Structural studies of antiviral inhibitor with HIV-1 protease bearing drug resistant substitutions of V32I, I47V and V82I

Abstract

HIV-1 protease inhibitors are effective in HIV/AIDS therapy, although drug resistance is a severe problem. This study examines the effects of four investigational inhibitors against HIV-1 protease with drug resistant mutations of V32I, I47V and V82I (PR_Tri) that model the inhibitor-binding site of HIV-2 protease. These inhibitors contain diverse chemical modifications on the darunavir scaffold and form new interactions with wild type protease, however, the measured inhibition constants for PR_Tri mutant range from 17 to 40 nM or significantly worse than picomolar values reported for wild type enzyme. The X-ray crystal structure of PR_Tri mutant in complex with inhibitor 1 at 1.5 Å resolution shows minor changes in interactions with inhibitor compared with the corresponding wild type PR complex. Instead, the basic amine at P2 of inhibitor together with mutation V82I induces two alternate conformations for the side chain of Arg8 with new interactions with inhibitor and Leu10. Hence, inhibition is influenced by small coordinated changes in hydrophobic interactions.

1. Introduction

HIV/AIDS is a pandemic disease with about 37 million people infected worldwide [1]. HIV infection can be controlled by antiviral drugs targeting different stages of viral replication, however, the genetic diversity of the virus and rapid selection of drug resistant strains pose a severe challenge [2,3]. The retrovirus HIV includes two types, HIV-1 and HIV-2, and HIV-1 is subdivided into three groups (M, O, and N) and subtypes with different geographical distribution. HIV-2 infections are common in West Africa, however, some drugs designed for HIV-1 are not effective for HIV-2 infections [4].

One important class of antiviral drugs targets the viral protease (PR), which is crucial for production of infectious virus. PR processes cleavage sites in Gag-Pol region during viral maturation to produce individual structural proteins [5]. This aspartic protease is catalytically active as a dimer of 99-residue monomers [6]. Clinical inhibitors bind in the active site cavity of the enzyme and block binding of substrates. Some HIV-1 PR inhibitors, including amprenavir (APV), are significantly less potent against HIV-2 infections probably because the amino acid sequences of HIV-1 and HIV-2 PRs show only about 40% sequence identity [7]. In addition to its natural genetic diversity, the virus has evolved resistance mutations for all clinical protease inhibitors (PIs) [8]. Single, major mutations decrease binding of inhibitor, however, they can be deleterious for viral replication. Resistance mutations can alter the catalytic activity, binding affinity and stability of PR [9,10]. The virus evolves additional mutations that compensate by restoring effective viral replication in the presence of inhibitor [11].

HIV-1 PR with drug resistant mutations V32I, I47V and V82I (PR_Tri) has been evaluated as a model for inhibition of HIV-2 PR to overcome the problem of autoproteolysis of HIV-2 PR. The three mutations in PR_Tri alter residues in the inhibitor-binding cavity (Fig. 1A) and represent the changes in the inhibitor binding site of HIV-2 protease. APV showed poor inhibition of both PR_Tri and HIV-2 PR at 15- and 19-fold worse than for HIV-1 PR, while darunavir (DRV) and saquinavir (SQV) were effective inhibitors of all three enzymes [12]. Moreover, the individual mutations of V32I, I47V and V82I are associated with resistance to one or more clinical inhibitors [8]. Hence, PR_Tri was chosen to evaluate the ef-ficacy of investigational inhibitors for drug-resistant mutants and HIV-2 PR.

Fig. 1. HIV PR dimer with sites of mutations and selected inhibitors.

A) Sites of mutation in HIV-1 PR dimer with inhibitor GRL-1111. PR is shown as green ribbons with sites of the three mutations, V32I, I47V and V82I, indicated as grey spheres. Inhibitor is shown in red sticks.

