Nonnucleoside inhibitors of reverse transcriptase, composite binding pocket and methods for use thererof

Modeling Procedure

Construction of the Composite NNI Binding Pocket

A novel model of the NNI binding pocket of RT was constructed by superimposing nine individual RT-NNI crystal structures and then generating a van der Waals surface which encompassed all of the overlaid ligands. This “composite binding pocket” surprisingly reveals a different and unexpectedly larger NNI binding site than shown in or predictable from any of the individual structures and serves as a probe to more accurately define the potentially usable space in the binding site (FIG. 2 A).

Modeling studies were based on the construction of a binding pocket which encompassed the superimposed crystal structure coordinates of all known RT-NNI complexes, including nine different structures of RT complexed with HEPT, MKC, TNK, APA, Nevirapine, N-ethyl Nevirapine derivative, 9-Cl TIBO (Ren, J. et al., Structure, 1995, 3, 915-926); 9-Cl TIBO (Das, K. et al., J. Mol. Biol., 1996, 264, 1085-1100) and 8-Cl-TIBO (PDB access codes rti, rt1, rt2, hni, vrt, rth, rev, tvr, and hnv, respectively).

The “thumb” region of RT complexes are relatively variable compared with the “palm” region. Therefore, a total of 117 C-alpha atoms of the residues from 97 to 213 (SEQ ID NO:1) which cover part of the NNI binding site and the “palm” region were used for a least-squares superimposing procedure within the program O (Jones, T. A. et al., Acta Crystallogr. A., 1991, 47, 110-119). Each coordinate set was superimposed onto the same initial coordinate set (RT/9-Cl TIBO). The distance between the pair was minimized by rotating and translating one coordinate set onto the other, minimizing distances between x, y, and z coordinates, according to the method of the program “O”. The root mean square (RMS) values of the coordinates of the atoms being superimposed are shown to be 1.00, 0.98, 0.99, 0.62, 0.80, 0.87, 0.94 and 0.65 Å for HEPT, MKC, TNK, APA, Cyclopropanyl Nevirapine, N-ethyl Nevirapine derivative and two 9-Cl TIBO compounds, respectively. Next, the coordinates of the corresponding inhibitor molecules were then transformed according to the same matrices derived from the superimposition. Lastly, the overlaid coordinates of all inhibitors were read into the program GRASP (Nicholls, A., GRASP 1992, New York), from which an overall molecular surface was generated providing a binding pocket encompassing all inhibitors.

As shown in FIG. 2A, the surface of the binding pocket was color coded to reflect characteristics of the overlaid inhibitors, such as hydrogen bonding, hydrophilic, and hydrophobic regions. The amide nitrogens on the uracil ring of HEPT and TIBO derivatives are color-coded red for hydrogen bonding atoms. Oxygen or sulfur atoms of carbonyl, thiocarbonyl, and ester groups, nitrogen atoms of amine groups, and halogen atoms are color-coded blue for polar (hydrophilic) groups. Carbon atoms are considered hydrophobic and are colored grey. This pocket, referred to as the composite binding pocket, was used as a basis for the analysis of inhibitor binding.

To generate the coordinates of the composite binding pocket using the InsightII program, each data point of the net defining the surface of the pocket was represented as a water molecule and was saved in Brookhaven Protein Databank (pdb) format. To provide a visual frame of reference, the coordinates have been superimposed on the pdb coordinates of an existing crystal structure having pdb access code hnv (HIV-1 RT/8-Cl TIBO complex). The coordinates of a composite binding pocket for HIV-1 RT generated by superimposing nine different NNI-RT complexes, are set forth in Table 9.

Docking and K i Prediction

A computer simulation of the binding of PETT, DABO, and HEPT compounds into the NNI binding site of RT was accomplished using a molecular docking procedure. Docking of the compounds into the NNI binding site required the use of X-ray coordinates of an RT-NNI complex (RT/9-Cl-TIBO complex was used for modeling PETT, and the RT/MKC-442 complex was used for modeling DABO and HEPT). Upon binding to RT, the compound can fit into a butterfly-shaped NNI binding site (described by Ding et. al), Ding, J. et al., Nat. Struct. Biol., 1995, 2, 407-415 (FIG. 1 B and 2 A). Once the final docked position of the molecule in the NNI site was determined, the molecule was assigned a score (LUDI), from which an estimation of the inhibition constant (K i value) was determined.

After docking and K i estimation was completed for the inhibitors, evaluation of the docked compounds in the active site of RT involved placing each compound into the composite binding pocket using the same orientation matrix utilized in construction of the pocket. The potentially flexible regions in the binding site were then readily identified as were atom sites for future derivatization of the compounds. Fixed docking in the Affinity program within InsightII (InsightII, Molecular Simulations Inc., 1996, San Diego, Calif.), was used for docking small molecules to the NNI binding site which was taken from a crystal structure (PDB code rev, RT/9-Cl-TIBO complex). The program has the ability to define a radius of residues within a 5 Å distance from the NNI molecule. As the modeling calculations progressed, the residues within the radius were allowed to move in accordance with the energy minimization. Ten final docking positions were initially chosen for each inhibitor modeling calculation but failed to reveal more than two promising positions. Later, only two calculated positions were set for the search target.

Calculations were carried out on a Silicon Graphics INIDIGO 2 using the CVFF force field in the Discover program and a Monte Carlo search strategy in Affinity (Luty, B. A. et al., J. Comp. Chem., 1995, 16, 454-464). No solvation procedures were used. Since the total number of movable atoms exceeds 200, Conjugated Gradient minimization was used instead of the Newton minimization method. The initial coordinates of the compounds were generated using the Sketcher module within InsightII. Each final docking position was then evaluated by a score function in LUDI. The top scoring model was then compared with the composite binding pocket and the known crystal structure of similar compounds and used for further analyses. The inhibitory constants (K i values) of the positioned NNI compounds were evaluated using the LUDI score function (Bohm, H. J., J. Comput. Aided. Mol. Des., 1994, 8, 243-256; Bohm, H. J., J. Comput. Aided. Mol. Des., 1992, 6, 593-606).

Several modifications were imposed during the calculation of inhibitory constants (K i values) of the positioned compounds using the LUDI score function (Bohm, H. J. 1994 supra; Bohm, H. J. 1992 supra). First, the molecular surface areas (MS) were directly calculated from the coordinates of the compounds in docked conformations using the MS program. Second, the number of rotatable bonds (NR), which was assessed inaccurately by INSIGHTII (rigidity imposed by hydrogen bonding was not accounted for in the program), was re-evaluated. Third, it was assumed that the conserved hydrogen bond with RT was assumed to not deviate significantly from the ideal geometry. This assumption was supported by the fact that in the known crystal structures of RT complexes, all hydrogen bonds between NNIs and RT are near the ideal geometry. Last, for the trovirdine compounds, an additional penalty was imposed for a charged group or halogen atoms when positioned near the ring plane of a protein residue such as tryptophan 229 because the interaction was not adequately accounted for in the LUDI score. The working modification of the LUDI scoring function for the PETT compounds included subtracting a score of P from the total LUDI score when the ring plane of the Trp229 was within 5 Å from a para substituent (R):

LUDI Score= MS*BS* 2.93+85(H-bond)− NR* 24.2−100− P;

where

P=200, when R=a hydrophilic group, e.g. —OH or —NO2;

P=100, when R=a para-halogen atom, e.g. —F, —Cl or —Br;

P=50, when R=a para-methoxy, e.g. —OMe;

P=0, when R=a hydrophobic group, e.g. H, CH3;

Consequently, the K i values for the modeled compounds were more predictable than they would be without such modification (Bohm, H. J. 1994 supra; Bohm, H. J. 1992 supra).

Contact Surface and Gap Analysis

Independent of the composite binding pocket and as a follow-up to the docking procedure, computer programs were used to analyze the surface complementarity between the compounds and the binding site residues. This analysis provided another useful way to examine binding interactions, based solely upon the structure that was used for docking (RT/9-Cl TIBO for PETT and RT/MKC-442 for DABO and HEPT) (Das, K. et al., J. Mol. Biol., 1996, 264, 1085-1100).

A number of computer programs were written to analyze the surface of the compounds in the NNI binding site of RT and to better visualize any spatial gaps between the compounds and nearby residues of the RT protein. The algorithm used in these programs was based on a series of cubic grids surrounding the compound, with a user-defined grid spacing. All cubes were coded based on the distance and the nature of the interaction with the protein residues and/or compound atoms. The cubes that overlap both protein and compound within the contact radius are displayed as spheres and were selected to represent the buried surface (user-defined contact radius was the van der Waals radius plus an uncertainty factor, dependent on the reliability of source coordinates). All other cubes that did not interact with protein residues and were within a certain distance from the compound were selected to represent the gap space (space unoccupied by compound or protein) and are displayed as rods.

