Evaluating the Role of Macrocycles in the Susceptibility of Hepatitis C Virus NS3/4A Protease Inhibitors to Drug Resistance

Synthesis of protease inhibitors with varied P2 moieties and macrocyclizations

The non-covalent NS3/4A protease inhibitors, such as asunaprevir (1), danoprevir (2), vaniprevir (3), MK-5172 (4), contain similar core structures, including an acylsulfonamide-linked P1 cysteine mimic, a substituted P2 proline, a hydrophobic P3 side chain, and a P4 capping group (Figure 1). However, these inhibitors are structurally distinct as heterocyclic moieties at P2 are different and inhibitors are either linear or macrocyclic; macrocyclization is either between the P1 side chain and the P3 side chain (mcP1P3) or the P2 heterocyclic moiety and the P4 capping group (mcP2P4). To evaluate the effect of P2 moieties and macrocylization status on inhibitor activity against drug resistant variants, we designed and synthesized linear and macrocyclic analogues of asunaprevir, danoprevir, vaniprevir, and MK-5172 (Figure 1). Danoprevir and Vaniprevir have many key components in common and their analogues are clustered together, hence three inhibitor series with the same P2 group but different macrocylization status.

Open in a separate window

Figure 1

Structures of NS3/4A protease inhibitors: (A) FDA approved drugs telaprevir and boceprevir; (B) asunaprevir (1) and P1–P3 macrocyclic analogue Asu-mcP1P3 (1a); danoprevir (2), linear analogue Dan-linear (2a), vaniprevir (3), and P1 olefin analogue Dan-mcP2P4 (3a); MK-5172 (4), linear analogue 5172-linear (4a) and P1–P3 macrocyclic analogue 5172-mcP1P3 (4b); the canonical nomenclature for drug moiety positioning is indicated using telaprevir and asunaprevir; arrows indicate corresponding analogues. Asunaprevir, danoprevir/vaniprevir and MK-5172 inhibitor series, which share the same P2 moiety but have different macrocylization status, are indicated by orange, green and blue, respectively.

For asunaprevir, a linear acylsulfonamide compound with an ether-linked isoquinoline moiety at P2 proline, a P1–P3 macrocyclic analogue Asu-mcP1P3 (1a) was designed. Danoprevir and vaniprevir are both macrocyclic compounds containing a carbamate-linked isoindoline moiety at P2 proline, but the location of the macrocycle is different: danoprevir contains a P1–P3 macrocycle while vaniprevir contains a P2–P4 macrocycle, and a saturated P1 side chain. For direct comparison with danoprevir, a danoprevir linear analogue Dan-linear (2a) was designed along with a vaniprevir P1 olefin analogue Dan-mcP2P4 (3a). For MK-5172, which is a P2–P4 macrocyclic compound containing an ether-linked quinoxaline moiety, a linear analogue 5172-linear (4a) and a P1–P3 macrocyclic analogue 5172-mcP1P3 (4b) were prepared; in both analogues the hydrocarbon chain of the P2–P4 macrocycle was replaced with an ethyl group at the quinoxaline.

The designed macrocyclization analogues were prepared using synthetic sequences illustrated in Figure 2. Analogous to asunaprevir synthesis, Asu-mcP1P3 (1a) was prepared from the intermediate 5, obtained by the reaction of Boc-trans-4-hydroxy-L-proline and 1,7-dichloro-4-methoxyisoquinoline (41). Coupling of 5 with the P1–P1′ fragment 6 (42) provided the P2–P1′ intermediate 7. Boc deprotection and coupling of the resulting free amine with the amino acid 9 provided the precursor bis-olefin 10 for ring-closing metathesis. Cyclization of the bis-olefin intermediate 10 was accomplished using a highly efficient ring-closing metathesis catalyst Zhan 1B, and provided the target compound Asu-mcP1P3 (1a) (Figure 2A).

Open in a separate window

Figure 2

Synthesis of designed protease inhibitors; reagents and conditions: (a) HATU, DIEA, DMF, CH₂Cl₂; (b) HCl, MeOH; (c) HATU, DIEA, DMF; (d) Zhan 1B catalyst, 1,2-DCE; (e) Potassium vinyltrifluoroborate, TEA, PdCl₂(dppf)-CH₂Cl₂, EtOH; (f) H₂, 10% Pd/C, MeOH, dioxane; (g) 4N HCl, dioxane; (h) Boc-Tle-OH, HATU, DIEA, DMF; (i) LiOH, THF, H₂O; (j) HATU, DMAP, DIEA, DMF, CH₂Cl₂.

