Fragment hopping approach directed at design of HIV IN-LEDGF/p75 interaction inhibitors

Abstract

We recently identified a series of indole derivatives as active inhibitors of IN-LEDGF/p75 interaction through structure-based pharmacophore models generated from the crystal structure of dimeric catalytic core domain (CCD) of HIV-1 IN in complex with the LEDGF integrase binding domain (IBD). In this paper we used the fragment hopping approach to design small molecules able to prevent the IN-LEDGF/p75 interaction. By means of the proposed approach, we designed novel non-peptidyl compounds that mimic the biological function of some IBD residues and in particular the LEDGF hot spot residues Ile365 and Asp366. The biological results confirmed the importance of several structural requirements for the inhibitory effects of this class of compounds.

Keywords::

• Indoles
• microwave assisted synthesis
• protein–protein interactions
• IN/LEDGF/p75
• HIV

Introduction

Drug discovery and development of new agents active against HIV, based on advances in the understanding of the viral life cycle, have transformed AIDS from a rapid and lethal infection into a chronic condition that can be controlled for many years through combination therapies with different classes of antiviral drugs, known as highly active antiretroviral therapy (HAART)^1,2. Despite progress in the treatment of HIV, the success of therapy is often limited by drug toxicity, development of multidrug-resistance phenotypes, drug–drug interactions, and more worryingly, by the fact that some newly HIV-infected patients carry viruses that are already resistant to the currently approved AIDS treatments. These drug-related side effects, poor patient tolerability and adherence as well as the presence of persistent reservoirs of virus replication, have highlighted the need to develop new anti-HIV drugs with acceptable toxicity and resistance profiles and, more importantly, with novel mechanisms of action³.

In particular, the search for new targets that could be exploited successfully in the design and development of new antiviral drugs has revealed that the inhibition of some protein-protein interactions (PPIs) in the HIV life cycle may provide an important new approach against AIDS and yield anti-HIV drugs with a broader activity spectrum and less chance of resistance^4–8.

Recent studies have evidenced the relevant role of the interactions between HIV-1 integrase (IN) and the cellular cofactor LEDGF/p75 (Lens Epithelium Derived Growth Factor). LEDGF/p75 is a cellular protein that has been identified and validated as a cellular cofactor of HIV integration and replication^9,10. It binds HIV-1 IN via a small (~80 residues) IN-binding domain (LEDGF_IBD, residues 347–429) within its C-terminal region. LEDGF_IBD is both necessary and sufficient for interaction with HIV-1 IN¹¹. The targeted disruption of direct cellular IN cofactors required for viral replication has become a heavily sought-after approach for the development of novel allosteric IN inhibitors for clinical development. So, the association between IN and the cellular cofactor LEDGF/p75 is currently the most promising IN interaction for the design of protein–protein disrupting therapeutics¹².

The availability of the dimeric catalytic core domain (CCD) crystal structure of HIV-1 IN in complex with the LEDGF_IBD enabled the application of different structure-based approaches for the discovery of novel PPIs inhibitors; particularly some peptides and small molecules have already been reported to be inhibitors of the PPI between IN and LEDGF/p75^13–15.

Our previous studies led to the discovery of a small molecule series of indoles (CHIBAs) able to inhibit the interaction between IN and LEDGF/p75^16,17, and among them we found CHIBA-3053 as the most active compound (Figure 1) showing IC₅₀ value of 3.5 µM in AlphaScreen Assay¹⁸.

Figure 1  . Chemical structure of derivative CHIBA-3053.

Herein, we used the “fragment hopping” approach that aims to substitute one fragment of the molecule, in our case the diketo acid portion, which seemed to be responsible for metabolic liability and/or cell toxicity, leaving the rest of the molecule as it is in the original template.

Material and methods

Fragment based pharmacophore generation

A previously reported structure-based pharmacophore model generated by LigandScout starting from the crystal structure of the dimeric CCD of HIV-1 IN complexed to LEDGF_IBD was considered as our starting point. This pharmacophoric hypothesis resulted in 14 chemical features: one H-bond donor (HBD), two H-bond acceptors (HBA1, HBA2), two hydrophobic groups (HY1, HY2) and nine excluded volumes¹⁶. By using Discovery Studio 2.5.5¹⁹ in this study we simplified this model to obtain a fragment-based pharmacophore which overlaps with the diketo acid chain of our CHIBA derivatives. This model consisted of two H-bond acceptors, 4 excluded volumes and a nitrogen atom that we located as anchor point in the exact 3D location of C3 of the diketo acid moiety (location 0.5); this atom has been used to define the linking point between two fragments. Moreover a shape restriction was added considering the space occupied from diketo acid chain.

