Skip to main content
NCBI home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2006 Nov 20;103(49):18464–18469. doi: 10.1073/pnas.0605809103

Crystal structure of HIV-1 protease in situ product complex and observation of a low-barrier hydrogen bond between catalytic aspartates

Amit Das †, Vishal Prashar †, Smita Mahale ‡, L Serre §, J-L Ferrer §, M V Hosur †,
PMCID: PMC1693685  PMID: 17116869

Abstract

HIV-1 protease is an effective target for designing drugs against AIDS, and structural information about the true transition state and the correct mechanism can provide important inputs. We present here the three-dimensional structure of a bi-product complex between HIV-1 protease and the two cleavage product peptides AETF and YVDGAA. The structure, refined against synchrotron data to 1.65 Å resolution, shows the occurrence of the cleavage reaction in the crystal, with the product peptides still held in the enzyme active site. The separation between the scissile carbon and nitrogen atoms is 2.67 Å, which is shorter than a normal van der Waal separation, but it is much longer than a peptide bond length. The substrate is thus in a stage just past the G'Z intermediate described in Northrop's mechanism [Northrop DB (2001) Acc Chem Res 34:790–797]. Because the products are generated in situ, the structure, by extrapolation, can give insight into the mechanism of the cleavage reaction. Both oxygens of the generated carboxyl group form hydrogen bonds with atoms at the catalytic center: one to the OD2 atom of a catalytic aspartate and the other to the scissile nitrogen atom. The latter hydrogen bond may have mediated protonation of scissile nitrogen, triggering peptide bond cleavage. The inner oxygen atoms of the catalytic aspartates in the complex are 2.30 Å apart, indicating a low-barrier hydrogen bond between them at this stage of the reaction, an observation not included in Northrop's proposal. This structure forms a template for designing mechanism-based inhibitors.

Keywords: AIDS, catalysis, reaction intermediate, x-ray crystallography

HIV-1 protease (PR) is a homodimeric, aspartyl PR containing the signature sequence Asp-Thr-Gly in each monomer. The enzyme, which cleaves the viral polyprotein at eight different sites during the maturation process of the virus, is an important target for structure-based drug design (1–4). Emergence of drug-resistant mutations presents a new challenge, and additional inputs, such as the mechanism and interactions of the active enzyme with plain peptide substrates rather than with analogs, are required to tackle drug resistance. To address this question, we have undertaken to solve the crystal structures of HIV-1 PR oligopeptide substrate complexes. The enzyme used in the present study is a single polypeptide chain where the C terminus of the first monomer is linked with the N terminus of the second monomer through a pentapeptide (-GGSSG-) linker. The enzyme activity of such a tethered dimer (TD) construct is comparable with that of the native homodimer (5). Previously (6, 7), we have reported structures of TD and its complex with the undecapeptide substrate that corresponds in sequence to the reverse transcriptase–integrase junction in the viral polyprotein. Here we report the crystal structure of TD complexed with a decapeptide of amino acid sequence matching the reverse transcriptase–RNase H junction of the polyprotein. The structure has been refined to 1.65 Å resolution against diffraction data collected on the FIP-BM30A beamline (8) at the European Synchrotron Radiation Facility. In this structure, the substrate is cleaved at the scissile bond, but the two product peptides have not separated out completely. The separation between the scissile carbon and nitrogen atoms has increased from 1.38 Å to 2.67 Å. The product peptides are still interacting with catalytic aspartates through hydrogen bonds, and the active-site center represents a stage just past the G'Z intermediate described by Northrop (9). The two in situ product peptides also interact between themselves through a hydrogen bond (2.46 Å) between scissile nitrogen and carboxyl oxygen atoms. In view of this hydrogen bond, one has to consider the possibility that the hydroxyl group attacking the scissile carbon atom protonates the scissile nitrogen during peptide bond cleavage. Another interesting feature is the presence of a low-barrier hydrogen bond (LBHB) between the inner oxygen atoms of catalytic aspartates in the complex. This feature was not incorporated in Northrop's kinetic isomechanism (9).

Results

Substrate in the Droplet.

As mentioned in Methods, a spectroscopic PR assay was used to determine the status of the substrate in the soak droplet. The absorbance at 310 nm only showed minor fluctuations around the initial reading of 0.510, and it did not decrease linearly as would happen if the p-nitrophenyl-containing chromogenic substrate were being cleaved by any enzyme present. Further, the absorption peak also did not shift to lower wavelengths as would happen if there were cleavage reaction taking place inside the cuvette. We therefore believe that no enzyme was released from the crystals into the soaking solution and that the decapaptide substrate is not cleaved outside the crystals in the droplet. This observation is also consistent with absence of any change in the appearance of the crystals on soaking.

