Abstract
The β-diketo acids (DKAs) represent a major advance for anti–HIV-1 integrase drug development. We compared the inhibition of HIV-1 integrase by six DKA derivatives using the wild-type enzyme or the double-mutant F185K/C280S, which has been previously used for crystal structure determinations. With the wild-type enzyme, we found that DKAs could be classified into two groups: those similarly potent in the presence of magnesium and manganese and those potent in manganese and relatively ineffective in the presence of magnesium. Both the aromatic and the carboxylic or tetrazole functions of DKAs determined their metal selectivity. The F185K/C280S enzyme was markedly more active in the presence of manganese than magnesium. The F185K/C280S integrase was also relatively resistant to the same group of DKAs that were potent in the presence of magnesium with the wild-type enzyme. Resistance was caused by a synergistic effect from both the F185K and C280S mutations. Molecular modeling and docking suggested metal-dependent differences for binding of DKAs. Molecular modeling also indicated that the tetrazole or the azido groups of some derivatives could directly chelate magnesium or manganese in the integrase catalytic site. Together, these experiments suggest that DKAs recognize conformational differences between wild-type and the double-mutant HIV-1 integrase, because they chelate the magnesium or manganese in the enzyme active site and compete for DNA binding.
Because actual treatments targeting HIV reverse transcriptase and protease do not suppress viral replication in all AIDS patients (Richman, 2001) and because HIV-1 integrase, a viral enzyme encoded by the pol gene, is critical for the HIV-1 viral cycle, HIV-1 integrase is a rational target for anti-AIDS therapy [for a recent review, see De Clercq (2001)]. After viral entry and reverse transcription, HIV-1 integrase catalyzes the integration of the viral DNA copies into host cell chromosomes [for reviews, see Pommier et al. (1997); Brown (1998); Skalka (1999)]. In the cytoplasm, integrase catalyzes the 3′-processing (3′-P) step consisting of the removal of a GT dinucleotide from the 3′-end of both extremities of the viral genome [U3 and U5 long terminal repeats (LTRs)] (see Fig. 1A). Then, integrase bound to the LTRs in the preintegration complex (Bowerman et al., 1989; Farnet and Haseltine, 1991; Miller et al., 1997) migrates to the nucleus, where the strand transfer (ST) step of the integration reaction occurs. In this reaction, the 3′-recessed viral ends (donor DNA) attack host chromosomes (acceptor DNA). Completion of the integration process requires excision of the two 5′-unpaired viral nucleotides, and gap filling, which is probably accomplished by cellular enzymes.
Integration can be monitored in vitro using recombinant HIV-1 integrase and a radiolabeled DNA substrate. In this assay, a 21-mer double-stranded DNA oligonucleotide corresponding to the last 21 base pairs of the U5 viral LTRs (Fig. 1A) is used as substrate to follow both the 3′-P and the ST reactions. 3′-P generates a 19-mer product, which can be integrated in another 21-mer double-stranded DNA oligonucleotide substrate during ST (Fig. 1A). Integration can occur at different positions on the acceptor DNA molecule, leading to several products migrating slower and faster than the original 21-mer substrate. Higher molecular weight species (Fig. 1B) can be used to quantify ST products [for review, see Marchand et al. (2001)].
β-Diketo acids (DKAs) are the first antiviral HIV-1 integrase inhibitors reported to be highly specific for integrase (Goldgur et al., 1999; Hazuda et al., 2000). Two DKAs are currently in clinical trial (Yoshinaga et al., 2002; Young et al., 2002; Billich, 2003) [for review, see Billich (2003); Johnson et al. (2003)]. In vitro, DKAs inhibit integration by selectively blocking ST (Hazuda et al., 2000; Marchand et al., 2002). DKAs have been proposed to bind to a divalent metal ion in the integrase catalytic site (Grobler et al., 2002; Pais et al., 2002) and this binding is metal dependent (Grobler et al., 2002).
