|Home | About | Journals | Submit | Contact Us | Français|
Non-peptide inhibition of fusion remains an important goal in anti-HIV research, due to its potential for low cost prophylaxis or prevention of cell–cell transmission of the virus. We report here on a series of indole compounds that have been identified as fusion inhibitors of gp41 through a structure-based drug design approach. Experimental binding affinities of the compounds for the hydrophobic pocket were strongly correlated to fusion inhibitory data (R2 = 0.91), and corresponding inhibition of viral replication confirmed the hydrophobic pocket as a valid target for low molecular weight fusion inhibitors. The most active compound bound to the hydrophobic pocket and inhibited cell-cell fusion and viral replication at sub-µM levels. A common binding mode for the inhibitors in this series was established by carrying out docking studies using structures of gp41 in the Protein Data Bank. The molecules were flexible enough to conform to the contours of the pocket, and the most active compound was able to adopt a structure mimicking the hydrophobic contacts of the D-peptide PIE7. The results enhance our understanding of indole compounds as inhibitors of gp41.
The continued HIV / AIDS epidemic worldwide and the absence of a vaccine to counter the spread of infection have sparked the development of an ever-expanding repertoire of anti-HIV drugs. New classes of drugs acting with a novel mechanism continue to be advantageous, due to the appearance of resistance to drugs used in the clinic. Entry inhibitors are of particular interest, since they could be used therapeutically to prevent the spread of infection among cells, or in prophylactic form to prevent transmission of the virus between individuals. Small molecule fusion inhibitors acting on conserved elements of the HIV-1 envelope glycoprotein gp41 are promising in this regard, since gp41 is universally required by all HIV strains. Gp41 facilitates the viral fusion process through a conformational switch involving association of three C-heptad repeat (CHR) helices along the conserved hydrophobic grooves of a central trimeric N-heptad repeat (NHR) coiled-coil.1–3 Many studies indicate that a fairly long-lived intermediate exists in which the coiled coil is exposed and susceptible as a drug target. However, the only currently available highly active fusion inhibitors targeting gp41 are peptides, including enfuvirtide (T20), a 36 amino acid peptide that was derived from the CHR and approved by the FDA in 2003,4 and D-peptide inhibitors which target a known hydrophobic pocket on the coiled coil revealed by crystal structures.5–8 Because of the lack of oral availability, emergence of drug resistance, and high cost of enfuvirtide, small molecule inhibitors of gp41 are highly desirable.
The hydrophobic pocket is critical for stability of the six-helix bundle and is considered a hotspot for small molecule drug design. It can accommodate a compound with a molecular weight of 500~600Da.9 Several studies have been carried out to identify small molecule compounds that target this pocket. These molecules include α-helical peptidomimetic compounds with ~15µM EC50 against HIV fusion,10 a series of phenylpyrroles11–14 and furan derivatives15, 16 with low to mid-µM EC50 anti-fusion activity and low nM antiviral activity, and several other compounds with anti-fusion activity in the tens of µM.17, 18
All known small molecule inhibitors have demonstrated no better than several µM inhibition of fusion, which represents a significant challenge to small molecule drug discovery against this target. It is compounded by the fact that no crystal structure of gp41 with a small molecule has been reported. Only recently, three NMR structures with weakly bound ligands have been obtained.19, 20, 21 Most known small molecule inhibitors of gp41 have been identified by screening typical small molecule libraries, which may not be sufficient for harvesting compounds that block the protein-protein interaction between the NHR and CHR of gp41. Most of the chemical entities in standard libraries have been generated for classical medicinal targets.22
We have previously reported eight indole compounds as inhibitors targeting the hydrophobic pocket with mid to low µM EC50 values against HIV fusion.23 These compounds were designed from an NMR structural study of a weakly bound inhibitor M1 (Figure 1) in the hydrophobic pocket20, 23. Compounds 3a and 6a (Figure 1) were described in the previous study. The optimization process described here has yielded compound 14g, with an extended structure more typical of a protein – protein interaction inhibitor, and with sub-µM hydrophobic pocket binding affinity and inhibitory activity against cell-cell fusion and viral replication. It is the first verified low molecular weight fusion inhibitor with < 1µM 50% inhibitory concentration. We report our lead optimization process and potential binding modes in the pocket. Overlapping binding poses of inhibitors of increased length suggested that they could adjust to the contours of the hydrophobic pocket, and mimic interactions typically seen with peptide inhibitors. Calculated poses gave a good prediction of observed rank order, with the exception of compounds with two polar ends.
The strategy for lead optimization was as follows: (1) retain the COOH group, which was confirmed to be essential for binding; (2) extend the length of the compound as is generally required to inhibit a protein – protein interaction; (3) consider shape complementarity by designing inhibitors having flexible groups at both ends to adapt to the contours of the groove. In addition, modifications were made that could “fill out” the area occupied by the molecule in the main pocket. 6-substituted indole compounds are described in Schemes 1 – 3 (see Experimental Methods). Three compounds with a 5-substituted indole were also prepared, namely 3g, 8c and 14i. These compounds are isomers of 6-indole variants 3a, 8b and 14f, respectively (see numbering in Figure 1 and Supporting Information Figure S1). Compounds were analyzed for biological activity in an assay for binding to the gp41 hydrophobic pocket, 24 in an HIV-1 HXB2 Env-mediated cell-cell fusion (CCF) assay 25 and in viral replication assays using HIV-1 Ba-L and HIV-1 IIIB 26, 27; examples are shown in Figure 2 and Figure S2, Supporting Information). Cell culture experiments were conducted at varying concentrations of serum, owing to our discovery that serum blunts the effect of many of the compounds, proportionately increasing the EC50’s observed in fusion, viral replication and cytotoxicity experiments (Table 1, Table 2).