Substrate in the Active Site.

Connected positive maxima of the Fo − Fc map (Fig. 1) in the active site region indicated presence of the substrate. Electron density for P2, P1, P1′, and P2′ residues of the peptide AETF*YVDGAA, were clearly visible at a contour level of 2.0σ. The shapes of omit densities at P1 and P1′ were different, and they were characteristic of Phe and Tyr residues, respectively. Nevertheless, because of the pseudo-2-fold symmetry of the free enzyme active site, we modeled the substrate in two orientations, and we treated their occupancies as variables during initial crystallographic refinement. It was found that the occupancy of one orientation increased from the originally assigned value of 0.5 while the other decreased to almost 0. The absence of difference electron densities corresponding to the second orientation further confirmed single orientation of the substrate in the active site. A similar single orientation has been observed in structures of decapeptide substrates complexed with the inactive D25N mutant of HIV-1 PR (10). The molecular motif refined was then changed to a 1:1 complex between the TD and the substrate oligopeptide in a single orientation. Three different types of linkage between P1 and P1′ residues were considered: (i) normal trans-peptide linkage; (ii) hydrated peptide linkage in which the scissile carbon becomes tetrahedrally bonded to a carbonyl oxygen and a hydroxyl oxygen; and (iii) no linkage, that is, a cleaved peptide in which the scissile carbonyl is converted into a carboxyl group. Table 1 gives details of the refinement statistics for the three models. The Rwork values for the three models are very comparable, whereas the Rfree is lowest for the cleaved peptide model. The Rfree for the tetrahedral model was higher by only 0.1% (26.0% vs. 25.9%). The possibility of a tetrahedral model and cleaved model existing simultaneously in different parts of the crystal was considered. The Rfree values obtained for different occupancy combinations, however, were all higher (data not shown).

Fig. 1.

Fig. 1.

Open in a new tab

Difference density in the active-site cavity of HIV-1 PR TD. The refined substrate molecule is also shown along with catalytic aspartates.

Table 1.

Refinement statistics

Statistic Regular peptide Cleaved peptide Tetrahedral peptide
Resolution range 50–1.65Å 50–1.65Å 50–1.65Å
    Rwork, % 21.4 (30.2) 21.4 (30.2) 21.5 (30.2)
    Rfree, % 26.3 (31.4) 25.90 (31.5) 26.0 (31.6)
    No. protein atoms 1,514 1,514 1,514
    No. solvent atoms 156 156 156
    No. ligand atoms 69 70 70
rmsd of bond lengths, Å 0.014 0.015 0.014
rmsd of bond angles, ° 1.56 1.55 1.54
Average B factor, Å2
    Protein atoms 31.2 31.1 31.1
    Substrate atoms 52.4 53.3 50.9
    Water molecules 47.2 47.2 47.2
Open in a new tab

Highest resolution bin, 1.693–1.650Å. The Rfree was calculated using 5% of reflections that were kept apart from the refinement during the whole process.

Fig. 2A presents the 2Fo − Fc and Fo − Fc maps for the P1–P1′ region, when the substrate AETF*YVDGAA is modeled as an uncleaved oligopeptide with standard geometry. It is seen from 2Fo − Fc map that, even when included in structure factor calculations and phasing, there is no density along the trans- peptide bond. Further, the scissile peptide bond is in the negative region of the Fo − Fc map, suggesting that the bond is either broken or does not have the standard trans- conformation. Fig. 2B presents the 2Fo − Fc omit map for the full substrate overlaid with the three models. In the tetrahedral model, the scissile C–N bond length is 2.27 Å, which is probably too long to have any covalence. In the cleaved peptide model, the distance between the scissile N and C atoms is 2.67 Å, which is shorter than a normal van der Waal separation. Such short distances could result if each active site was only half-occupied systematically by either the P product or the Q product. The strengths of electron density suggested that the two halves of the active site are equally occupied. Therefore, the structure was refined by assigning an occupancy of 0.5 to both the P and Q products. The crystallographic Rfree value for this refinement was higher (27.10% vs. 25.90%). Further, there was residual difference electron density along the input product peptides, indicating that the occupancy of the two products was higher than the assigned value of 0.5. This observation established that, in the cleaved peptide model, both fragments were present in the active site with full occupancy. Real-space R factor and correlation coefficients calculated over the substrate, modeled in the three different ways, are listed in Table 2. The correlation coefficient averaged over all of the nine residues is higher for the cleaved peptide model. Similarly, the R factor is the lowest for the cleaved peptide model. The structure therefore is of a complex between TD and the two product peptides. The O2 oxygen of the newly formed carboxyl group superposes to within 1.6 Å the water molecule hydrogen bonding to catalytic aspartates in the native structure (7), and therefore O2 could have been the nucleophile, while the other carboxyl oxygen (O1), would be the peptide oxygen.