In vitro, the wild-type HIV-1 integrase requires manganese (Mn2+) or magnesium (Mg2+) for activity (Engelman and Craigie, 1995; Deprez et al., 2000; Leh et al., 2000). In 12 of the studies published in 2003 using recombinant retroviral integrases, four were performed with Mg2+ as a divalent metal cofactor (Cherepanov et al., 2003; Singh et al., 2003a,b; Zhuang et al., 2003), whereas five studies used Mn2+ (Bao et al., 2003; Priet et al., 2003; Zagarian et al., 2003; Zhang et al., 2003; Zhao et al., 2003). Three other studies were performed with Mg2+ as well as Mn2+ (Bischerour et al., 2003; King et al., 2003; Podtelezhnikov et al., 2003). Thus, there is no consensus on the metal to be used for in vitro integrase catalytic assays. In addition, the relevant metal used by HIV-1 integrase in vivo still remains to be identified. Therefore, the initial aim of the present study was to investigate the metal and DKA structure requirements for HIV-1 integrase inhibition by a series of structurally different DKAs.
A double-mutant of HIV-1 integrase bearing two mutations, F185K and C280S, was designed to improve the enzyme solubility for crystallization purposes (Jenkins et al., 1996). This enzyme has been reported to be as catalytically active as the wild-type enzyme in vitro (Jenkins et al., 1996). The mutation F185K is localized in the enzyme core domain at the dimer interface (i.e., located on the surface of one monomer that contacts another monomer) and is present in all crystal structures (Dyda et al., 1994; Goldgur et al., 1998, 1999; Maignan et al., 1998; Chen et al., 2000a; Wang et al., 2001). The mutation F185K is deleterious to virion assembly in vivo (Jenkins et al., 1996) and decreases the in vitro activity of the enzyme in the presence of Mg2+ (Podtelezhnikov et al., 2003). The mutation C280S is located in the enzyme C-terminal domain and is present in the crystal structure involving both core and C-terminal domains (Chen et al., 2000a). This mutation abolishes the disulfide-related oligomerization of integrase but does not affect viral replication as well as in vitro catalytic activity (Bischerour et al., 2003).
We studied the potency of a series of DKAs for inhibition of the wild-type enzyme in the presence of Mn2+ and Mg2+ and compared the results with those obtained using the double-mutant F185K/C280S enzyme in the presence of Mn2+ (Marchand et al., 2002). We report that the selected DKAs exhibit different metal requirements for the inhibition of the wild-type enzyme and that these differences are determined both by the aromatic and the acidic portions of the DKAs. Molecular modeling was performed to interpret these observations. We also found that the double-mutant integrase is relatively resistant to some of the DKAs, suggesting that caution should be used in the interpretation of structural studies involving mutations F185K and C280S, as well as inhibition data obtained with this integrase mutant.
Materials and Methods
DNA Oligonucleotides and Drugs. Oligonucleotides were purchased from IDT Inc. (Coralville, IA) and purified on 20% (19:1) denaturing polyacrylamide gels using UV shadow. Purified oligonucleotides were 5′-end labeled by T4-polynucleotide kinase (New England Biolabs, Inc. Beverly, MA) as described previously (Marchand et al., 2001). The synthesis of DKA derivatives has been described in detail separately (Pais et al., 2002; Zhang et al., 2002).
HIV-1 Integrase Expression and Purification. The double-mutant F185K/C280S integrase was expressed and purified as described previously (Marchand et al., 2001). The wild-type enzyme was expressed and purified accordingly with the following modifications: 10% glycerol was added to all the purification buffers and 50 μM of ZnCl2 was added to the elution buffer (25 mM HEPES, pH 6.7, 0.5 M NaCl, 2 mM β-mercaptoethanol, and 750 mM of imidazole) (Leh et al., 2000). Dialysis was performed in a buffer containing 50 mM HEPES, pH 6.7, 1 M NaCl, 4 mM EDTA, 2 mM dithiothreitol, and 50% glycerol.
