Formation of a Quaternary Complex of HIV-1 Reverse Transcriptase with a Nucleotide-competing Inhibitor and Its ATP Enhancer*

Background: Nucleotide-competing RT inhibitors (NcRTIs), like INDOPY-1, show antiviral activity against HIV. Results: HIV-1 RT, the nucleic acid substrate, INDOPY-1, and its enhancer ATP can form a quaternary complex in the post-translocational state. Conclusion: The bound ATP compromises dissociation of the inhibitor from the RT complex. Significance: These findings reveal a novel mechanism of inhibition of HIV-1 RT. Nucleotide-competing reverse transcriptase inhibitors were shown to bind reversibly to the nucleotide-binding site of the reverse transcriptase (RT) enzyme of human immunodeficiency virus type 1 (HIV-1). Here, we show that the presence of ATP can enhance the inhibitory effects of the prototype compound INDOPY-1. We employed a combination of cell-free and cell-based assays to shed light on the underlying molecular mechanism. Binding studies and site-specific footprinting experiments demonstrate the existence of a stable quaternary complex with HIV-1 RT, its nucleic acid substrate, INDOPY-1, and ATP. The complex is frozen in the post-translocational state that usually accommodates the incoming nucleotide substrate. Structure-activity relationship studies show that both the base and the phosphate moieties of ATP are elements that play important roles in enhancing the inhibitory effects of INDOPY-1. In vitro susceptibility measurements with mutant viruses containing amino acid substitutions K70G, V75T, L228R, and K219R in the putative ATP binding pocket revealed unexpectedly a hypersusceptible phenotype with respect to INDOPY-1. The same mutational cluster was previously shown to reduce susceptibility to the pyrophosphate analog phosphonoformic acid. However, in the absence of INDOPY-1, ATP can bind and act as a pyrophosphate donor under conditions that favor formation of the pre-translocated RT complex. We therefore conclude that the mutant enzyme facilitates simultaneous binding of INDOPY-1 and ATP to the post-translocated complex. Based on these data, we propose a model in which the bound ATP traps the inhibitor, which, in turn, compromises its dissociation.

The reverse transcriptase (RT) 4 of human immunodeficiency virus type 1 (HIV-1) is an RNA-and DNA-dependent DNA polymerase that remains a prime target for antiretroviral intervention. The two approved classes of RT inhibitors are nucleoside analog (NRTI) and non-nucleoside analog (NNRTI) RT inhibitors. Phosphorylated NRTIs compete with natural deoxynucleoside triphosphates (dNTPs), bind to the polymerase active site, and cause chain termination of the growing DNA strand. Binding of NNRTIs to a hydrophobic pocket near the polymerase active site results in inhibition of nucleotide incorporation without directly competing with the substrate (1)(2)(3)(4)(5).
Indolopyridone-derived compounds (6, 7) represent a novel, investigational class of RT inhibitors that suppress DNA polymerization through a mechanism that is distinct from both NRTIs and NNRTIs. INDOPY-1 (Fig. 1), the prototype compound of this class, was shown to be active against NNRTIresistant HIV strains (6,7). However, certain NRTI resistanceconferring mutations in the vicinity of the polymerase active site confer resistance to this inhibitor (6,7). Mutation M184V that confers high level resistance to the NRTI lamivudine (3TC) was shown to reduce susceptibility to INDOPY-1. This mutation discriminates against both inhibitors at the level of binding (8 -11). In contrast, Y115F in HIV-1 RT appears to enhance nucleotide binding at the cost of inhibitor binding. As such, both mutations together amplify the level of resistance. These findings represent strong evidence to show that despite distinct structures, INDOPY-1 may share some common properties with NRTIs.
Different classes of RT inhibitors can bind to different conformations of the RT⅐nucleic acid complex. Following incorporation of the nucleotide substrate and release of the pyrophosphate (PPi) product, the RT enzyme can rapidly shuttle between pre-and post-translocational states (12)(13)(14)(15)(16)(17). Footprinting studies revealed that NRTIs and natural nucleoside triphosphates (dNTPs) can only bind to and trap the RT complex in its post-translocated state (13). INDOPY-1 behaves identically in these assays (6). Conversely, the PPi analog foscarnet (phosphonoformic acid, PFA) traps the complex in the pre-translocational state (15). However, unlike NRTIs and dNTPs, INDOPY-1 does not require the templated nucleotide for base pairing. It rather seems that the compound recognizes the ultimate base pair. "Hot spots" for inhibition of DNA synthesis are seen when the primer 3Ј terminus contains a pyrimidine base (6,7). Together, these data suggest that the binding sites for INDOPY-1 and the dNTP substrate partially overlap. Hence, RT inhibitors with the aforementioned properties are collectively referred to as "nucleotide-competing RT inhibitors" (NcRTIs) (8,18).