Fig. 2.

Fig. 2.

Open in a new tab

Stereo diagrams of electron density maps. (A) 2Fo − Fc map in blue and Fo − Fc map in red when the substrate model refined is of a regular peptide. The peptide bond is in the negative density of the Fo − Fc map. (B) 2Fo − Fc omit map overlaid with the three models: regular peptide model (cyan), tetrahedral hydrated peptide model (brown), and cleaved peptide model (atomic color). Omit density for P1, P3, P1′, and P3′ residues defines the single orientation of the substrate.

Table 2.

Real-space R factor (R) and correlation coefficient (CC) for residues in the three models for the substrate

Residue no. Cleaved peptide model
 Regular peptide model
 Tetrahedral intermediate model
CC R CC R CC R
1 0.573 0.292 0.536 0.288 0.483 0.284
2 0.599 0.258 0.528 0.276 0.585 0.267
3 0.706 0.133 0.646 0.154 0.720 0.128
4 0.591 0.184 0.570 0.194 0.605 0.196
5 0.727 0.119 0.651 0.128 0.715 0.121
6 0.840 0.092 0.840 0.094 0.849 0.093
7 0.699 0.234 0.696 0.238 0.680 0.235
8 0.607 0.207 0.627 0.203 0.595 0.215
9 0.482 0.277 0.529 0.251 0.470 0.275
All res 0.647 0.199 0.625 0.203 0.634 0.202
Open in a new tab

Hydrogen Bonding and Protonation State of Aspartates.

The pattern of hydrogen bonding in the catalytic center would help our understanding of the mechanism of the cleavage reaction. In the absence of hydrogen positions, the presence of a hydrogen bond has been inferred from interatomic distance considerations. As shown in Fig. 3, four hydrogen bonds form a ring involving P1, P1′, Asp-25, and Asp-25′ residues. The carboxyl oxygen (O1) of P1 forms a hydrogen bond (2.51 Å) to OD2 of Asp-25. The other carboxyl oxygen (O2) is hydrogen-bonded with the scissile N atom on the Q product. This N atom forms a second hydrogen bond with the OD2 atom of Asp-25′ (2.90 Å). Finally, the inner oxygens (OD1) of the catalytic aspartates are also engaged in a hydrogen bond with each other. In the structures of substrate complexes with the inactive D25N mutant, the scissile nitrogen does not form any hydrogen bond, whereas the scissile carbonyl oxygen is hydrogen bonded to the Asn-25 nitrogen (10). The hydrogen bonds at the catalytic center also help decide the ionization state of the aspartates. There are four hydrogens (one from the catalytic aspartates in the free enzyme, two from the lytic water molecule, and one from the scissile peptide), whose positions need be determined in the present complex. One hydrogen atom has to be shared by the inner oxygens of the catalytic aspartates. Because the peptide bond is broken, there have to be two hydrogen atoms bonded to the nitrogen. The fourth hydrogen has to form part of the hydrogen bond (2.51 Å) between the P product carboxyl (O1 − the peptide oxygen) and the OD2 of Asp-25. This hydrogen atom could be attached either to the aspartate oxygen or to the carboxyl oxygen. If it is attached to the aspartate oxygen, the aspartates will be diprotonated, and the P product carboxyl group will be anionic corresponding to the F'PQ intermediate in the mechanism proposed by Northrop (9) (Fig. 4). If the proton is on the carboxyl oxygen, the P product will be neutral, and the aspartates will be monoprotonated corresponding to something just past the G'Z intermediate. From the present structure, it is not possible to distinguish between these two scenarios. However, analysis of x-ray structures of HIV-1 PR inhibitor complexes has led to a correlation between protonation state of the aspartates and their hydrogen-bonding pattern (11). Whenever the aspartates are monoprotonated, they form strong hydrogen bonds to donor groups from the substrate/inhibitor or to water molecules. In the diprotonated state, there are no strong hydrogen bonds from aspartates. Therefore, according to this hypothesis, the catalytic aspartates in the present structure should be monoprotonated. The structure would then correspond to a state just past the G'Z intermediate envisaged by Northrop. However, although the scissile peptide oxygen does not interact with aspartates in Northrop's model, that oxygen (O1) is seen in a strong hydrogen bond with OD2 of an aspartate in the present structure.

Fig. 3.

Fig. 3.

Open in a new tab

Hydrogen-bonding interactions at the catalytic center are shown by dotted lines.

Fig. 4.