HIV-1 Integrase Site Mutagenesis. F185K and C280S single-mutants were generated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA), according to the manufacturer's recommendations. Two sets of oligonucleotides, F185K+ (34-mer, 5′-GGCAGTATTCATCCACAATAAAAAAAGAAAAGGG-3′), F185K– (34-mer, 5′-CCCTTTTCTTTTTTTATTGTGGATGAATACTGCC-3′) and C280S+ (27-mer, 5′-GGCAGGTGATGATAGTGTGGCAAGTAG-3′), C280S– (27-mer, 5′-CTACTTGCCACACTATCATCACCTGCC-3′) embedding the 185 and 280 mutations, respectively, were purchased from IDT (Coralville, IL). Polymerase chain reaction was performed with each set of primers and the pET-15b-IN1–288 plasmid (Engelman and Craigie, 1992) as a starting template. Mutants were selected by sequencing and were used for protein expression as described above.
HIV-1 Integrase Inhibition Assay. The HIV-1 integrase assays were performed as described previously (Marchand et al., 2002). Briefly, the integrase-DNA complexes were preformed (Hazuda et al., 1994) by mixing 400 nM of HIV-1 integrase with 20 nM concentrations of the 5′-end 32P-labeled, 21-mer, double-stranded DNA template for 15 min on ice in a reaction buffer containing 25 mM MOPS, pH 7.2, 50 mM NaCl, 7.5 mM MnCl2 or MgCl2, and 14.3 mM β-mercaptoethanol. Inhibitors were then added to reactions in a final volume of 10 μl, and integration reactions were carried out for 1 h at 37°C. Reactions were quenched by adding 10 μl of denaturing loading dye (99% formamide, 1% SDS, 0.2 mg/ml bromphenol blue, and 0.2 mg/ml xylene cyanol). Samples were loaded onto 20% (19:1) denaturing polyacrylamide gels. Gels were dried, exposed overnight and analyzed using a PhosphorImager (Amersham Biosciences, Piscataway, NJ). The densitometric analysis was performed using ImageQuant (Amersham Biosciences). Each lane was quantified for 3′-P and ST products, which were expressed as a fraction of the total DNA per lane. Percentage of inhibition was computed using the integrase control lanes as a reference.
Molecular Modeling and Docking Studies. Docking studies were performed on the structural model of the full-length HIV-1 integrase complexed with donor DNA using the program GLIDE, part of the First Discovery suite (Schrödinger Inc., Portland, OR). The coordinates of the full-length HIV-1 integrase tetramer complexed with donor and acceptor DNA were the results of modeling, using all available experimental evidence (Karki et al., 2002). Known crystal structures (Lodi et al., 1995; Cai et al., 1997; Maignan et al., 1998; Goldgur et al., 1999; Chen et al., 2000a; Wang et al., 2001) were used as templates for structure building. All missing residues were added using the Biopolymer module in Sybyl 6.8 (Tripos Inc., St. Louis, MO), and the conformation was adjusted by the loop search method. The whole complex was refined by molecular dynamics simulation using the program CHARMM (Brooks et al., 1983), version c27b3, with the CHARMM all-atom force field (MacKerell et al., 1998; Foloppe and MacKerell, 2000) in vacuo followed by molecular dynamics simulation in a 10-Å shell of explicit solvent. The transferable intermolecular potential-3-point water model was used to simulate the solvent (Jorgensen et al., 1983). Each monomer consists of residues 1 to 288. This structural model shows differences from models reported earlier (Heuer and Brown, 1998; Chen et al., 2000a,2000b; Gao et al., 2001; Podtelezhnikov et al., 2003), and a detailed account will be published elsewhere. For the purpose of this study, a dimer of the full-length integrase complexed with donor DNA was extracted from the modeled tetramer complex. This structural model was refined further by running a series of restrained impact energy minimizations using the Impact utility, part of the first discovery suite (Schrödinger Inc., Portland, OR), until the average root-mean-square deviation of the nonhydrogen atoms reached 0.3 Å. Grid files representing shape and properties of the receptor were generated separately, with either Mg2+ or Mn2+ in the active site, respectively. The three-dimensional structures of the inhibitors used for docking were constructed using the 3D-Sketcher tool available through the Maestro interface of the program GLIDE. The geometry of the structures was optimized using the OPLS-AA force field (Jorgensen et al., 1996) to perform 1000 steps of conjugate gradient minimization. The optimized structures were then docked using the precomputed grid files; for each of them, 100 poses were saved.