The predicted orientation of small indole compounds in the pocket was based on early docking studies 23 as well as two NMR structures obtained on compound M1 20 and on 3g 21, the 5-substituted indole isomer of 3a. The results indicated that the carboxylate group of the ligand pointed toward Lys574. This implies that the aromatic ring supporting the carboxylate lies in a strongly hydrophobic part of the pocket lined by residues Leu565’, Leu568’, Trp571’ and Val570. The bend provided by the CH2 group between aromatic moieties was essential for shape complementarity, and absence of this hinge led to a 10-fold reduction in potency (compound 15). Substitution on the benzyl ring of 3a was studied to potentially improve affinity through 1) additional hydrophobic interactions with the surrounding pocket (3b, 3c), or 2) through polar functionality (3d–f), including the possibility of accessing the Leu568’.O or Val570.O carbonyl group for an additional hydrogen bond (3d–e) (Figure 4A,C). However, all of the substitutions decreased binding affinity to varying extents compared to 3a (Figure 2), and led to lower biological activity as well (Table 1).
Addition of a third and fourth heterocyclic or aromatic ring was investigated for its potential to generate higher hydrophobic content and an extended interaction surface typical of protein – protein interaction inhibitors. The data in Tables 1 and and22 indicate that this strategy was successful, resulting in significant improvement in binding affinity in some cases. A dimeric form of 3a, compound 10, showed improved activity.23 6a, contained an aryl group attached at the indole nitrogen. It had higher binding and anti-fusion activity than 3a. Compound 8a, which contained a methoxybenzyl group attached at the indole nitrogen, did not have improved activity over 3a. 14c, an isomer of 6a, had similar binding affinity but slightly lower biological activity. Due to ready extension of the scaffold on both ends to match the shape of the groove, several variants based on 14c were prepared. 14d, in which the m-methoxy group was replaced by a hydroxyl group, had slightly lower activity, suggesting that a hydrophobic moiety at the phenyl m-position may be preferred. 14f was designed to increase the hydrophobicity and occupy more space inside the hydrophobic pocket, by replacing the phenyl ring of 14c with a 6-chloroquinoline. 14i was made as the 5-substituted isomer of 14f. Improved binding was seen for 14i and greater than 2-fold improved inhibition of viral replication. 14f, however, showed both weaker binding and reduced biological activity compared to 14c. A difference in activity between isomers was also observed for 3a and 3g, although in that case, the 6-substituted indole 3a was more active (Table 1).21 This difference indicated that a very specific interaction was occurring between the ligands and the residues in the pocket. 14b and 14e contained an additional 1 or 2 benzyloxy groups to make 4 or 5 ring systems respectively, and both of these showed higher binding affinity. These two compounds have a large number of torsional degrees of freedom (10 and 13 respectively), which likely results in an entropic penalty upon binding. Inspired by the improved activity, we designed 14g and 14h, with the potential for shape complementarity to 14e but with substantially reduced flexibility (8 torsional degrees of freedom). 14g demonstrated sub-µM binding, antifusion and antiviral activity. 14h was also active in the sub- to low µM range. 14g and 14h showed increased toxicity in 0% serum compared to earlier compounds in the series, although their anti-fusion activity could clearly be discerned (Figure S2, Supporting Information) and erum had the effect of protecting the cells without substantially reducing the activity of the compounds.
A strong correlation was obtained between the hydrophobic pocket binding affinity (KI) and the CCF EC50, with R2 = 0.91 (Figure 3, Table 1, Table 2). This result suggested that inhibition of CCF was attributable to binding inside the hydrophobic pocket. This validates the use of the hydrophobic pocket as a target, and verifies that the compounds are fusion inhibitors. The data indicated a 3 – 4 fold reduction in biological activity compared to binding affinity, and suggested an upper limit of ~ 5 µM KI to get substantial biological activity.
A very good agreement between fusion inhibition and inhibition of viral replication was observed for the indole compounds, barring the effect of serum concentration. Binding to serum is significant for many of the compounds, due to the propensity for serum albumin to bind to small lipophilic anionic compounds.28, 29 While this can improve ADME properties, it can also reduce the apparent efficacy of inhibitors that have µM activity, including many of the compounds described here. A clear and steady decrease in activity occurred with increased serum concentration in the assay (Tables 1, ,2).2). Fortunately the effect of serum binding was mitigated for inhibitors with higher affinity for their intended target, observed here for 14g and 14h. The equilibrium appears to be shifted in favor of pocket binding for these inhibitors. It is significant that these molecules have equivalent inhibitory activity in binding, fusion and viral replication, suggesting a clear mechanism of action with no off-target effects.
In the absence of experimental structural information on almost all of the compounds, we have analyzed compound activity by using the experimentally verified orientation of bound 3g 21, and by looking for common binding modes that explained the observed structure-activity relationships. We docked the ligands into two receptor structures used in the NMR study, namely 3p7k, an apo-gp41 structure (NHR, 2.3Å resolution), and 2R5D, a bound D-peptide structure (IQN17 / PIE7, 1.6Å resolution).5 A single dihedral angle rotation on Lys574 Cδ-Cε was applied to 2r5d to rotate the side chain amine towards the pocket and make it accessible to small molecules binding in the pocket. This modification is in line with previous studies which showed the importance of the lysine in pocket interactions with both linear peptides30 and small molecules.15
Most of the compounds were observed to dock in one of two common binding modes with low energy and large cluster size. Examples of the poses are shown in Figure 4A and 4B (receptor 3p7k) and Supporting Information Figure S3 (receptor 2r5d). Very similar binding modes were obtained in both receptors. Compounds with 2 ring systems (3a–g) clustered tightly in pose 4A. They conformed to the expected orientation inferred from the position of 3g determined by NMR experiments. The carboxylate group of 3g formed an electrostatic interaction with Lys574, and the indole NH pointed towards the pocket.21 Based on this structure, we reasoned that the 6-substituted isomer 3a would dock similarly, with the NH pointing away from the pocket. This pose was found universally for 3a – f with 3p7k and 2r5d. Hydrogen bonds were predicted between 3e (p-OH) and the carbonyl oxygen of Leu568’ in 3p7k, and between 3d (m-OH) and the carbonyl oxygen of Val570 in 2r5d. In reality, 3d and 3e have weaker activity than 3a, suggesting that these hydrogen bonds do not form or do not contribute substantially to the binding energy.