Fig. 4.

Open in a new tab

Chemical and kinetic isomechanism of an aspartic PR proposed by Northrop. The order of product release is not designated, and release of products is shown as a single step. [Reproduced with permission from ref. 9 (Copyright 2001, American Chemical Society).]

Discussion

Substrate in the State of a Reaction Intermediate.

Attempts have been made earlier to determine structures of HIV-1 and simian immunodeficiency virus (SIV) PRs complexed with substrate oligopeptides (10, 12). In these attempts, the crystals of the complex were prepared by using the method of cocrystallization, in the hope that crystal formation preceded product release. However, in the crystals obtained, HIV-1 PR was found to be complexed with only the N-terminal (P) product peptide, whereas SIV PR was found to be complexed with a C-terminal (Q) product peptide. No structures with both products bound simultaneously were obtained (12). Through careful superpositions of the two product complexes, these authors have concluded that unacceptably close separations between scissile carbon and nitrogen atoms (1.3–2.2 Å) preclude the presence of both products in the active site during cocrystallization. In contrast, we have attempted to prepare the complex by the soaking method. Because, as has been shown here, soaking does not result in cleavage of the substrate outside the crystal, the whole substrate would diffuse into the active site through the solvent channels of the crystal. Our earlier work (6, 13) had shown that the active-site tunnel is quite wide, and it can be accessed through the solvent channels of the crystal by molecules as large as acetylpepstatin and an undecapeptide of sequence very similar to the one used here. Thus, the cleavage reaction has to happen inside the crystals, and both the products generated in situ are expected to be within the enzyme active site. This expectation is also borne out by the crystallographic refinement. Thus, the structure presented here is that of an aspartyl PR with both cleavage products, P and Q, bound in the active site. The shorter than van der Waal separation between scissile carbon and nitrogen atoms is an indication of the reaction having been stalled before completion. Complete separation and release of products P and Q require loss of hydrogen bonding and other interactions between the products and the enzyme (14). The residues from the flap and the catalytic aspartates are involved in a majority of these interactions, and opening of the flexible flaps, therefore, could disrupt these hydrogen bonds. However, flap opening is not possible in the present crystals because of intermolecular contacts, so the system is left in the intermediate stage after the bond-cleavage event (the stage just past the intermediate G'Z in refs. 9 and 14). The detailed interatomic interactions in this state would give, by extrapolation, an insight into the reaction mechanism. In Fig. 5we have compared the present structure with experimental structures assumed to represent the beginning, the precleavage intermediate, and the product-release stages of the cleavage reaction. Fig. 5A shows the comparison with a regular peptide bound to inactive D25N mutant enzyme (10). This mutant complex structure is assumed to represent the binding of the substrate in the active site before the onset of the reaction. It may be seen that, as might be expected, just after bond cleavage (the present structure), the scissile nitrogen and carboxyl group are found closer to and forming hydrogen bonds with the catalytic aspartates. This conformational rearrangement is achieved without making large changes in the positions of the side-chain atoms inside subsites, especially on the C-terminal side. Fig. 5B is a comparison with the structure of a tetrahedral reaction intermediate formed after initial substrate binding but before bond cleavage (6). Interestingly, the P1′ residue has not moved at all, whereas the P product has been displaced away from the Q product, as the reaction has proceeded toward cleavage. This selective mobility of the P product is also consistent with the observation that it is the P product that is released first (9). The scissile nitrogen and the gem-hydroxyl derived from the attacking water molecule have almost identical positions just before and after bond cleavage. Because the hydrogen bond between the scissile nitrogen and the aspartate OD2 is preserved throughout the reaction, this hydrogen bond may be playing a crucial role in proper positioning of the substrate. Fig. 5C shows the superposition with the P product complex prepared by Rose et al. (12) by cocrystallization. There are significant deviations between superposed Cα atom pairs for residues P1, P2, and P3, the maximum shift (1.45 Å) being for the P1 residue. The direction of movement is toward the nonexistent Q product rather than away from it. The distance between scissile carbon atom of Rose structure and the nitrogen atom from the present structure is only 1.03 Å, which is unacceptably short for a nonbonded separation. Such a short contact was also found by Rose et al. when they superposed the structures of individual P and Q product complexes of HIV-1 PR and SIV PR. Rose et al. have ascribed the presence of such short contacts to be the reason why they have failed to prepare, by cocrystallization, a complex containing both product peptides. In all Fig. 5 the aspartate side chains of the present structure have moved closer to each other because of the LBHB between inner oxygen atoms.

Fig. 5.

Fig. 5.