Virtual screening on 3D building block databases

The fragment-based pharmacophore was used as a 3D query to search suitable amine derivatives into SigmaAldrich database of building blocks (bicyclic systems were excluded), in order to identify matching reagent hits.

A multi-conformation database of 13298 fragments was created in Discovery Studio 2.5.5¹⁹ using Build 3D Database protocol with the default parameter of 100 conformations. The same software was used to perform a 3D pharmacophore search (Search 3D Database protocol). The search method was set to “Best” (that performs a flexible fit of the ligand conformations against the pharmacophore) and Hits were limited to “Best N” (return the N ligands with the best fit values). The 138 hits, obtained as output, were sorted by fit value and analysed. The best fragments were then combined with the indole scaffold to form an amide compound library for docking studies.

Docking studies

All inhibitors used for these studies were constructed using Discovery Studio 2.5 (Accelrys, San Diego, CA)¹⁹ and energy minimized using the Smart Minimizer protocol (1000 steps) which combines the Steepest Descent and the Conjugate Gradient methods. Docking calculations into the LEDGF/p75_IBD binding pocket in the CCD of IN (entry code 2B4J)¹¹ were carried out by GOLD software package version 5.0²⁰ and using the same protocol that we successfully applied in our previous papers^16–18,21.

Chemistry

All commercially available reagents and solvents were used without any further purification. The microwave-assisted reactions were carried out in a CEM Focused Microwave Synthesis System, Model Discover, working at the power necessary for refluxing under atmospheric conditions. Melting points were determined on a BUCHI Melting Point B-545 apparatus and are uncorrected. Elemental analyses (C, H, N) were carried out on a Carlo Erba Model 1106 Elemental Analyzer and the results are within ± 0.4% of the theoretical values. Merck silica gel 60 F₂₅₄ plates were used for analytical TLC; column chromatography was performed on Merck silica gel 60 (230–400 mesh) and Flash Chromatography (FC) on a Biotage SP₁ EXP. ¹H-NMR spectra were recorded in CDCl₃ with TMS as internal standard or [D₆]DMSO on a Varian Gemini-300 spectrometer. Chemical shifts were expressed in δ (ppm) and coupling constants (J) in hertz (Hz). All the exchangeable protons were confirmed by addition of D₂O.

General procedure for the synthesis of 2-(1H-indole-3-carboxamido)acetic acids (2a-b) and 1-(1H-indole-3-carbonyl)pyrrolidine-2-carboxylic acids (3a-b)

Method i: A mixture of 1H-indol-3-carboxylic acid (0.161 mg, 0.001 mol) or 4-methoxy-1H-indol-3-carboxylic acid (0.191 mg, 0.001 mol) and N,N,N′,N′-tetramethyl-O-(1H-benzotriazol-1-yl)uronium hexafluorophosphate (HBTU) (0.001 mol) in dimethylformamide (DMF) (2 mL) was stirred for 30 min at r.t. Successively, glycine (0.075 g, 0.001 mol) or proline (0.115 g, 0.001 mol) and DIPEA (0.129 g, 0.001 mol) were added, and the mixture was stirred overnight. The reaction mixture was then quenched with water (10 mL) and extracted with ethyl acetate (3 × 8 mL). The combined extracts were dried with dry Na₂SO₄ and concentrated in vacuo. The crude product was purified by chromatography using a mixture of chloroform/methanol/acetic acid (9.0/0.8/0.2) and crystallized from ethanol/diethyl ether (3/7).

Method ii: A mixture of 1H-indol-3-carboxylic acid (0.161 mg, 0.001 mol) or 4-methoxy-1H-indol-3-carboxylic acid (0.191 mg, 0.001 mol), N,N,N′,N′-tetramethyl-O-(1H-benzotriazol-1-yl)uronium hexafluorophosphate (HBTU) (0.001 mol) and DIPEA (0.129 g, 0.001 mol) in dimethylformamide (DMF) (2 mL) was placed in a cylindrical quartz tube (diam. 2 cm), stirred at r.t. and irradiated in a microwave oven at the following conditions: 100 W, 50°C, 10 min. Successively, glycine (0.075 g, 0.001 mol) or proline (0.115 g, 0.001 mol) was added and the mixture irradiated at the following conditions: 100 W, 80°C, 10 min. The reaction mixture was then quenched with water (10 mL) and purified following the same procedure of method i.