Compounds with three ring systems including series 6 and 8 as well as compounds 14c and 14d could also dock in the pose shown in Figure 4A. However, an alternative low energy conformation of series 14 compounds (Table 2) was found in both 3p7k and 2r5d (Figure 4B, Figure S3B) that extended to four and five ring compounds. In this conformation, strong hydrophobic and electrostatic interactions occurred between the benzylcarboxylate and the pocket lysine, and the compounds curved around the pocket. Interestingly, these putative poses of the extended indole ligands followed the same S-shaped contour that is formed by the hydrophobic residues dY6, dW9, dW11 and dL12 of PIE7 in the pocket of 2r5d.31. The overlap is shown for a predicted pose of the longest and one of the most potent indole compounds 14g in Figure 4C. The indole rings occupied the same part of the pocket as dW9 and dL12, the carboxybenzyl group made similar hydrophobic contacts as dW11, and the methoxybenzyl group occupied the same channel as dY.
We observed tight clustering at the N-terminal end of the pocket and a larger variation in the adopted poses of ligand moieties at the C-terminal end. This correlated with a greater degree of variation in pocket side chains at the C-terminal end in different structures, and lower hydrophobic contact with the ligands. It is clear that residue flexibility around the hydrophobic pocket required the inhibitors to adjust to the groove shape. Reciprocal compatibility between inhibitor and side chain flexibility is an important property of gp41 pocket inhibitors.
The orientation in Figure 4A explained the poor activity of compounds 9 and 11, which have a polar substituent at position 3 of the indole ring. This position points towards the pocket, and polar substitution would be likely to destabilize the complex. An orientational rearrangement may occur. However, we did find for these molecules, as well as for molecules with three aromatic moieties and having polar groups at opposite ends (6b, 8b, 8c), that docking predicted a much higher affinity than that observed, which is likely due to the fact that flexibility of Gln and Arg residues at the C-terminal edge of the pocket reduced the contribution of electrostatic interactions calculated for a rigid model.
In Figure 4D, a linear fit of calculated vs. observed binding affinities is shown, excluding molecules that have polar groups at both ends (shown in red). A good correlation was observed, with a correlation coefficient R = 0.85. Calculated energies were not able to distinguish between the poses shown in 4A and 4B for 14c and 14d.
In this study, we have designed, synthesized and tested a series of indole compounds as inhibitors against gp41-mediated HIV-1 fusion. We have observed a strong correlation between hydrophobic pocket binding affinity and CCF fusion inhibition in the sub-µM to hundreds of µM range. The CCF results correlate well with the ability of the compounds to inhibit viral replication. This implies that the hydrophobic pocket is the target of these molecules. The set of compounds has enabled us to establish a direction for optimization. Specifically, we observed that elongating the compound caused significant improvement. Adding defined flexibility so that the compound could conform to the grooves of the coiled coil improved its activity. In general, adding polar groups at both ends reduced activity.
Computational docking suggested common binding modes for groups of inhibitors, and AutoDock4.2 calculations gave a good correlation with the observed relative rank order. Two- and three-ring compounds with polar groups at both ends were an exception, since they were predicted to have high affinity, but they actually were poor binders and fusion inhibitors.
Compounds 14g and 14h with a long hydrophobic interface showed sub-µM binding and fusion inhibition, and 14g also had sub-µM inhibition of viral replication. To our knowledge, this is the best activity yet reported for a verified low molecular weight fusion inhibitor. A putative binding mode mimics the hydrophobic interactions of the D-peptide inhibitor PIE7. The binding models of 14g and other indole compounds in Figure 4 explained the observation that addition of hydrophobic groups while maintaining shape complementarity was required to improve activity. It appears to be a more general property that amphiphilic structures, with separated polar and non-polar ends, provide better inhibition of the gp41 hydrophobic pocket.32 We will conduct lead optimization on this basis. Compounds with improved activity have elevated cLogP values and should be modified to increase solubility and reduce toxicity in consideration of ADMET issues. 29, 33 In addition, improved binding affinity to the hydrophobic pocket must be accompanied by reduced serum binding propensity.29, 34.
Chemical reagents and solvents were used as received. HPLC was performed on a Waters Breeze HPLC with multi λ fluorescence detector and dual λ UV detector, using a water/acetonitrile or water/methanol gradient with 0.05% TFA. Mass spectra were determined by LCMS (Waters micromass ZQ: H2O/acetonitrile gradient, with 0.1% formic acid). 1H NMR spectra were determined at 400MHz in DMSO-d6, unless otherwise stated, using TMS as a reference.
The organic synthesis was carried out according to Schemes 1 – 3. Reaction conditions i – vii are described below. The main scaffold of the compounds (2a–f, 13a–g) was synthesized by Suzuki-Coupling.35 Ullmann reaction was employed to add the phenyl ring at the 1 (NH) position of the indole ring 36. Benzyl bromide intermediates for Suzuki coupling were synthesized by radical substitution using N-bromosuccinimide.37 All compounds were purified by preparative HPLC, which resulted in ≥95% purity for assay testing.