Results
Because DKAs have been proposed to bind a divalent metal ion in the catalytic site of the enzyme (Grobler et al., 2002; Pais et al., 2002) in a metal-dependent manner (Grobler et al., 2002), we studied the potency of DKAs for the in vitro inhibition of wild-type integrase in the presence of Mn2+ versus Mg2+ and compared the effects of DKAs against the wild-type and the soluble double-mutant enzymes (Marchand et al., 2002). Even though the double-mutant has been reported to be active in the presence of Mg2+ (Jenkins et al., 1996), we found in our laboratory that the activity level of the double–mutant in the presence Mg2+ was not suitable for inhibitor testing. In the presence of Mn2+, the catalytic activity of the double-mutant enzyme was equivalent to that of wild-type integrase (Marchand et al., 2002). The drug inhibition data are shown as dose response curves in Figs. 2 and 3, and IC50 values are summarized in Table 1.
The Carboxylic Function Enhances Inhibition of Wild-Type Integrase in the Presence of Mg2+. To study the metal selectivity for inhibition of wild-type HIV-1 integrase by DKAs, we compared the activity of 5CITEP and DKA1, bearing a carboxylic function compared with the tetrazole group (Table 1) (Goldgur et al., 1999; Pais et al., 2002). We previously reported these two inhibitors to be equivalent for inhibition of the double-mutant integrase in the presence of Mn2+ (Marchand et al., 2002), but they were recently described to exhibit differential binding efficiencies depending on the metal present in the reaction (Grobler et al., 2002). Using wild-type HIV-1 integrase, 5CITEP inhibited ST in the presence of Mn2+ (Figs. 1B, top gel, and 2A) with an IC50 value of 0.93 μM (Table 1) but was approximately 100-fold less active in the presence of Mg2+ (Figs. 1B, bottom gel, and 2A; Table 1). DKA1 was more potent than 5CITEP in the presence of Mn2+ (Figs. 1B, top gel, and 2B; Table 1) and was even more potent than 5CITEP in the presence of Mg2+ (Figs. 1B, bottom gel, and 2B; Table 1). Thus our data confirm that the tetrazole and carboxylic functions are not equivalent for inhibition of the wild-type integrase (Grobler et al., 2002). The lack of potency of 5CITEP in the presence of Mg2+ may be explained by the reduced binding of the tetrazole group in the catalytic site of the enzyme in the presence of Mg2+ (Grobler et al., 2002). Together, these results demonstrate that potent inhibition of wild-type integrase in the presence of Mg2+ requires a carboxylic function rather than a tetrazole in the DKAs.
The Carboxylic Group Increases Selectivity for the Wild-Type Enzyme.Figure 2 demonstrates that in the presence of Mn2+, the potency of 5CITEP for ST was almost equivalent against the wild-type and the double-mutant enzymes (compare Fig. 2, A and D). By contrast, DKA1, which is also a Mn2+-selective drug (Fig. 2B), was more potent against the wild-type enzyme than the double mutant when tested in the presence of Mn2+ (compare Fig. 2, B and E). These results demonstrate that the carboxylic group is not only required for potency in the presence of Mg2+ but is also globally more efficient than the tetrazole function for inhibition of the wild-type enzyme.
