a and 1b (Fig. 3 and Table 1). Genotype 2a NIH and gt 3a replicon cells were treated with DCV, Syn-690, or the combination for a period of 1 to 7 days. The ability of Syn-690 (100 nM) to suppress colony formation of gt 2a NIH replicon cells was minimal, with few colonies eliminated compared to a DMSO control. DCV was more efficient than Syn-690 at suppressing colony formation but was not able to eliminate the replicon, even after 7 days. The combination of DCV and Syn-690 is more efficient than either compound alone, with approximately 8 colonies remaining on each plate after 7 days (Fig. 3, top panel).

FIG 3.

Time course of gt 2a NIH and gt 3a replicon elimination. Genotype 2a NIH and 3a replicon cells (250,000 cells/plate) were treated with DCV (250 nM), Syn-690 (100 nM), or the combination in the presence of G418 for 1, 3, 5, or 7 days (D1, D3, D5, or D7, respectively). After selection with G418, the plates were stained to reveal the remaining replicon colonies.

Colony elimination of gt 3a by DCV–Syn-690 was more effective than the same combination on gt 2a NIH, decreasing the number of surviving colonies (Fig. 3, bottom panel). This likely reflects the difference in potency of Syn-690 on 2a NIH versus that of 3a (EC₅₀, 39 and 4 nM, respectively). DCV alone does not eliminate gt 3a, and the pattern of elimination is unchanged after day 1 exposure. The combination of DCV–Syn-690 efficiently suppresses gt 3a replicon colony formation after 3 days of exposure, with few colonies remaining. These inhibition patterns suggest that NS5A-Syn combined with DCV effectively suppresses most gt3a NS5A-resistant variants that arise during compound exposure or that are present in the HCV quasispecies.

NS5A-Syn activity enhances the pangenotype coverage of DCV.

Colony elimination studies have been used to determine the relative resistance barrier of DCV on multiple genotypes: 1b > 4a ≥ 5a > 6a ≅ 1a > 2a JFH > 3a > 2a NIH (3, 25, 26). DCV combined with a protease inhibitor (MK-5172) effectively eliminated the genotype with the lowest barrier to resistance (2a NIH), indicating that DCV used in combination with other DAAs can be highly effective on all genotypes. To determine if DCV combined with a synergistic compound provides similar genotype coverage, DCV, Syn-690, and Syn-535 were titrated versus replicon gts 1a, 2a NIH, 3a, 4a, 5a, and 6a. The EC₅₀s for the NS5A synergists vary significantly across genotypes (Table 1). Most EC₅₀s are in the nanomolar range (79 nM for Syn-690 on gt 1a to 0.3 nM for Syn-535 on gt 6a); however, both of these synergists have EC₅₀s of >1,000 nM on gt 4a. The EC₅₀s for DCV versus the same genotypes range from 54 nM for gt 2a NIH to <15 pM for gt 5a. The profiles indicate that, like DCV, the NS5A-Syn compounds can interact with NS5A proteins from all genotypes, and that the level of activity is genotype specific.

A colony elimination assay with DCV–Syn-690 was used to compare the activities of the DCV–Syn-690 combination on different genotypes (Fig. 4). DCV alone (250 nM) is efficient for replicon elimination on gts 4a, 5a, and 6a. The ability of DCV to eliminate replicons in this assay format (4a ≅ 5a > 6a > 1a > 3a ≅ 2a NIH) is in agreement with the results from previous experiments (3). Syn-690 alone (150 nM) appears most active on gt 3a but has poor activity overall. The few colonies remaining suggest the DCV–Syn-690 combination has the lowest resistance barrier versus that of gts 2a NIH and 3a.

FIG 4.

Replicon elimination studies with gts 1 to 6. Hybrid replicons expressing the HCV NS5A protein from gts 1a, 2a NIH, 3a, 4a, 5a, and 6a were treated for 7 days with DCV (250 nM), Syn-690 (150 nM), or DCV–Syn-690. After selection with G418, the plates were stained to reveal the remaining replicon colonies.

NS5A-Syn activity is highly effective on infectious HCV containing NS5A from gt 1a.

A more biologically relevant system was used to investigate the impact of NS5A-Syn on the DCV resistance barrier. A Renilla luciferase reporter incorporated into hybrid viruses (gts 1a, 2a L31M, and 3a) with a JFH-1 backbone provided an indirect measure of virus replication that could be monitored over several weeks to characterize the ability of compounds to suppress virus and/or eliminate (cure) the virus from host cells (Huh-7.5).