Indole-6-boronic acid 386 mg (2.4 mmol) and methyl-3-(bromomethyl) benzoate 460 mg (2.0 mmol) were added into a 100 ml round-bottomed flask containing THF (15 ml), then 115 mg Pd(PPh3)4 was added, followed by 3 ml 2M K2CO3. The mixture was stirred and heated to reflux under N2 for 4 hours. The reaction was monitored by TLC. After the reaction was complete, the mixture was cooled to room temperature, 10 ml H2O was added and the product was extracted with ethyl acetate (10 ml×3). The organic solvent was combined, dried (anhydrous Na2SO4), filtered and evaporated. The crudeproduct was purified by chromatographic column using hexane: ethyl acetate (7:1) as eluent. 360 mg methyl 3-(1H-indol-6-ylmethyl) benzoate (2a) was obtained as pale yellow solid, yield 67%. MS calcd for C17H15NO2: 265; LCMS:266.01 (M+H)+; 1H NMR δ 10.99 (s, <1H, exch), 7.83(s, 1H), 7.78 (d, J = 7.3 Hz, 1H), 7.55 (d, J =7.3 Hz, 1H), 7.45 (s, 1H), 7.43 (s, 1H), 7.28 (t, J = 3.0 Hz, 1H), 7.22 (s, 1H), 6.87 (dd, J = 7.9, 1.2 Hz, 1H), 6.36 (s, 1H), 4.10 (s, 2H), 3.82 (s, 3H).
20 mg of 2a was dissolved into 4ml THF: methanol (4:1), and 1ml 25% NaOH in H2O was added. The mixture was stirred for 3 hours at room temperature, then adjusted to pH 3.0 using 2M HCl. The solution was extracted with CH2Cl2 (15 ml×3). The organic solvent was combined and dried, then evaporated. The final product was purified by HPLC using an acetonitrile/H2O gradient. After lyophilization, 11.2 mg target compound was obtained as grey powder, yield 60%. 3-(1H-indol-6-ylmethyl)benzoic acid (3a): MS calcd for C16H13NO2: 251; LCMS:252.6 (M+H)+; 1H NMR δ 10.97 (s, < 1H, exch), 7.78 (s, 1H), 7.74 (d, J = 7.8 Hz, 1H), 7.50 (d, J = 7.3 Hz, 1H), 7.38 (m, 2H), 7.29 (m, 2H), 6.95 (d, J = 8.5Hz, 1H), 6.35 (s, 1H), 4.05 (s, 2H).
MS calcd for C17H15NO3: 281; LCMS: 282.6 (M+H)+; 1H NMR δ 10.98 (s, <1H, exch), 7.44 (d, J = 8.5 Hz, 1H), 7.39 (s, 1H), 7.26 (m, 2H), 7.21 (s, 1H), 7.09 (t, J = 1.8 Hz, 1H), 6.88 (dd, J = 8.5, 1.2 Hz, 1H), 6.35 (s, 1H), 4.05 (s, 2H), 3.77 (s, 3H).
MS calcd for C17H15NO3: 281; LCMS: 282.8 (M+H)+; 1H NMR δ 10.93 (s, <1H, exch), 7.80 (m, 1H), 7.65 (s, 1H), 7.42 (m, 1H), 7.25 (s, 1H), 7.17 (s, 1H), 7.08 (d, J = 7.3 Hz, 1H), 6.87 (m, 1H), 6.35 (s, 1H), 4.00 (s, 2H), 3.88 (s, 3H).
MS calcd for C16H13NO3: 267; LCMS: 268.6 (M+H)+; 1H NMR δ 10.97 (s, <1H exch), 9.63 (s, <1H, exch), 7.44 (d, J = 7.9 Hz, 1H), 7.26 (m, 2H), 7.19 (s, 1H), 7.13 (d, J = 1.8 Hz, 1H), 6.85 (m, 2H), 6.35 (t, J = 1.8 Hz, 1H), 3.98 (s, 2H).
MS calcd for C16H13NO3: 267; LCMS: 268.5 (M+H)+; 1H NMR δ 10.85 (s, < 1H, exch), 10.36 (s, < 1H, exch), 7.58 (m, 2H), 7.41 (d, J = 7.9 Hz, 1H), 7.32 (d, J = 7.9 Hz, 1H), 7.13 (d, J = 1.8 Hz, 1H), 7.03 (t, J = 7.9 Hz, 1H), 6.90 (m, 2H), 3.96(s, 2H).
MS calcd for C16H12N2O4: 296.3; LCMS: 297.8 (M+H)+; 1H NMR δ 11.00 (s, < 1H, exch), 7.98 (m, 3H), 7.45 (d, J = 7.9 Hz, 1H), 7.29 (t, J = 3.0 Hz, 1H), 7.13 (s, 1H), 6.80 (dd, J = 7.9, 1.2Hz, 1H), 6.36 (s, 1H), 4.34(s, 2H).
MS calcd for C16H13NO2: 251; LCMS: 252.6 (M+H)+; 1H NMR δ 11.00 (s, <1H, exch), 7.78 (s, 1H), 7.74 (d, J = 7.9 Hz, 1H), 7.51 (d, J = 7.3 Hz, 1H), 7.38 (m, 2H), 7.29 (m, 2H), 6.95 (dd, J = 8.5, 1.2 Hz, 1H), 6.35 (s, 1H), 4.05 (s, 2H).