B) Structures of darunavir and investigational inhibitors. The P2-P2^’ groups are indicated for DRV. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

The four investigational antiviral inhibitors were designed on the DRV scaffold (Fig. 1B) with the aim of introducing new interactions of the P2 group with PR. They exhibited potent antiviral activity on wild type and drug resistant strains of HIV-1. Crystal structures confirmed the presence of new interactions of wild type PR with the P2 groups of these inhibitors compared to DRV. Inhibitor 1 has a novel P2 bis-THF group with a basic amine that forms direct and water-mediated hydrogen bond interactions with the main chain carbonyl oxygen and amide of Gly48 in the PR flap [13]. Inhibitor 2 contains tris-tetrahydrofuranylurethane (tris-THF) as the P2 ligand instead of bis-THF in DRV [14–16]. The third THF group introduces new water-mediated hydrogen bond interactions with Gly27 and Arg87 in the PR dimer interface. Inhibitor 3 has a novel tricycle cyclohexyl-bis-tetrahydrofuranylurethane at P2, and resembles inhibitor 2 in its new water-mediated interactions [17]. This inhibitor showed favorable penetration of the central nervous system (CNS) in an in vitro model [18]. Inhibitor 4 has a 3-(S)eN-methoxycarbonyl amino substituted cyclopentyltetrahydrofuranyl (Cp-THF) at the P2 group [19]. The carbamate NH of the ligand forms hydrogen bonds with the main chain of Gly48 similar to those of inhibitor 1 and the carbamate carbonyl has a water-mediated interaction with the guanidinium group of Arg8. Therefore, we hypothesized they would show good inhibition of PR_Tri. Inhibition constants were measured for the four compounds, and a high resolution X-ray structure was determined for PR_Tri in complex with inhibitor 1.

2. Materials and methods

2.1. Inhibitors

Amprenavir (APV) (HPLC purity of 99.7%) was obtained from the AIDS Reagent Program, Division of AIDS, NIAID, NIH. Compounds 1–4 (>95.0% purity by HPLC) were provided by Dr. Arun Ghosh at Purdue University.

2.2. Protein purification

The clone for triple mutant PR_Tri (V32I, I47V and V82I) includes optimizing mutations of Q7K, L33I, and L63I to decrease autoproteolysis, and C67A, C95A to eliminate cysteine-thiol oxidation [20]. Protein was expressed in E. coli BL21 and purified from inclusion bodies as described previously [21,22] using gel filtration followed by reverse phase chromatography and refolding. Samples were concentrated to 5.0 mg/mL for crystallization or diluted for kinetic assays.

2.3. Protein crystallography

PR_Tri was mixed with inhibitor (dissolved in dimethylsulfoxide) at a molar ratio of 1:5. Crystals were grown at room temperature by hanging drop vapor diffusion. Each drop contained 1 mL protein and 1 mL reservoir solution. Crystals of PR_Tri in complex with compound 1 were grown from 0.1 M sodium acetate, pH 4.6, and 2 M NaCl. Crystals were cryo-protected in 30% glycerol and flash frozen in liquid nitrogen. X-ray diffraction data were collected on the SERCAT 22BM beamline, Advanced Photon Source, Argonne National Laboratory (Argonne, IL), and processed using HKL-2000 [23]. The structure was solved by molecular replacement with PR complex with APV (3NU3) [24] using CCP4i Phaser [25,26]. The structure was refined with SHELX-2014 [27], followed by REFMAC5 [28]. COOT [29] was used for visualization and refitting. Alternate conformations were modeled according to the electron density maps. Anisotropic B factors were applied in the refinement. Structural figures were made using PyMOL [30]. Atomic coordinates and structure factors for PR_Tri/1 have been deposited in the PDB [31] with ID: 6OTG.

2.4. Enzyme kinetic assays

Kinetic parameters of PR_Tri were determined in 3e5 replicate runs by monitoring hydrolysis of fluorescence substrate derived from the HIV-1 p2/NC cleavage site: Abz-Thr-Ile-Nle-p-nitro-Phe-Gln-Arg-NH₂ (BACHEM H-2992) (where Abz is anthranilic acid, Nle is norleucine, and p-nitro-Phe is p-nitrophenylalanine). Samples were equilibrated at 37 °C for 5 min prior to initiating the reactions. Enzyme activity was measured at 37 °C using a PolarStar Optima microplate reader (BMG Labtech) with excitation wavelength at 340 nm and emission wavelength at 420 nm, as described previously [24,32]. To determine catalytic efficiency, 10 μL of PR_Tri (final well concentration of 40–100 nM determined by active site titration with APV) was mixed with 100 μL reaction buffer (100 mMMES pH 5.6, 400 mM NaCl, 1 mM EDTA, and 5% glycerol). Reaction was initiated by adding 100 μL substrate (12–96 μM final concentration). Initial velocities (V₀) were determined using MARS software (BMG Labtech). The K_m and k_cat were determined by fitting data to the Michaelis-Menten plot of V₀ vs [substrate].