2-(1H-Indole-3-carboxamido)acetic acid (2a)

Yield: method i: 21%, method ii: 52%; mp 135–137°C; ¹HNMR (DMSO-d₆), δ ppm, 3.78 (d, J = 5.3, 2H, CH₂), 7.05–7.15 (m, 2H, ArH), 7.41 (d, J = 8.5, 1H, ArH), 7.93 (t, J = 5.3, 1H, NH), 8.00 (s, 1H, ArH), 8.09 (d, J = 6.9, 1H, ArH), 11.59 (bs, 1H, NH). Anal. calcd for C₁₁H₁₀N₂O₃: C 60.55, H 4.62, N 12.84, found C 60.83, H 4.48, N 12.66.

2-(4-Methoxy-1H-indole-3-carboxamido)acetic acid (2b)

Yield: method i: 35%, method ii: 52%; mp 178–180°C; ¹HNMR (DMSO-d₆), δ ppm, 4.01 (s, 3H, OCH₃), 4.75 (d, J = 4.7, 2H, CH₂), 6.71 (d, J = 6.4, 1H, ArH), 7.08–7.15 (m, 2H, ArH), 7.88 (d, J = 2.5, 1H, ArH), 9.14 (t, J = 4.7, 1H, NH), 11.76 (bs, 1H, NH), 12.82 (bs, 1H, COOH). Anal. calcd for C₁₂H₁₂N₂O₄: C 58.06, H 4.87, N 11.29, found C 58.22, H 4.61, N 11.07.

1-(1H-Indole-3-carbonyl)pyrrolidine-2-carboxylic acid (3a)

Yield: method i: 69%, method ii: 81%; mp 185–187°C; ¹HNMR (DMSO-d₆), δ ppm, 1.90–2.16 (m, 3H, CH), 2.19–2.25 (m, 1H, CH), 3.65–3.82 (m, 2H, CH), 4.38–4.59 (m, 1H, CH), 7.05–7.17 (m, 3H, ArH), 7.42 (d, J = 7.9, 1H, ArH), 8.05 (s, 1H, ArH), 11.65 (bs, 1H, NH), 12.42 (bs, 1H, COOH). Anal. calcd for C₁₄H₁₄N₂O₃: C 65.11, H 5.46, N 10.85, found C 65.36, H 5.68, N 10.71.

1-(4-Methoxy-1H-indole-3-carbonyl)pyrrolidine-2-carboxylic acid (3b)

Yield: method i: 57%, method ii: 70%; mp 198–200°C; ¹HNMR (DMSO-d₆), δ ppm, 1.75–1.93 (m, 3H, CH), 2.15–2.21 (m, 1H, CH), 3.19–3.38 (m, 1H, CH), 3.42–3.55 (m, 1H, CH), 3.79 (s, 3H, OCH₃), 4.34–4.38 (m, 1H, CH), 6.53 (d, J = 6.9, 1H, ArH), 6.98–7.08 (m, 2H, ArH), 7.35 (s, 1H, ArH), 11.37 (bs, 1H, NH), 12.40 (bs, 1H, COOH). Anal. calcd for C₁₅H₁₆N₂O₄: C 62.49, H 5.59, N 9.72, found C 62.32, H 5.73, N 9.57.

General procedure for the synthesis of 2-(4-hydroxy-1H-indole-3-carboxamido)acetic acid (2c) and 1-(4-hydroxy-1H-indole-3-carbonyl)pyrrolidine-2-carboxylic acid (3c)

Derivatives 2c and 3c were obtained from compounds 2b and 3b, respectively. The starting material (0.001 mol), under nitrogen atmosphere, was dissolved in methylene chloride (DCM) (5 mL), treated with BBr₃ (1 M in DCM), (6 mL, 0.006 mol) and stirred overnight. Successively, methanol (7 mL) was carefully added at 0°C and the solvents removed under reduced pressure. The residue was dissolved in ethyl acetate (10 mL) and washed with water (10 mL × 3). The organic layer was dried (Na₂SO₄), combined and concentrated in vacuo. The crude product was crystallized from ethanol/diethyl ether (3/7).

2-(4-Hydroxy-1H-indole-3-carboxamido)acetic acid (2c)

Yield: 76%; mp 235–237°C; ¹HNMR (DMSO-d₆), δ ppm, 3.96 (d, J = 5.8, 2H, CH₂), 6.37 (d, J = 8.0, 1H, ArH), 6.85 (d, J = 7.9, 1H, ArH), 6.99 (t, J = 8.0, 1H, ArH), 8.15 (s, 1H, ArH), 8.95 (t, J = 5.8, 1H, NH), 11.71 (bs, 1H, NH), 12.31 (s, 1H, OH). Anal. calcd for C₁₁H₁₀N₂O₄: C 56.41, H 4.30, N 11.96, found C 56.64, H 4.22, N 11.78.