20 mg 2a was dissolved into 4 ml DMSO, and 18 mg 3-iodoanisole, 12 mg potassium hydroxide and 5 mg Cu2O catalyst were added. The mixture was stirred for 24 hours at 135°C under N2. After cooling to room temperature, the mixture was treated as in general procedure ii. The final product was purified by HPLC using acetonitrile/H2O as eluent. After lyophilization, 3.5 mg of the target compound was obtained as grey powder, yield 13%. MS calcd for C23H19NO3: 357; LCMS: 358.7(M+H)+; 1H NMR δ 7.81 (s, 1H), 7.74 (d, J = 6.8 Hz, 1H), 7.61 (d, J = 3.0 Hz, 1H), 7.58 (d, J = 7.8 Hz, 1H), 7.49 (m, 3H), 7.39 (t, J = 6.8 Hz, 1H), 7.14 (d, J = 7.8 Hz, 1H), 7.08 (s, 1H), 7.01 (d, J = 7.8 Hz, 1H), 6.96 (d, J = 7.8 Hz, 1H), 6.64 (d, J = 2.4 Hz, 1H), 4.12 (s, 2H), 3.82(s, 3H).
MS calcd for C23H17NO4: 371; LCMS: 372.6 (M+H)+; 1H NMR δ 8.02 (s, 1H), 7.95 (br, 2H), 7.86 (d, J = 8.8 Hz, 1H), 7.78 (s, 1H), 7.67 (m, 1H). 7.59 (d, J = 7.8 Hz, 1H), 7.49 (br, 2H), 7.39 (d, J = 7.8 Hz, 1H), 6.68 (br, 2H), 6.56 (s, 1H), 4.11(s, 2H).
32 mg of 2a was dissolved into 4ml anhydrous DMF, and 15 mg sodium hydride in oil (60%) was added. The mixture was cooled in an ice-water bath and stirred for 1 hour at room temperature. 35 mg 4-methoxybenzylbromide was added, and the mixture was stirred at room temperature overnight. TLC indicated no starting material remained, and the reaction was quenched by adding 10 ml water. The solution was extracted with ethyl acetate (15 ml×3) or dichloromethane (15 ml×3). The organic solvent was combined and evaporated, and the product directly used for the next step without further purification. The ester group was saponified, following the general procedure ii to yield the target compound, crude yield 95%. 5.0 mg purified target compound was obtained by HPLC as an off-white powder. MS calcd for C24H21NO3: 371; LCMS: 372 (M+H)+; 1H NMR δ 7.77 (s, 1H), 7.72 (d, J = 7.7 Hz, 1H), 7.49 (d, J = 7.0 Hz, 1H), 7.43 (d, J = 2.5 Hz, 1H), 7.34 (br, 3H), 7.14 (d, J = 8.3 Hz, 2H), 6.95 (d, J = 7.7 Hz, 1H), 6.82 (d, J = 8.3 Hz, 2H), 6.37 (d, J = 2.6 Hz, 1H), 5.27 (s, 2H), 4.04 (s, 2H), 3.67 (s, 3H).
MS calcd for C24H14NO4: 385; LCMS: 386.5 (M+H)+; 1H NMR δ 7.78 (s, 1H), 7.72 (br, 2H), 7.68 (m, 1H), 7.60 (d, J = 7.9 Hz, 1H), 7.45 (d, J = 3.0 Hz, 1H), 7.34 (m, 3H), 7.30 (br, 2H), 7.08 (d, J = 7.9 Hz, 1H), 6.57 (d, J = 2.6 Hz, 1H), 5.42 (s, 2H), 4.09 (s, 2H).
MS calcd for C24H14NO4: 385; LCMS:386.7 (M+H)+; 1H NMR δ 7.78 (d, J = 7.9 Hz, 3H), 7.73 (d, J = 7.9 Hz, 1H), 7.49 (br, 2H), 7.40 (br, 4H), 7.35 (d, J = 8.5 Hz, 1H), 6.97 (d, J = 8.5 Hz, 1H), 6.43 (d, J = 3.0 Hz, 1H), 5.46 (s, 2H), 4.04 (s, 2H).
3 ml anhydrous DMF was cooled in ice-water bath, then 50µl POCl3 was added, the solution was stirred for about 15 minutes, then 32 mg of 2a in 2ml DMF was added. The mixture was heated to 40°C for 3 hours. After cooling to room temperature, the mixture was treated by cold water, following the general procedure ii; 11.2mg target compound was purified as a brown solid, yield 34%. MS calcd for C17H13NO3: 279; LCMS: 280.6 (M+H)+; 1H NMR δ 9.89 (s, 1H), 8.24 (d, J = 3.0Hz, 1H), 7.98 (d, J = 7.3 Hz, 1H), 7.80 (s, 1H), 7.76 (d, J = 7.9 Hz, 1H), 7.52 (d, J = 7.9 Hz, 1H), 7.41 (t, J = 7.9 Hz, 1H), 7.34 (s, 1H), 7.12 (d, J = 7.3 Hz, 1H), 4.13 (s, 2H).
32 mg of 2a was dissolved into 3 ml DMSO, then 56 mg KOH was added;the mixture was stirred at room temperature overnight under N2. Then 50 µl CH2Cl2 was added, the mixture was stirred for 8 hours at room temperature. Then following the general procedure ii, 3.6 mg target compound was purified as orange powder, yield 12%. MS calcd for C33H26N2O4: 514; LCMS: 515.1 (M+H)+ 1H NMR δ 7.84 (s, 2H), 7.79 (s, 2H), 7.74 (d, J = 7.7 Hz, 2H), 7.62 (d, J = 3.2 Hz, 2H), 7.48 (d, J = 7.7 Hz, 2H), 7.39 (d, J = 1.9 Hz, 2H), 7.37 (d, J = 1.9 Hz, 2H), 6.90 (d, J = 8.3 Hz, 2H), 6.57 (s, 2H), 6.35 (d, J = 2.6 Hz, 2H), 4.07 (s, 4H).