For inhibition studies, 10 μL PR_Tri was mixed with 98 μL reaction buffer and 2 μL inhibitor in DMSO (final well concentration 0–40 mM). Reaction was initiated with 90 μL substrate (final well concentration 60 mM). IC₅₀ values were determined using Sigma-Plot (Systat Software) by non-linear regression curve fitting to a dose-response plot of V₀ vs [inhibitor]. K_i values were calculated using the equation for tight-binding inhibitor of K_i= (IC₅₀ - 0.5[E])/ (1 þ [S]/K_m) [33].

For the urea denaturation assay, 10 mL PR_Tri was mixed with 100 mL reaction buffer containing 8 different urea concentrations (0–4 μM). Reaction was initiated with 90 μL substrate (final well concentration 60–72 μM) in 0–4 M urea. The urea concentration in each well remained the same throughout the experiment. A plot of V₀ vs [urea] was constructed using SigmaPlot (Systat Software) and sigmoidal curve fitting used to determine the urea concentration for 50% maximum velocity (UC₅₀).

3. Results

3.1. Kinetic parameters, inhibition and stability of PR_Tri mutant

Kinetic parameters were determined for PR_Tri hydrolysis of a fluorescent substrate analog at 37 °C prior to assaying the effects of compounds 1–4 under the same conditions. PR_Tri mutant had k_cat of 299 ± 48 μM/min and K_m of 72 ± 17.7 μM. The catalytic efficiency (k_cat /K_m) of 4.2 μM⁻¹ min ¹ is similar to that value of 6.5 μM ¹ min ¹ for wild type enzyme at 26° C [34].

The measured K_i values were 31.4 ± 3.9 nM for compound 1, 38.4 ± 1.2 nM for compound 2, 16.7 ± 0.8 nM for compound 3, and 39.5 ± 6.1 nM for compound 4. These K_i values were significantly worse for the mutant compared to K_i values of 2 to 10 pM reported for the wild type enzyme [13,15,17,19]. Therefore, these antiviral inhibitors are unlikely to be effective on HIV-1 with these mutations or on HIV-2.

The stability of the mutant and wild type proteases were assessed by measuring enzyme activity under urea denaturation. A UC₅₀ value of 0.97 ± 0.05 M was obtained for PR_Tri mutant. In comparison, the wild type enzyme gave a UC₅₀ value of 0.70 ± 0.07 M under the same conditions. These values suggest the mutant dimer is somewhat more stable than the wild type protease.

3.2. Crystallographic analysis of PR_Tri complex with inhibitor 1

Co-crystallization experiments were performed for the four compounds, however, crystals grew only for PR_Tri with compound 1. The crystal structure of PR_Tri/1 was solved at 1.50 Å resolution in the P2₁2₁2 space group, and refined to R/R_free values of 13.4/17.2%. Crystallographic statistics are listed in Table 1. The asymmetric unit of the crystal structure contained a dimer of PR_Tri, and the inhibitor was bound at the active site in two alternate orientations with an occupancy ratio of 0.5/0.5. The solvent was modeled with 133 water, 1 glycerol and 4 formic acid molecules from the crystallization and cryo-protectant solutions. The crystal structure of PR_Tri/1 was compared with the corresponding complex of wild type PR (PDB ID: 5BRY), which was determined at 1.34 Å resolution in the same space group with isomorphous unit cell dimensions and contained PR dimer with inhibitor bound in two orientations with 0.6/0.4 relative occupancy [13]. The two dimer structures superimposed with a low RMSD value of 0.23 Å on Cα atoms.

Table 1. Crystallographic Data Collection and Refinement Statistics.

(Values in parentheses are given for the highest resolution shell.).