Dan-linear (2a) was synthesized using the reaction sequence analogous to danoprevir (37). Vaniprevir (3) and its P1 olefin analogue Dan-mcP2P4 (3a) were synthesized following the synthetic methods reported by McCauley et al (43).

The MK-5172 analogues 5172-linear (4a) and 5172-mcP1P3 (4b) were synthesized using the reaction sequences shown in Figure 2B. The intermediate 11 was prepared from 3-chloro-7-methoxyquinoxalin-2-ol and the bosylated Boc-cis-4-hydroxy-L-proline, using Cs₂CO₃-mediated S_N2 displacement reaction (24). Palladium-catalyzed vinylation followed by hydrogenation of the double bond provided the common P2 intermediate 12. The acyclic analogues 5172-linear (4a) was assembled from 12 by a four step reaction sequence, including Boc deprotection, P2–P3 amide coupling, ester hydrolysis, and coupling with the P1–P1′ fragment 6 to provide the target compound 5172-linear (4a). The precursor for 5172-mcP1P3, bis-olefin 18, was synthesized from the P2 fragment 12 in an analogous reaction sequence. Finally, cyclization of the bis-olefin intermediate 18 using ring-closing metathesis catalyst Zhan 1B provided the 5172-mcP1P3 (4b).

Macrocyclization enhances inhibitor potency against wild-type and drug resistant variants

The enzyme inhibition constants (K_i) of asunaprevir, danoprevir, vaniprevir, MK-5172, and their respective analogues were determined against genotype 1a full-length NS3/4A, the isolated protease domain, and drug-resistant variants R155K, V36M/R155K, A156T, and D168A of both full-length enzyme and protease (Figure 3a, Tables S1 and S2, Figure S1). In addition, half-maximal inhibitory concentrations (IC₅₀) were determined using viral replicon-based inhibition assays with wild-type genotype 1b HCV and drug resistant variants R155K, V36M/R155K, A156T and D168A; telaprevir was used as a control (Figure 3b, Table S3, Figures S2–S6).

Open in a separate window

Figure 3

(a) Enzyme inhibition constants for HCV genotype 1a full-length NS3/4A and isolated protease domain (dark and light bars) and (b) replicon-based half maximal inhibitory concentrations for genotype 1b HCV NS3/4A and drug resistant variants for (first row) telaprevir and asunaprevir series, (second row) danoprevir/vaniprevir series, and (third row) MK-5172 series. Error bars represent standard errors of the mean (n = 4); * indicates Ki and IC₅₀ values are greater than the highest inhibitor concentration tested.

In each inhibitor series, macrocyclic compounds exhibited lower K_i’s than the corresponding linear analogues against wild-type protease and, in most cases, drug-resistant variants. The P2–P4 macrocyclic compounds were highly active against wild-type protease, with K_i values in the mid to high pM range (K_i = 0.46 nM for vaniprevir, K_i = 0.22 nM for Dan-mcP2P4 and K_i = 0.06 nM for MK-5172, respectively), and retained low to high nanomolar activity against resistant variants. The P1–P3 macrocyclic compounds exhibited slightly lower enzymatic activities against wild-type, but overall retained better activities against drug-resistant variants. Notably, 5172-mcP1P3 (A156T K_i = 74.3 nM) retained 3-fold better enzymatic activity against A156T variant than the parent compound MK-5172 (A156T K_i = 251.3 nM). Interestingly, compared to vaniprevir, Dan-mcP2P4, though only slightly better against wild-type protease, was significantly more active against drug-resistant variants. This signifies the importance of the terminal olefin in the P1 moiety where potential π stacking interactions with F154 are possible (25, 37), likely reducing the off-rate of the P1 olefin analogue compared to vaniprevir.