Open in a new tab

Superposition of the present postcleavage intermediate structure (yellow) with structures representing the beginning, precleavage intermediate, and product-release stages of the cleavage reaction. (A) With D25N mutant substrate complex (magenta). (B) With the precleavage tetrahedral intermediate (blue). (C) With the P product just before release (green). Only the P1 residue of the P product peptide is shown along with catalytic aspartates.

LBHBs in HIV-1 PR.

The role of hydrogen bonds in enzymatic reactions is a well known fact, but the proposal that LBHBs drive and contribute to enzymatic catalysis is currently gaining importance, and it has been demonstrated experimentally in a number of enzymatic systems (15–17). In the case of HIV-1 PR, Northrop has invoked formation of an LBHB between the inner oxygens of catalytic aspartates to account for all experimental observations on catalysis by aspartyl PRs. Theoretical calculations of varying degrees of sophistication also point to existence of LBHBs in the active site of HIV-1 PR (18–22). Recent computer simulation studies on oligopeptide hydrolysis by HIV-1 PR have led to the conclusion that electrostatic stabilization, such as hydrogen bonding, is more important for enhancing enzyme reaction rates than dynamic effects (23). In x-ray structures, hydrogen bonds shorter than 2.5 Å are classified as LBHBs, and there are >250 structures of HIV-1 PR inhibitor complexes available in the literature (24). Examination of these structures reveals that the length of the hydrogen bond between the inner carboxylate oxygen atoms of the two catalytic aspartate residues ranges from 2.59 Å to 2.86 Å. The separation of 2.30 Å observed in the present structure (Fig. 6) is the shortest observed so far, and therefore it is an LBHB in the active site of HIV PR. It is perhaps significant that this LBHB is observed when the ligand is a true substrate rather than a substrate analog.

Fig. 6.

Fig. 6.

Open in a new tab

Simulated annealing omit map to show positions of inner oxygens of catalytic aspartates. The difference density in the (Fo − Fc) map is contoured at 3.0σ level. The two oxygens are separated by 2.30 Å, indicating a strong hydrogen bond between them.

Reaction Mechanism.

Because discoveries in the mechanistic details of aspartic PRs provide new approaches to design of inhibitors, there has been considerable interest to know the correct mechanism. Biochemical, structural, and theoretical studies have led to several proposals for the mechanism of HIV-1 PR (9, 12, 25–28). However, there is not yet proof for the sequence of events that actually happen at the atomic level. We examine below these proposals in relation to the present structure.

Protonation of scissile nitrogen.

The class of a general acid–general base mechanism for peptide bond hydrolysis consists of two processes: (i) nucleophilic attack on the scissile carbon atom, and (ii) protonation of the scissile nitrogen atom. It is generally accepted that the nucleophilic attack is by a water molecule hydrogen-bonded to the catalytic aspartates. Polarization of scissile carbonyl through hydrogen bonding between the carbonyl oxygen and OD2 atom of a catalytic aspartate has been suggested in some mechanism proposals (27, 28). Although in the majority of proposals the protonation of scissile nitrogen is through the outer aspartate oxygen, Stroud and coworkers (12) have suggested that the protonation is through the hydroxyl of the gem-diol. In the present structure of the in situ enzyme–product complex, the scissile N forms a stronger hydrogen bond (2.46 Å) with a scissile carboxyl oxygen (O2) rather than with the outer oxygen of a catalytic aspartate (2.9 Å) (Fig. 3). If we make the assumption that no new interproduct hydrogen bonds are formed during product separation after bond cleavage, the observed interproduct hydrogen bond could be a remnant of proton transfer to scissile nitrogen from the gem-diol hydroxyl group.

Interaspartate hydrogen bond.

Based on crystal structures, many researchers have proposed mechanisms for aspartyl PRs that involve a series of proton transfers, but they do not require formation of a hydrogen bond between the inner oxygens of the catalytic aspartates. More recently, Northrop has proposed an isomechanism that is claimed to resolve all uncertainties about pH profile, transpeptidation activity, and anomalous isotope effects in aspartyl PRs. The central feature of Northrop's kinetic isomechanism is the presence of a hydrogen bond between OD1 atoms of the aspartates and its changing character during the reaction. This hydrogen bond, which is present in all steps of the reaction only in the mechanism proposed by Northrop, is an LBHB when the enzyme is in state E, ready to bind the substrate, but it ceases to be an LBHB when the enzyme is in state F, postcleavage. In the proposals by Jaskolski et al. (26), Hyland et al. (27) and Silva et al. (28), although there is no inner oxygen hydrogen bond at the start of the reaction, there is one postcleavage. On the other hand, in the proposal by Stroud and coworkers, although there is an inner oxygen hydrogen bond at the start of the reaction, there is no such bond postcleavage. Similarly, theoretical calculations have also led to different predictions regarding the interaspartate hydrogen bond at different stages of the cleavage process (18–22). In the most recent computer simulation of peptidolysis by HIV-1 PR, the inner oxygens of catalytic aspartates do not form a hydrogen bond when the substrate is in the state of a tetrahedral intermediate (23). In the present structure of the complex, the inner oxygens form a hydrogen bond even when product peptides, P and Q, are bound in the active site, an observation supporting the ideas of Jaskolski et al. (26), Hyland et al. (27), Silva et al. (28), and Northrop (9). Because this hydrogen bond is very short (2.30 Å), it is likely to be an LBHB (Fig. 6), a feature not included in the proposal of Northrop (9). However, the two carboxyl groups are not coplanar, as envisioned by Northrop. Interestingly, in the structure of the precleavage reaction intermediate, which represents an earlier stage of the reaction, this hydrogen bond is slightly longer (2.6 Å) (6).