1-(4-Hydroxy-1H-indole-3-carbonyl)pyrrolidine-2-carboxylic acid (3c)

Yield: 71%; mp 253–255°C; ¹HNMR (DMSO-d₆), δ ppm, 1.86–1.93 (m, 1H, CH), 2.01–2.12 (m, 2H, CH), 2.21- 2.27 (m, 1H, CH), 3.91–3.97 (m, 2H, CH), 4.48–4.52 (m, 1H, CH), 6.37 (d, J = 7.4, 1H, ArH), 6.84 (d, J = 8.0, 1H, ArH), 7.00 (d, J = 8.0, 1H, ArH), 7.99 (s, 1H, ArH), 11.89 (bs, 1H, NH), 12.42 (s, 1H, OH), 12.57 (bs, 1H, COOH). Anal. calcd for C₁₄H₁₄N₂O₄: C 61.31, H 5.14, N 10.21, found C 61.54, H 5.01, N 10.39.

General procedure for the synthesis of 2-(1-benzyl-1H-indole-3-carboxamido)acetic acids (4-5a-c) and 1-(1-benzyl-1H-indole-3-carbonyl)pyrrolidine-2-carboxylic acids (6-7a-c)

2-(1H-Indole-3-carboxamido)acetic acids (2a-c) or 1-(1H-indole-3-carbonyl)pyrrolidine-2-carboxylic acids (3a-c), DMF (2 mL) and K₂CO₃ (0.276 g, 0.002 mol) were stirred for 5 min. The appropriate benzyl bromide (0.0015 mol) was added dropwise and the resulting mixture stirred at r.t. for 2 h. A saturated aqueous NaHCO₃ solution was added (10 mL). The reaction was extracted with ethyl acetate (3 × 8 mL) and dried over Na₂SO₄. After removal of the solvent under reduced pressure, the residue was powdered by treatment with diethyl ether and crystallized from ethanol or, in some cases, purified by flash chromatography (FC) eluting with dichloromethane/methanol: 9.8/0.2.

2-[1-(4-Fluorobenzyl)-1H-indole-3-carboxamido]acetic acid (4a)

Yield: 40%; mp 139–141°C; ¹HNMR (DMSO-d₆), δ ppm, 4.05 (d, J = 6.3, 2H, CH₂), 5.15 (s, 2H, CH₂), 7.08–7.20 (m, 4H, ArH), 7.43–7.48 (m, 3H, ArH), 8.04 (s, 1H, ArH), 8.11(d, J = 6.9, 1H, ArH), 8.36 (t, J = 6.3, 1H, NH), 11.59 (bs, 1H, COOH). Anal. calcd for C₁₈H₁₅FN₂O₃: C 66.25, H 4.63, N 8.58, found C 66.49, H 4.49, N 8.71.

2-[1-(4-Fluorobenzyl)-4-methoxy-1H-indole-3-carboxamido]acetic acid (4b)

Yield: 60%; mp 71–73°C; ¹HNMR (DMSO-d₆), δ ppm, 3.98 (s, 3H, OCH₃), 4.20 (d, J = 5.3, 2H, CH₂), 5.17 (s, 2H, CH₂), 6.71 (d, J = 6.2, 1H, ArH), 7.08–7.23 (m, 4H, ArH), 7.44–7.48 (m, 2H, ArH), 7.89 (s, 1H, ArH), 9.13 (t, J = 5.3, 1H, NH), 11.79 (bs, 1H, COOH). Anal. calcd for C₁₉H₁₇FN₂O₄: C 64.04, H 4.81, N 7.86, found C 64.28, H 4.69, N 7.98.

2-[1-(4-Fluorobenzyl)-4-hydroxy-1H-indole-3-carboxamido]acetic acid (4c)

Yield: 43%; mp 189–191°C; ¹HNMR (DMSO-d₆), δ ppm, 4.12 (d, J = 5.5, 2H, CH₂), 5.16 (s, 2H, CH₂), 6.39 (d, J = 7.2, 1H, ArH), 6.86 (d, J = 8.3, 1H, ArH), 7.00 (t, J = 8.3, 1H, ArH), 7.15–7.21 (m, 2H, ArH), 7.42–7.47 (m, 2H, ArH), 8.11 (s, 1H, ArH), 9.06 (bs, 1H, NH), 11.72 (bs, 1H, COOH), 12.19 (s, 1H, OH). Anal. calcd for C₁₈H₁₅FN₂O₄: C 63.16, H 4.42, N 8.18, found C 63.41, H 4.33, N 8.34.