In this study, we demonstrate that the presence of adenosine triphosphate (ATP) can enhance the inhibitory effects of INDOPY-1. Previous studies have shown that ATP can also act as a PPi donor that promotes excision of incorporated NRTIs (19). This reaction provides an important mechanism for resistance to this class of inhibitors, in particular to zidovudine (AZT). However, in this context, ATP traps the pre-translocated RT complex (20). Here, we utilized a combination of cellbased and cell-free assays to characterize the mode of binding of ATP in conjunction with INDOPY-1. We demonstrate that ATP and INDOPY-1 can form a quaternary complex with RT and its nucleic acid substrate. These findings point to a highly flexible ATP-binding site that can accommodate this ligand in various orientations in both pre-and post-translocational conformations. We conclude that the bound ATP compromises dissociation of INDOPY-1 from a post-translocated complex.

EXPERIMENTAL PROCEDURES
Materials-The heterodimeric (p66/p51) HIV-1 RT enzyme was expressed in Escherichia coli and purified as described previously (21). Site-directed mutagenesis was performed using the Stratagene QuikChange procedure. WT RT refers to the wild-type enzyme (HXB2 strain). TAM1 RT contains mutations M41L, D67N, L210W, and T215Y. TAM2 RT contains mutations D67N, K70R, T215F, and K219Q. "Remodeled TAMs" RT harbors mutations K70G, V75T, L228R, and K219R. Oligonucleotides were chemically synthesized and purchased from IDT with the exception of RNA template PBS-250 that was synthesized through in vitro transcription with T7 RNA polymerase (22). The DNA primer was either synthesized with a fluorescent Cy5 or Alexa488 label at its 5Ј-end or 32 P-radiolabeled with [␥-32 ]ATP and T4 polynucleotide kinase (Fermentas) (13). Alexa555 quencher dUTP was purchased from Invitrogen. The RT inhibitor INDOPY-1 was obtained through the AIDS Research and Reference Reagent Program (Division of AIDS NIAID, National Institutes of Health), and Tibotec BVBA is the reagent contributor. Natural and modified nucleotides were purchased from Fermentas and TriLink, respectively (chemical structures are provided in supplemental Fig. 1).
Single Nucleotide Incorporation Assay-A 3-fold excess of DNA template (C1-A, 5Ј-GTAACTAGAGATCCCTCAGAC-CCTTTTAGTCAGAAT) was hybridized to 50 nM of a fluorescently labeled DNA primer (8aT_Cy5, 5Ј-Cy5-TTCTGACTA-AAAGGGTCTGAGGGAT). DNA⅐RNA hybrids were formed in a buffer containing 50 mM Tris-HCl, pH 7.8, and 50 mM NaCl. Samples were heated to 95°C for 3 min followed by a gradual decrease to room temperature. The hybrid was then incubated with 50 nM HIV-1 RT in a buffer containing 50 mM Tris-HCl, pH 7.8, 50 mM NaCl, as well as varying amounts of ATP and increasing concentrations of INDOPY-1 (0 -18 M). Nucleotide incorporation was initiated with the addition of 6 mM MgCl 2 and 100 nM dCTP at 37°C and was allowed to proceed for 5 min. Samples were resolved on a 12% denaturing polyacrylamide gel and analyzed with a phosphorimager instrument (Amersham Biosciences) using Quantity One and ImageQuant software.