Conclusions

The structure reported here is that of an in situ complex between HIV-1 PR and product peptides. The 1.65 Å x-ray structure shows that the decapeptide substrate is cleaved into products, which are then not completely separated. The separation between scissile carbon and nitrogen atoms is 2.67 Å. There is a hydrogen bond (2.46 Å) between a carboxyl oxygen from the P product peptide and the scissile nitrogen of the Q product peptide. Because of this interaction in the in situ complex, we propose that the protonation of scissile nitrogen, needed for peptide bond cleavage, is being done by the gem-hydroxyl. The inner oxygens of the catalytic aspartates are 2.30 Å apart, indicating that they are held together in an LBHB at this stage of the reaction. It is significant that such an LBHB has been observed when the ligand is a substrate peptide rather than a substrate analog. These interactions at the active site provide important insight into the reaction mechanism. The structure of the complex will also provide a template for design of mechanism-based inhibitors of HIV-1 PR (9).

Methods

Crystallization and Soaking.

Production and crystallization of TD has been reported earlier (6, 7, 13, 29). The 10-residue peptide of amino acid sequence AETF*YVDGAA, which acts as a substrate for HIV-1 PR, was synthesized at the National Institute for Research in Reproductive Health, Parel, Mumbai, by using an automatic peptide synthesizer. The peptide was dissolved in reservoir buffer containing 10% acetic acid to prepare a 5 mM stock solution. This stock solution was diluted 10-fold into the reservoir solution to prepare the soaking drop, into which a PR crystal was transferred by using a cryoloop. The coverslip was inverted and sealed over the same reservoir well in which crystals had been grown. To ascertain whether during soaking any enzyme was released from the crystal into the soaking solution, a spectroscopic assay was carried out. Four crystals, rather than one, of size eventually used for data collection were soaked into a 10-μl drop made of the soaking solution mentioned above but without the substrate. At the end of 72 h, the crystals were still intact, and only the solution was extracted from the drop and added to a reaction mixture containing a chromogenic substrate at a concentration of 95 μM. Absorption spectra were recorded daily at regular intervals for >4 days by using Jenway 6500 UV spectrophotometer (Barloworld Scientific Ltd., Essex, U.K.).

X-Ray Data Collection and Refinement.

At the end of 72 h of soaking at room temperature, the crystal was equilibrated for 5 min in the cryoprotectant (25% glycerol and 75% reservoir buffer) before flash freezing under the liquid nitrogen stream, for exposure to x-rays on the FIP-BM30A beamline. Images (150) were collected, each of 1.0o oscillation and 2 min of exposure. The crystals diffracted to 1.65 Å resolution. The diffraction data were indexed, integrated, and scaled by using the computer program XDS (x-ray detector software; Max-Planck-Institute for Medical Research, Heidelberg, Germany) (30). Crystal and intensity data statistics are given in Table 3. The structure was refined in Crystallography and NMR System (CNS; Yale University, New Haven, CT) by using standard simulated annealing protocols and the amplitude-based maximum likelihood target function (31). Five percent of the reflections were set aside for cross-validation. All reflections in the resolution range 50–1.65 Å were included in the refinement. Because the crystals of the present complex are almost isomorphous to those of the unliganded protein structure (Protein Data Bank ID code 1LV1), protein coordinates extracted from the Protein Data Bank were used as the starting model for refinement. When the substrate was being treated as a tetrahedral reaction intermediate, the scissile Phe residue was modified by conversion of the carbonyl carbon atom into a gem-diol carbon atom. CNS parameter and topology files for this modified Phe residue, designated as PHT, were generated manually. The force constants in the parameter file for bonds and angles involving the gem-diol carbon atom were set at comparatively low values so that the actual geometry was not tightly restrained but was dictated by the diffraction terms. Water molecules were added manually by examining the environment around the electron densities that were present in both Fo − Fc and 2Fo − Fc maps. During the final stages, the models were subjected to 10 cycles of TLS refinement followed by 10 cycles of maximum likelihood restrained refinement, by using the macromolecular refinement program REFMAC5 (University of York and CLRC, Daresbury Laboratory, U.K.) (32). In the TLS refinement, the two protein subunits together and the substrate were defined as two independent domains, and the B factors of all atoms belonging to these domains were set at a starting value of 40. Composite omit and other electron-density maps were calculated by using CNS. The entire model building and structural superposition were carried out by using the software O (Uppsala University, Uppsala, Sweden) (33). Real-space correlation coefficients and real-space R factors were calculated by using O. For structural comparisons with other substrate–inhibitor complexes, only the protein Cα atoms were used in the superposition.