A graphic interface was then used to examine whether the “gap” spheres could be connected with the compounds without intersecting the “contact” spheres. If the criterion was met, the points that stemmed from the surface of the compound were defined as an expandable region (eligible for synthetic modification). The spheres generated by the programs (shown in FIG. 3) represent the sites buried by protein residues, indicating regions of the compound which are probably not available for derivatization.

FIG. 4 shows the binding pocket embellished with a grid of red rods which represent unoccupied space between the compound and active site residues, providing a complementary view to that shown by the spheres. The grid illustrates the candidate sites for derivatization of the compound and, when used as a distance scale (the length of one rod represents 1 Å), also indicates the volume available for new functional groups.

One example of a useful program is the “SeeGap” program, whose code is listed below in Example 11, together with instructions for its use.

Composite NNI Binding Pocket of RT Reveals Protein Flexibility and Future Inhibitor Modification Sites

The integrated structural information surprisingly revealed a much larger binding site than any shown in individual structures and served as a probe to define the potentially usable space in the binding site (FIG. 1 ). The three-dimensional binding site can be used as a reference point for the analysis of compounds which have been positioned by a docking procedure.

Upon inspection of the pocket it was apparent that although there are no large-scale conformational changes within the NNI binding site, a number of RT protein residues in contact with the inhibitors are relatively flexible and vary from structure to structure. These residues include Tyr188, Tyr181, Tyr318, Try319, Phe227, Leu234, Trp229, Pro95, and Glu138 latter from p51 subunit of RT).

As shown in FIG. 2B, the surface of the composite binding pocket which is overlaid with the RT-TIBO binding site is a short distance (<1.5 Å) away from or even extends past RT residues 234-236, Y188, F227, and the backbone of K101 (SEQ ID NO:1). This indicates that these residues are flexible and can be displaced by the right substituent on an inhibitor.

The composite binding pocket of the invention, unlike a single crystal structure, is able to integrate the nature and extent of the flexibility of the active site residues in the NNI binding site of RT. This uniquely permits prediction of potential modification sites on PETT, DABO, and HEPT derivatives after positioning the compounds in the NNI active site of RT. The method for designing new NNI compounds was particularly useful given the fact that no known crystal structures exist for RT-PETT and RT-DABO complexes, a fact which in this case would prevent the successful application of traditional structure-based drug design methods. Importantly, the model was validated by experimentally demonstrating the superior potency of newly designed agents, predicted to have strong RT inhibitory activity, based upon the low K i values estimated.

Predicted Efficacy of HEPT Derivatives

Compounds listed in Table 1 have been modeled into the NNI binding site of RT (RT/MKC 422 complex) using the docking procedure. The modeled positions were compared with the composite binding pocket of the invention, having the coordinates set forth in Table 9. Modeling was followed by analysis with the LUDI score function.

All of the positions of the compounds with top scores fall into the butterfly-shaped binding site, with the benzyl ring residing in wing 1 and the thymine ring in the wing 2 (FIG. 2 ). For all compounds tested, the benzyl ring is near Trp229 and the N-1 group is near Pro236, a typical position observed in crystal structures (FIG. 1 B). The trend of calculated values listed in Table 1 shows that the K i value decreases as a result of three factors: para substituents (R2) removed from the benzyl ring, larger alkyl groups added to the thymine ring (R 1 ), and sulfur atoms substituted for oxygen (at X and/or Y). The modeling calculations, along with the application of the composite NNI binding pocket, provided a guideline for the synthesis of lead compounds designed to have potent anti-HIV activity. The choice of compounds was also based on synthetic feasibility.

Synthesis of HEPT Derivatives

The compounds listed in Table 1 above can be synthesized by reaction of substituted aryl acetonitriles and appropriately functionalized 2-bromo ethyl esters, for example in the presence of zinc in refluxing tetrahydrofuran. Products of the reaction are purified by gel chromatography. Generated 3-oxo esters are next converted into 5-alkyl-6-(arylmethyl)-2-thiouracils with chloroacetic acid, e.g., overnight to yield 5-alkyl-6-(arylmethyl)uracils. The final step in the synthesis is reaction of the uracil with hexamethyldisilazane (HMDS) in the presence of ammonium sulfate. Subsequent treatment with acetals and trimethyl silyl triflate in acetonitrile leads to the formation of N-′substituted uracil and thiouracil derivatives.

These and other known methods can be used to synthesize the compounds of the invention.

DABO Derivatives

Chemical Synthesis

All chemicals were used as received from Aldrich Chemical Company (Milwaukee, Wis.). All reactions were carried out under nitrogen. Column chromatography was performed using EM Science silica gel 60 and one of the following solvents: ethyl acetate, methanol, chloroform, hexane, or methylene chloride. Nuclear magnetic resonance (NMR) spectra were recorded on a Varian (Palo Alto, Calif.) 300 MHz instrument (Mercury 2000 model) and chemical shifts (δ) are reported in parts per million (ppm) relative to tetramethylsilane as an internal standard at 0 ppm. 13 C NMR spectra were recorded at 75 MHz in CDCl 3 on the same instrument using a proton decoupling technique. The chemical shifts reported for 13 C NMR are referenced to the chloroform triplet at 77 ppm. Melting points were measured using a Mel-Temp 3.0 (Laboratory Devices Inc., Holliston, Mass.) melting apparatus and are uncorrected. UV spectra were recorded from a Beckmann (Fullerton, Calif.) model DU 7400 UV/Vis spectrometer using a cell path length of 1 cm and methanol solvent. Fourier Transform Infrared spectra were recorded using an FT-Nicolet (Madison, Wis.) model Protege 460 instrument. Mass spectrum analysis was performed using a Hewlett-Packard (Palo Alto, Calif.) Matrix Assisted Laser Description time-of-flight (MALDI-TOF) spectrometer (model G2025A) in the molecular ion detection mode (matrix used was cyanohydroxycinnamic acid). Some samples were analyzed using a Finnigan (Madison, Wis.) MAT 95 instrument. Elemental analysis was performed by Atlantic Microlabs (Norcross, Ga.).

General Procedure for the Synthesis of DABO Compounds 3a-d

The 5-alkyl-2-[(methylthiomethyl)thio]-6-(benzyl)-pyrimidin-4-(1H)-one derivatives 3a-d were prepared as shown in Scheme 1.

Ethyl-2-alkyl-4-(phenyl)-3-oxobutyrates 1a-d were obtained from commercially available phenyl acetonitrile. The β-ketoesters were condensed with thiourea in the presence of sodium ethoxide to furnish the corresponding thiouracils 2a-d. Compounds (1a-d and 2a-d) were produced by a methods previously described (Danel, K. et al., Acta Chemica Scandinavica, 1997, 51, 426-430; Mai, A. et al., J. Med Chem., 1997, 40, 1447-1454; Danel, K. et al., J. Med. Chem., 1998, 41, 191-198).

Subsequent reaction of thiouracil with methylchloromethyl sulfide in N,N-dimethylformamide (DMF) in the presence of potassium carbonate afforded compounds 3a-d in moderate yields A mixture of thiouracil compound 2 (1 mmol), methylchloromethyl sulfide (1 mmol), and potassium carbonate (1 mmol) in anhydrous DMF (5 ml) was stirred overnight at room temperature. After treatment with water (50 ml), the solution was extracted with ethyl acetate (3×50 ml). The combined extracts were washed with saturated NaCl (2×50 ml), dried (MgSO 4 ), filtered and concentrated in vacuo to give the crude products 3a-d which were purified by column chromatography (hexane: ethyl acetate eluent).

X-ray Crystallography

Yellow rectangular plates of compound 3b were grown from tetrahydrofuran by slow evaporation at room temperature. X-ray diffraction data for a 0.5×0.2×0.08 mm plate crystal of compound 3b was collected at room temperature using a SMART CCD X-ray detector (Bruker Analytical X-ray Systems, Madison, Wis.). Structure solution and refinement was performed using the SHELXTL suite of programs (Bruker Analytical X-ray Systems, Madison, Wis.). All nonhydrogen atoms were refined using anisotropic displacement parameters. Hydrogen atoms were placed at ideal positions and refined as riding atoms with relative isotropic displacement parameters.

The refined small molecule X-ray crystal structure of compound 3b is shown as an Oak Ridge Thermal Ellipsoid Program (ORTEP) drawing in FIG. 8 . Table 2 lists the crystal data and structure refinement statistics for compound 3b. Data was collected at room temperature (λ=0.71073 Å), refined using full-matrix least-squares refinement on F 2 , and corrected for absorption using semi-empirical psi-scans.