	PRTri/GRL-1111
Space group	P21212
Unit cell dimensions: (Å)
A	58.57
B	86.52
C	45.32
Resolution range (Final Shell)(Å)	50–1.50 (1.55–1.50)
Unique reflections	37,188 (2,865)
Rmerge (%) overall (final shell)	5.9 (49.2)
I/σ(I) overall (final shell)	26.6 (3.1)
Completeness (%) overall (final shell)	96.8 (76.4)
Redundancy (final shell)	6.7 (4.1)
R (%)	13.4
Rfree (%)	17.2
No. of solvent atoms	151
RMS deviation from ideality
Bonds (Å)	0.020
Angle distance (degree)	2.29
Average B-factors (A2)
Wilson B-factor	17.7
Main-chain atoms	17.0
Side-chain atoms	23.6
Inhibitor	15.7
Solvent	28.3

The interactions of PR_Tri with inhibitor were analyzed in comparison to those in wild type PR complex. Inhibitor 1 forms six direct hydrogen bonds with PR_Tri, excluding the interactions of the central hydroxyl with the catalytic Asp25 and 25’ (Fig. 2). Additionally, it forms water-mediated interactions with flap residues Gly48, Ile50 and Ile50’, while one orientation of inhibitor (designated as the “major” conformation) has a water-mediated interaction with the carboxylate side chain of Asp29. PR-inhibitor hydrogen bond interactions are conserved in the mutant and wild type complexes with differences of no greater than 0.2 Å in length for the major inhibitor conformations. This inhibitor was designed to incorporate an amine on bis-THF at P2 that forms a new direct hydrogen bond interaction with the carbonyl oxygen of Gly48, a water-mediated interaction with the amide of Gly48 in the flap region, and a second water-mediated interaction with the side chain of Asp29. These interactions of compound 1 cannot occur with clinical inhibitor DRV, which lacks the basic amine. However, the minor conformation of 1 in PR_Tri complex has lost the second water and its interaction with Asp29.

Fig. 2. Hydrogen bond interactions of PR_Tri with inhibitor 1.

Protein is shown as grey sticks and inhibitor in green sticks. Dotted lines indicate hydrogen bond interactions with interatomic distances in Å. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

The hydrophobic side chains of residues 32, 47 and 82 are important components of the PR binding site for substrates or inhibitors. In wild type PR/1, Val82 interacts with the P1 and P1^’ groups of inhibitors, while Val32 and Ile47 contribute to the binding site for P2 and P2^’ groups. Ile82 in the mutant has similar interactions with inhibitor P1 and P1^’ as seen for Val82 of wild type enzyme. Val32 and Ile32 show similar van der Waals contacts with the P2 group of 1, while Val47 in the mutant has shifted to form new interactions with P2 compared to Ile47 in the flap. The P2^’ group has gained van der Waals interactions with Ile32^’ and lost contacts with Val47’ relative to those in the wild type structure.

In addition to the contacts with inhibitors, these residues interact with adjacent PR side chains. The side chains of residue 32 and 47 form hydrophobic contacts with each other and with neighboring side chains of Asp30, Lys45, Ile54, Val56, Leu76, Thr80, Ile84 and Ile50’ (Fig. 3). The majority of these internal hydrophobic contacts are retained in the PR_Tri mutant. The most significant change is a shift of about 0.5 Å in the Ca atoms of both Val47 and 47^’ relative to their locations in wild type PR. This change is illustrated in Fig. 3 for Val47/Ile47. This shift enables Val47 to form 3 new hydrophobic contacts with the P2 group of inhibitor, while losing contacts with neighboring side chains of Asp30, Lys45, Ile54 and Ile50’.

Fig. 3. Hydrophobic interactions of residues 32 and 47.