2-[1-(3,5-Dimethylbenzyl)-1H-indole-3-carboxamido]acetic acid (5a)

Yield: 43%; mp 156–158°C; ¹HNMR (CDCl₃), δ ppm, 2.32 (s, 6H, CH₃), 4.34 (d, J = 4.9, 2H, CH₂), 5.17 (s, 2H, CH₂), 6.55–6.59 (m, 2H, ArH), 6.99 (s, 2H, ArH), 7.24 (s, 1H, ArH), 7.40–7.43 (m, 1H, ArH), 7.78 (s, 1H, ArH), 8.01–8.04 (m, 1H, ArH), 8.82 (bs, 1H, NH). Anal. calcd for C₂₀H₂₀N₂O₃: C 71.41, H 5.99, N 8.33, found C 71.62, H 5.78, N 8.05.

2-[1-(3,5-Dimethylbenzyl)-4-methoxy-1H-indole-3-carboxamido]acetic acid (5b)

Yield: 95%; mp 207–209°C; ¹HNMR (DMSO-d₆), δ ppm, 2.24 (s, 6H, CH₃), 3.99 (s, 3H, OCH₃), 4.21 (d, J = 5.4, 2H, CH₂), 5.11 (s, 2H, CH₂), 6.71–7.13 (m, 6H, ArH), 7.90 (s, 1H, ArH), 9.14 (t, J = 5.4, 1H, NH), 11.80 (bs, 1H, COOH). Anal. calcd for C₂₁H₂₂N₂O₄: C 68.84, H 6.05, N 7.65, found C 68.69, H 6.24, N 7.43.

2-[1-(3,5-Dimethylbenzyl)-4-hydroxy-1H-indole-3-carboxamido]acetic acid (5c)

Yield: 20%; mp 150–152°C; ¹HNMR (DMSO-d₆), δ ppm, 2.23 (s, 6H, CH₃), 4.13 (d, J = 5.6, 2H, CH₂), 5.10 (s, 2H, CH₂), 6.40 (d, J = 7.7, 1H, ArH), 6.88–7.04 (m, 5H, ArH), 8.13 (s, 1H, ArH), 9.12 (t, J = 5.6, 1H, NH), 11.77 (bs, 1H, COOH), 12.26 (s, 1H, OH). Anal. calcd for C₂₀H₂₀N₂O₄: C 68.17, H 5.72, N 7.95, found C 68.34, H 5.58, N 7.77.

1-[1-(4-Fluorobenzyl)-1H-indole-3-carbonyl]pyrrolidine-2-carboxylic acid (6a)

Yield: 82%; mp 175–177°C; ¹HNMR (CDCl₃), δ ppm, 1.92–2.08 (m, 3H, CH), 2.28–2.35 (m, 1H, CH), 3.76–3.78 (m, 2H, CH), 4.76–4.78 (m, 1H, CH), 5.14 (s, 2H, CH₂), 6.97–7.42 (m, 8H, ArH), 8.10 (d, J = 6.4, 1H, ArH), 8.76 (bs, 1H, OH). Anal. calcd for C₂₁H₁₉FN₂O₃: C 68.84, H 5.23, N 7.65, found C 68.59, H 5.46, N 7.51.

1-[1-(4-Fluorobenzyl)-4-methoxy-1H-indole-3-carbonyl]pyrrolidine-2-carboxylic acid (6b)

Yield: 89%; mp 78–80°C; ¹HNMR (CDCl₃), δ ppm, 1.90–2.07 (m, 3H, CH), 2.18–2.33 (m, 1H, CH), 3.44–3.46 (m, 2H, CH), 3.84 (s, 3H, OCH₃), 4.72–4.86 (m, 1H, CH), 5.20 (s, 2H, CH₂), 6.54 (d, J = 8.0, 1H, ArH), 6.93-–7.41 (m, 7H, ArH), 8.70 (bs, 1H, COOH). Anal. calcd for C₂₂H₂₁FN₂O₄: C 66.66, H 5.34, N 7.07, found C 66.45, H 5.51, N 7.32.