Multisite Incorporation Assay on a Heteropolymeric Template-A 2-fold molar excess of PBS-250 RNA template was hybridized to 50 nM of 5Ј-radiolabeled PBS-28 DNA primer (5Ј-CTTTCAGGTCCCTGTTCGGGCGCCACTG). The hybrid was then incubated with 125 nM RT in a buffer containing 50 mM Tris-HCl, pH 7.8, 50 mM NaCl, and 10 M of each dATP, dTTP, dGTP, and dCTP. DNA synthesis was initiated by the addition of 6 mM MgCl 2 in the absence or presence of inhibitor and NTPs at 37°C. DNA synthesis was stopped at defined time points by the addition of formamide loading dye. Samples were subsequently resolved on a 12% denaturing polyacrylamide gel. Similar conditions were employed for monitoring AZT-MP excision whereby 50 nM DNA primer⅐template were incubated with 250 nM RT and 10 M AZP-TP in the presence of 0.5 M dNTPs and 3.5 mM pyrophosphatase-treated ATP. Reactions were preincubated at 37°C for 5 min and then initiated with 6 mM MgCl 2 . AZT-MP excision and DNA synthesis rescue was  Band Shift Experiments-50 nM DNA⅐DNA hybrid (primer, PPT-24Cy5 5Ј-Cy5-ACTTTTTAAAAGAAAAGGGGGGAT; template, PPT-57 5Ј-CGTTGGGAGTGAATTAGCCCTTCC-AGTCCCCCCTTTTCTTTTAAAAAGTGGCTAAGA) was incubated with 5-fold molar excess of WT RT enzyme in the presence of 50 mM Tris-HCl, pH 7.8, 50 mM NaCl, and increasing concentrations of INDOPY-1 (0 -4 M). The stability of enzyme binding to the nucleic acid substrate was assessed in the absence or presence of ATP by the addition of 1.5 g/l heparin trap. Samples were incubated for 15 min at 37°C and visualized on a nondenaturing 6% polyacrylamide gel.
In Vitro IC 50 Assay-40 nM of the DNA template containing an Alexa-488 fluorescent dye at the 5Ј terminus (5Ј-Alexa488-AGTCCCCCCTTTTCTTTTAAAAAGTGGCTAAGA) was hybridized to a 3-fold molar excess of DNA primer (5Ј-TTAA-AAGAAAAGGGGGG). The hybrid was then incubated with 1 nM RT in a buffer containing 50 mM Tris-HCl, pH 7.8, 80 mM KCl, 0.025% CHAPS, 6 mM MgCl 2, 0.1 mM EGTA, and 1 mM DTT. 200 nM of nucleotide mix containing dATP, dCTP, dGTP, and a modified dUTP were added along with INDOPY-1 (0 -30 M). The modified dUTP contains an Alexa555 quencher, which upon incorporation quenches excitation by Alexa488 present on the DNA template (23). The signal was monitored over a period of 60 min using SPECTRAmax M5 fluorometer where reductions in Alexa488 emission signals are indicative of increased DNA polymerization.
Site-specific Footprinting-Translocation of the enzyme relative to the nucleic acid substrate was monitored using Fe 2ϩ site-specific footprinting techniques as described previously (13,24). Briefly, 150 nM of primer (PPT-20 5Ј-TTAAAA-GAAAAGGGGGGACT) was hybridized to 50 nM DNA template PPT-57 (5Ј-CGTTGGGAGTGAATTAGCCCTTC-CAGTCCCCCCTTTTCTTTTAAAAGTGGCTAAGA) containing a radiolabel at the 5Ј terminus. The DNA hybrid was then incubated with 750 nM RT (WT or mutant) in a buffer containing 120 mM sodium cacodylate, pH 7, 20 mM NaCl, and 10 mM MgCl 2 with and without 1 mM ATP as indicated. The samples were then incubated with a concentration gradient of INDOPY-1 and treated with Fe 2ϩ after a 10-min incubation at 37°C. Samples were resolved on a 12% denaturing polyacrylamide gel.
Standard deviations for all datasets were calculated through GraphPad Prism software. Briefly, the sum of the square of the difference between each value and the mean was divided by n Ϫ 1. The square root of this value represents the standard deviation as calculated by the software.
In Silico Docking of INDOPY-1 and ATP-Binding models were generated using the crystal structure of the HIV-1 RT/chain-terminated primer⅐template⅐dTTP ternary complex (PDB code 1RTD) (25). The dTTP substrate present at the active site of the 1RTD structure was removed prior to INDOPY-1 docking. ATP was subsequently docked to the structural model with INDOPY-1 present in the active site. The structure of INDOPY-1 was constructed in ChemDraw (Cam-bridgeSoft Corp.), energy-minimized for 100 cycles under default parameters using Chimera, and saved as a PDB file. The structures of the protein and the small molecule were prepared for docking using AutoDockTools 1.5.4. Subsequently, the docking was performed using AutoDock Vina. Modeling of ATP binding in the context of INDOPY-1 was performed using the same protein crystal structure as above; however, the coordinates for the best candidate pose for INDOPY-1 binding were this time introduced into the PDB file of the protein. Molecular graphics images were produced using the UCSF Chimera package from the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco (supported by National Institutes of Health Grant P41 RR-01081).