32 mg of 2a was treated by 2M oxalyl dichloride in methylene chloride for 4 hours at room temperature. Then following the general procedure ii, part of the product was purified to yield 3.4 mg target compound as a yellow powder, yield 9%. MS calcd for C18H13NO5: 323; LCMS: 324.6( M+H)+; 1H NMR δ 11.52 (s,<1H, exch), 8.36 (d, J = 3.0 Hz, 1H), 8.07 (d, J = 7.9 Hz, 1H), 7.80 (s, 1H), 7.77 (d, J = 7.9 Hz,1H), 7.54 (d, J = 7.9 Hz, 1H), 7.44 (br, 1H), 7.40 (s, 1H), 7.17 (d, J = 7.9 Hz, 1H), 4.13 (s, 2H).
180 mg 6-chloro-2-hydroxy quinoline was treated with 3ml POCl3, the mixture was refluxed for 3 hours; the remaining POCl3 was evaporated away. After cooling to room temperature, the residue was treated with cold water; the solid was filtered and dried, giving a green powder, yield: 90%. MS calcd for C9H5Cl2N: 198; LCMS: 200.1 (M+2)+; 1H NMR in CDCl3 δ 8.04 (d, J = 7.8 Hz, 1H), 7.97 (d, J = 8.8 Hz, 1H), 7.81 (d, J = 1.9 Hz, 1H), 7.69 (dd, J = 9.8, 2.9 Hz, 1H), 7.42 (d, J = 8.8 Hz, 1H)
1H NMR δ 7.73 (s, 1H), 7.51 (br, 2H), 7.17 (d, J = 8.5 Hz, 2H), 7.12 (dd, J = 8.5, 1.2 Hz, 1H), 6.87 (d, J = 8.5 Hz, 2H), 6.49 (d, J = 3.0 Hz, 1H), 5.34 (s, 2H), 3.70 (s, 3H).
MS calc for C14H11NO: 209; LCMS: 210.5 (M+H)+; 1H NMR δ 9.46 (s, <1H, exch), 7.58 (m, 2H), 7.38(t, J = 2.9 Hz, 1H), 7.23 (m, 2H), 7.07 (d, J = 7.8 Hz, 1H), 7.04 (d, J = 1.9 Hz, 1H), 6.71 (dd, J = 7.8, 1.9 Hz, 1H), 6.44 (s, 1H).
MS calcd for C15H13NO: 223; LCMS: 224.6 (M+H)+; 1H NMR δ 11.16 (s, <1H, exch), 7.63 (s, 1H), 7.59 (d, J = 7.9 Hz, 1H), 7.37 (t, J = 3.0 Hz, 1H), 7.34 (d, J = 7.9 Hz, 1H), 7.30 (dd, J = 7.9, 1.8 Hz, 1H), 7.23 (d, J = 7.9 Hz, 1H), 7.17 (br, 1H), 6.89 (dd, J = 8.5, 1.8 Hz, 1H), 6.44 (s, 1H), 3.82 (s, 3H).
1H NMR (400MHz) δ ppm 7.49 (s, 1H), 7.38 (s, 1H), 7.34 (s, 1H), 7.33 (d, J = 1.9 Hz, 1H), 7.28 (t, J = 7.8 Hz, 1H), 7.17 (d, J = 8.8 Hz, 1H), 7.11 (d, J = 2.9 Hz, 2H), 7.07 (d, J = 5.9 Hz, 1H), 6.96 (d, J = 9.8 Hz, 1H), 6.90 (d, J = 8.8 Hz, 2H), 6.87 (dd, J = 8.8, 1.9Hz, 1H), 5.04 (s, 2H), 3.74 (s, 3H).
MS calcd for C11H17ClN2: 278; LCMS: 280.8 (M+2)+; 1H NMR in CDCl3 δ 8.33 (s, 1H), 8.12 (t, J = 8.8 Hz, 2H), 7.99 (d, J = 8.8 Hz, 1H), 7.90 (dd, J = 7.8, 1.9 Hz, 1H), 7.81 (d, J = 1.9 Hz, 1H), 7.78 (d, J = 7.8 Hz, 1H), 7.65 (dd, J = 8.8, 1.9 Hz, 1H), 7.32 (t, J = 2.9 Hz, 1H), 6.62 (s, 1H).
MS calcd for C24H20N2O: 352; LCMS: 335.8 (M+H-H2O); H NMR δ 7.72 (s, 1H), 7.68 (m, 1H),7.62 (s, 1H), 7.61 (d, J = 3.6 Hz, 1H), 7.59 (d, J = 4.3 Hz, 1H), 7.48 (d, J = 3.0 Hz, 1H), 7.34 (s, 1H), 7.31 (d, J = 7.3 Hz, 1H), 7.21 (d, J = 8.5 Hz, 2H), 6.87 (d, J = 8.5 Hz, 2H), 6.47 (d, J = 3.0 Hz, 1H), 6.43 (s, 1H), 5.42 (s, 2H), 3.68 (s, 3H).
MS calcd for C11H17ClN2: 278; LCMS: 279.1 (M+H)+; H NMR in CDCl3 δ 8.45 (s, 1H), 8.11 (m, 3H), 7.99 (d, J = 8.8 Hz, 1H), 7.91 (dd, J = 11.0, 7.8 Hz, 1H), 7.80 (d, J = 1.9Hz, 1H), 7.64 (dd, J = 8.8, 1.9 Hz, 1H), 7.54 (d, J = 8.8 Hz, 1H), 7.19 (t, J = 2.9 Hz, 1H), 6.69 (s, 1H).