The antiviral activities of macrocyclic compounds in replicon assays against wild-type HCV and resistant variants correlate with protease binding affinities, and further demonstrate enhanced potency of macrocyclic compounds compared to linear analogues (Figure 3). In general, the P1–P3 macrocyclic compounds exhibited better antiviral potencies against wild-type and drug-resistant variants than the corresponding P2–P4 analogues. Against wild-type virus, P1–P3 macrocyclic inhibitors danoprevir, Asu-mcP1P3 and 5172-mcP1P3 exhibited antiviral potencies in the sub-nanomolar range with IC₅₀ of 0.24, 0.23 and 0.26 nM, respectively. The antiviral potency of Asu-mcP1P3 was significantly better than the parent drug, showing a 4-fold enhancement against wild-type and 4–16-fold against resistant variants. Interestingly, 5172-mcP1P3 (WT IC₅₀ = 0.26 nM) was almost equipotent to MK-5172 (WT IC₅₀ = 0.29 nM) against wild-type virus and R155K, V36M/R155K, and D168A variants, but exhibited 12-fold improved potency against A156T variant (A156T IC₅₀ = 3.96 nM for 5172-mcP1P3 versus 46.6 nM for MK-5172).

The macrocyclic inhibitors are more potent compared to their linear analogues likely due to the restriction of flexibility of the molecule core around the P2–P3 proline amide bond, which is expected to reduce the conformational entropic penalty for binding thus increasing the overall binding energy (39, 44). The P1–P3 macrocycle restricts the P1 terminus and rotation around the P2–P3 amide bond, which skews the cis-trans equilibrium, favoring the trans geometry which is a conserved feature in all protease structures in complex with viral substrate peptides (37, 45). In contrast, the P2–P4 macrocycle restricts the flexibility of the large heterocyclic moiety attached to P2 proline, and probably restricts the rotation around the P2–P3 amide bond, in concert with the bulky tert-butyl group at P3. In addition, the long hydrophobic chain of the macrocycles reduces the overall polarity of the molecule and makes additional interactions with the protease. Thus, macrocyclization provides several advantages, including pre-organization of the molecule in a binding-competent conformation and restriction of the inhibitor geometry to stabilize protease-inhibitor interactions (44).

P2 moiety also determines inhibitor susceptibility to drug resistance

The susceptibilities of linear and macrocyclic protease inhibitors to drug resistance were evaluated by normalizing K_i and IC₅₀ values with respect to the corresponding wild-type enzyme (Figure 4) to better ascertain the relative changes. The K_i’s against drug-resistant variants generally correlated with IC₅₀’s, but the fold potency losses in replicon assays were in most cases lower than those observed from in vitro enzymatic assays. However, we focused on the fold-changes in enzymatic inhibition constants as these reflect the direct effects of drug resistance mutations on inhibitor binding, while antiviral potencies in cellular replicon assays may reflect changes in inhibitor binding when the protease is part of an active replication complex.

Open in a separate window

Figure 4

Resistance profiles of protease inhibitors in (a) enzyme inhibition and (b) cell-based replicon assays for (first row) asunaprevir series, (second row) danoprevir/vaniprevir series, and (third row) MK-5172 series. Enzyme inhibitory and antiviral activities against mutants were normalized with respect to the wild-type full-length HCV NS3/4A protease, NS3/4A protease domain or wild-type HCV clone. Error bars represent propagated standard errors.

Within each inhibitor series of same P2 moiety but different macrocyclization (Figure 4 – individual rows) the drug resistance profiles are similar, indicating that the identity of P2 moiety largely determines the susceptibility to drug resistance. Notably, activity losses for the linear and P1–P3 macrocyclic analogues were similar against all resistant variants. The P2–P4 macrocyclic inhibitors displayed higher fold-losses compared to linear and P1–P3 macrocyclic compounds against A156T variants and, to a small extent, D168A variants, but the overall resistance profiles are similar in each series. Thus, regardless of the macrocyclization, inhibitors with different P2 moieties exhibit distinct resistance profiles.

The P2–P4 macrocyclic inhibitors are more susceptible to drug resistance possibly because the P2 moiety is constrained to allow stronger interactions with the protein, reducing the inhibitor’s ability to adapt to changes in binding environment due to mutations. In all inhibitors, the large heterocyclic P2 moieties bind to the protease in the S2 sub-site, but vary in how they interact with the primary sites of drug resistance mutations: R155, A156, and D168 (37, 38). These mutations lead to disruption of stacking interactions (R155K), pronounced steric clashes (A156T), or disruption of the electrostatic network around the S2 sub-site (D168A), but the rigidity of the P2–P4 macrocycle prevents the inhibitor from adjusting to these changes (38). In contrast, the P1–P3 macrocyclic inhibitors are less susceptible to drug resistance as the location of the macrocyclization is distal from the sites of primary drug resistance mutations. Thus, the P1–P3 macrocyclization, although stabilizing the inhibitor’s interactions with the protease, allows the P2 moiety to be flexible.