Table 3.

Crystal and intensity data statistics

Crystal HIV-1 PR TD product complex
Unit cell, Å a = b = 62.03, c = 81.78
Wavelength, Å 0.97945
Resolution range, Å 50–1.65 (1.75–1.65)
Space group P61
No. reflections measured 112,519
    No. unique reflections 20,957
    Completeness 97.5 (92.8)
    Rmerge, % 8.1 (60.2)
    I/σ, I 10.16 (2.9)
Open in a new tab

The numbers in parentheses indicate the value in the outer resolution shell.

Acknowledgments

We thank Drs. R. Chidambaram, K. K. Kannan, Bindu Pillai, and Mukesh Kumar for scientific discussions and S. R. Jadhav for technical help. We thank the National Facility for Macromolecular Crystallography, Solid State Physics Division, Bhabha Atomic Research Centre (BARC), for the x-ray diffraction and biochemistry equipment. This work was carried out under a memorandum of understanding signed between BARC and the Commissariat à l'Energie Atomique (CEA) for cooperation in the field of life sciences.

Abbreviations

HIV-1 PR

HIV-1 protease

LBHB

low-barrier hydrogen bond

SIV

simian immunodeficiency virus

TD

tethered dimer.

Footnotes

The authors declare no conflict of interest.

This article is a PNAS direct submission.

Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID code 2NPH).