1-[1-(4-Fluorobenzyl)-4-hydroxy-1H-indole-3-carbonyl]pyrrolidine-2-carboxylic acid (6c)

Yield: 25%; mp 164–166°C; ¹HNMR (DMSO-d₆), δ ppm, 1.97–2.11 (m, 3H, CH), 2.28–2.31 (m, 1H, CH), 3.67–3.76 (m, 2H, CH), 4.67–4.72 (m, 1H, CH), 5.18 (s, 2H, CH₂), 6.64–7.37 (m, 8H, ArH), 8.97 (bs, 1H, COOH), 12.15 (s, 1H, OH). Anal. calcd for C₂₁H₁₉FN₂O₄: C 65.96, H 5.01, N 7.33, found C 65.81, H 5.26, N 7.09.

1-[1-(3,5-Dimethylbenzyl)-1H-indole-3-carbonyl]pyrrolidine-2-carboxylic acid (7a)

Yield: 53%; mp 186–188°C; ¹HNMR (CDCl₃), δ ppm, 1.94–2.09 (m, 3H, CH), 2.28 (s, 6H, CH₃) 2.29–2.37 (m, 1H, CH), 3.77–3.82 (m, 2H, CH), 4.78–4.82 (m, 1H, CH), 5.11 (s, 2H, CH₂), 6.96 (d, J = 7.4, 2H, ArH), 7.18–7.43 (m, 5H, ArH), 8.12 (s, 1H, ArH), 8.73 (bs, 1H, COOH). Anal. calcd for C₂₃H₂₄N₂O₃: C 73.38, H 6.43, N 7.44, found C 73.26, H 6.60, N 7.52.

1-[1-(3,5-Dimethylbenzyl)-4-methoxy-1H-indole-3-carbonyl]pyrrolidine-2-carboxylic acid (7b)

Yield: 87%; mp 184–186°C; ¹HNMR (DMSO-d₆), δ ppm, 1.78–1.90 (m, 4H, CH), 2.21 (s, 3H, CH₃), 2.24 (s, 3H, CH₃), 3.26–3.32 (m, 2H, CH), 3.75 (s, 3H, OCH₃), 4.46–4.72 (m, 1H, CH), 5.08 (s, 2H, CH₂), 6.53 (d, J = 7.4, 2H, ArH), 6.89–7.05 (m, 4H, ArH), 7.34 (s, 1H, ArH), 11.41 (bs, 1H, COOH). Anal. calcd for C₂₄H₂₆N₂O₄: C 70.92, H 6.45, N 6.89, found C 70.78, H 6.58, N 6.70.

1-[1-(3,5-Dimethylbenzyl)-4-hydroxy-1H-indole-3-carbonyl]pyrrolidine-2-carboxylic acid (7c)

Yield: 64%; mp 204–206°C; ¹HNMR (DMSO-d₆), δ ppm, 1.91–2.05 (m, 3H, CH), 2.19 (s, 6H, CH₃), 2.28–2.32 (m, 1H, CH), 3.91–3.98 (m, 2H, CH), 4.64–4.69 (m, 1H, CH), 5.09 (s, 2H, CH₂), 6.40 (d, J = 7.7, 1H, ArH), 6.85–7.05 (m, 5H, ArH), 8.03 (s, 1H, ArH), 11.93 (bs, 1H, COOH), 12.32 (s, 1H, OH). Anal. calcd for C₂₃H₂₄N₂O₄: C 70.39, H 6.16, N 7.14, found C 70.51, H 6.29, N 7.01.

LEDGF/p75-HIV-1 Integrase interaction screening (AlphaScreen technology)

The AlphaScreen Assay was performed as previously described²². Reactions were performed in 25 µL final volume in 384-well Optiwell™ microtiter plates (Perkin–Elmer). The reaction buffer contained 25 mM Tris–HCl (pH 7.4), 150 mM NaCl, 1 mM MgCl₂, 0.01% (v/v) Tween-20 and 0.1% (w/v) bovine serum albumin. His6-tagged integrase (300 nM final concentration) was incubated with the compounds at 4°C for 30 min. The compounds were added in varying concentrations from 1 up to 100 nM. Afterward 100 nM of recombinant flag-LEDGF/p75 was added and incubation was extended by another hour at 4°C. Subsequently, 5 µL of Ni-chelate-coated acceptor beads and 5 µL of anti-flag donor beads were added to a final concentration of 20 µg/mL of both beads. Proteins and beads were incubated at 30°C for 1 h in order to allow association to occur. Exposure of the reaction to direct light was prevented as much as possible and the emission of light from the acceptor beads was measured in the EnVision plate reader (Perkin–Elmer, Benelux) and analyzed using the EnVision manager software.