The exposure of cells infected with gt 1a to either DMSO or Syn-690 (150 nM, estimated minimum concentration [C_min] for humans) had no detectable effects on the virus, leaving the virus to exert an extensive cytopathic effect (CPE) (Fig. 5). These cultures were discontinued due to decreasing cell viability and luciferase activity after 11 days. DCV exposure (250 nM) yielded a sharp drop in gt 1a viral titer from ∼2.8E6 to 880 luciferase units by day 4, with subsequent rebound to baseline by day 25 (2.5E6 luciferase units, Fig. 5). DCV–Syn-690 mirrored the initial effect of DCV alone but caused a further decline to background levels (∼100 luciferase units) by day 11. Removal of the combination and passage of cells for 7 additional days indicated that the DCV–Syn-690 combination, with enhanced resistance barrier versus naive virus, was sufficient for a cure. This is identical to the results seen with authentic gt 1a (H77) replicon (Fig. 4).

FIG 5.

Genotype 1a, time course of naive and DCV rebound viruses. DMSO (dark-blue diamond, days 0 to 11) and Syn-690 (150 nM, light-blue square, days 0 to 11) did not inhibit gt 1a virus. Treatment with DCV (250 nM, red square, days 0 to 4) caused a decline, followed by rebound to baseline (day 25). Treatment with DCV–Syn-690 (green triangle, days 0 to 18) caused a decline to background levels. Retreatment of the DCV rebound virus with no compound (blue +, days 25 to 36), DCV (orange circle, days 25 to 36), or Syn-690 (purple x, days 25 to 36) had no effect. Retreatment of DCV rebound virus with DCV–Syn-690 caused a decline, followed by rebound at day 30 (blue x, days 25 to 37). Retreatment of DCV rebound virus with DCV–Syn-690 plus either ASV (100 nM, green dot, days 25 to 39) or BCV (100 nM, purple diamond, days 25 to 39) caused a decline to background levels (days 37 to 44). The up arrows indicate the removal of compound, and the down arrows indicate addition of compound.

Cells infected with gt 1a and treated with DCV were monitored during the rebound phase (days 4 to 25), and compound was removed on day 23 (Fig. 5). The rebounding virus-infected cells were subcultured in compound-free medium to reduce cellular compound retention; cultures that were not “washed out” retained a slight but detectable synergistic effect (data not shown). Cultures infected with rebound virus were subsequently treated with compounds (days 25 to 44) and monitored for inhibition (Fig. 5). Retreatment with DCV, Syn-690, or DMSO caused no detectable drop in luciferase activity, as expected. Treatment with DCV–Syn-690 (days 25 to 37) resulted in a rapid multilog drop (∼2.5 log₁₀), followed by rebound, similar to the profile observed after the initial exposure of the naive virus to DCV alone (days 1 to 4).

The treatment of DCV rebound virus with DCV–Syn-690–ASV or BCV (days 25 to 38) caused a viral decline to background levels (Fig. 5); all samples were negative for viral RNA by RT-PCR (data not shown). No viral rebound occurred after removal of the triple combination and continued passage from days 38 to 44. The addition of a third DAA from either inhibitor class effectively cured the cells of DCV-resistant virus, confirming the results of replicon elimination experiments previously reported for the synergist BMS-395 (22).

The sequences of rebound virus collected on day 25 from DCV-treated cultures and on day 38 from DCV–Syn-690-treated cultures show that highly resistant substitutions emerged at tyrosine 93 (Y93) (Table 3). Asparagine (N) and arginine (R) substitutions at Y93 (∼50% each) cause an ∼40,000–fold increase in EC₅₀ versus that of DCV (31). The treatment of DCV-resistant virus with DCV–Syn-690 selected an additional change at serine (S) 38 (Table 3). The S38 change to phenylalanine (F) has not been reported for DCV. S38 is highly conserved across HCV genotypes, providing evidence that Syn-690, along with DCV, may bind to the NS5A protein to cause replication inhibition; however, direct binding of NS5A-synergists to NS5A has not been documented.

TABLE 3.