Physical Data of Synthesized Compounds

5-methyl-2-[(methylthiomethyl)thio]-6-benzyl-pyrimidin4-1H-one (3a)

Yield 62%; mp 148-149° C.; 1 H NMR(CDCl 3 ): δ2.10 (s, 3H), 2.14 (s, 3H), 3.91 (s, 2H), 4.29 (s, 2H), 7.29-7.26 (m, 5H), 12.20 (s, 1H); 13 C NMR(CDCl 3 ): δ10.7 (CH 3 ), 15.5 (SCH 3 ), 36.6 (CH 2 Ph), 41.0 (SCH 2 ), 116.7 (C-5), 137.6-126.4 (Ph), 155.2 (C-6), 162.0 (C-4), 165.1 (C-2); Cl—MS: 293.1 (M+1).

5-ethyl-2-[(methylthiomethyl)thiol-6-benzyl-pyrimidin-4-1H-one (3b)

Yield 65%; mp 124-126° C.; 1 H NMR(CDCl 3 ): δ1.08 (t, 3H), 2.12 (s, 3H), 2.58 (q, 2H), 3.91 (s, 2H), 4.26 (s, 2H), 7.28-7.26 (m, 5H), 12.30 (s, 1H); 13 C NMR(CDCl 3 ): δ13.1 (CH 3 ), 15.4 (SCH 3 ), 18.7 (CH 2 ), 36.4 (CH 2 Ph), 40.3 (SCH 2 ), 122.4 (C-5), 138.0-126.3 (Ph),155.4 (C-6), 161.5 (C-4), 165.2 (C-2); CI—MS: 307.1 (M+1).

5-isopropyl-2-[(methylthiomethyl)thio]-6-benzyl-pyrimidin-4-1H-one (3c)

Yield 57%; mp 116-117° C.; 1 H NMR(CDCl 3 ): δ1.22 (d, 6H), 2.07 (s, 3H), 3.03 (q, 1H), 3.88 (s, 2H), 4.21 (s, 2H), 7.24-7.13 (m, 5H), 12.43 (s, 1H); 3 C NMR(CDCl 3 ): δ15.4 (SCH 3 ), 19.6 (CH 3 ), 28.0 (CH), 36.3 (CH 2 Ph), 40.9 (SCH 2 ), 125.3 (C-5), 138.3-126.3 (Ph), 155.5 (C-6), 161.1 (C-4), 164.5 (C-2); CI—MS 321.1 (M+1).

5-isopropyl-2-[(methylthiomethyl)thio]-6-(3,5-dimethylbenzyl)-pyrimidin-4-1H-one (3d)

Yield 67%; mp 116-120° C.; 1 H NMR(CDCl 3 ): δ1.28 (d, 6H), 2.15 (s, 3H), 2.27 (s, 6H), 3.10 (q, 1H), 3.88 (s, 2H), 4.31 (s, 2H), 6.84 (s, 3H), 12.42 (s, 1H); 13 C NMR(CDCl 3 ): δ15.3 (SCH 3 ), 19.6 (CH 3 ), 21.2 (CH 3 ), 28.0 (CH), 36.3 (CH 2 Ph), 40.8 (SCH 2 ), 125.2 (C-5), 138.0-126.5 (Ph), 155.4 (C-6), 161.3 (C-4), 164.7 (C-2); CI—MS: 349.2 (M+1).

Modeling and Design of DABO Compounds

The calculated molecular coordinates of DABO compounds which were energy-minimized and docked into the NNI binding site adopted a conformation remarkably similar to that of the crystal structure of compound 3b. FIG. 7B shows the modeled coordinates superimposed on the crystal structure coordinates of 3b and illustrates their conformational similarity, suggesting that the final docked positions of the DABO compounds in the NNI pocket were energetically favorable and quite suitable for these studies. Multiple sterically allowed unoccupied spatial gaps in the binding site were identified from the docking studies which could be filled by strategically designed functional groups (FIG. 7 B).

The docked DABO molecule (compound 3a) unexpectedly showed significant space surrounding the benzyl ring and the 5th position of the thymine ring, which led to design of compounds 3b, 3c and 3d. The inhibition constants of the docked molecules were calculated based on a LUDI score function and are listed in Table 3. The calculated K i values suggested that compounds 3c and 3d would be particularly active inhibitors of RT.

Compound 3d, which differs from compound 3c by the addition of two methyl groups to the benzyl ring, provides more hydrophobic contact with the NNI binding pocket and was predicted to be more potent than compound 3c, based on the modeling studies. Calculations indicate that compounds 3a-3d have progressively larger molecular surface areas but still maintain approximately the same percentage of the molecular surface area in contact with the protein residues. Consequently, the calculated contact surface area between the protein and the compound increases in the following order: compound 3a, 3b, 3c, and 3d. This increased surface area in turn dictates a decrease in calculated K i values, with 3a having the worst value and 3d the best.

The Tyr183 residue of the HIV RT is located in the catalytic region which has a conserved YMDD motif characteristic of reverse transcriptases. Therefore, the displacement of this tyrosine residue can interfere with catalysis and render the HIV-1 RT protein inactive. Bulky substituents at the 5th position of the thymine ring could indirectly accomplish such inactivation by displacing Tyr181 which is near Tyr183 (Ding, J. et al., Nat. Struct. Biol., 1995, 2, 407-415). The composite binding pocket shows sufficient room for at least a 3-carbon group at the 5th position. The addition of a methyl, ethyl or isopropyl group at the 5th position of the thymine ring is expected to lead to higher affinity for the relatively hydrophobic environment at this location of the binding pocket. The favorable hydrophobic contact increases as the hydrophobic group at the 5th position gets bulkier. As the DABO derivative binds to the site, the ethyl or isopropyl group can also cause the nearby Tyr181 residue to rotate away from the inhibitor.

Modeling studies showed that this change in conformation in turn affects the positions of neighboring Tyr183 and Tyr188 which may contribute to the inactivation of HIV-1 RT. The benzyl ring of compounds 3a-3d is located near a region surrounded by the hydrophobic ring planes of residues Trp229, Pro95, Y188 and Y181. The analysis of compounds 3a-3c in the composite binding pocket suggests that the benzyl ring would be located on the boundary of the pocket, near residue Y188. A para substituent of the ring is situated perpendicular to the ring plane of nearby Trp229, within van der Waals contact, and leaves a lot of space unfilled between the compound and Pro95. With a slight conformational rotation of the benzyl ring, compound 3d, with the addition of two methyl groups, was found to better fill the composite binding pocket (FIG. 7 B). Such observations indicate that further modifications to the benzyl ring could lead to even more potent inhibitors.

Predictable Activities

The trend of the calculated K i values based on the modeling and on the use of the composite binding pocket, with surprising accuracy, predicted the trend of the experimentally determined IC 50 values from HIV replication assays. Compounds 3a-3d were tested for RT inhibitory activity in cell-free assays using purified recombinant HIV RT (listed as IC 50 [rRT] in Table 4), as well as by in vitro assays of anti-HIV activity in HTLVIIIB-infected peripheral blood mononuclear cells (IC 50 [p24] in Table 4) (Zarling, J. M. et al., Nature, 1990, 347, 92-95; Erice, A. et al., Antimicrob. Ag. Chemother., 1993, 37, 835; Uckun, F. M. et al., Antimicrobial Agents and Chemotherapy, 1998, 42, 383).

Larger compounds which better fill the composite binding pocket and have lower calculated K i values showed better IC 50 [rRT] values. This is reflected by the enhancement of the inhibitory activity with the addition of progressively larger groups such as methyl (3a), ethyl (3b), and isopropyl (3c) at the C-5 position of the thymine ring (see Table 4). The same trend was also observed for IC 50 [p24] values.

The lead DABO derivative, 5-isopropyl-2-[(methylthiomethyl)thio]-6-(benzyl)pyrimidin-4-(1H)-one (compound 3c), elicited potent anti-HIV activity with an IC 50 value less than 1 nM for inhibition of HIV replication (measured by p24 production in HIV-infected human peripheral blood mononuclear cells) and showed no detectable cytotoxicity (inhibition of cellular proliferation was >100 μM as measured by MTA) (Table 4). In contrast to all previously published data for DABO and S-DABO derivatives which were less active than AZT and MKC-442 (Danel, K. et al., Acta Chemica Scandinavica, 1997, 51, 426-430; Mai, A. et al., J. Med. Chem., 1997, 40, 1447-1454; Danel, K. et al., J. Med. Chem., 1998, 41, 191-198) and showed selectivity indices of <1,000, the novel compound 3c was more than 4-fold more active than AZT and MKC-442, and abrogated HIV replication in peripheral blood mononuclear cells at nanomolar concentrations with an unprecedented selectivity index (=IC 50 [MTA]/IC 50 [p24]) of >100,000.

The X-ray crystal structure of 3b was determined to compare its conformation to that of the compound after docking into the NNI binding site. The refined small molecule X-ray crystal structure of compound 3b is represented as an ORTEP drawing in FIG. 8 . The calculated molecular coordinates of DABO compounds which were energy-minimized and docked into the NNI binding site adopted a conformation remarkably similar to that of the crystal structure of compound 3b. FIG. 7B shows the modeled coordinates superimposed on the crystal structure coordinates of 3b and illustrates their conformational similarity, suggesting that the final docked positions of the DABO compounds in the NNI pocket were energetically favorable.