MS calcd for C30H23NO5: 477; LCMS: 478.1 (M+H)+; 1H NMR δ 8.07 (s, 1H), 7.91 (d, J = 7.8 Hz, 1H), 7.79 (m, 3H), 7.73 (d, J = 7.8 Hz, 1H), 7.62 (d, J = 7.8 Hz, 1H), 7.56 (d, J = 3.9 Hz, 1H), 7.53 (d, J = 7.8 Hz, 1H), 7.44 (m, 2H), 7.33 (m, 2H), 7.30 (d, J = 1.9 Hz, 1H), 7.25 (d, J = 7.8 Hz, 1H), 6.97 (dd, J = 8.8,1.9 Hz, 1H,), 6.53 (d, J = 2.9 Hz, 1H), 5.60 (s, 2H), 5.26 (s, 2H).
MS calcd for C23H19NO3: 357; LCMS: 358.7 (M+H)+; 1H NMR δ 7.82 (s, 1H), 7.81 (s, 1H),7.77 (s, 1H), 7.62 (d, J = 8.5 Hz, 1H), 7.57 (d, J = 2.4 Hz, 1H), 7.46 (m, 2H), 7.34 (m, 2H), 7.23 (d, J = 7.3 Hz, 1H), 7.18 (s, 1H), 6.88 (dd, J = 7.9, 1.9 Hz, 1H), 6.53 (d, J = 2.9 Hz, 1H), 5.60 (s, 2H), 3.82 (s, 3H).
MS calcd for C22H17NO3: 343; LCMS: 344.1 (M+H)+; 1H NMR δ 8.07(s, 1H), 7.90 (d, J = 7.8 Hz, 1H),7.74 (d, J = 6.8 Hz, 1H), 7.63 (s, 1H), 7.59 (d, J = 8.8 Hz 1H), 7.54 (t, J = 7.8 Hz, 1H), 7.36 (m, 2H), 7.29 (br, 2H), 7.25 (d, J = 7.8 Hz, 1H), 6.97 (d, J = 7.8 Hz, 1H), 6.44 (s, 1H), 5.28 (s, 2H).
MS calcd for C39H31NO6: 597; LCMS: 598.1 (M+H)+; 1H NMR δ 8.02 (s, 1H), 7.89 (d, J = 7.8 Hz, 1H), 7.78 (d, J = 6.8 Hz, 1H), 7.75 (s, 1H), 7.68 (d, J = 7.8 Hz, 1H), 7.57 (br, 2H), 7.50 (t, J = 7.8 Hz, 1H), 7.37 (m, 2H), 7.31 (s, 1H), 7.07 (d, J = 8.8 Hz, 1H), 6.92 (m, 2H), 6.84 (d, J = 2.9 Hz, 1H), 6.69 (m, 5H), 6.54 (d, J = 2.9 Hz, 1H), 5.45 (s, 2H), 5.17 (s, 2H), 3.71 (s, 2H), 3.66 (s, 3H).
MS calcd for C25H17ClN2O2: 412.8; LCMS: 415.1 (M+2)+; 1H NMR δ 8.53 (s, 1H), 8.39 (d, J = 8.5 Hz, 1H), 8.24 (d, J = 8.5 Hz, 1H), 8.10 (dd, J = 8.5, 1.8 Hz, 2H), 8.05 (d, J = 9.2 Hz, 1H), 7.83 (dd, J = 6.7, 1.8 Hz, 1H), 7.79 (s, 1H), 7.75 (m, 1H), 7.62 (m, 2H), 7.45 (m, 2H), 6.68 (d, J = 3.0 Hz, 1H), 5.58 (s, 2H).
MS calcd for C32H26N2O3: 486; LCMS: 487.8 (M+H)+; 1H NMR δ 7.83 (br, 2H), 7.69 (d, J = 8.5 Hz, 2H), 7.60 (t, J = 8.5 Hz, 2H), 7.55 (t, J = 3.0 Hz, 2H), 7.46 (br, 2H), 7.35 (m, 2H), 7.20 (d, J = 8.5 Hz, 2H), 6.84 (d, J = 8.5 Hz, 2H), 6.51 (d, J = 3.0 Hz, 1H), 6.45 (d, J = 3.0 Hz, 1H), 5.60 (s, 2H), 5.39 (s, 2H), 3.66 (s, 3H).
MS calcd for C31H23FN2O2: 474.2; LCMS: 473 (M-H)−; 1H NMR δ 7.82 (br, 2H), 7.69 (d, J = 4.8 Hz, 2H), 7.60 (t, J = 8.8 Hz, 2H), 7.54 (d, J = 2.9 Hz, 1H), 7.51 (d, J = 2.9 Hz, 1H), 7.45 (m, 2H), 7.35 (m, 2H), 7.27 (dd, J = 6.8, 5.8 Hz, 2H), 7.12 (t, J = 8.8 Hz, 2H), 6.51 (d, J = 2.9 Hz, 1H), 6.48 (d, J = 2.9 Hz, 1H), 5.59 (s, 2H), 5.48 (s, 2H).
MS calcd for C25H17ClN2O2: 413; LCMS: 415.3 (M+2)+; 1H NMR δ 8.42 (s, 1H), 8.40 (d, J = 9.2Hz, 1H), 8.27 (d, J = 8.5 Hz, 1H), 8.10 (d, J = 2.4 Hz, 1H), 8.04 (d, J = 8.5 Hz, 2H), 7.81 (br, 2H), 7.72 (m, 2H), 7.67 (d, J = 3.0 Hz, 1H), 7.48 (m, 2H), 6.59 (d, J = 2.4 Hz, 1H), 5.67 (s, 2H).