Substitutions identified by population sequencing of rebound viruses

Treatment	Virus type	Amino acid changes detected in NS5A by genotypea
		1a	2a L31M	3a
Syn-690	Naive	ND	ND	S38F
Syn-535	Naive	ND	S38T	S38T
DCV	Naive	Y93N/R (∼50)	F28S	Y93H
Syn-690	DCV rebound	ND	ND	R56H (∼60) + E92K
DCV–Syn-690	DCV rebound	S38F+Y93N/R	ND	S38T+E92K
DCV–Syn-535	DCV rebound	ND	F28S+S38F	S38F+E92K

^aVirus populations (gt 1a, 2a L31M, and 3a) that rebounded on treatment were sequenced to determine the identity of substitutions that permitted viral escape. The substitutions observed were as follows: asparagine (N), arginine (R), and histidine (H) replacing tyrosine (Y) at amino acid (aa) 93; phenylalanine (F) and threonine (T) replacing serine (S) at aa 38; S replacing F at aa 28; and lysine (K) replacing glutamic acid (E) at aa 92. The numbers in parentheses are the percent estimated from the sequencing chromatogram. ND, not done.

DCV–Syn-690 versus DCV–Syn-535 on infectious HCV containing NS5A from gt 2a L31M.

Replicon elimination studies indicate that the barrier to resistance emergence for DCV and DCV–Syn-690 is slightly lower for gts 2a NIH and 3a than for other genotypes (Fig. 4). To further investigate the impact of NS5A synergy on the resistance barrier, the activities of Syn-690 and the more active synergist Syn-535 in combination with DCV were profiled, using the recombinant virus 2a L31M (Fig. 6). The exposure of 2a L31M-infected cells to DMSO had no detectable effect on the virus, and extensive CPE was observed during a 9-day treatment period before discontinuation. The exposure of 2a L31M virus-infected cells to DCV (250 nM) caused a sharp reduction in virus titer (2.1E6 to 1,160 luciferase units by day 5) and a subsequent rebound that was ∼4-fold higher than that at baseline (8.2E6 luciferase units by day 18). The treatment of 2a L31M-infected cells with Syn-690 (150 nM) caused a small drop from baseline (2.1E6 to 2.8E5 luciferase units), followed by a slow rebound (3.6E6 luciferase units) by day 14 (Fig. 6). This effect mirrored the gt 2a NIH replicon results (Fig. 4), in which the number of colonies remaining after treatment with Syn-690 alone was slightly lower (decreased staining intensity) than that with DMSO. Syn-535 (100 nM) was more effective on 2a L31M than Syn-690 (150 nM) and almost as effective as DCV (Fig. 6). The replicon EC₅₀s for Syn-535 (21 nM), Syn-690 (39 nM), and DCV (54 nM) are similar, suggesting that NS5A inhibitor characteristics other than potency impact virus reduction.

FIG 6.

Genotype 2a L31M, time course of naive and DCV or Syn-535 rebound virus. DMSO (dark-blue diamond, days 0 to 9) did not inhibit gt 2a L31M virus. Treatment with Syn-690 (green triangle, days 0 to 14), Syn-535 (purple x, days 0 to 21), and DCV (red square, days 0 to 14) caused a decline, followed by rebound. DCV–Syn-690 (blue x, days 0 to 11) and DCV–Syn-535 (orange circle, days 0 to 11) caused a decline to background levels. Retreatment of Syn-535 rebound virus with Syn-535 (blue line, days 14 to 21) had no effect. Retreatment of Syn-535 rebound virus with DCV (green line, days 14 to 21), or DCV–Syn-535 (red line, days 14 to 21) caused a decline, followed by rebound. Retreatment of DCV rebound virus with DCV (orange triangle, days 18 to 23) had no effect. Retreatment of DCV rebound virus with DCV–Syn-535 (blue square, days 18 to 25) caused a decline, followed by rebound. The up arrows indicate the removal of compound, and the down arrows indicate the addition of compound.

The treatment of 2a L31M virus with DCV–Syn-690 or DCV–Syn-535 mirrored the initial decline of DCV alone, with both combinations continuing to decline to background levels (∼200 luciferase units) by day 7. The removal of each combination on day 11 and passage of cells for 5 days in the absence of compounds confirmed the ability of either combination to cure the virus-infected cells (Fig. 6). The results compare favorably with the 2a NIH replicon exposed to DCV–Syn-690 (Fig. 4).