Table 4: Inhibitory Activity of DABO Compounds

Synthesis of PETT Derivatives

Chemical Synthesis

All chemicals were used as received from Aldrich Chemical Company (Milwaukee, Wis.). All reactions were carried out under nitrogen. Column chromatography was performed using EM Science silica gel 60 and one of the following solvents: ethyl acetate, methanol, chloroform, hexane, or methylene chloride. Nuclear magnetic resonance (NMR) spectra were recorded on a Varian (Palo Alto, Calif.) 300 MHz instrument (Mercury 2000 model) and chemical shifts (δ) are reported in parts per million (ppm) relative to tetramethylsilane as an internal standard at 0 ppm. 13 C NMR spectra were recorded at 75 MHz in CDCl 3 on the same instrument using a proton decoupling technique. The chemical shifts reported for 13 C NMR are referenced to the chloroform triplet at 77 ppm. Melting points were measured using a Mel-Temp 3.0 (Laboratory Devices Inc., Holliston, Mass.) melting apparatus and are uncorrected. UV spectra were recorded from a Beckmann (Fullerton, Calif.) model DU 7400 UV/Vis spectrometer using a cell path length of 1 cm and methanol solvent. Fourier Transform Infrared spectra were recorded using an FT-Nicolet (Madison, Wis.) model Protege 460 instrument. Mass spectrum analysis was performed using a Hewlett-Packard (Palo Alto, Calif.) Matrix Assisted Laser Desorption time-of-flight (MALDI-TOF) spectrometer (model G2025A) in the molecular ion detection mode (matrix used was cyanohydroxycinnamic acid). Some samples were analyzed using a Finnigan (Madison, Wis.) MAT 95 instrument. Elemental analysis was performed by Atlantic Microlabs (Norcross, Ga.).

General Procedure for Synthesis of PETT Derivatives

Compounds I-1, I-3, and I-4 were synthesized as described in Scheme 3. Trovirdine (I-2) was synthesized according to the literature procedure (Bell, F. W., et al., J. Med. Chem., 1995, 38, 4929-4936).

Physical Data of Synthesized Compounds

N-[2-(2-pyridylethyl)]-N′-[2-(pyridyl)]-thiourea (I-1)

white solid (1 g, 49%); mp 98-100° C.; UV(MeOH) λmax: 293, 265, 247 and 209 nm; IR(KBr Disc) ν3415, 3222, 3050, 2360, 1600, 1533, 1479, 1436, 1315, 1240, 1151 and 775 cm −1 ; 1 H NMR (CDCl 3 ) δ11.90 (s, 1H), 8.8 (s, 1H), 8.60-8.58 (d, 1H), 8.03-8.01 (d, 1H), 7.65-7.56 (m, 2H), 7.27-7.14 (m, 2H), 6.93-6.89 (d, 1H), 6.80-6.77 (d, 1H) 4.23-4.15 (q, 2H), and 3.41-3.20 (t, 2H); 13 C NMR(CDCl 3 ) δ179.2, 158.9, 153.0, 149.2, 145.5, 138.5, 136.4, 123.5, 121.4, 117.7, 111.8, 44.9, and 36.9; MALDI-TOF mass found, 257.1(M−1), calculated, 258.3; Anal. (C 13 H 14 N 4 S) C, H, N, S.

N-[2-(1-piperidinoethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (I-3)

white solid (2 g, 74%); mp 150-152° C.; UV (MeOH) λmax: 306, 275 and 205 nm; IR(KBr Disc) ν3155, 3077, 2935, 2850, 2360, 1591, 1525, 1465, 1319, 1226, 1095, 827 and 756 cm −1 ; 1 H NMR (CDCl 3 ) δ8 11.53 (br s, 1H), 9.72 (br s, 1H), 8.22 (d, 1H), 7.72-7.68 (dd,1H), 6.95-6.92 (d, 1H), 3.84-3.78 (q, 2H), 2.61-2.57 (t, 2H), 2.45 (br s, 4H), 1.64-1.48 (m, 6H); 13 C NMR(CDCl 3 ) δ178.1, 151.8, 146.3, 140.8, 113.5, 112.6, 56.1, 54.0, 43.0, 26.3, and 24.3, MALDI-TOF mass found, 343.5, calculated, 343.3; Anal. (C 13 H 19 BrN 4 S) C, H, N, S, Br.

N-[2-(2,5-dimethoxyphenylethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (I-4)

white solid (2 g, 67%); mp 133-138° C.; UV (MeOH) λmax: 202, 205, 231, 276 and 300 nm; IR(KBr Disc) ν3209, 3152, 3078, 3028, 2951, 2831, 1595, 1533, 1468, 1306, 1227, 1095, 1059, 1022, 862, 825, 796, 707 cm −1 ; 1 H NMR(CDCl 3 ) δ11.24 (br s, 1H), 9.30 (br s, 1H), 8.10-8.09 (d, 1H), 7.65 (dd, 1H), 6.82-6.76 (m, 4H), 4.03-3.97 (q, 2H), 3.77 (s, 3H), 3.00-2.96 (t, 2H); 13 C NMR(CDCl 3 ) δ178.7, 153.1, 151.8, 151.7, 146.5, 140.9, 128.1, 117.7, 113.3, 112.6, 111.2, 110.9, 55.7, 55.5, 45.6, and 29.9; MALDI-TOF mass found, 394.0 (M−1), 396.0 (M+1), calculated, 395.0; Anal. (C 16 H 18 BrN 3 O 2 S) C, H, N, S, Br.

Chemical Synthesis II

Compounds II-1-9 were synthesized according to Scheme 4. In brief, 2-amino-5-bromopyridine was condensed with 1,1-thiocarbonyl diimidazole to firnish the precursor thiocarbonyl derivative (A). Further reaction with appropriately substituted phenylethyl amine gave the target PETT derivatives in good yields.

General Procedure for Synthesis

Thiocarbonyldiimidazole (8.90 g, 50 mmol) and 2-amino-5-bromo pyridine (8.92 g, 50 mmol) were added to 50 mL of dry acetonitrile at room temperature. The reaction mixture was stirred for 12 h and the precipitate filtered, washed with cold acetonitrile (2×25 mL), and dried under vacuum to afford (11.40 g, 80% ) of compound A. To a suspension of compound A (0.55 eqv) in dimethyl formamide (15 mL) an appropriate amine (0.50 eqv) was added. The reaction mixture was heated to 100° C. and stirred for 15 hours. The reaction mixture was poured into ice-cold water and the suspension was stirred for 30 minutes. The product was filtered, washed with water, dried, and further purified by column chromatography to flimish the target compounds 1-9 in good yields.

Peagents and conditions: 1) 1,1-thiocarbonyldiimidazole, acetonitrile, RT, 12h.

Physical Data of Synthesized Compounds

N-[2-(2-methoxyphenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-1)

yield: 65%; mp 143-145° C.; UV (MeOH) λmax: 202, 205, 275 and 306 nm; IR(KBr) ν3211, 3153, 3036, 2956, 2835, 1593, 1533, 1462, 1242, 1186, 1036, 1007, 862, 812, 756, 708 cm −1 ; 1 H NMR (CDCl 3 ) δ11.22 (br s, 1H), 9.37 (br s, 1H), 8.02-8.01 (d, 1H), 7.69-7.65 (dd, 1H), 7.28-7.18 (m, 2H), 6.94-6.80 (m, 3H), 4.04-3.98 (q, 2H), 3.81 (s, 3H), 3.04-2.99 (t,2H), 13 C NMR(CDCl 3 ) δ178.7, 157.6, 151.7, 146.3, 141.0, 130.7, 127.9, 126.8, 120.3, 113.5, 112.5, 110.3, 55.2, 45.6, 29.8; Maldi Tof found: 366.0 (M+1), calculated: 365.0; Anal. (C 15 H 16 BrN 3 OS) C, H, N, S.

N-[2-(2-fluorophenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-2)

yield: 71%; mp 156-157° C.; UV (MeOH) λmax: 209, 256, 274 and 305 nm; IR(KBr) ν3446, 3234, 3163, 3055, 2935, 1672, 1595, 1560, 1531, 1466, 1390, 1362, 1311, 1265, 1227, 1169, 1136, 1089, 1003, 864, 825, 756 cm −1 ; 1 H NMR (CDCl 3 ) δ11.36 (br s, 1H), 9.47 (br s, 1H), 8.05-8.04 (d, 1H), 7.72-7.68(dd, 1H), 7.30-7.03 (m, 4H), 6.87-6.84 (d, 1H), 4.06-3.99 (q, 2H), 3.10-3.05 (t, 2H); 13 C NMR(CDCl 3 ) δ179.1, 163.1, 151.7, 146.2, 141.1, 131.2, 131.1, 128.5, 128.4, 124.1, 115.5, 115.2, 113.6, 112.2, 45.8 and 28.2; 19 F NMR(CDCl 3 ) δ−42.58 & −42.55 (d); Maldi Tof found: 355.0 (M+1), calculated: 354.0; Anal. (C 14 H 13 BrFN 3 S) C, H, N, S.