MS calcd for C15H11NO2: 237; LCMS: 220.1 (M+H-H2O)+; 1H NMR δ 11.21 (s, 1H), 8.21 (s, 1H), 7.93 (d, J = 7.3 Hz, 1H), 7.89 (d, J = 7.3 Hz, 1H), 7.67 (s, 1H), 7.64 (d, J = 7.3 Hz, 1H), 7.58 (t, J = 7.9 Hz, 1H), 7.41 (br, 1H), 7.33 ( d, J = 7.9 Hz, 1H), 6.47 (s, 1H).
Inhibition constants KI for binding in the hydrophobic pocket was determined using a fluorescence intensity assay as previously described.23, 24 Briefly, a metallopeptide receptor structure FeII(env2.0)3 was used to mimic the hydrophobic pocket in the gp41 coiled coil. Env2.0 contains 17 hydrophobic pocket residues flanked by 5 residues on either side. Mixing FeII(env2.0)3 with a fluorescein labeled pocket-binding C-peptide C18-FL caused quenching of fluorescence, which could be reversed in the presence of a competitive inhibitor. Typically, 7µM binding sites (three per receptor trimer) and 7.5 nM C18-FL were used in the assay.
Cell-cell fusion was measured following a published procedure,25 and using cell lines obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH. TZM-bl cells (#8129, contributed by J.C. Kappes, X. Wu and Tranzyme Inc.) expressing CD4, CCR5 and CXCR4, 38 and containing an integrated reporter gene for firefly luciferase under control of HIV-1 LTR 39 were used as target cells. They were grown overnight in 96 well plates in DMEM supplemented with 10% FBS, using 25,000 cells per well. The following day, the medium was exchanged with reduced serum medium (Gibco), and 1µl of compound in DMSO was added to each well, using 6 – 10 serial dilutions to obtain dose response curves. HL2/3 effector cells (#1294, contributed by B.K. Felber and G.N. Pavlakis) which produce HXB2 Env, Tat and Rev40 were added, using 50,000 cells per well, to a total well volume of 100µl. After 6 hours, luciferase expression was measured using Luciferase Assay Reagent (Promega). Controls containing 1µl DMSO with and without HL2/3 cells were measured for each compound, and experiments were performed in duplicate.
Inhibition of HIV-1 replication was determined in CCR5- and CXCR4-tropic MAGI antiviral assays as previously described. 26,27 HIV-1 isolates and cells were obtained from the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH, as follows: HIV-1 Ba-L from Suzanne Gartner, Mikulas Popovic, and Robert Gallo.26,41 HIV-1 IIIB from Robert C. Gallo 26,42 MAGI-CCR5 cells from Dr. Julie Overbaugh 43,44. Briefly, MAGI-CCR5 cells were grown overnight in 96 well plates in DMEM supplemented with 10% FBS, using 10,000 cells per well. The following day the medium was removed and compounds diluted in medium were added (6 dilutions in triplicate at each dilution), followed by the addition of either HIV-1 Ba-L (CCR5-tropic assay) or HIV-1 IIIB (CXCR4-tropic assay) at approximately ten 50% tissue culture infective doses per well (~10 TCID50/well). Assay plates were incubated for 48 hours, after which medium was removed and HIV-1 Tat-induced β-Gal enzyme expression was determined by chemiluminescence using Tropix Gal-Screen (Applied Biosystems) according to the manufacturer’s instructions. For the evaluation of serum effects on compound efficacy, assays for HIV-1 Ba-L were conducted at serum concentrations of 2%, 5%, and 10%.
The cytotoxic effect of the compounds was determined using the identical cell culture procedure to that described above for fusion, but measuring cell death using Cytotox Glo (Promega) instead of luciferase expression or using an Alamar Blue cell viability reagent (Invitrogen). Similarly, the cytotoxic effect of compounds evaluated in the viral replication assay was determined in uninfected MAGI-CCR5 cell cultures prepared and incubated identically to the infected cultures (virus inoculum replaced with media), with cell viability determined using 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H–tetrazolium (MTS; CellTiter®96 Reagent, Promega) following the manufacturer’s protocol. For the evaluation of serum effects on compound cytotoxicity, assays in MAGI-CCR5 cells were conducted at serum concentrations of 2%, 5%, and 10%.
The 3D coordinates of each ligand were generated from their SMILES strings using OMEGA2 and SZYBKI (OpenEye, Inc.). AutoDock4.2 (Scripps Research Institute) was used for docking, according to the procedure provided by the authors.45 The default atom types and parameters supplied with AutoDock4.2 were used throughout this study. The grid dimensions were chosen to be between 15~26Å in each dimension with 0.375Å spacing between grid points, ensuring that the grid was large enough to cover the entire pocket, and fine enough to sample the molecular characteristics of the pocket. 100 – 200 docked conformations were calculated for each inhibitor and clustered according to default settings. Docking of the indole series was performed with two receptor structures, 2r5d and 3p7k, keeping each structure rigid during the docking process. In 2r5d, a dihedral angle change of −126° was applied to the Cδ-Cε bond of the Lys574, rotating the lysine side chain amine to be accessible to small molecules bound in the pocket.
This work was supported by NIH grants NS066469 and AI093243. 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 NIH P41 RR-01081). The authors thank Dr. Eric Springman at Locus Pharmaceuticals (currently Ansaris) for providing the coordinates of 3p7k prior to publication. The authors thank M. Anderson for helpful suggestions on computational simulations and structure comparisons.
Supporting Information Available: This material is available free of charge via the Internet at http://pubs.acs.org.