Generation of Recombinant Viruses-The N terminus of RT (amino acids 25-314) of HIV-1 clones containing the TAM1, TAM2, and remodeled TAM mutations (26) were amplified, digested, and ligated using the previously described NRT-vector system (27). Clones were verified by sequence analysis. To generate recombinant viruses, RT molecular clones were transfected in 293T cells. For this, 5-6 ϫ 10 6 293 T cells were seeded the day prior to transfection to achieve 90 -95% confluence on the day of transfection. For transfection, 10 g of plasmid DNA and Lipofectamine 2000 (Invitrogen) were used according to the manufacturer's protocol. Two days after transfection, recombinant virus was harvested. 50% tissue culture infective dose was determined by end point dilution in MT2 cells or human donor peripheral blood mononuclear cells (PBMCs).
Phenotypic Drug Susceptibility-Drug susceptibility was determined by a multiple cycle cell-killing assay (28). MT-2 cells (5 ϫ 10 4 in 200 l of RPMI 1640 medium and 10% FBS per well) were plated in 96-well microplates. Recombinant virus and reference virus were inoculated for 5 days on a single 96-well plate in the presence of drug dilutions. All viruses were inoculated at multiple multiplicities of infection to adjust for any differences in viral replication capacity (RC). Fold change values were calculated by dividing the mean 50% effective concentration for a recombinant virus by that of the HXB2 reference strain. Fold changes are averages of at least two separate experiments. Drug susceptibility was also analyzed in donor PBMCs. Here, activated PBMCs were infected with a recombinant virus or reference strain using a multiplicity of infection of 0.001 (PBMC titration) incubated for 2 h at 37°C, after which cells were washed twice. Subsequently, 0.2 ϫ 10 6 cells/well were plated into a 96-well plate with increasing drug concentrations. p24 was analyzed on days 0 and 7. We next asked whether the enhancement of inhibition was specific to the nucleotide base of ATP. To address this issue, we monitored multiple nucleotide incorporations on a heteropoly-meric template. In the absence of inhibitor, the full-length DNA product appeared within 7 min (Fig. 2B). Saturating concentrations of 6 M INDOPY-1 reduced the amount of DNA product observed over the same period of time. When DNA synthesis was compared in the presence of INDOPY-1 and 3 mM ATP, UTP, GTP, or CTP, we found that ATP showed the strongest enhancement of inhibition. GTP showed subtle but visible increases in inhibition, whereas pyrimidines UTP and CTP do not seem to act as enhancers. In the absence of INDOPY-1, the presence of NTPs showed only subtle effects on DNA synthesis (data not shown). These findings suggest that the base moiety represents an important structural element in enhancing the inhibitory effects of NcRTIs.

ATP Enhances INDOPY-1-mediated Inhibition of DNA
Structure-Activity Relationship Studies for ATP Enhancement-We monitored inhibition of DNA synthesis and its enhancement in the presence of derivatives of ATP with modifications in base, sugar, and phosphate moieties, respectively (for structural details, see supplemental Fig. 1). The loss of the 6-amino group of the purine (benzimidazole and 2-aminopurine), as well as modifications at this position (N6-methyl) severely compromised the enhancement of inhibition (Fig. 3A). Note that in contrast to 2-aminopurine, the related GTP and 2-amino-ATP retain their ability to increase inhibition (Figs. 2B and 3B). Hence, it appears that the presence of a hydrogen bond donor or acceptor is essential at position 6. Modifications at the smaller ring did not show any significant effect (Fig. 3B). In the absence of INDOPY-1, each of these compounds had no significant effect on polymerization under the same conditions (data not shown).
We also studied several ATP derivatives with modifications primarily at the 2Ј-and/or 3Ј-position of the sugar ring. Neither of these compounds showed any significant reductions in enhancement of inhibition. Modifications at the phosphate moiety show a stronger effect. Whereas 1-thio-ATP retains the ability to enhance the inhibitory effect of INDOPY-1, marked decreases are seen in the following order: ATP Ͼ ADP Ͼ AMP Ͼ A (Fig. 3B). Taken together, an intact triphosphate moiety together with a hydrogen bond donor or acceptor at position 6 of the purine base seem to be important structural elements that facilitate binding of the ATP enhancer.