References

  • 1.Heagarty MC. AIDS: The Battle Rages On. Issues in Science and Technology Online. (Summer) 2003 Available at www.nap.edu/issues/19.4/heagarty.html#top.
  • 2.Kohl NE, Emini EA, Schleif WA, Davis LJ, Heimbach JC, Dixon RA, Scolnick EM, Sigal IS. Proc Natl Acad Sci USA. 1988;85:4686–4690. doi: 10.1073/pnas.85.13.4686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Wolfenden R. Bioorg Med Chem. 1999;7:647–652. doi: 10.1016/s0968-0896(98)00247-8. [DOI] [PubMed] [Google Scholar]
  • 4.Wlodawer A, Vondrasek J. Annu Rev Biophys Biomol Struct. 1998;27:249–284. doi: 10.1146/annurev.biophys.27.1.249. [DOI] [PubMed] [Google Scholar]
  • 5.Cheng YSE, Yin FH, Foundling S, Blomstrom D, Kettner CA. Proc Natl Acad Sci USA. 1990;87:9660–9664. doi: 10.1073/pnas.87.24.9660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Kumar M, Prashar V, Mahale S, Hosur MV. Biochem J. 2005;389:365–371. doi: 10.1042/BJ20041804. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Kumar M, Kannan KK, Hosur MV, Bhavesh NS, Chatterjee A, Mittal R, Hosur RV. Biochem Biophys Res Commun. 2002;294:395–401. doi: 10.1016/S0006-291X(02)00482-5. [DOI] [PubMed] [Google Scholar]
  • 8.Roth M, Carpentier P, Kaikati O, Joly J, Charrault P, Pirocchi M, Kahn R, Fanchon E, Jacquamet L, Borel F, et al. Acta Crystallogr D. 2002;58:805–814. doi: 10.1107/s0907444902003943. [DOI] [PubMed] [Google Scholar]
  • 9.Northrop DB. Acc Chem Res. 2001;34:790–797. doi: 10.1021/ar000184m. [DOI] [PubMed] [Google Scholar]
  • 10.Jeyabalan MP, Nalivaika E, Schiffer CA. Structure (London) 2002;10:369–381. doi: 10.1016/s0969-2126(02)00720-7. [DOI] [PubMed] [Google Scholar]
  • 11.Piana S, Sebastiani D, Carloni P, Parrinello M. J Am Chem Soc. 2001;123:8730–8737. doi: 10.1021/ja003145e. [DOI] [PubMed] [Google Scholar]
  • 12.Rose RB, Craik CS, Douglas NL, Stroud RM. Biochemistry. 1996;35:12933–12944. doi: 10.1021/bi9612733. [DOI] [PubMed] [Google Scholar]
  • 13.Pillai B, Bhat SV, Kannan KK, Hosur MV. Acta Crystallogr D. 2004;60:594–596. doi: 10.1107/S0907444903029676. [DOI] [PubMed] [Google Scholar]
  • 14.Dunn BM. Chem Rev. 2002;102:4431–4458. doi: 10.1021/cr010167q. [DOI] [PubMed] [Google Scholar]
  • 15.Cleland WW, Frey PA, Gerlt JA. J Biol Chem. 1998;273:25529–25532. doi: 10.1074/jbc.273.40.25529. [DOI] [PubMed] [Google Scholar]
  • 16.Frey PA, Whitt SA, Tobin JB. Science. 1994;264:1927–1930. doi: 10.1126/science.7661899. [DOI] [PubMed] [Google Scholar]
  • 17.Guthrie JP. Chem Biol. 1996;3:163–170. doi: 10.1016/s1074-5521(96)90258-6. [DOI] [PubMed] [Google Scholar]
  • 18.Trylska J, Grochowski P, McCammon JA. Protein Sci. 2004;13:513–528. doi: 10.1110/ps.03372304. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Piana S, Carloni P. Proteins Struct Funct Genet. 2000;39:26–36. doi: 10.1002/(sici)1097-0134(20000401)39:1<26::aid-prot3>3.0.co;2-n. [DOI] [PubMed] [Google Scholar]
  • 20.Lee H, Darden TA, Pedersen LG. J Am Chem Soc. 1996;118:3946–3950. [Google Scholar]
  • 21.Liu H, Müller-Plathe F, van Gunsteren WF. J Mol Biol. 1996;261:454–469. doi: 10.1006/jmbi.1996.0476. [DOI] [PubMed] [Google Scholar]
  • 22.Park H, Suh J, Lee S. J Am Chem Soc. 2000;122:3901–3908. [Google Scholar]
  • 23.Bjelic S, Aqvist Y. Biochemistry. 2006;43:7709–7723. doi: 10.1021/bi060131y. [DOI] [PubMed] [Google Scholar]
  • 24.Vondrasek J, Wlodawer A. Proteins Struct Funct Genet. 2002;49:429–431. doi: 10.1002/prot.10246. [DOI] [PubMed] [Google Scholar]
  • 25.Rodriguez EJ, Angeles TS, Meek TD. Biochemistry. 1993;32:12380–12385. doi: 10.1021/bi00097a015. [DOI] [PubMed] [Google Scholar]
  • 26.Jaskolski M, Tomasselli AG, Sawyer TK, Staples DG, Hendrickson RL, Schneider J, Kent SBH, Wlodawer A. Biochemistry. 1991;30:1600–1609. doi: 10.1021/bi00220a023. [DOI] [PubMed] [Google Scholar]
  • 27.Hyland LJ, Tomaszek TA, Jr, Meek TD. Biochemistry. 1991;30:8454–8463. doi: 10.1021/bi00098a024. [DOI] [PubMed] [Google Scholar]
  • 28.Silva AM, Cachau RE, Sham HL, Erickson JW. J Mol Biol. 1996;255:321–346. doi: 10.1006/jmbi.1996.0026. [DOI] [PubMed] [Google Scholar]
  • 29.Pillai B, Kannan KK, Hosur MV. Proteins. 2001;43:57–64. doi: 10.1002/1097-0134(20010401)43:1<57::aid-prot1017>3.0.co;2-d. [DOI] [PubMed] [Google Scholar]
  • 30.Kabsch W. J Appl Crystallogr. 1993;26:795–800. [Google Scholar]
  • 31.Brünger AT, Adams PD, Clore GM, DeLano WL, Gros P, Grosse-Kunstleve RW, Jiang JS, Kuszewski J, Nilges M, Pannu NS, et al. Acta Crystallogr D. 1998;54:905–921. doi: 10.1107/s0907444998003254. [DOI] [PubMed] [Google Scholar]
  • 32.Murshudov GN, Vagin AA, Dodson EJ. Acta Crystallogr D. 1997;53:240–255. doi: 10.1107/S0907444996012255. [DOI] [PubMed] [Google Scholar]
  • 33.Jones TA, Zou JY, Cowan SW, Kjeldgaard M. Acta Crystallogr A. 1991;47:110–119. doi: 10.1107/s0108767390010224. [DOI] [PubMed] [Google Scholar]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES