Three-dimensional structure of a mutant HIV-1 protease displaying cross-resistance to all protease inhibitors in clinical trials.

Analysis of mutational effects in the human immunodeficiency virus type-1 (HIV-1) provirus has revealed that as few as four amino acid side-chain substitutions in the HIV-1 protease (M46I/L63P/V82T/I84V) suffice to yield viral variants cross-resistant to a panel of protease inhibitors either in or being considered for clinical trials (Condra, J. H., Schleif, W. A., Blahy, O. M., Gadryelski, L. J., Graham, D. J., Quintero, J. C., Rhodes, A., Robbins, H. L., Roth, E., Shivaprakash, M., Titus, D., Yang, T., Teppler, H., Squires, K. E., Deutsch, P. J., and Emini, E. A. (1995) Nature 374, 569-571). As an initial effort toward elucidation of the molecular mechanism of drug resistance in AIDS therapies, the three-dimensional structure of the HIV-1 protease mutant containing the four substitutions has been determined to 2.4-Å resolution with an R factor of 17.1%. The structure of its complex with MK639, a protease inhibitor of the hydroxyaminopentane amide class of peptidomimetics currently in Phase III clinical trials, has been resolved at 2.0 Å with an R factor of 17.0%. These structures are compared with those of the wild-type enzyme and its complex with MK639 (Chen, Z., Li, Y., Chen, E., Hall, D. L., Darke, P. L., Culberson, C., Shafer, J., and Kuo, L. C.(1994) J. Biol. Chem. 269, 26344-26348). There is no gross structural alteration of the protease due to the site-specific mutations. The C tracings of the two native structures are identical with a root-mean-square deviation of 0.5 Å, and the four substituted side chains are clearly revealed in the electron density map. In the MK639-bound form, the V82T substitution introduces an unfavorable hydrophilic moiety for binding in the active site and the I84V substitution creates a cavity (unoccupied by water) that should lead to a decrease in van der Waals contacts with the inhibitor. These changes are consistent with the observed 70-fold increase in the K value (2.5 kcal/mol) for MK639 as a result of the mutations in the HIV-1 protease. The role of the M46I and L63P substitutions in drug resistance is not obvious from the crystallographic data, but they induce conformational perturbations (0.9-1.1 Å) in the flap domain of the native enzyme and may affect the stability and/or activity of the enzyme unrelated directly to binding.

The emergence of drug resistance remains a major bottleneck in the pursuit of a long lasting, antiviral treatment against AIDS (1). When faced with selective pressure of an inhibitor, some 20 of the 99 amino acid residues of the HIV protease undergo mutations (2)(3)(4). Sequence analyses of virus isolates from patients participating in clinical trials have revealed that various amino-acid substitutions in the HIV 1 protease, in combination with as many as 10 or more residues, are associated with the decrease in antiviral efficacy of the protease inhibitor MK639 upon its prolonged usage in the course of 1 year (2) (Structure 1).
MK639 (previously L-735,524) is an inhibitor of peptide cleavages catalyzed by HIV-1 and HIV-2 proteases with K i values of 0.38 and 2.48 nM, respectively. It is effective against HIV replication in cell culture and is efficacious in reducing viral load in carriers of HIV-1 (5). Analyses of mutations in the HIV provirus have revealed (2) that as few as four amino acid substitutions in the HIV-1 protease (M46I/L63P/V82T/I84V) 2 suffice to yield cross-resistance to a panel of protease inhibitors either in or being considered for clinical trials, including MK639. To aid in the design of inhibitors able to evade viral resistance, it is important to understand the structural basis of resistance. We have previously determined (6) the x-ray crystallographic structures of HIV-1 and HIV-2 proteases complexed to MK639 at 2.0-and 1.9-Å resolution, respectively. Although four active-site residues (V32I, I47V, L76M, V82I) are altered between the two enzymes, there are only subtle changes in the binding mode of the inhibitor, suggesting that the structural basis of resistance may not be discernible until the degree of resistance becomes greater.
In this work, the native and bound structures of a mutant HIV-1 protease, with a 70-fold change in affinity (K i ϭ 26 nM) toward MK639, have been determined. The observed structural features of the MK639-bound mutant and wild-type proteases are consistent with the extent of resistance raised against MK639.
Crystallization of the mutant protease was accomplished at 4°C with use of the vapor diffusion method. The enzyme was prepared at 15 mg/ml in a pH 5 solution containing 10 mM MES, 1 mM EDTA, 1 mM DTT. The reservoir solution contained 0.6 M NaCl, 1 mM DTT, 3 mM * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. NaN 3 , and 0.1 M imidazole at pH 7. Hanging drops were prepared by mixing the protein and reservoir solutions in 1:1 (v/v) ratio. Tetragonal bipyrimidal crystals (0.9 mm ϫ 0.4 mm ϫ 0.4 mm) of the mutant protease were obtained in 5-10 days. The cell constants were a ϭ b ϭ 49.98 Å, c ϭ 108.1 Å with 1 dimer/asymmetric unit. A 2.4-Å resolution diffraction data set was collected on a RAXIS II imaging plate using CuK ␣ irradiation (Rigagu RU200 rotating anode) generated at 50 kV and 100 mA. The data set, showing clearly that the space group was P4 1 and not P4 1 22, encompassed 29,001 measurements with 8,651 unique reflections (82.3% completeness) and an R merge ([(⌺(I Ϫ ͗I͘) Ϭ ⌺I]) of 3.3%. The coordinates (10) of the wild-type HIV-1 protease were used as the molecular model. The initial R factor, defined as ⌺(͉F o Ϫ F c ͉) Ϭ ⌺͉F o ͉, was 0.308. Using the X-PLOR program package (11), rigid body refinements of the two monomers alternatively of the protease reduced the R value to 0.239 using the data in the 8-Å to 3.5-Å resolution range. Another cycle of X-PLOR refinement further reduced the R value to 0.226 at 2.4-Å resolution. A 2F o Ϫ F c map was calculated, and a mutant protease model was built by employing the program CHAIN on a Silicon Graphics system. The model was further refined; the final model included 82 solvent molecules with an R factor of 0.171 for a total of 8,101 reflections. The R F was 0.312, and the r.m.s. deviations, from ideal values of bond and angles, were 0.016 Å and 1.9°, respectively.
Crystals of the 4X protease complexed with MK639 were obtained at ambient temperature in hanging drops under vapor diffusion conditions against a solution of 0.1 M NaAc, pH 5.4, 0.5 M NaCl, 1 mM DTT, and 3 mM NaN 3 . The cell constants of the crystals were a ϭ 60.06 Å, b ϭ 86.90 Å, and c ϭ 46.68 Å in the space group P2 1 2 1 2. There was one protease dimer per asymmetric unit. The diffraction data, collected as described for the native enzyme and extending to 2.0-Å resolution, included 41,085 measurements with 13,590 independent reflections (86% completeness) and an R merge of 5.02%. The structure of the inhibitor-bound HIV-1 mutant protease was determined with the difference Unlike the wild-type HIV-1 proteases, the two subunits of the 4X mutant are not related by a 2-fold crystallographic axis in the asymmetric unit. Nonetheless, the r.m.s. deviation for the C ␣ atoms of the two subunits is relatively small (0.37 Å). The greatest difference between the subunits lies mainly in the flap loop and in the peptide chains containing residues 15-20 and 64 -72. 5

Structure of a Cross-resistant HIV-1 Protease Mutant 21434
64 -72, the r.m.s. deviation for the C ␣ atoms in the two subunits is further reduced to 0.25 Å. For the four substituted residues between the subunits, the displacements of the C ␣ atoms are 0.2 Å for Met-46, 0.31 Å for Leu-63, 0.2 Å for Val-82, and 0.18 Å for Ile-84. Fig. 2 illustrates the electron density of the quadruple mutant HIV-1 protease surrounding residues 46 and 63 as defined by x-ray diffraction. It reveals clearly that residues 46 and 63 of the mutant enzyme are isoleucine and proline as dictated by its DNA sequence. Similar unambiguous electron density definitions are found for V82T and I84V (electron density not shown). and Ile-84Ј are symmetry-related by a two-fold rotation but Val-82 and Val-82Ј are only pseudo symmetrically related because their propyl side chains are oriented differently about the C ␣ -C ␤ bond. The dashed lines mark the distances, within van der Waals radii, between atoms of the inhibitor and the CD1 atoms of Ile-84/Ile-84Ј, as well as the CG2 and CG1 atoms of Val-82 and Val-82Ј, respectively. Related by a pseudo two-fold rotation, the t-butyl and indanyl groups of MK639 are bound in the S2 and S2Ј pockets of the enzyme. Correspondingly, the pyridyl methyl piperidine and benzyl rings of the inhibitor are situated in the S3/S1 and S1Ј pockets. A bound water molecule The mutant OG1 atom of Thr-82 points away from the plane of the stereoview and is shielded in this diagram; but its CG2 atom is still within van der Waals distance (3.58 -3.72 Å) from the phenyl group of MK639 as opposed to a distance of 3.54 -3.70 Å seen for the corresponding wild-type atom.

FIG. 4. Comparison of the binding conformation of MK639 as seen in the active sites of (A) the HIV-1 (green) and HIV-2 (yellow) proteases (see Ref. 6 for details) and (B) the HIV-1 protease (green) and the 4X mutant (purple) as defined by x-ray diffraction data at 2-Å resolution.
The inhibitors were superimposed with the matrices derived by the least square fitting of all C ␣ atoms of each pair of the proteases. For clarity, the proteins are omitted in the stereoviews. is seen to be cushioned tetrahedrally between the tips of the flap loops (Ile-50 and Ile-50Ј) and the 2 amide oxygen atoms of the inhibitor. (A second bridging water molecule is found between the N2 amide nitrogen of the inhibitor and the carboxyl oxygen of Asp-29Ј.) Not seen in Fig. 3 are interactions of the hydroxyethylene group of the inhibitor hydrogen bonded to Asp-25 and Asp-25Ј (beneath the inhibitor structure). Altogether, seven atoms of MK639 are within hydrogen bonding distances from atoms of the enzyme, either directly or indirectly. These interactions have been reported in detail (6) for the HIV-2 protease. For comparison, the conformation of MK639 (as seen in the active sites of HIV-1 and HIV-2 protease) are shown in Fig. 4A.
In Fig. 3 are also shown thick lines colored in yellow representing the positions of the side chains of Thr-82, Thr-82Ј, Val-84, and Val-84Ј in the MK639-complexed 4X mutant protease, as dictated by its electron density and by super-positioning its C ␣ backbone onto that of the MK639-complexed wildtype HIV-1 protease. Both mutations in the active site introduce small alterations (ϳ0.3 Å) in their side chain positions. Most importantly, the V82T substitution, while isosteric, introduces an unfavorable hydroxyl moiety (OG1) at what was previously (wild type) the CG1 position of Val-82Ј in the S1 pocket, within van der Waals distance (3.39 Å) to the pyridyl methyl piperidine group of MK639. The corresponding substitution (OG1 of V82T) in the S1Ј site points away from the inhibitor and thus does not directly impact on binding because its CG2 atom is still within van der Waals distance (3.58 -3.72 Å) from the phenyl moiety. The I84V and I84VЈ substitutions, on the other hand, symmetrically create in both the S1 and S1Ј pockets a small void in bulk (ϳ25 Å 3 ) that can be expected to lead to a decrease in van der Waals interactions with the piperidine group and with the benzyl moiety of the inhibitor, respectively. 6 In an apparently unsuccessful attempt to fill this void, the CD1 atom of Ile-50Ј from the tip of the flap domain relocates by 1.29 Å toward this pocket by rotating Ϫ35°about the C ␣ -C ␤ bond and 170°about the C ␤ -C ␥ bond; however, a similar change is not seen for Ile-50. The structures of the bound MK639 in the active sites of the wild-type and 4X mutant proteases are shown in Fig. 4B, revealing that the binding mode of MK639 is essentially unchanged in the active site of the two enzymes. Together, these results suggest that resistance against inhibitor binding, in the case of MK639, is caused by subtle changes of the substituted side chains (the introduction of an isosteric but unfavorable hydrophilic group and the introduction of a smaller side chain to decrease van der Waals contacts) rather than repositioning the bound inhibitor (due to the introduction of a spatially hindering, larger side chain). The side-chain positions of Thr-82 and Val-84 are also not different in the open and closed forms of the mutant protease (data not shown), indicating that these residues are not perturbed, within a r.m.s. deviation of ϳ0.2 Å, upon binding of MK639. The observed changes in binding interactions of MK639 in the active sites of the 4X protease are consistent with an attenuated affinity of this inhibitor by a factor 70 (ϳ2.5 kcal/mol), 7 a value sufficient to render a potent protease inhib-itor ineffective as an antiviral agent against the HIV (2).
Whereas structural contributions of V82T and I84V toward resistance against MK639 are accountable with the diffraction data shown here, the role of the M46I and L63P substitutions is not obvious. These mutations away from the immediate vicinity of the active site induce significant changes in the C ␣ positions of residues in the flap domain of the enzyme (see text above related to Fig. 1 and Footnote 4), but the significance of these changes is unclear. However, preliminary kinetic results from our laboratory suggest that the combination of M46I and L63P mutations affords the protease greater catalytic efficiency than the native enzyme, and the V82T/I84V double mutations render the HIV-1 protease a very poor enzyme. 8 Thus, it may be that the M46I and L63P modifications, by introducing fine adjustments to the protease conformation, compensate dynamically for deleterious effects of the mutations (V82T and I84V) in the active site.
Our observations are in contrast to those reported by Baldwin et al. (15). These authors have observed rearrangements of the HIV-1 protease C ␣ backbone around residues 81-84 by up to 0.6 Å on binding of a symmetrical inhibitor, A-77003, when Val-82 is substituted by an alanine. These changes lead to a repacking of enzyme and inhibitor atoms in the S1 but not S1Ј subsite in a manner that would diminish the potential loss of binding affinity. The remainder of the mutant protease complex is little altered from that of the wild-type enzyme (supporting our interpretation that the changes seen in the flap domain of the 4X mutant are due to the M46I and L63P substitutions). It is difficult to generalize at this time the effect of resistance mutations of the protease based on limited available data thus far. To further extend our understanding of the structural basis of drug resistance in the HIV-1 protease, we are pursuing determination of additional inhibitor-bound structures of the 4X HIV-1 protease.