The Aromatic Portion of DKAs Contributes to the Metal Selective Inhibition of HIV-1 Integrase. To evaluate the influence of the aromatic portion of DKAs, we also tested the Merck compound L-708,906 (Table 1), which differs from DKA1 by the replacement of the chloro-indole group with a bis-benzyloxy phenyl group (see Table 1). With the wild-type enzyme, L-708,906 showed the same IC50 value of 0.06 μM for ST inhibition with either Mg2+ or Mn2+ (Fig. 2C; Table 1). Thus, exchanging the chloro-indole group of DKA1 for the bis-benzyloxy phenyl group of L-708,906 selectively increased potency against ST in the presence of Mg2+ (30-fold; 1.93 versus 0.06 μM) compared with Mn2+ (2-fold; 0.12 versus 0.06 μM) (Table 1). Moreover, the potency of L-708,906 in the 3′-P reaction was 6- to 14-fold greater in the presence of Mg2+ than in the presence of Mn2+ (Fig. 2C, Table 1). The nitrogen atom present in the indole ring of DKA1 does not play any significant role in the metal selectivity because a naphthalene-containing DKA lacking this nitrogen was also Mn2+-selective (data not shown). DKA1 is more potent in the presence of Mn2+, whereas L-708,906 is equally active for ST with either Mg2+ or Mn2+ and more potent for 3′-P in the presence of Mg2+ than Mn2+, which demonstrates that the structure of the aromatic moiety of the DKAs contributes to their metal selectivity.
The Aromatic Portion of DKAs Can Be Modified without Abrogating Mg2+-Dependent Inhibition of HIV-1 Integrase. To further investigate the influence of the aromatic portion of DKAs for their metal selectivity, we extended our studies to the novel azido-containing DKAs (Table 1). We previously reported that these compounds are selective inhibitors of ST, with potency similar to that of L-708,906 in assays using the double-mutant integrase (Zhang et al., 2002). Figure 3 shows that the di-azido DKA (DA-DKA) and the mono-azido DKA (MA-DKA) exhibit potency similar to that of L-708,906 when tested with the wild-type enzyme in the presence of Mn2+ (Fig. 2C, Table 1). They also exhibit submicromolar IC50 values for ST inhibition in the presence of Mg2+. DA-DKA and MA-DKA were also 20- to 30-fold less potent when tested against the double-mutant in the presence of Mn2+ (Fig. 3, D and E). These results demonstrate that the aromatic portion of L-708,906 can be modified while retaining the activity of these compounds with either Mg2+ or Mn2+.
A Second Diketo Acid Side Chain on the Central Aromatic Ring Confers Mn2+Selectivity. We reported previously that the bifunctional DKA (DKA2) (Table 1), which contains a second diketo acid side chain appended to the central aromatic ring, is more efficient for 3′-P inhibition than DKAs with a single diketo function, when tested against the double-mutant integrase in the presence of Mn2+ (Marchand et al., 2002) (Fig. 3F). When tested against the wild-type enzyme, DKA2 showed weaker inhibition in the presence of Mg2+ than Mn2+ (Fig. 3C), as was observed for 5CITEP and DKA1 (Fig. 2). The IC50 values of DKA2 for ST/3′-P in the presence of Mn2+ and Mg2+ were approximately 0.6/5 and 10/110 μM, respectively (Fig. 3C, Table 1). These experiments indicate that the introduction of a second diketo acid side chain in the DKA2 molecule, rendering it symmetrical, reduces potency, especially in the presence of Mg2+ compared with the other DKAs: L-708,906, DA-DKA, and MA-DKA. In the presence of Mn2+, DKA2 had almost equivalent potency against wild-type (Fig. 3C) and double-mutant enzymes (Fig. 3F). Thus, among the DKAs examined, DKA2, DKA1, and 5CITEP (see Fig. 2) showed similar activity in the double-mutant and wild-type integrases.