Specific interactions of the P2 moiety with known positions of drug resistant mutations explain the mechanism behind their distinct resistance profiles. In the danoprevir/vaniprevir series, the aromatic ring of P2 isoindoline stacks on R155 side chain, making favorable cation-π stacking interactions (38). As R155 conformation is stabilized by hydrogen bond interactions with D168, mutations at R155 and D168 destabilize this packing thereby lowering the affinity of these inhibitors (38). For asunaprevir and analogues, although there is no reported co-crystal structure, the substituted isoquinoline moiety is expected to interact similarly with R155 and D168. Future co-crystal structures will shed more light on the molecular mechanisms of resistance for these inhibitors.

The in vitro resistance profiles of NS3/4A protease inhibitors generally correlate with resistance observed in patients failing protease inhibitor therapy. Asunaprevir resistance-associated variants mainly include protease variants D168A/E/V/Y (46), in agreement with significant activity loss against D168A variant in vitro. Danoprevir resistance is primarily associated with R155K variant (29), while vaniprevir selects for both R155 and D168 variants (47). In in vitro assays, danoprevir was highly susceptible to R155K mutation but maintained relatively better activity against A156T and D168A variants, while vaniprevir was highly susceptible to all three variants.

We have previously shown that MK-5172 retains relatively better potency against R155K and D168A variants, but is highly susceptible to A156T variant, due to steric clashes with the P2–P4 macrocycle (38). Both the linear and P1–P3 macrocyclic analogues of MK-5172 retain better potency against drug resistant variants compared to the parent compound. In fact, the P1–P3 macrocyclic analogue 5172-mcP1P3 displays single digit nanomolar antiviral potency against all four drug-resistant HCV variants. Thus compared to MK-5172, 5172-mcP1P3 exhibits a flatter resistance profile. Inhibitors with similar or slightly reduced potency, but flatter resistance profiles may prove to be better than those with high potency against wild-type and pivotal resistant variants but reduced activity against other variants such as A156T. These variants may be less competent for replication, but could select for secondary mutations that restore viral fitness, rendering the inhibitors highly susceptible to these variants, e.g. R155K versus R155K/V36M.

The activity profile of MK-5172 analogues suggest that even in the absence of P2–P4 macrocycle, the P2 quinoxaline moiety maintains the unique binding conformation observed in MK-5172 (38). In fact, the flexible P2 quinoxaline moiety lacking P2–P4 macrocycle can be easily modified at 3-position for structure-activity exploration, and together with changes at the P4 capping group, provide opportunities for improving potency against drug-resistant variants and for optimizing pharmacokinetic properties.

Expression and Purification of NS3/4A Protease Constructs

The expression and purification scheme for the isolated protease domain is detailed elsewhere (37). Briefly, transformed BL21(DE3) E. coli expression cells were grown to an A₆₀₀ of 0.6, induced with 1 mM isopropyl-1-thio-β-D-galactopyranoside (IPTG) and incubated with shaking for 5 hours at 37 °C. Cells expressing the protein were harvested by centrifugation and stored at −80 °C. Frozen pellets were resuspended in resuspension buffer (50 mM phosphate buffer, 500 mM NaCl, 10% glycerol, 2 mM β-mercaptoethanol (β-ME), pH 7.5), lysed and centrifuged to pellet the cell debris and the resulting supernatant was applied to a nickel column (Qiagen). The column was washed with resuspension buffer and the protein was eluted with resuspension buffer, supplemented with 200 mM imidazole. The eluate was supplemented with thrombin and dialyzed overnight to cleave the His-tag and remove the imidazole. The purified protein was flash frozen in liquid nitrogen and stored at −80 °C.