Residues in PR_Tri are shown in magenta sticks superimposed on corresponding residues of wild type PR in grey. Mutations Ile32 and Val47 are labeled in red, and P2 group of inhibitor is green. Hydrophobic interactions between residues are shown as dashed arrows with the number of contacts indicated. Black arrows indicate identical contacts in both structures, red arrows mark fewer contacts, and magenta arrows show more contacts in mutant than in wild type PR. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Mutation of Val82 to the longer Ile introduces new van der Waals interactions and significant changes in the side chain conformation of Arg8. In the wild type PR complex, the side chain of Arg8 has a single conformation that forms a key intersubunit ion pair with Asp29^’ and no contacts with inhibitor. In the PR_Tri/1 structure, the longer side chain of Ile82/82^’ appears to induce two alternate conformations of Arg8/8^’ with equivalent changes in both subunits (Fig. 4A). The guanidinium group of one conformation of Arg8 retains the ion pair with Asp29^’ and also forms a new van der Waals contact with the P2 group of inhibitor. The second conformation of Arg8 is rotated away to form van der Waals contacts with Leu10 instead of the intersubunit ionic interaction with Asp29’. The second conformation of Arg8/8^’ is not observed in the corresponding complex of PR_Tri/DRV, probably because DRV lacks the extra amine of compound 1 (Fig. 4B). These changes may be due to the local environment, since a multiple mutant PR20 has similar alternative conformations of Arg8/8’ related to the substitution of the larger Phe side chain instead of Leu10 [35].

Fig. 4. Interactions of alternative conformations of Arg8′ with neighboring side chains and inhibitor.

A) Mutant PR_Tri is shown in magenta sticks with alternate conformation of Arg8^’ in cyan and inhibitor 1 in green. Wild type PR and inhibitor are grey. Dotted lines indicate ionic interaction, and dashed lines indicate hydrophobic interactions. CH-p interactions with the aromatic P1 group of inhibitor are indicated by a single dashed line. B) Mutant PR_Tri in peach bonds with DRV in yellow sticks. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

4. Discussion

The four tested investigational antiviral compounds were poor inhibitors (K_i values of 16.7–39.5 nM) of HIV-1 PR_Tri despite their excellent pM inhibition of wild type enzyme [13,15,17,19]. Analysis of protease-inhibitor interactions in the crystal structure of PR_Tri/1 revealed the loss of one water-mediated polar interaction in one orientation of inhibitor and small changes in hydrophobic contacts. In particular, Val47 loses hydrophobic contacts with flap residues Lys45, Ile50^’ and Ile54 compared to those of wild type complex. The most substantial change compared to the wild type complex occurred for mutated residue Ile82/82^’, where the larger side chain produces two alternate conformations for the side chain of Arg8/8’. One conformation of the guanidinium group of Arg8/8^’ forms new van der Waals contacts with Leu10 and the P2 group of inhibitor 1 instead of its typical intersubunit ion pair with Asp29’/29. We previously reported the crystal structures of PR_Tri complexes with APV, DRV and SQV [12]. These structures exhibited a single conformation for Arg8/8^’, except for the APV complex which had two alternate conformations for Arg8^’ side chain in one subunit and new van der Waals contacts with Ile82^’ and Leu10^’ similar to those seen in the PR_Tri/1 structure. Moreover, only a single conformation was observed for Arg8/8^’ in our structures of HIV-2 PR with different inhibitors [12,36]. This change in conformation seen for Arg8/8^’ in PR_Tri/1, comprising partial loss of its intersubunit ion pair and new intra-subunit interactions, might be expected to alter the stability of the mutant, however, the UC50 for urea denaturation remains close to the value for wild type PR. In contrast, PR mutant with the single substitution of R8Q, which completely eliminated the ion pair with Asp29^’, showed decreased stability with UC₅₀ of 0.7 relative to wild type enzyme [37]. Overall, the loss of internal contacts among flap residues and disruption of the ion pair between Arg8 and Asp29’ are consistent with molecular dynamics simulations suggesting substitutions V32I, I47V and V82I in HIV-2 PR decrease the hydrophobic interactions with APV and DRV [38].

This new structure of PR_Tri/1 suggests how drug resistant mutations of V32I and I47V on opposite sides of the S2 and S2^’ subsites can partially compensate for altered hydrophobic interactions with inhibitors and other protease residues, while mutation V82I induces alternate conformations of Arg8/8’ and introduces new interactions with Leu10 and the P2 group of 1. Structural analysis suggests these changes in hydrophobic interactions are specific for this combination of compound 1 and mutant PR_Tri and act to decrease the inhibition.