ATP and INDOPY-1 Form a Quaternary Complex with RT and Its Nucleic Acid Substrate-The effect of ATP on INDOPY-1-mediated inhibition of DNA synthesis suggest the formation of a quaternary complex. We employed band shift experiments to test this hypothesis. In this assay, INDOPY-1 can bind to and stabilize the RT⅐DNA⅐DNA complex (Fig. 4A)  (6, 8). The resulting ternary complex is resistant to the challenge with heparin that traps the dissociated enzymes. Increas- ing concentrations of INDOPY-1 correlate with an increased stability of the complex (Fig. 4A). In the absence of INDOPY-1, ATP alone has no effect on complex stabilization. In the presence of ATP, stable complex formation is observed at lower concentrations of INDOPY-1. The K d(app) value for INDOPY-1 is 216.5 nM (Ϯ56.2). In the presence of ATP, this value is decreased to 15.4 nM (Ϯ1.8) (Fig. 4B). Thus, the enhancing effect of ATP on INDOPY-1-mediated inhibition of RT correlates with the formation of a stable quaternary complex.
Binding Modes of ATP-ATP is able to act as a PPi donor that removes incorporated NRTIs from the 3Ј-end of the primer. Like PFA, ATP binds to and exerts this function through the pre-translocational complex (13, 16, 17, 19, 20, 30 -33). However, INDOPY-1 was shown to trap the post-translocated complex (6). To address this apparent disconnect, we examined the potency of INDOPY-1 in the context of a series of RT mutations that are known to affect binding of ATP. Mutations that confer resistance to the NRTI AZT, referred to as thymidine-associated mutations (TAM1 or TAM2), facilitate binding of ATP in an orientation that promotes excision (19,20,31,34,35). We therefore determined EC 50 values for INDOPY-1, in MT-2 cells in the presence of HIV-1 virus containing either TAM1 (M41L, D67N, L210W, and T215Y) or TAM2 (D67N, K70R, T215F, and K219Q) mutational clusters (Table 1). TAM1 does not affect susceptibility to INDOPY-1, whereas TAM2 appears to confer low levels of resistance to the inhibitor (Table 1). We have also tested a mutational cluster, referred to as remodeled TAMs (K70G, V75T, L228R, and K219R) (36). This pattern emerges in NRTI experienced patients who are infected with TAM-containing HIV variants prior to PFA salvage therapy. Remodeled TAMs were shown to reduce resistance to AZT and confer resistance to PFA. Here, we confirmed this phenotype in biochemical assays. Whereas TAM1 and TAM2 show subtle increases in sensitivity to PFA, the RT enzyme containing remodeled TAMs is only inhibited at markedly higher concentrations of PFA (Fig. 5). This effect is in part mediated through K70G, which is in agreement with previous reports that analyzed K70G in a different mutational context (37, 38). Moreover, DNA synthesis in the presence of AZT-TP and the PPi  donor ATP showed reduced excision of the incorporated chain terminator when remodeled TAM was compared with TAM1or TAM2-containing RT enzymes (Fig. 6). Hence, remodeled TAMs may directly affect the binding site of ATP, although several other mutations also confer resistance to PFA (39 -42). As a control, we also measured susceptibility to these compounds in cell culture, and in agreement with Mathiesen et al. (36), remodeled TAMs conferred resistance to PFA, whereas decreased susceptibility to AZT was only observed with strains harboring TAM1 and TAM2 (Table 1). Remodeled TAMs cause hypersusceptibility to INDOPY-1 (0.2-fold change in EC 50 when compared with WT HIV-1). The presence of remodeled TAMs in the purified RT also caused a decrease in IC 50 values for INDOPY-1, which is in agreement with the cell culture data (Table 2). This effect is more pronounced in the presence of ATP, suggesting that this mutational cluster can promote binding of INDOPY-1 and ATP, respectively ( Table 2).
An improved binding of INDOPY-1 and ATP against a backbone of remodeled TAMs is difficult to reconcile with the findings that these mutations diminish resistance to AZT (36). However, the excision-competent complex with ATP exists in the pre-translocated conformation, whereas INDOPY-1 traps the post-translocated complex. To establish that the quaternary complex with INDOPY-1 and ATP remains in the post-translocated conformation, we performed site-specific footprinting experiments that allow us to distinguish between both pre-and post-translocational states (Fig. 7A) (13,15). As the concentration of INDOPY-1 increases, the cleavage pattern with WT RT is shifting from pre-to post-translocation, as described previously (6). Here, the data further show that the presence of ATP lowers the concentration of INDOPY-1 required to cause a complete shift toward post-translocation (Fig. 7). These findings demonstrate that ATP indeed stabilizes the complex with INDOPY-1 in the post-translocated conformation. The effect of INDOPY-1 on the translocational state is slightly increased against a backbone of remodeled TAMs (Fig.  7, A and B), which is in agreement with the hypersusceptible phenotype.
Together, these data suggest that remodeled TAMs confer a post-translocational bias that reduces PFA susceptibility and ATP-dependent excision of AZT-MP, both of which occur in the pre-translocated state. At the same time, the increased access to the post-translocated complex results in INDOPY-1 hypersusceptibility.

DISCUSSION
NcRTIs, like the prototype compound INDOPY-1, represent a novel class of small molecules that inhibit HIV-1 RT at con-  centrations in the submicromolar range. Enzyme kinetics and in vitro selection experiments provided strong evidence to show that the binding site of INDOPY-1 overlaps with the nucleotide-binding site (6,8). Site-specific footprinting experiments further revealed that the inhibitor, like the natural dNTP substrate, traps the RT complex in its post-translocational state (6). Here, we show that the presence of ATP can significantly enhance the inhibitory effect of INDOPY-1. We studied the underlying molecular mechanism with particular focus on the binding mode and orientation of the ATP enhancer molecule. A limited structure-activity relationship analysis revealed that the phosphate and base moieties of ATP are required to promote this effect. The enhancement of inhibition decreases in the following order: ATP Ͼ ADP Ͼ AMP. Pyrimidines UTP and CTP are inactive, whereas purines ATP and, to a lesser extent, GTP are active enhancer molecules. In attempts to provide a better understanding of these findings, we also included several ATP derivatives in this analysis. The data suggest that the presence of a hydrogen bond donor or acceptor at position 6 of the purine ring system is crucial for enhancing NcRTImediated inhibition. The specificity for purine analogs also points to a unique feature of the ATP-binding site that is distinct from the dNTP-binding site, which needs to accommodate both purines and pyrimidines. These results are not unexpected, given that ribonucleotides bind to this site with low affinity due to a steric conflict between the 2Ј-OH group and residue Tyr-115 in HIV-1 RT (43,44). Moreover, INDOPY-1 competes with dNTP substrates, which suggests that the nucleotide-binding site is at least in part occupied by the inhibitor.
Like the natural dNTP substrate, INDOPY-1 can form a stable complex with HIV-1 RT and its primer⅐template substrate (6,8,45). These complexes are resistant to challenges with an enzyme trap, such as heparin. Crystal structures of HIV-1 RT with and without a bound dNTP substrate have shown that the fingers subdomain close down and trap the incoming dNTP, which stabilizes the complex (25,46). A similar scenario has been postulated for the ternary complex with INDOPY-1 (6,8), although the particular conformation of the "closed" structure likely differs from the ternary RT⅐DNA⅐DNA⅐dNTP complex. Here, we show that concentrations of ATP in the low millimolar range can reduce the amount of INDOPY-1 required to stabilize the RT complex. Together, these data provide strong evidence to suggest that RT, its nucleic acid substrate, the inhibitor, and the ATP enhancer molecule form a stable quaternary complex.
Binding of ATP to HIV-1 RT has previously been demonstrated in a different context (19,20,35). ATP can bind in an orientation that facilitates the excision of the ultimate nucleotide from the primer terminus. This reaction provides an important mechanism for resistance to NRTIs. TAM1 or TAM2 mutational clusters were shown to increase rates of exci-  sion of incorporated AZT-MP, and other inhibitors of this class (19). Structures that contain the dinucleotide tetraphosphate product suggest that the resistance-conferring mutations can affect the orientation of the base moiety of the bound ATP thereby facilitating binding and its function as a PPi donor (20). Site-specific footprinting and crystal structures of a related polymerase have shown that the PPi analog PFA stabilizes the pre-translocational conformation though closure of the fingers subdomain (15,47). By contrast, INDOPY-1 traps the posttranslocational state, which raised the question whether ATP can bind to both translocational states of the RT complex. Our footprinting experiments provide strong support for the notion that the quaternary complex with INDOPY-1 and ATP is indeed stabilized in the post-translocational state.