For the expression and the purification of the full-length NS3/4A, transformed BL21(DE3) E. coli expression cells were grown to an A₆₀₀ of 0.6 at 37 °C, transferred to 20 °C, induced by 0.5 mM IPTG and incubated with shaking for 4 hours. Cells were then harvested via centrifugation; cell pellets were washed with 1X phosphate-buffered saline (PBS), repelleted and stored at −80°C. Frozen pellets were resuspended in Buffer HT (25 mM HEPES, 500 mM NaCl, 10% glycerol, 0.1% OβG, 2 mM TCEP and 20 mM imidazole, pH 8.0) supplemented with DNAse I (Roche) and homogenized using a cell disruptor (Micro Fluidics). Lysed cells were centrifuged to clear the cell debris and applied to a 1 mL HisTrap HP column (GE Life Sciences) using an AKTA Purifier (GE Life Sciences). The protein was washed with Buffer HT supplemented with 40 mM imidazole and eluted with Buffer HT supplemented with 250 mM imidazole. The eluate was dialysed overnight against Buffer S (25 mM MES, 150 mM NaCl, 10% glycerol, 0.1% OβG, 2 mM TCEP, pH 6.0). Dialyzed protein was applied to a Mono S Column (GE Life Sciences) and eluted with a linear gradient of NaCl up to 1 M. The eluate was judged >90% pure by polacrylamide gel electrophoresis, concentrated, flash frozen and stored at −80 °C.

Determination of Michaelis-Menten constants

Protease cleavage assays were performed to determine Michaelis-Menten constants (K_M), which were then used to obtain inhibition constants through Morrison’s equation (see below). The assays were done in a final volume of 60 μL containing protease assay buffer (50 mM Tris, 2.5% glycerol, 0.1% OβG, 5 mM TCEP, 1% DMSO, pH 7.5) and up to 20 μM of a modified HCV NS3/4A protease substrate Ac-DE-Dap(QXL-520)-EE-Abu-ψ-[COO]AS-C(5-FAMsp)-NH2 (Anaspec) in black 96-well flat bottom non-binding surface half-area plates (Corning) at room temperature. The reaction was initiated by the rapid injection of 10 μL of HCV NS3/4A protease to a final concentration of 20 nM. The fluorescence output from the depsipeptide cleavage product was measured kinetically for at least an hour using an EnVision plate reader (Perkin Elmer) with excitation wavelength at 485 nm and emission at 530 nm. End point readings were also taken after 6 hours. For each construct, three independent experiments were performed.

Inner filter effect in higher concentrations of substrate was corrected using the following scheme: End point readings were subtracted from the background fluorescence values to obtain reaction amplitudes which were plotted against substrate concentration. Initial linear portion of this curve was fitted with linear regression and extrapolated to maximum substrate concentration to obtain the ideal amplitude line. Corresponding amplitude values in the ideal amplitude line were divided by the actual amplitudes to obtain inner filter effect correction parameters which were then multiplied by initial velocities to obtain corrected velocities. Corrected velocities were plotted against substrate concentrations and all the replicates were fitted to Michaelis-Menten equation globally to obtain Michaelis-Menten constants (K_M), sharing K_M between different data sets. Except full-length A156T mutant, K_M values for all the constructs were similar, which signify that the substrate recognition was not altered significantly with respect to different multi-drug resistant mutations and helicase mutations.

Enzyme Inhibition Assays

All enzyme inhibition assays were performed in non-binding surface 96-well black half-area plates (Corning) in a reaction volume of 60 μL. The NS3/4A protease (~1 nM) was pre-incubated with increasing concentration of drugs in protease reaction buffer for an hour. The reaction was initiated by the rapid injection of 5 μL of HCV NS3/4A protease substrate to a final concentration of 200 nM and kinetically monitored using a Perkin Elmer EnVision plate reader (excitation at 485 nm; emission at 530 nm) for an hour with 30 sec between data points. At least four independent datasets were collected for each inhibitor with each protease construct. Each inhibitor titration included at least 12 inhibitor concentration points, which were globally fit to the Morrison equation to obtain the Ki value.

Initial velocities were obtained from the progress curves and plotted against inhibitor concentrations to get inhibition curves. Resulting curves were fitted to Morrison’s equation:

Where [E]_T is the total enzyme concentration, [I]_T is the total inhibitor concentration, [S] is the substrate concentration, K_M is the Michaelis-Menten constant (obtained from protease cleavage assays), V₀ is the initial velocity when there is no drug, V_i is the initial velocity at [I]_T, Kiapp is the apparent inhibition constant and K_i is the inhibition constant. Non-linear regression analyses were performed where independent replicates from different protease constructs with different drugs were fitted globally, sharing K_i.

This work was supported by the National Institute of Allergy and Infectious Diseases of the NIH (R01-AI085051). We thank members of the Schiffer laboratory for helpful discussions and N. Kurt-Yilmaz for editorial assistance.