Results and discussion

In previous papers we have disclosed a series of indole derivatives as active inhibitors of IN-LEDGF/p75 interaction in the AlphaScreen Assay^{16–18,23,24}. This class of PPIIs has been identified through structure-based pharmacophore models generated by LigandScout starting from the crystal structure of the dimeric CCD of HIV-1 IN complexed to LEDGF_IBD. Computational tools together with mutagenesis results were used to highlight structural features that are relevant for the interaction between the two proteins. In particular, focusing our interest on the LEDGF/p75 hot spot residues Ile365 and Asp366, we hypothesized that a small molecule could mimic this dipeptide thus inhibiting IN–LEDGF/p75 interaction. The generated pharmacophoric hypothesis consisted of 14 features: one H-bond donor (HBD), two H-bond acceptors (HBA1-HB2), two hydrophobic groups (HY1-HY2) and nine excluded volumes. These features were able to describe the main contacts between the peptide Ile365-Asp366 of LEDGF/p75 and key amino acid residues of IN. In Figure 2 is reported the pharmacophore model superimposed with the docking pose of the first active indole CHIBA-3003 (IC₅₀ value of 35 µM), that was retrieved through in silico screening in our “in-house database” (CHIME). The docking studies confirmed that the prototype CHIBA-3003 was able to mimic the two residues Ile365 and Asp366 in the IN pocket.

Figure 2.  Chemical structure of derivative CHIBA-3003; Docking pose of derivative CHIBA-3003 (A) superimposed to the pharmacophore model (HBD, magenta; HBA1-HBA2, green; HY1-HY2, cyan; excluded volumes, gray) and (B) into the LEDGF/p75_IBD binding pocket of IN.

To acquire more information about the IN-LEDGF/p75 interaction and to identify new inhibitors in the present study we decided to perform the search of new analogs by using a virtual screening strategy.

In order to substitute the 1,3-diketo acid moiety of the prototype CHIBA-3003 with a different functionality we used the fragment hopping approach. This approach aims to substituting one fragment of the molecule, leaving the rest of the structure as it is in the original template and may be useful for better understanding of the structure-activity relationships (SARs).

This method follows the same steps usually applied in classical virtual screening studies: (i) query preparation (in this case, a fragment-based pharmacophore); (ii) generation of the virtual fragment library and (iii) scoring and ranking the possible solutions using also docking simulations. Moreover, in the case of a fragment substitution, one of the key factors to consider is the synthetic feasibility of the suggested fragment replacement. Therefore, any such method needs an accurate definition of the query fragment.

The fragment-based pharmacophore was built considering:

1. Our previously reported pharmacophore model suggestions

2. The docking pose of the compound CHIBA-3003

3. The synthetic pathway of indole derivatives

Therefore we decided to maintain the 1H-indole-3-carbonyl fragment as a fixed portion and to delete all other chemical features that display fitting to the pharmacophore model. Moreover, the two H-bond acceptor features overlapping the carboxylic group were kept and a further shape was added, considering the space occupied from the 1,3-diketo acid chain.

Finally, a nitrogen atom was defined as anchor point in an exact 3D location, obtaining a fragment-based pharmacophore that consists of two H-bond acceptors, 4 excluded volumes and a nitrogen atom (Figure 3).

Figure 3.  Fragment-based pharmacophore generation.

In order to identify matching fragment hits, the obtained pharmacophore model was used as a query to screen the SigmaAldrich database of building blocks containing 13298 molecules.

Among the retrieved 138 hits only the fragments with a fit value higher than 2.9 were considered. Then the filtered 33 fragments were combined with the original indole scaffold to build an amide compound library and a docking procedure was applied to obtain information about the binding mode of these new derivatives. At the end we selected two series of CHIBA-3003 analogues bearing the chain containing a glycine- and proline-derived moiety; compounds 2c and 3c were considered the prototypes of these two series of indole-based derivatives (Figure 4).

Figure 4.  Chemical structure of CHIBA-3003 analogues 2c and 3c.

Figure 5 shows the ability of these two amino acid moieties to fit the fragment-based pharmacophore (panels A and B); the binding modes of the designed indole derivatives 2c and 3c are shown in panels C-D. Noteworthy 2c and 3c displayed a very similar binding mode when compared with prototype CHIBA-3003 (see Figure 2B). The results of these docking studies prompted us to carry out the synthesis of both molecules. Moreover, in an attempt to expand SAR information we synthesized a small class of indole derivatives (2a-b, 3a-b, 4a-c, 5a-c, 6a-c, and 7a-c) bearing the glycine- or proline-derived moiety as well as the same chemical features that characterized our previously reported active inhibitors containing an indole nucleus^16–18. In particular our choice was based on the following assumptions: (i) the substitution at the indole nitrogen atom revealed that the most active inhibitors were characterized by the presence of the 4′-fluorobenzyl or 3′,5′-dimethylbenzyl group; (ii) the 4-OMe, 4-OH as well as 4-unsubstituted derivatives exhibited relevant activity.