The Double-Mutant Integrase Is Resistant to L-708,906 and the Azido-Containing DKAs.Fig. 2, C and F, shows that L-708,906 is more potent against wild-type integrase than the double-mutant by approximately 40-fold for 3′-P and 7-fold for ST (Table 1). Similarly, we found that the azido-containing DKAs (DA-DKA and MA-DKA) were markedly more potent against the wild-type integrase than the double mutant. The double mutant is thus relatively resistant to L-708,906 and to the azido-containing DKAs. These experiments demonstrate that the DKAs studied fall into two groups. The first group (L-708,906, DA-DKA, and MA-DKA) shows resistance in the double-mutant enzyme; and the second group (5CITEP, DKA1, and DKA2) has comparable activities against both enzymes.
The ability of wild-type enzyme to discriminate for ST between these two groups of compounds, given that all of them had similar activity against the double-mutant enzyme, suggests that structural differences between the wild-type and the double-mutant integrases are recognized by the DKAs.
Mutations F185K and C280S Have a Synergistic Effect on the Resistance of the Double-Mutant Integrase to L-708,906. To determine the effect of each mutation on the resistance to DKAs of the double-mutant integrase, we expressed single-mutant proteins bearing either the F185K or the C280S mutation and tested L-708,906 against each of these mutant enzymes in the presence of Mn2+. These two single mutant proteins had similar catalytic activities compared with the wild-type enzyme (data not shown). The ST inhibition curves are presented in Fig. 4. The inhibition curve of the F185K mutant is almost superimposable upon the one of the wild-type enzyme (Fig. 4), suggesting that this mutation alone cannot be held responsible for the observed resistance of the double-mutant to L-708,906. The C280S mutant was more resistant to L-708,906 than the wild-type enzyme, but this resistance was almost 10-fold lower than that of the double-mutant. These experiments demonstrate that the resistance observed with the double-mutant integrase is caused by a synergistic effect of both the F185K and the C280S mutations.
Molecular Modeling and Docking Studies. Molecular modeling was performed by docking the various inhibitors into a structural model of full-length HIV-1 integrase dimer complexed with donor DNA. DKAs can bind in the catalytic site of HIV-1 integrase in the absence of DNA substrate as shown by the crystal structure of 5CITEP with the enzyme core domain (Goldgur et al., 1999). Resistance mutations to DKAs are located in the enzyme active site in the vicinity of catalytic residues D64, D116, and E152 (Hazuda et al., 2000; Witvrouw et al., 2002) [for reviews, see Neamati (2002); Johnson et al. (2003)]. It has been speculated that HIV-1 integrase could require binding of two divalent metal ions in its active site to achieve integration and that DKAs inhibit integrase via sequestration of these metals (Grobler et al., 2002; Pais et al., 2002).
Figure 5 shows differences in the binding of 5CITEP and DKA1 in the presence of Mg2+ (Fig. 5, A and C). DKA1 chelates one of the Mg2+ ions by its diketo group and interacts via hydrogen bonding of its carboxylic group with Asn120 and Gly118 (Fig. 5C). On the other hand, 5CITEP interacts with the divalent metal via its tetrazole group and binds in a different orientation (Fig. 5A). This orientation is slightly different from the one observed in the crystal structure (Goldgur et al., 1999), which may be related to crystallographic environment (crystal packing). Thus, 5CITEP could retain considerable mobility and adopt different orientations in the binding site of integrase (Sotriffer et al., 2000a,b). In the case of DKA1, our model suggests that the presence of the carboxylic acid group could provide an optimal positioning of the diketo group for chelation of the divalent metal, which globally would increase binding efficiency of the drug. In the presence of Mn2+, the binding of 5CITEP and DKA1 are more similar (Fig. 5, B and D) and the hydrogen bonding with the enzyme is now brought about by the tetrazole hydrogen in the case of 5CITEP and the enolic hydroxy group in case of DKA1.
L-708,906 adopts an orientation similar to that of DKA1 in the presence of Mg2+ or Mn2+ (Fig. 5, E and F). For these two carboxylate-containing compounds, the chelation of the Mg2+ metal ion is performed by the two planar oxygens of the diketo function, whereas the chelation of Mn2+ is shared by one oxygen of the carboxylate and one of the diketo function (compare Fig. 5, C and E with D and F). The sliding of the molecule toward the carboxylate in the presence of Mn2+ could be explained by the existence of a slightly more flexible coordination sphere for Mn2+ than for Mg2+ (Bock et al., 1999). Moreover, L-708,906 fills the catalytic pocket more efficiently with its aromatic portion than DKA1 (compare Fig. 5, C and D with E and F). One aromatic arm of L-708,906 reaches and interacts with the donor DNA, which may increase binding efficiency. Biochemical experiments also suggest an interaction between DKAs and the terminal unpaired dinucleotide present in the donor DNA (Marchand et al., 2002).
The docking of the azido-containing DKAs reveals that DA-DKA and MA-DKA could bind in two optimal orientations. One is similar to DKA1 and L-708,906 (not shown). Surprisingly, in the other orientation (shown in Fig. 6, A and D), the diketo groups do not participate in divalent metal chelation. The chelation is instead performed by one of the azido arms, and the orientation of the azido-containing DKAs is metal selective (compare Fig. 6, A and C with B and D; note the orientation of the diketo group). A search of the Cambridge Structural Database (CSD version 5.23) (Allen and Kennard, 1993) for organic compounds containing nitrogen and magnesium using the program Quest (version 5) resulted in 321 hits for which nitrogens chelate the metal ion (data not shown). This suggests that an azido group can potentially replace the diketo acid side chain of the DKA as the chelating moiety.
The docking of the bifunctional DKA2 indicates an interaction of the diketo group of one side chain with the divalent cation (Fig. 6, E and F) and suggests a conservation of the orientation of the molecule in the presence of either Mg2+ or Mn2+. Our molecular modeling and docking studies demonstrate the importance of the DKAs' carboxylic function for Mg2+ chelation by the diketo group of DKAs. They also suggest that DKAs bind to the active site of integrase differentially in the presence of Mg2+ and Mn2+. Finally, they reveal a potential chelation of either the Mg2+ or Mn2+ by the azido group of the azido-containing DKAs and by the tetrazole of 5CITEP.
Discussion
The potency of DKAs against wild-type HIV-1 integrase is divalent metal-dependent (Grobler et al., 2002; current study). The present study further demonstrates that this metal-dependence depends on the chemical structure of DKAs (Grobler et al., 2002). Novel azido-containing DKAs (MA-DKA and DA-DKA) and the Merck derivative L-708,906 show almost similar inhibition in the presence of either metal, primarily because these compounds exhibit potent inhibition in the presence of Mg2+. On the other hand, some DKAs, such as 5CITEP, DKA1, or DKA2, are markedly less potent in the presence of Mg2+ than Mn2+. We find that both the aromatic and acidic portions of the DKAs contribute to their metal selectivity. In the set of compounds examined, a single carboxylic acid function was required for potency in the presence of Mg2+. Molecular docking demonstrated differences in drug binding in the presence of Mg2+ versus Mn2+; the most striking difference was observed with 5CITEP. We also found that metal chelation can involve not only the diketo function but also the nitrogens contained in the azido DKAs or the tetrazole of 5CITEP.
The double-mutant F185K/C280S and the wild-type enzymes are not equivalent for inhibition by some of the DKAs. The double-mutant integrase was resistant to L-708,906 and the two azido DKAs, whereas 5CITEP, DKA1, and DKA2 were similarly active against the two enzymes. This resistance was caused by a synergistic effect of both mutations. The relative potency of the compounds against the wild-type enzyme versus the double-mutant can be expressed as the ratio of the ST IC50 values of the double-mutant compared with those obtained with the wild-type enzyme (Table 1, Fig. 7). Figure 7 shows that the selectivity of a given compound for the wild-type enzyme (versus the double-mutant) is directly correlated with the relative activity of that compound in the presence of Mg2+ (versus Mn2+). For example, compounds such as L-708,906 or the azido-containing DKAs (DA-DKA and MA-DKA) that are selective for the wild-type enzyme are also potent in the presence of either metal. Conversely, 5CITEP, DKA1, and DKA2 show no cross-resistance against the double-mutant integrase and are relatively inactive against the wild-type integrase in the presence of Mg2+. The differences in drug sensitivity may be caused by structural differences between the integration complexes formed by the wild-type and the double-mutant enzymes. These differences could involve the enzyme active site because we find that the double-mutant integrase requires Mn2+ rather than Mg2+ for 3′-P and ST activities and because it has been speculated that the active site of integrase is considerably less “tight” in the presence of Mn2+ (Esposito and Craigie, 1998). The Mg2+- and wild-type enzyme-selective compounds such as L-708,906 and the azido-containing DKAs may bind preferentially to the enzyme active site in its wild-type configuration in the presence of Mg2+. By contrast, DKA1, DKA2, and 5CITEP may bind poorly to the normal enzyme catalytic site in the presence of Mg2+. The versatility of the Mn2+ coordination chemistry (Bock et al., 1999) might allow chelation of DKA1, DKA2, or 5CITEP. Thus, DKAs reveal themselves as molecular probes for differences in the structures of integrase-DNA complexes formed by wild-type and mutant integrase.
The correlation observed between relative drug potency in the presence of Mg2+ and relative potency in the wild-type enzyme suggests that assays performed using wild-type enzyme and Mg2+ represent the most stringent conditions with respect to drug inhibition. Wild-type integrase in the presence of Mg2+ gives the greatest discrimination among DKAs. The ST IC50 values range from 0.06 μM for L-708,906 to 97 μM for 5CITEP (Table 1). DKAs may not be the only family of drugs that show a differential potency between wild-type and double-mutant integrases. Caution should therefore be used when selecting an HIV-1 integrase enzymatic system for drug screening. It should also be kept in mind when interpreting structural studies that at least one of the two mutated residues (F185K) is present in every crystal structure obtained so far (Dyda et al., 1994; Goldgur et al., 1998, 1999; Maignan et al., 1998; Chen et al., 2000a; Wang et al., 2001).
Finally, relative activity in the presence of Mg2+ and relative activity in the wild-type enzyme seem also to correlate with antiviral activity. DA-DKA (Zhang et al., 2002), MA-DKA (Zhang et al., 2002), and L-708,906 (Hazuda et al., 2000; Pais et al., 2002) are potent antiviral compounds, whereas 5CITEP, DKA1, and DKA2 (Pais et al., 2002) are not antiviral. If this trend is confirmed for larger compound sets, the sensitivity of a compound against wild-type enzyme versus the F185K/C280S double-mutant, and the selectivity toward Mg2+ versus Mn2+, may be an outright indicator for in vivo activity.
Footnotes
-
ABBREVIATIONS: 3′-P, 3′-processing; ST, strand transfer; DKA, β-diketo acids; MOPS, 3-(N-morpholino)propanesulfonic acid; 5CITEP, 1-(5-chloroindol-3-yl)-3-(tetrazolyl)-1,3-propanedione enol; L-708,906, 4-[3,5-bis(benzyloxy)phenyl]-2-hydroxy-4-oxobut-2-enoic acid; DA-DKA, diazido β-diketo acids; MA-DKA, mono-azido β-diketo acids; LTRs, long terminal repeats.
- Received February 24, 2003.
- Accepted May 30, 2003.
- The American Society for Pharmacology and Experimental Therapeutics