7-Azaindolequinuclidinones (7-AIQD): A novel class of cannabinoid 1 (CB1) and cannabinoid 2 (CB2) receptor ligands

Abstract

A series of N-benzyl-7-azaindolequinuclidinone (7-AIQD) analogs have been synthesized and evaluated for affinity toward CB1 and CB2 cannabinoid receptors and identified as a novel class of cannabinoid receptor ligands. Structure–activity relationship (SAR) studies indicate that 7-AIQD analogs are dual CB1/CB2 receptor ligands exhibiting high potency with somewhat greater selectivity towards CB2 receptors compared to the previously reported indolequinuclidinone (IQD) analogs. Initial binding assays showed that 7-AIQD analogs 8b, 8d, 8f, 8g and 9b (1 μM) produced more that 50% displacement of the CB1/CB2 non-selective agonist CP-55,940 (0.1 nM). Furthermore, Ki values determined from full competition binding curves showed that analogs 8a, 8b and 8g exhibit high affinity (110, 115 and 23.7 nM, respectively) and moderate selectivity (26.3, 6.1 and 9.2-fold, respectively) for CB2 relative to CB1 receptors. Functional studies examining modulation of G-protein activity demonstrated that 8a acts as a neutral antagonist at CB1 and CB2 receptors, while 8b exhibits inverse agonist activity at these receptors. Analogs 8f and 8g exhibit different intrinsic activities, depending on the receptor examined. Molecular docking and binding free energy calculations for the most active compounds (8a, 8b, 8f, and 8g) were performed to better understand the CB2 receptor-selective mechanism at the atomic level. Compound 8g exhibited the highest predicted binding affinity at both CB1 and CB2 receptors, and all four compounds were shown to have higher predicted binding affinities with the CB2 receptor compared to their corresponding binding affinities with the CB1 receptor. Further structural optimization of 7-AIQD analogs may lead to the identification of potential clinical agents.

Cannabinoid (CB) receptors are seven transmembrane receptors belonging to the large family of rhodopsin-like class A G-protein-coupled receptors (GPCRs)¹. CB receptors are activated by three major groups of ligands; endocannabinoids, produced endogenously; plant cannabinoids, present in the Cannabis plant; and synthetic cannabinoids, compounds specifically designed and synthesized with high affinity for CB receptors to modulate endocannabinoid function.² Endocannabinoids produce actions via binding and activation of two well-characterized G-protein coupled CB1 and CB2 receptors and share 44% overall homology and 68% homology in their transmembrane domain. CB1 receptors are expressed in the central nervous system (CNS) and are also found in various peripheral tissues including the gastrointestinal (GI) tract, pancreas, liver, kidney, prostate, testis, uterus, eye, lungs, adipose tissue and heart.^{3, 4} CB2 receptors are expressed mainly in the immune system and hematopoietic cells, although these receptors have also recently been observed in the CNS. ^{5, 6}

CB1 receptors are known to play an important role in modulation of analgesia, memory impairment, regulation of appetite, spasmolysis, inhibition of nausea, and lipolysis. ^{4, 7} CB2 selective agonists appear to have potential clinical usefulness for treatment of a range of health conditions, including reduction of acute pain, chronic inflammatory pain, and neuropathic pain⁸, without producing any CB1-mediated CNS side effects.^{9, 10}

The CB1 subtype was first described in 1993 as the major target receptor for the natural product Δ⁹-tetrahydrocannabinol (Δ⁹- THC, Fig. 1), the main psychoactive component of the herbal drug, marijuana, derived from the plant Cannabis sativa.³ In recent decades, a broad range of potent synthetic ligands targeting CB1 receptors have been developed for potential treatment of various diseases, including spasticity and neuropathic pain.^{11, 12} However, drugs in this class also produce undesirable psychotropic effects¹³, including dizziness, dry mouth, tiredness/fatigue, muscle pain and palpitations.^{9, 10} These adverse effects are major limitations for development of therapeutic compounds directly activating CB1 receptors.

Figure 1.