The multilog decline in virus titer caused by DCV alone and Syn-535 alone returned to baseline on days 18 and 14, respectively (Fig. 6). Cells infected with the rebound viruses were washed and passaged in the absence of compound before retreatment. Cells infected with virus that rebounded on Syn-535 treatment were retreated on day 14 with Syn-535, DCV, or DCV–Syn-535 (Fig. 6). As expected, retreatment with Syn-535 had no effect, and cultures were discontinued on day 14. Treatment of the Syn-535 rebound virus with DCV alone or DCV–Syn-535 caused an ∼2-log₁₀ or ∼2.5-log₁₀ reduction in viral titer, respectively (Fig. 6), indicating that the Syn-535 rebound virus is sensitive to DCV. Rebound of the Syn-535-resistant virus during treatment with DCV and DCV–Syn-535 was more rapid than rebound of the naive virus during treatment with DCV or Syn-535, suggesting that the rebound virus contains amino acid substitutions that enable the rapid selection of additional mutations that reduce sensitivity to DCV and DCV–Syn-535 (Fig. 6).

The sequence of the rebound virus from DCV treatment showed the emergence of F28S (Table 3). The F28S substitution in the 2a L31M background has been shown to arise with DCV selection and confer a high level of resistance to DCV (EC₅₀, >1.8 μM) (3). The sequence of the rebound virus from Syn-535 treatment showed the emergence of threonine at S38 (S38T). S38 is the same amino acid mutated in gt 1a by DCV–Syn-690 treatment; however, the substitution in gt 2a L31M is a polar residue (T) rather than the hydrophobic residue (F) observed for gt 1a. Syn-535 rebound virus exposed to DCV–Syn-535 yielded the emergence of linked variants F28S and S38F, which suggests that the S38 position is influenced primarily by synergistic compounds modulated by combination with DCV.

NS5A-Syn activity on infectious HCV containing NS5A from gt 3a.

Since gt 3a has proven to be difficult to treat with currently available therapies (32) and represented a challenge for DCV (Fig. 4), it was evaluated extensively with both synergistic compounds (Syn-690 and Syn-535). DMSO treatment of gt 3a-infected cells resulted in a rise in virus titer and a subsequent decrease in cells due to CPE (Fig. 7). DCV treatment (250 nM) of gt 3a virus resulted in a rapid decline and rebound, similar to the pattern on gt 1a and 2a L31M; however, the magnitude of the decline (∼2 log₁₀) was smaller, and the rate of rebound was faster (Fig. 7). The gt 3a DCV rebound virus returned to baseline by day 12, whereas gt 2a L31M and gt 1a required 18 and >20 days, respectively. This pattern suggests the DCV-resistant gt 3a variants in the quasispecies are more robust than gt 1a and 2a L31M-resistant variants, emerging more quickly under DCV pressure. These patterns are reflected in the colony profile observed in replicon cells exposed to DCV (Fig. 4). The gt 3a virus levels gradually declined with Syn-690 (150 nM) treatment. Suppression of the virus continued through day 14, when its numbers increased slightly; the delay in rebound suggests the virus adapts to Syn-690 pressure with difficulty. The virus was treated with Syn-690 for >30 days, with low but detectable activity persistently present and more efficient recovery of activity occurring at the end of passage (Fig. 7, day 38). The removal of compound resulted in a rapid recovery of luciferase activity, consistent with continued inhibition in the presence of Syn-690 (data not shown). Treatment results with DCV–Syn-690 were identical to the results for gt 1a and 2a, with efficient suppression of naive gt 3a virus and curing of the virus-infected cells in ∼12 days. No rebound was observed during subsequent passages (Fig. 7).

FIG 7.

Genotype 3a, time course of naive and rebound virus. DMSO treatment (purple x, days 0 to 10) did not inhibit gt 3a virus. DCV treatment (green triangle) caused a decline, followed by rebound to baseline by day 12. Syn-690 treatment (blue diamond) caused a gradual decline continuing up to 13 days. DCV–Syn-690 treatment cleared virus by day 12 (red squares). Retreatment of DCV rebound virus with the Syn-690 (orange circle) caused a decline, followed by a second rebound. Retreatment of DCV rebound virus with the DCV–Syn-690 combination (blue + sign) caused a decline, followed by a second rebound. The addition of a 3rd DAA (light-purple dot, NS3; or light-blue square, NS5B) to the DCV-Syn combination cured DCV-resistant cells. The up arrows indicate the removal of compound, and the down arrows indicate the addition of compound.

DCV-treated viral rebound was detected on day 7. Washout of DCV was followed by retreatment on day 12 (Fig. 7, down arrow). Continued treatment versus washout resulted in a 2-fold difference in titer (2.7E6 versus 4.3E6 luciferase units, respectively) on day 12 (Fig. 7). Retreatment with DCV after washout yielded no detectable inhibitory effect, similar to with the DMSO control. Syn-690 treatment caused an ∼1.2-log₁₀ decline in the DCV rebound virus on day 19, followed by rebound by day 28. The combination of DCV–Syn-690 suppressed the DCV-resistant variants, yielding a multilog decline in titer, followed by a return to baseline. The inclusion of a third DAA (NS3 or an NS5B inhibitor) with the DCV–Syn-690 combination cured the DCV-resistant virus infection, as indicated by the absence of rebound when the combination was removed and untreated cells were monitored from days 25 to 33 (Fig. 7). Although the NS3 and NS5B proteins in the gt 2a L31M-NS5A hybrid virus are from the gt 2a background, the activities of the NS3 and NS5B inhibitors on authentic gt 3a enzymes and replicons were similar to (for NS3) or better than (for NS5B) the activity on gt 2a. Thus, we expect the impact on authentic gt 3a virus to be similar to the impact on the hybrid virus used in the experiment (33, 34).

Sequencing of rebound virus yielded a histidine substitution (H) at Y93 for DCV-exposed virus (Table 3). Y93H has been shown to be highly resistant to DCV in gt 3a (3). A phenylalanine (F) substitution at S38 was identified in the Syn-690 rebound virus. The low level of luciferase activity that persisted for several weeks in this population suggests the S38F variant may have a low fitness for replication (Fig. 7). DCV rebound virus exposed to Syn-690 yielded R56H (∼60%) with E92K. The phenotype of R56H is under investigation; however, it has been observed that a K56R-L31F variant in gt 5a causes ∼8,000-fold resistance to DCV (3). The linked substitutions E92K and S38T emerged when DCV rebound virus was treated with DCV–Syn-690 (Table 3). E92K was shown previously to emerge under high-DCV-selective pressure (3). The selection by Syn-690 of the substitution at S38 in the gt 3a virus supports the hypothesis that S38 variants are a signature of synergistic compounds.

Dose-related changes in NS5A synergy.

Syn-535 is approximately 5-fold more potent than Syn-690 versus gt 3a (Table 1). Since gt 3a appears to be more difficult than other genotypes to treat in the clinic, Syn-535 was further evaluated on naive and DCV rebound gt 3a virus with lower concentrations (24 and 8 nM) than the concentrations used to treat gt 2a L31M (Fig. 6). DMSO treatment of infected cells had no inhibitory effect, and cultures were discontinued on day 8 due to CPE (Fig. 8a). Virus exposed to DCV (250 nM) rebounded rapidly beginning at day 4 and was above baseline by day 7. Surprisingly, Syn-535 at 24 nM suppressed naive gt 3a virus more efficiently than DCV at 250 nM, even though the EC₅₀s were similar (0.95 and 0.67 nM, respectively). This may reflect different binding interactions and is similar to the observations with Syn-690 versus DCV on gt 3a. Virus exposed to Syn-535 (24 nM) rebounded after day 4. Sequencing revealed an S38T change in this population. The combination of DCV–Syn-535 (8 and 24 nM) was effective at curing the gt 3a virus from the Huh-7.5 cells.

FIG 8.

Genotype 3a, time course and dose response of Syn-535 with and without DCV. (a) DMSO (blue-green x, days 0 to 7) had no effect on the naive gt 3a virus. DCV (250 nM, purple x, days 0 to 9) and Syn-535 (24 nM, blue diamond, days 0 to 11) caused a decline, followed by rebound. Both the DCV–Syn-535 (red square, 24 nM) and DCV–Syn-535 (green diamond, 8 nM) combinations cleared the virus by day 11. (b) DMSO (purple x, days 0 to 5) and DCV (250 nM, days 0 to 5, green triangle) had no effect on DCV rebound virus. Syn-535 (purple +, 24 nM) and DCV+Syn-535 (red square, 8 nM; blue diamond, 24 nM) caused a decline, followed by rebound. DCV–Syn-535–NS3 (blue x, 100 nM) or NS5B (orange circle, 300 nM) cleared the DCV rebound virus by day 12. The up arrows indicate the removal of compound.

To ensure that gt 3a DCV rebound virus was fully infectious, cells were infected with virus released from gt 3a cultures treated with DCV (2.8E6 luciferase units; Fig. 8b). Exposure of the infected cells to DCV or to DMSO had no effect on the virus, and cultures were discontinued after two passages. Treatment of the cells with Syn-535 (24 nM) had modest inhibitory effects before the virus rebounded toward baseline (days 5 to 9). DCV–Syn-535 yielded efficient inhibition at Syn-535 concentrations of 8 nM (∼7.5E4 luciferase units, day 2) and 24 nM (7.5E3 luciferase units, day 5) before the virus rebounded on days 5 and 7. Sequencing revealed changes at both S38, the synergistic compound signature variant, and E92. E92K likely reflects the increased selective pressure of the combinations (Table 3). The triple combination of DCV, Syn-535 (24 nM), and NS3 (100 nM) or NS5B (300 nM) cured the cells of DCV rebound virus (Fig. 8b).

DISCUSSION

Clinical resistance observed with DAA therapies, including NS3, NS5B, and NS5A inhibitors, has necessitated the exploration of combinations that suppress resistance and promote rapid suppression of all HCV genotypes. DAAs with profiles that inhibit mutations associated with first-generation HCV inhibitors provide highly effective combinations, decreasing the need for pretreatment screening and forming the cornerstones of emerging therapies (35). The profile of HCV NS5A synergists suggests they can be added to the list of effective HCV DAAs capable of inhibiting clinically significant resistance variants associated with first-generation NS5A inhibitors. NS5A synergy restores the effectiveness of DCV versus DCV-resistant variants to a level that is at least as effective as the level observed for WT viruses (22).

The strategy used to discover the NS5A synergistic compound class was counterintuitive and employed a combination of compounds placed onto replicon cell lines that were minimally affected by either compound when tested singly. The inhibitory potency of the NS5A synergists was revealed only by applying compounds in combination with DCV onto DCV-resistant replicon cell lines. DCV combined with synergistic compounds achieved antiviral potency reaching femtomolar levels in some cell lines. This level of potency is rare for antiviral compounds reported in the literature (36, 37).

The synergistic compounds Syn-690 and Syn-535, similar to DCV, have different EC₅₀s on gts 1 to 6. With the exception of gt 4a, Syn-535 was consistently more potent than Syn-690 on all genotypes tested; comparative titrations on a gt 4a L30H variant extended the trend to include gt 4a (38 and data not shown). The characterization of DCV–NS5A-Syn demonstrated the highly cooperative nature of the functional interaction and the ability of the functional interaction to enhance potency on gts 1 to 6. DCV–NS5A-Syn combinations are effective in both replicon and virus elimination studies, yielding comparable results. In this study, the most difficult genotypes for DCV to eliminate (gts 1a, 2a, and 3a) were used to challenge DCV–NS5A-Syn. Even though neither DCV alone nor NS5A-Syn alone could overcome this challenge, DCV–NS5A-Syn eliminated wild-type virus populations and rebound viruses containing highly DCV-resistant variants (25, 26).

DCV–NS5A-Syn also enhanced the resistance barrier. DCV is an effective synergy partner, and the HCV-inhibitory activity of DCV–NS5A-Syn is enhanced further by the inclusion of a DAA directed at a different target; triple combinations were consistently effective cures for HCV populations consisting exclusively of DCV-resistant variants. Other NS5A RCIs are also capable of synergy, strongly suggesting that the applicability of NS5A synergy extends beyond DCV (22 and data not shown); however, the potential clinical value of synergy remains to be tested.

Mechanisms that would enable DCV+NS5A-Syn to enhance inhibitory potencies have been proposed (22). It is hypothesized that in the proposed mechanisms, the binding of DCV or NS5A-Syn to an NS5A polymer can induce a conformational change that does not lead to inhibition but enhances the inhibitory binding of the second inhibitor. Higher-order structures and protein polymers have been postulated in pathways important for the amplification and modulation of signal transduction in cells and viruses (39). Viruses may employ polymers to enhance replication via assembly line arrays of components, such as viral proteins, to provide a cellular factory for viral replication (40). Polymeric forms of adenovirus E4-ORF3 consist of dimers, which interact to form higher-order oligomeric structures via flexible dimeric forms, providing large surfaces that can bind to and interrupt normal cellular functions while maintaining efficient virus replication (41). HCV NS5A protein is also reported to disrupt diverse cellular functions important for cell survival while maintaining viral replication (42, 43). Inhibitors directed at interfering with or disrupting the ability of capsid proteins to self-assemble into semipolymeric forms provide another example of inhibitors capable of targeting protein-protein interactions important for the assembly of multimeric polymers (43, 44).