N-[2-(2-chlorophenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-3)

yield: 72%; mp 137-139° C.; UV (MeOH) λmax: 208, 213, 256, 275 and 305 nm; IR(KBr) ν3433, 3221, 3157, 3089, 3037, 2922, 2866, 1668, 1597, 1535, 1466, 1338, 1263, 1209, 1188, 1130, 1095, 1053, 1001, 864, 823, 750 cm −1 ; 1 H NMR (CDCl 3 ) δ11.41 (br s, 1H), 9.54 (br s, 1H), 8.17-8.16 (d, 1H), 7.83-7.79 (dd, 1H), 7.50-7.30 (m, 4H), 6.97-6.94 (d, 1H), 4.19-4.13 (q, 2H), 3.30-3.26(t, 2H); 13 C NMR(CDCl 3 ) δ179.2, 151.7, 146.3, 141.2, 136.3, 134.2, 131.1, 129.6, 128.1, 126.8, 113.6, 112.7, 45.2, and 32.5; Maldi Tof found: 371.8 (M+l), calculated: 371.0; Anal. (C 14 H 13 BrClN 3 S) C, H, N, S, Br.

N-[2-(3-methoxyphenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-4)

yield: 68%; mp 155-156° C.; UV (MeOH) λmax: 208, 274 and 306 nm; IR(KBr) ν3454, 3236, 3147, 3030, 2951, 2869, 2827, 1591, 1545, 1525, 1466, 1304, 1265, 1229, 1188, 1151, 1095, 1051, 1024, 980, 860, 825, 789, 698 cm −1 ; 1 H NMR (CDCl 3 ) δ11.30 (br s, 1H), 9.25 (br s, 1H), 8.05-8.04 (d, 1H), 7.71-7.67 (dd, 1H), 7.29-7.24 (t, 1H), 6.89-6.78 (m, 4H), 4.05-3.99 (q, 2H), 3.81 (s, 3H), 3.00-2.96 (t, 2H); 13 C NMR(CDCl 3 ) δ178.9, 159.7, 151.6, 146.4, 141.1, 140.3, 129.6, 121.2, 115.0, 113.4, 112.7, 111.6, 55.1, 47.1 and 34.8; Maldi Tof found: 367.0 (M+2), calculated: 365.0; Anal. (C 15 H 16 BrN 3 OS) C, H, N, S.

N-[2-(3-fluorophenethyl)]-N′-12-(5-bromopyridyl)]-thiourea (II-5)

yield: 73%; mp 171-172° C.; UV (MeOH) λmax: 202, 208, 258, 275 and 306 nm; IR(KBr) ν 3213, 3155, 3084, 3028, 2866, 1595, 1533, 1477, 1336, 1308, 1229, 1211, 1173, 1136, 1092, 1026, 935, 870, 827, 791, 740 cm −1 ; 1 H NMR (CDCl 3 ) 611.33 (br s, 1H), 9.46 (br s, 1H), 8.05-8.04 (d, 1H), 7.73-7.69 (dd, 1H), 7.31-7.26 (m, 1H), 7.08-6.97 (m, 3H), 6.87-6.83 (d, 1H), 4.06-3.99 (q, 2H), 3.05-3.00 (t, 2H); 13 C NMR (CDCl 3 ) δ179.1, 163.1, 151.7, 146.2, 141.2, 130.1, 129.9, 124.5, 115.9, 115.6, 113.7, 113.5, 113.4, 112.8, 46.7 and 34.6; 19 F NMR(CDCl 3 ) δ−37.30 & −37.33 (d); Maldi Tof found: 354.0 (M + ), calculated: 354.0; Anal. (C 14 H 13 BrFN 3 S) C, H, N, S.

N-[2-(3-chlorophenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-6)

yield: 72%; mp 163-165° C.; UV (MeOH) λmax: 202, 213, 258, 276 and 305 nm; IR(KBr) ν 3242, 3161, 3043, 2929, 1593, 1579, 1547, 1527, 1466, 1313, 1227, 1167, 1095, 997, 889, 827, 812, 785, 700 cm −1 ; 1 H NMR (CDCl 3 ) δ11.33 (br s, 1H), 9.37 (br s, 1H), 8.09-8.08 (d, 1H), 7.73-7.69 (dd, 1H), 7.28-7.15 (m, 4H), 6.85-6.82 (d, 1H), 4.04-3.98 (q, 2H), 3.02-2.97 (t, 2H), 13 C NMR (CDCl 3 ) δ179.1, 151.6, 146.3, 141.2, 140.7, 134.2, 129.8, 129.0, 127.0, 126.8, 113.4, 112.8, 46.7 and 34.5; Maldi Tof found: 371.8 (M+1), calculated: 371.0; Anal. (C 14 H 13 BrClN 3 S) C, H, N, S.

N-[2-(4-methoxyphenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-7)

yield: 85%; mp 178-179° C.; UV (MeOH) λmax: 205, 226, 275 and 305 nm; IR(KBr) ν3221, 3159, 3042, 2931, 2827, 1587, 1510, 1464, 1311, 1225, 1165, 1088, 1034, 820, 773, 708 cm −1 ; 1 H NMR (CDCl 3 ) δ11.30 (br s, 1H), 9.87 (br s, 1H), 8.00-7.99 (d, 1H), 7.67-7.63 (dd, 1H), 7.21-7.18 (d, 2H), 6.95-6.85 (m, 3H), 4.00-3.93 (q, 2H), 3.81 (s, 3H), 2.96-2.92 (t, 2H), 13 C NMR (CDCl 3 ) δ179.1, 158.0, 151.9, 145.8, 140.7, 130.6, 129.6, 113.8, 113.7, 112.1, 55.1, 46.9 and 33.8; Maldi Toffound: 366.0 (M+1), calculated: 365.0; Anal. (C 15 H 16 BrN 3 OS) C, H, N, S.

N-[2-(4-fluorophenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-8)

yield: 69%; mp 177-178° C.; UV (MeOH) λmax: 208, 211, 274 and 306 nm; IR(KBr) ν3456, 3213, 3155, 3086, 3028, 2868, 1595, 1560, 1533, 1477, 1336, 1308, 1238, 1211, 1173, 1136, 1092, 1026, 933, 869, 827, 791, 741, 694 cm −1 ; 1 H NMR (CDCl 3 ) δ11.29 (br s, 1H), 9.27 (br s, 1H), 8.04-8.03 (d, 1H), 7.73-7.69 (dd, 1H), 7.27-7.22 (m, 2H), 7.04-6.99 (m, 2H), 6.83-6.79 (d, 1H), 4.03-3.96 (q, 2H), 3.02-2.97 (t, 2H); 13 C NMR(CDCl 3 ) δ179.1, 163.2, 151.6, 146.3, 141.2, 134.3, 130.3, 130.2, 115.4, 115.2, 113.5, 112, 47.0, and 34.1; 19 F NMR (CDCl 3 ) δ−40.55 (m); Maldi Tof found: 354.8 (M+1), calculated: 354.0; Anal. (C 14 H 13 BrFN 3 S) C, H, N,S.

N-[2-(4-chlorophenethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (II-9)

yield: 71%; mp 180-183° C.; UV (MeOH) λmax: 206, 209, 219, 256, 275 and 305 nm; IR(KBr) ν 3221, 3153, 3086, 3022, 2931, 1674, 1593, 1562, 1533, 1473, 1406, 1340, 1304, 1265, 1227, 1169, 1138, 1092, 1016, 820, 752, 714 cm −1 ; 1 H NMR (CDCl 3 ) δ11.40 (br s, 1H), 9.34 (br s, 1H), 8.15-8.14 (d, 1H), 7.84-7.80 (dd, 1H), 7.46-7.30 (m, 4H), 6.92-6.89 (d, 1H), 4.10-4.07 (q, 2H), 3.13-3.08 (t, 2H); 13 C NMR (CDCl 3 ) δ179.2, 151.6, 146.3, 141.3, 137.1, 130.2, 128.6, 113.5, 112.8,46.8 and 34.2; Maldi Tof found: 372.0 (M+1), calculated: 371.0; Anal. (C 14 H 13 BrClN 3 S) C, H, N, S.

Chemical Synthesis III

Compounds III-1-3 were prepared as illustrated in Scheme 5. The synthesis involved condensing 2-amino-5-bromopyridine with 1,1-thiocarbonyl diimidazole to furnish the required thiocarbonyl derivative. Further reaction of this thiocarbonyl derivative with an appropriate amine gave 1-3 in good yields.

Structure-based Design and Docking of Novel PETT Derivatives into Composite NNI Binding Pocket I

A novel model of the NNI binding pocket of RT was constructed by carefully superimposing the coordinates of 9 individual RT-NNI crystal structures and then generating a van der Waals surface which encompassed all of the overlaid ligands. The integrated structural information of this “composite binding pocket” revealed an unexpectedly different and much larger NNI binding site than shown in or predictable from any of the individual structures and served as a probe to more accurately define the potentially usable space in the binding site (FIG. 2 a ). A number of protein residues in contact with the inhibitors are relatively flexible and vary from structure to structure. These residues include Tyr188, Tyr181, Tyr318, Try319, Phe227, Leu234, Trp229, Pro95, and Glu138 (from p51subunit of RT) (SEQ ID NO:1). As shown in FIG. 2 b, the surface of the composite binding pocket is a short distance away from (<1.5 Å) or even extends past RT residues 234-236, Y188, F227, and the backbone of K101. This indicates that these residues are flexible and can be displaced by the right inhibitor. The composite binding pocket, unlike an individual crystal structure, is able to summarize the nature and extent of the flexibility of the active site residues. This allowed prediction of potential modification sites on the PETT derivatives I after positioning the compounds in the RT active site (see Methods).

A computer simulation of the binding of PETT compounds into the NNI binding site of RT was accomplished using a molecular docking procedure. Docking of PETT and trovirdine into the NNI binding site required the use of X-ray coordinates of an RT/NNI complex (in this case the RT/9-Cl-TIBO complex).

Upon binding to RT, the compound can fit into a butterfly-shaped NNI binding site (described by Ding, J., et al., Nat. Struct. Biol., 1995, 2, 407-415) (FIGS. 1 B and 2 ). PETT and its derivatives such as compounds I-1-4 could be viewed as two chemical groups linked together by a thiourea group (Table 5). One half of the molecule is composed of a 2-aminothiazole group (PETT) or a pyridylthiourea group (compounds I-1-4) which forms an intramolecular hydrogen-bonded heterocyclic ring. The other half of the molecule is a phenyl or heterocyclic ring separated from the thiocarbonyl group by an ethyl linker.

Once the final docked position of the molecule in the NNI site was determined, the molecule was assigned a score, from which an estimation of the inhibition constant (K i value) was determined (Table 5). When trovirdine was docked into the NNI binding site of RT it had a higher binding score than PETT and fit into the butterfly-shaped binding region with one part residing in Wing 1 and the other in Wing 2 (FIG. 1 B). The ring closest to the thiocarbonyl group resided near the Lys(K)101 loop and the other pyridyl ring was near Trp(W)229.

After docking and K i estimation was completed for the PETT inhibitors, evaluation of the docked compounds in the active site of RT involved placing each compound into the composite binding pocket using the same orientation matrix utilized in its construction. The potentially flexible regions in the binding site were then readily identified as were atom sites for future derivatization of the compounds. The area within Wing 2 and the residues near the thiourea group seemed to be the most forgiving regions in the binding site of RT. This observation was also supported by the analysis of gaps in atom-to-atom contact between the protein and the inhibitor.

Analysis of the molecular surface of the compounds in the NNI binding site of RT included visualization of spatial gaps between the compounds and nearby residues of the RT protein, as described above for Example 1. The spheres generated are shown in FIG. 3, and indicate regions of the compound which are probably not available for derivatization. FIG. 4 shows the binding pocket embellished with a grid of red rods which represent unoccupied space between the compound and active site residues, providing a complementary view to that shown by the spheres in FIG. 3 . The grid illustrates the candidate sites for derivatization of the compound and, when used as a distance scale (the length of one rod represents 1 Å), also indicates the volume available for new functional groups. After the docked PETT compounds were subjected to the grid (gap) analysis, a number of gaps in the binding site were identified (FIGS. 3 - 4 ), some of which could be filled by strategically designed functional groups on new PETT derivatives. It was postulated that a more efficient use of such sterically allowed unoccupied spatial gaps in the binding site could be achieved by replacing the 2-pyridyl ring of trovirdine with a 1-piperidinyl (compound I-3) or 2,5-dimethoxyphenyl moiety (compound I-4) and yield potentially more active PETT compounds with larger molecular surface areas, higher Ludi scores, and lower K i values (Table 5).

Compounds I-1, I-3 and I-4 were subjected to the same docking procedure and K i calculation used to analyze the parent compounds PETT and trovirdine (compound I-2). The molecular surface area of the compounds calculated after docking increased in the following order: PETT, compound I-1, I-2 (trovirdine), I-3, and I-4. At docked positions, the atom surface area in contact with the protein residues constituted an average of 84% of the entire molecular surface (FIG. 3 ). We used this average value in the calculation of the inhibitory constant (K i ) based on the Ludi score function. Calculated Ki values for I-3 and I-4 predicted that these compounds would have potency superior to that of trovirdine. The calculated K i values of our compound I-3 (0.6 μM), and compound I-4 (0.2 μM) were better than those of known compounds PETT (2.4 μM), compound I-1 (1.7 μM) and trovirdine (0.7 μM).

In virto Assays of Anti-HIV Activity Using PETT Derivatives I

The HIV-1 strain HTLVIIIB (kindly provided by Dr. Neal T. Wetherall, VIROMED Laboratories, Inc.), was propagated in CCRF-CEM cells, and used in in vitro assays of the anti-HIV-1 activity of the synthesized novel derivatives. Cell-free supernatants of HTLVIIIB-infected CCRF-CEM cells were harvested, dispensed into 1 ml aliquots, and frozen at −70° C. Periodic titration of stock virus was performed by examining its cytopathic effects in MT-2 cells following the procedures described in (Erice, et al., Antimicrob. Ag. Chemother., 1993, 37, 835).

Normal human peripheral blood mononuclear cells (PBMNC) from HIV-negative donors were cultured 72 hours in RPMI 1640 supplemented with 20%(v/v) heat-inactivated fetal bovine serum (FBS), 3% interleukin-2, 2 mM L-glutamine, 25 mM HEPES, 2 g/L NaHCO 3 , 50 μg/ml gentamicin, and 4 μg/ml phytohemagglutinin prior to exposure to HIV-1. The incubated cells were then exposed to HIV-1 at a multiplicity of infection (MOI) of 0.1 during a one-hour adsorption period at 37° C. in a humidified 5% CO 2 atmosphere. Subsequently, infected cells were cultured in 96-well microtiter plates (100 μl/well; 2×10 6 cells/ml) in the presence of test compounds, including AZT as a control. Aliquots of culture supernatants were removed from the wells on the 7th day after infection for p24 antigen assays. The methods used in the P24 assay were as previously described in Uckun, et al., Antimicrobial Agents and Chemotherapy, 1998, 42, 383; Zarling, et al., Nature, 1990, 347, 92-95; Erice, et al., Antimicrob. Ag. Chemother., 1993, 37, 835.

The applied p24 enzyme immunoassay (EIA) was the unmodified kinetic assay commercially available from Coulter Corporation/Immunotech, Inc. (Westbrooke, Me.). In the assay, a murine monoclonal antibody against HIV core protein is coated onto microwell strips. Antigen (HIV core protein) present in the test culture supernatant samples binds the antibody and the bound antibody-antigen complex is quantitated. Percent viral inhibition was calculated by comparing the p24 values from the test substance-treated infected cells with p24 values from untreated infected cells (i.e., virus controls).

In addition, the activity of the test compounds to inhibit recombinant HIV-1 reverse transcriptase (rRT) activity was determined using the Quan-T-RT assay system (Amersham, Arlington Heights, Ill.), which utilizes the scintillation proximity assay principle. The assay method is described in Bosworth, N., et al., Nature, 1989, 341, 167-168. Data for both bioassays is reported as IC 50 values.

In parallel with the bioactivity assays, the effects of the test compounds on cell viability was also examined, using the Microculture Tetrazolium Assay (MTA) described in Darling, et al., Nature, 1990, 347, 92-95; Erice, et al., Antimicrob. Ag Chemother., 1993, 37, 835. In brief, non-infected PBMNC were treated with test compounds or controls for 7 days under identical experimental conditions and 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino)carbonyl]-2H-tetrazolium hydroxide (XTT), was added to quantitative cellular proliferation.

An energy-minimized model of compound I-4 in the RT binding site had the largest molecular surface area in contact with the protein and thus achieved the highest lipophilicity score. The docking studies indicated that the 2-methoxy group of compound I-4 is situated beneath the ethyl linker and fits favorably into a cavity of the binding pocket, providing contact with protein residues that cannot be achieved by trovirdine. Likewise, the 5-methoxy group of compound I-4 provides close contact with residues Pro95 and Trp229. The trend of the calculated K i values accurately predicted the trend of the experimentally determined IC 50 values from HIV replication assays, as shown in Table 5, thereby providing conclusive evidence of the practical utility of the composite model.

The lead compound, I-4 with the lowest calculated K i values of the series, was 8-times more potent than trovirdine against purified recombinant HIV-RT using the cell-free Quan-T-RT system (IC50[rRT] was 0.1 μM for I-4 versus 0.8 μM for trovirdine). Compound I-4 also elicited potent anti-HIV activity with IC 50 values of less than 0.001 μM in 3 of 3 independent experiments which was consistently lower than the IC 50 values for trovirdine (0.007 μM) and AZT (0.008 μM). None of the PETT derivatives were cytotoxic at concentrations as high as 100 μM. Therefore, the calculated selectivity index (IC 50 [MTA]/IC 50 [p24]) of compounds I-3 and I-4 were greater than 10 5 .

All active PETT compounds listed in Table 5 are able to form an intramolecular hydrogen bond between the nitrogen atom of pyridine or thiazole and an amide hydrogen of the thiourea group, as shown in Wing 1 of FIG. 1 B. The intramolecular hydrogen bond was also observed in our small molecule crystal structure of compound I-3 (data not shown). The energy gained by the formation of such a hydrogen bond has been estimated to be about 5 kcal/mol (Bell, et al., J. Med. Chem., 1995, 38, 4929-4936). Our docking results showed that the internal hydrogen bond keeps the pyridyl thiourea (or thiazolylthiourea) in a more rigid conformation and allows the molecule to adopt the appropriate geometry to occupy Wing 1 of the binding site, and at the same time maintain a hydrogen bond with a backbone carbonyl of residue Lys101 (FIG. 1 B).

Compounds I-3 and I-4 differ from trovirdine at the proposed Wing 2 binding region of the molecule. Compound I-3 has a heterocyclic ring which replaces the pyridyl ring and compound 4 has two methoxy groups added at meta and ortho positions of the phenyl ring. The molecular surface areas of compounds I-3 and I-4 are larger than that of trovirdine, as calculated from the coordinates of the predicted active conformation obtained from docking. This larger surface area results in a better lipophilic score and lower calculated K i value (Table 5). Both pyridylethyl and piperidinylethyl groups occupy the same region of Wing 2 near Trp229 (FIG. 2 and 5 ). Our composite binding pocket shows a space large enough to accommodate a group larger than the pyridyl ring of trovirdine. Docking results and analyses of gaps indicate that the pyridyl ring of trovirdine has multiple sites which can be used for incorporation of larger groups. As shown in FIG. 5, there is sufficient space surrounding the pyridylethyl ring for the addition of a two- to four-atom substituent at any of the ring positions. Both sides of the pyridylethyl ring plane of trovirdine are relatively exposed in the pocket (FIG. 3A) and can accommodate additional substituents (FIG. 4 A). This prediction was confirmed by the potency of compound I-4 (which contains ortho, meta-dimethoxy substituents), in inhibiting HIV replication.

The piperidinyl group of 1-3 is puckered and therefore occupies a larger overall volume than the planar pyridyl ring of trovirdine and is in close contact with residues Leu234 and Leu100, the latter of which can mutate to isoleucine, frequently found in a drug-resistant RT mutant strain. In contrast to previously reported extensive attempts at expanding within the pyridyl ring plane (Bell, et al., J. Med Chem., 1995, 38, 4929-4936; Cantrell, A. S., et al., J. Med. Chem., 1996, 39, 4261-4274; Ahgren, C., et al., Antimicrob. Agents Chemotherapy, 1995, 39, 1329-1335), the success of our efforts at modification perpendicular to the ring plane introduces new possibilities to develop more potent inhibitors which combine both modifications. The piperidinyl ring is conformationally more flexible than an aromatic ring has the advantage of fitting an uncompromising binding pocket more effectively, despite the expense paid for loss of entropy upon binding. The analysis shown in FIGS. 3, 4 , and 5 provides new insights for modifications which are different from those of trovirdine derivatives. Various combinations of double substitutions at axial or equatorial positions of the piperidinyl ring generate derivatives with a broader range of curvatures than trovirdine derivatives and better fit Wing 2 which itself contains some curvature.

In summary, a composite binding pocket was constructed which integrated all available crystal structure information about the NNI binding site of RT. This novel computer-generated model was an unexpectedly effective tool that helped to much better comprehend the flexible nature of the binding pocket and to identify specific areas for structural improvements of the inhibitors. Nine lead NNI compounds from published crystal structures were analyzed. With all strategies combined, a number of previously unknown candidate sites for developing more potent derivatives of PETT were identified, such as substituting a bulkier piperidinyl group or an ortho/meta substituted phenyl group in place of an unsubstituted ring which resulted in enhanced inhibitory activity. The presented experimental results demonstrate that two novel PETT derivatives which resulted from our structure-based design efforts using the composite binding pocket are remarkably potent and noncytotoxic anti-HIV agents with unprecedented selectivity indices of >10 5 . The superior activity of these designed PETT compounds would not have been predictable from existing information about trovirdine alone, or from any single crystal structure of RT complexed with an NNI.

Structure-based Design and Docking of PETT Derivatives into Composite NNI Binding Pocket II

The PETT derivatives II, synthesized as described above for Example 4, were analyzed for fit into the NNI binding pocket. Target compounds were also analyzed for anti-viral activity in p24 enzyme immunoassays and also for the ability to inhibit HIV reverse transcriptase activity, using rRT. Methods for these biological assays are described above for Example 6.

A computer simulation of the binding of the target PETT derivatives into the NNI binding site of RT was accomplished using a molecular docking procedure. Docking of the compounds into the NNI binding site required the use of X-ray coordinates of an RT/NNI complex (in this case the RT/9-Cl-TIBO complex).

Trovirdine derivatives could be viewed as two chemical groups linked together by a thiourea group (Table 6). One half of the molecule is composed of a pyridylthiourea group (compounds II-1-9) which forms an intramolecular hydrogen-bonded heterocyclic ring (shown in trovirdine structure). The other half of the molecule is a pyridyl ring separated from the thiocarbonyl group by an ethyl linker.

When trovirdine was docked into the NNI binding site of RT, it fit into the butterfly-shaped binding region (described by Ding, et al., Nat. Struct. Biol., 1995, 2, 407-415) with one part of the molecule residing in Wing 1 and the other in Wing 2. The ring closest to the thiocarbonyl group resided near the Lys(K)101 loop and the other pyridyl ring was near Trp(W)229.

Compounds II-1-9 were positioned according to this binding mode into the RT/9-Cl-TIBO active site by a docking procedure described above for Example 1. The results are shown in FIG. 6 . One of the NH groups of the thiourea part of these compounds consistently formed a hydrogen bond with the backbone of K101.

Once the final, energetically favored docked position of the molecule in the NNI site was determined, a LUDI score was assigned, from which an estimation of the inhibition constant (K i value) was determined (Table 6). The calculated K i values, ranging from 0.4 μM to 0.8 μM suggested that compounds II-2 to II-7 would be active inhibitors of RT. The modeling data, shown below in Table 6, predicted that compounds II-2 to II-7 would be as potent as or more potent than trovirdine for inhibiting RT. The data for the bioassay of RT inhibition follows this prediction.

In virto Assays of PETT Derivatives II

Methoxy Substitutions

The estimated K i values accurately predicted the trend of the measured IC 50 [rRT] values for the inhibition of recombinant HIV RT. Compound II-4 had the lowest K i value. The docking results showed that the meta-methoxy group of II-4 is situated near Pro95 and Trp229 in the binding site, providing contact with these protein residues which cannot be achieved by trovirdine (FIG. 5 ). Based on the IC 50 [rRT] values of all methoxy compounds, the meta-methoxy substituted compound II-4, which had a K i value of 0.4 μM, showed greater inhibitory activity against recombinant HIV RT and it was approximately 2-fold more potent than trovirdine (IC 50 [rRT] was 0.4 μM for compound II-4 versus 0.8 μM for trovirdine). Compound II-4 abrogated HIV replication in human peripheral blood mononuclear cells at nanomolar concentrations with an IC 50 value of 3 nM and a selectivity index (SI) of >3×10 4 (Table 6).

Fluorine Substitutions

Among the fluorine (F) substituted compounds II-2, II-5, and II-8, both meta and ortho fluoro compounds were at least 7-fold more active than trovirdine (IC 50 [p24]<1 nM) (Table 6). Based on the IC 50 [rRT] values, compounds with F substitutions at the meta and ortho positions had nearly the same inhibitory activity against recombinant HIV RT but the para-F substituted compound was 10-fold less active. The color-coded composite binding pocket (FIG. 5) also shows that Wing 2 is mostly hydrophobic except for the region near the ortho positions on both sides of the phenyl ring where polar groups such as halogen atoms would be compatible. Trovirdine, however, lacks such ring substitutents which could provide favorable interactions with these regions of the binding site based on our modeling. Substitutions at the meta position could be on the polar region or the hydrophobic region depending on the chemical group and its consequent conformational change (FIG. 5 ). The m-F substituent of compound II-5 is probably exposed to the polar (blue) region and therefore is as active as the o-F group which would also be exposed to the polar region according to our modeling. The trend in IC 50 [rRT] values observed for F-substituted compounds may reflect such a preference. The p-F atom, which is small in size but electronegative, may not be compatible with the location of the ring plane of nearby hydrophobic Trp229 and could contribute to the lower activity. We postulate that this same incompatibility should be observed for any other highly hydrophilic group at the para position, and that an additional binding penalty be imposed to better quantitate such features when undertaking modeling studies.

Chlorine Substitutions

Chlorine (Cl) substituted compounds II-3, II-6, and II-9 show a trend of observed biological activities which differs from that of both the fluorine and methoxy compounds. Like the p-F substituted compound which was less active than other F-substituted compounds, the p-Cl compound was less active than the o-Cl compound based on the IC 50 [rRT] values. Unlike the m-F substituted compound which was as active as the o-F substituted compound, the m-Cl compound was not as active as the o-Cl substituted compound. According to our modeling, o-Cl is the most likely substituent to be situated near a limited polar region at Wing 2, an interaction which would be favorable. The o-Cl compound, like the o-F compound discussed above, was in fact more active than trovirdine, as was predicted by the modeling procedure and by the use of the composite binding pocket.

Hydrophobic Group Preferred at the Para Position

When IC 50 [rRT]values of all compounds with para substitutions are compared (II-7-9), a distinct trend is evident: the p-methoxy (OMe) compound (7) is favored over the p-halogen group compounds (II-8 and II-9) (Table 6). Only the p-OMe substituted PETT derivative, compound II-7, is comparable to trovirdine in its inhibitory activity against recombinant HIV RT. Compound II-7 inhibited HIV replication in peripheral blood mononuclear cells with an IC 50 value of 15 nM (Table 6). This p-OMe preference is consistent with the understanding of the color-coded composite binding pocket at Wing 2, where the binding pocket residues near the para position are relatively hydrophobic. One can reasonably assume, based on chemical intuition and the available inhibition data which is consistent with the modeling, that para substituted hydrophobic groups positioned near a hydrophobic region of the pocket are most preferred, followed by halogens, and finally hydrophilic groups.

Conclusions

In summary, the data revealed the following structure-activity relationships affecting the potency of PETT derivatives with substitutions on various positions of the phenyl ring:

1) methoxy substitution is more favorable at the meta position than at the ortho or para positions;

2) fluorine substitution is favorable at ortho and meta positions but not at the para position;

3) chlorine substitution is favorable only at the ortho position;

4) a hydrophobic group is more desirable than a polar group or hydrophilic group at the para position. These results were generally consistent with predictions made during modeling.

The use of the composite NNI binding pocket allowed the identification and structure-based design of at least 3 promising PETT derivatives with ortho-F (II-2), ortho-Cl (II-3), and meta-F (II-5) substituents on the phenyl ring. These novel PETT derivatives were more active than trovirdine (as predicted) or AZT and showed potent anti-HIV activity with IC 50 [p24] values <1 nM and selectivity indices (SI) of >100,000 (Table 6).

Design of Heterocyclic PETT Derivatives III

In the course of the search for potent NNIs, a computer model has been developed in which a composite binding pocket was constructed from nine individual crystal structures of RT-NNI complexes. Modeling studies lead to the identification of a number of NNIs with IC 50 values beyond 1 nM for the inhibition of HIV replication (measured by p24 production in HIV-infected human peripheral blood mononuclear cells) and showed no detectable cytotoxicity against human T-lymphocytes (inhibition of cellular proliferation was >100 μM as measured by MTA).

The detailed analysis of trovirdine, a potent PETT derivative, revealed multiple sites which can be used for the incorporation of larger functional groups. In the composite binding pocket, the docked trovirdine molecule showed a lot of usable space surrounding the pyridyl ring, the ethyl linker and near the 5-bromo position (shown in structure of PETT derivative). It was proposed that efficient use of this space by strategically designed functional groups would lead to high affinity binding and ultimately result in better inhibitors.

The effect of systematic substitutions of the phenyl ring of trovirdine with various heterocyclic rings was studied. This provides an alternative strategy to fit the compound into the relatively flexible and spacious Wing 2 region (as illustrated by the composite binding pocket). In the subsequent modeling studies these heterocyclic rings, which have a larger volume than the pyridyl ring of trovirdine, were shown to better fill the Wing 2 region of the composite binding pocket.

The piperidinyl, piperzinyl and morpholinyl rings of compounds III-1-3 are puckered and therefore occupy a larger overall volume than the planar pyridyl ring of trovirdine and are in close contact with residues Leu234 and Leu100, the latter of which can mutate to isoleucine, frequently found in a drug-resistant RT mutant strain. The encouraging results from efforts to make modifications perpendicular to the ring plane introduces new possibilities to develop more potent inhibitors of RT.

The heterocyclic rings which are conformationally more flexible than an aromatic ring may have the advantage of fitting an uncompromising binding pocket more effectively, despite the expense paid for loss of entropy upon binding. Various combinations of double substitutions at axial or equatorial positions of these heterocyclic rings would generate derivatives with a broader range of curvatures than trovirdine derivatives and would serve to better fit Wing 2 which itself contains some curvature.

In virto Assays of Anti-HIV-1 Activity Using PETT Derivatives III

Compounds III-1 to III-3 were tested for anti-HIV activity in HTLVIIIB-infected peripheral blood mononuclear cells. Anti-HIV activity was tested using the p24 immunoassay described above for Example 6. Cytotoxicity was also analyzed using a Microculture tetrazolium Assay (MTA), as described above for Example 6.

The data below in Table 7 show the inhibitory effects of PETT derivatives (compounds III-1-3) on p24 production in HIV-infected peripheral blood mononuclear cells and on viability of peripheral blood mononuclear cells. IC 50 values represent the concentration required to inhibit by 50% the activity of HIV replication as measured by assays of p24 production (IC 50 [p24]) or the concentration required to decrease cellular proliferation by 50% as measured by MTA (IC 50 [MTA]) (Uckun, et al., Antimicrobial Agents and Chemotherapy, 1998, 42, 383; Zarling, et al., Nature, 1990, 347, 92-95; Erice, et al., Antimicrob. Ag. Chemother., 1993, 37, 835)

All three compounds III-1-3 were more potent than trovirdine for inhibitition of HIV. Our lead heterocyclic PETT derivatives, N-[2-(1-piperidinylethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (compound III-1) and N-[2-(1-morpholinylethyl)]-N′-[2-(5-bromopyridyl)]-thiourea (compound 3) elicited potent anti-HIV activity with IC 50 values less than 1 nM for the inhibition of HIV replication (measured by p24 production in HIV-infected human peripheral blood mononuclear cells) and showed no detectable cytotoxicity (inhibition of cellular proliferation was >100 μM as measured by MTA) (Table 7).

“SeeGap” Program for Analysis of Gap Space

To analyze the gap space between the binding pocket and complexed NNI, the “SeeGap” program was developed. The following instructions are for use of the program, whose code is listed below in Table 8:

Preparation

1. Extract the source codes at the lines indicated. The first program is a C-shell command file and should be named as “SeeGap”; the second program should be named as “pdbmax.f”; the third “gridbox.f” and fourth “chgcolor.f”.

2. Compile the source codes: for the first, chmod +x SeeGap; the second, third, and fourth by “f77-o file file.f”.

3. You should now have the executive versions of the programs named as “SeeGap”, “pdbmax”, “gridbox” and “chgcolor”. The preparation is ready.

Use the Program

1. Open “insightII” window, and read in the coordinates of the protein and the coordinates of the ligand. Next, assign the potential to both coordinates by builder module within “insightII” (see insight II manual).

2. Position the ligand in the binding site by a docking procedure, if the position of the ligand is unknown.

3. Using subset/interface command, determine the coordinates of the protein that immediately surround the ligand by a defined distance, e.g., 7 angstroms. Write out the coordinates and name it as “bind.pdb”; write out the coordinates of the ligand and name it as “ligand.pdb”.

4. Adjust the input parameters in the command file “SeeGap” as appropriate.

5. Run the program by typing “SeeGap ligand.pdb bind.pdb>out&”.

6. The results should be in three files: contact.pdb, which represents the grid points on the surface of the ligand and in contact with the protein residues; gap.pdb, which represents the grid points available for modification; and lig.pdb, which represents the grid points covering the ligand.

7. Use a molecular graphics software to display these coordinates.

Table 9: Coordinates of Composite Binding Pocket

These coordinates can be entered into a molecular graphics program to generate a molecular surface representation of the composite binding pocket, which then can be used to design and evaluate inhibitors of RT.