Figure 5.  Glycine (A) and proline (B) fitting the fragment-based pharmacophore. Docking poses of (C) derivative resulting from glycine and (D) derivatives resulting from proline (R and S stereoisomers).

The synthetic pathway to obtain indole derivatives 2-7a-c is shown in Scheme 1.

Scheme 1.  Reagents and conditions: (i) HBTU, DIPEA, DMF, Gly or L-Pro, r.t. over-night; (ii) HBTU, DIPEA, DMF, Gly or L-Pro, MW: 80°C, 10 min, 100 W; (iii) BBr₃, N₂, r.t., over-night; (iv) DMF, K₂CO₃, 4-fluorobenzyl bromide or 3,5-dimethylbenzyl bromide, r.t., 1 h.

1H-Indol-3-carboxylic acid (1a) or 4-methoxy-1H-indol-3-carboxylic acid (1b) was treated with N,N,N′,N′-tetramethyl-O-(1H-benzotriazol-1-yl)uronium hexafluorophosphate and N,N-diisopropylethylamine (DIPEA) in dimethylformamide (DMF); then a suitable amino acid derivative (glycine or L-proline) was added to the reaction mixture. To obtain derivatives 2a-b and 3a-b in the first step of the synthetic pathway we initially employed the conventional thermal heating (method i). However, in order to achieve reduction in reaction times, a better yield and a cleaner reaction than the conventional synthetic processes, the same reactions were performed by the application of the eco-friendly microwave-assisted organic synthesis (MAOS), employing a MW reactor equipped with an efficient control system. According to our previously reported findings^25,26, the MAOS approach furnished significant improvement of yields. Moreover, the coupling reaction was carried out using HBTU to minimize the chance of the partial proline racemization that could occur during the formation of amide 3 employing both classical approach and in MAOS conditions. However, no detection of eventual proline racemization process was made following the pharmacophore hypothesis that suggested that both enantiomers could display a very similar interaction through the carboxylic moiety (see Figure 5D). Under an atmosphere of nitrogen, 4-methoxy-substituted derivative 2b and 3b were subsequently treated with boron tribromide (BBr₃) to give corresponding 4-hydroxyl analogs 2c and 3c. Finally, by reaction of the obtained compounds 2a-c and 3a-c with the suitable benzyl bromide in alkaline conditions we prepared the corresponding N-alkyl derivatives 4-7a-c. The structures of all obtained compounds were supported by elemental analyses and spectroscopic measurements.

In order to screen their capability to block the interaction between HIV-1 IN and the cellular cofactor LEDGF/p75, this series of 18 indole derivatives 2-7a-c was tested in an AlphaScreen Assay and the results are shown in Table 1.

Table 1.  Inhibition of IN-LEDGF/p75 interaction.

N°	% Inhibition at 100 µM	N°	% Inhibition at 100 µM
2a	NA	5b	NA
2b	26%	5c	34%
2c	19%	6a	25%
3a	NA	6b	34%
3b	NA	6c	34%
3c	NA	7a	NA
4a	NA	7b	39%
4b	NA	7c	NA
4c	21%	CHIBA-3003	71%
5a	NA	CHIBA-3053	94%

NA, not active.

We found that the newly synthesized compounds generally were not able to produce significant inhibitory effects. The most interesting molecules were derivatives 2b-c, 4-5c, 6a-c and 7b with a percentage of inhibition ranging from 19% to 39% at a fixed dose of 100 µM. However this value was significantly lower than that of prototype CHIBA-3003 (71%). The highest activity was measured for the proline-derivative 7b bearing a N-3,5-dimethylbenzyl-substituent and a methoxy substituent at position C-4 of the indole system, thus confirming that the combination of these two chemical features could positively influence the activity of this class of compounds. In fact these results were in agreement with our previously reported findings for other indole analogs^17,18. Furthermore, even though the computational approach suggested that the carboxylic group of the amino acid residue (glycine or proline moiety) could furnish conceivable interactions for the disruption of IN-LEDGF/p75, the biological results suggested that the 1,3-diketo acid moiety represents a better structural requirement for the occurrence of inhibitory effects of this class of molecules.

Declaration of interest

This work was supported by the European Commission (HEALTH-F3-2008–201032) (THINC project).