NS5A-Syn binding is proposed to be similar to DCV binding, since previous compounds related to this series have been shown to bind directly, although indirect effects cannot be ruled out (6, 22). Located in a polymeric structure consisting of alternating NS5A dimers of differing conformations, one dimer form is postulated to interact with NS5A synergistic compounds, while another dimer form interacts with DCV (8, 22). The polymer model allows communication between dimer forms, and independent interactions might be modulated by either NS5A RCI or NS5A-Syn binding. This model is supported by the distinct pattern of resistance for DCV and NS5A-Syn. S38 changes emerged with NS5A-Syn molecules singly and when combined with DCV, implying that there is an interaction within the NS5A protein network that is unique to synergists. Yet, the synergistic functional interaction suggests the NS5A-Syn binding promotes perturbations outside the binding pocket(s). Cross-linking studies with NS5A inhibitors also indicate the presence of NS5A multimers inside cells (6). The formation of higher-order structures might endow NS5A with the ability to usurp cellular factors for viral advantage, similar to adenovirus ORF3, but it also reveals an Achilles' heel amenable to small-molecule targeting.

Resistance selection with NS5A synergistic compounds (22) caused changes at a highly conserved position in the NS5A protein (S38), which is adjacent to a highly conserved cysteine residue (aa 39) implicated in NS5A dimerization (14), HCV replication (17), and zinc coordination (45). Saturation mutagenesis studies have suggested that aa 38 may be associated with DCV resistance (46); however, data from clinical studies have yet to report aa 38 as a major site of resistance. NS5A mutations emerging from exposure to DCV alone have been well characterized and are consistent with variants selected by other NS5A RCIs (3, 25, 26, 31). Novel mutations observed following exposure to DCV–NS5A-Syn are less well characterized due to technical challenges caused by their low replication capacity (data not shown). Preliminary investigations using transient assays suggest that the S38F substitution does not significantly change the sensitivity of NS5A to DCV, with an estimated EC₅₀ of ∼61 pM on the gt 1a replicon with the S38F substitution. However, the linked substitutions S38F and Y93H yielded an EC₅₀ estimate of >2 μM for DCV (C. Wang, unpublished data). Efforts to fully define the EC₅₀s of DCV, Syn-690, and Syn-535 and combinations toward these mutants are continuing.

DCV has been shown to compete with direct compound binding to the NS5A protein (8). Differences in genotype DNA sequences, similar to what was seen for a resistant mutant, likely modulate the binding interactions to the NS5A protein, the NS5A protein dimers, and an NS5A oligomer (6, 8, 15). It is therefore not surprising that the synergy class of compounds would also have gt-dependent differences in activities if it is assumed they also interact with the NS5A protein (22). Evidence for these gt-dependent interactions includes (i) aa 38 changes that occur only in the presence of the synergistic compound class, (ii) the gt-dependent differences seen with the synergy compound class in both potency and replicon elimination results, and (iii) the inhibition of diverse DCV-resistant variants with DCV-Syn combinations. Combining all of the data, the broad gt inhibitory activity of DCV–NS5A-Syn combinations may be explained as being due to conserved structures (protein, protein-dimer, and oligomer) that can be conformationally adjusted by individual compounds and by combinations.

NS5A synergistic compounds expand the number of HCV protein-compound interactions that can be targeted to inhibit the virus. NS3 protease inhibitors bind the catalytic site or the NS4A activator site, defining 2 targets that inactivate protease function to inhibit replication (47). NS5B inhibitors bind to 1 of 4 allosteric inhibitor binding sites or the catalytic site, defining 5 targets that inactivate the polymerase to inhibit virus replication (33, 48, 49). Displacement of bound radiolabeled compound suggests the less well-characterized inhibitors of NS4B bind at least one site that inhibits replication (50). NS5A synergy strongly suggests that the NS5A protein has at least 2 sites of inhibitor binding that can communicate to shift the NS5A oligomer to a replication-incompetent form, significantly enhancing the potency and resistance barrier of NS5A RCIs, such as DCV. Whether other proteins capable of forming higher-order assemblies can be targeted for synergistic modulation remains speculative but is a worthy subject for exploration.

Funding Statement

This study was supported by Bristol-Myers Squibb.