To provide a more detailed picture of the ATP binding mode and orientation, we conducted in vitro susceptibility measurements with TAM1 (M41L, D67N, L210W, and T215Y) and TAM2 (D67N, K70R, T215F, and K219Q) and the remodeled TAMs cluster that contains mutations K70G, V75T, L228R, and K219R. Given that TAM1 and TAM2 confer resistance to AZT and facilitate ATP binding, one would expect increases in susceptibility to INDOPY-1. However, TAM1 does not appear to show any significant effect, and TAM2 shows subtle decreases in INDOPY-1 susceptibility. Conversely, given that remodeled TAMs reduce resistance to AZT and increase resistance to PFA (36), one would expect decreases in susceptibility to INDOPY-1. However, in this case we observed a hypersusceptible phenotype, and biochemical studies show that the mutations increase the effects of both INDOPY-1 and ATP. A possible explanation is that remodeled TAMs can shift the translocational equilibrium toward post-translocation, which facilitates binding of INDOPY-1. Binding of ATP and PFA to the pre-translocated conformation is at the same time reduced under these circumstances. Essentially, the diminished excision of AZT by remodeled TAMs indicates that binding of ATP to the pre-translocated complex is compromised. This effect passively promotes binding of ATP instead to the post-translocated state of the enzyme to stabilize a quaternary complex with INDOPY. This model helps to reconcile the apparent contradiction of ATP-mediated enhancement of INDOPY-1 for remodeled TAMs and diminished resistance to AZT for the same enzyme. In this context it is also important to note that TAM enzymes show the opposite effect on the translocational equilibrium and confer a pre-translocational bias (48).
Model for INDOPY-1 and ATP Binding to the Polymerase Active Site of HIV-1 RT-We employed in silico approaches in attempts to identify possible binding pockets for INDOPY-1 and its ATP enhancer. We docked a flexible structure of INDOPY-1 to the RT crystal structure whereby the crystallized dTTP substrate was removed (Fig. 8). To validate our docking strategy, we redocked dTTP back into the polymerase active site and confirmed a high degree of similarity with the original crystal structure. In our model, INDOPY-1 binds to the nucleotide-binding site of a post-translocated complex. An important feature of this model is that the benzene ring of the inhibitor can stack with the ultimate base of the primer (Fig. 8A). Binding of INDOPY-1 is severely compromised with an abasic residue at the primer terminus (data not shown), which is consistent with this model. The nitro group of the benzene ring is positioned within close proximity to residues Met-184 and Tyr-115, and changes at these positions are associated with resistance to this compound (Fig. 8B). Finally, the methyl group present on the pyrrole ring appears to protrude into the "unoccupied pocket" that exists between the fingers and the thumb subdomains of HIV-1 RT (Fig. 8, C and D).
We next docked ATP to the model of the ternary RT⅐ DNA⅐DNA⅐INDOPY-1 complex (Fig. 8, C-E). ATP appears to bind in an orientation that is similar to the excision complex (20), although the two complexes exist in post-and pre-translocational states, respectively. The phosphate moiety of ATP interacts with the same residues as seen with the crystal structure of ATP bound to RT (20), although the base of ATP bends toward the thumb subdomain of RT and the sugar backbone of the DNA⅐DNA substrate. Consistent with our structure-activity relationship data, the amino group at position 6 of ATP makes possible contacts with other residues in the quaternary complex (Fig. 8, C-E). Finally, amino acid residues that are mutated as part of remodeled TAMs are in close proximity to the phosphate and base moiety of the docked ATP (Fig. 8E). In summary, according to this model, ATP can bind to the posttranslocational state with INDOPY-1 bound at the active site (Fig. 8, C-E). It is possible that both ligands ATP and INDOPY-1 stabilize the closed complex more effectively. However, it should be noted that this docking model does not reflect the intrinsic flexibility of the fingers and thumb subdomains. As such, crystallographic data are warranted to delineate subtle differences in subdomain movements in the context of INDOPY-1/ATP versus dNTP binding.
Overall, the results of this study point to a highly flexible ATP-binding site of HIV-1 RT. Although previous studies have described and characterized binding of ATP to the pre-translocated complex as a PPi donor, this work describes its binding to the post-translocated complex as an enhancer of the inhibitory effects of INDOPY-1. Together, the data identify a new binding pocket for small molecule inhibitors that can be exploited in drug discovery and development efforts to improve the properties of NcRTIs.