Structures of Δ⁹-THC and Δ⁹-THC analogs Nabilone and Nabiximol

Many studies to develop synthetic cannabinoids for clinical use have been directed towards optimizing the structure of Δ⁹-THC (also known as dronabinol).^{14, 15} Drugs in this class include nabilone,¹⁵ and nabiximol¹⁷ (Fig. 1) which have emerged as potent cannabinoid drugs used clinically.

In addition to Δ⁹-THC and its structural modifications, recent studies have led to the synthesis of many other classes of cannabinoid ligands, including non-classical^{18, 19} or synthetic compounds with affinity for both CB1 and CB2 receptors^20-23, preclinically developed to treat a broad range of diseases and pathological conditions, including neuroinflammation,²⁴ pain,⁹ bone disorders,²⁵ cardiovascular disorders,^26-29 gastrointestinal tract distress,³⁰ and cancer.³¹

The indole nucleus was found to be a key structural element in many synthetic cannabinoids, and was important for conferring high binding affinity for both CB1 and CB2 receptors. Examples of such compounds include IPDN,³² AM1241,²³ GW-405833,³³ JWH-015,³⁴ JWH-081,³⁵ JWH-122,³⁵ JWH-267,³⁶ A796260³⁷ and the cannabimimetic compounds JWH-250,³⁸ and AM-2233,³⁸ (Fig. 2) and other indole derivatives.^{11, 39, 40} In previous studies, our laboratory has reported a novel class of CB ligands possessing an indole nucleus connected to a quinuclidinone moiety.^{41, 42} These compounds were shown to bind to both CB1 and CB2 receptors with high affinity in the nanomolar (nM) range.⁴¹ Furthermore, we have functionally characterized these indolequinuclidinone (IQD) compounds, and have shown that most IQD analogs act as neutral CB1 antagonists or CB2 agonists.⁴²

Figure 2.

Potent indole-containing structural scaffolds that exhibit high affinity for CB1 and CB2 receptors

The azaindole nucleus is another structural element which has been reported to play a key role in improving affinity and activity at CB receptors, with examples of drugs in this class being GSK-554418A,⁴³ 5-fluoro-PCN,⁴⁴ and 3-naphthoyl-N-pentyl-7-azaindole (Fig. 3). The 5-, 6- and 7-azaindole bioisosteres of indole have been incorporated into drug molecules to reportedly improve the physicochemical properties of drugs containing indole moieties.⁴⁵ Azaindoles possess advantageous physicochemical properties due to the combination of an electron-deficient pyridine ring fused with a pyrrole ring, putatively leading to advantages in druglikeness and bioavailability.^{46, 47} In this respect, 3-naphthoyl-N-pentyl-7-azaindole has been reported as a potent cannabinoid agent which inhibits excitatory synaptic transmission in an electrophysiological brain slice model of cannabinoid receptor-modulated neurotransmission (Fig. 3).⁴⁵ Based on these reports, the current study explored the synthesis and evaluation of novel series of N-benzyl-7-azaindolequinuclidinone (7-AIQD) analogs as potent ligands for CB1 and CB2 receptors.

Figure 3.

Potent azaindole scaffolds that interact with CB1 and CB2 receptors

The synthetic route for the preparation of 7-AIQD analogs (7a, 7b and 8a-8g, 9a, 9b) is given in Scheme 1. Initially, 1-acetyl azaindole-3-carboxaldehyde (2) was prepared by N-acetylation of 7-azaindole-3-carboxaldehyde (1) with acetic anhydride in presence of triethylamine in dichloromethane under reflux conditions. Aldol condensation of N-acetyl-7-azaindole-3-carboxaldehyde (2) with quinuclidin-3-one hydrochloride (3) in the presence of freshly prepared lithium diiso-propyl amide (LDA) in tetrahydrofuran at −78°C afforded the (Z)-2-((1H-pyrrolo[2,3-b]pyridin-3-yl)methylene)quinuclidin-3-one derivative (4). In situ N-deacetylation was also observed under these aldol condensation reaction conditions. Utilizing this synthetic procedure a series of 7-AIQD analogs, 7a, 7b and 8a-8g were prepared in 76-85% yield by reaction of 4 with a variety of alkyl (5a, 5b) and benzyl bromides (6a-6g) under catalytic phase-transfer conditions (PTC) using benzyltriethylammonium chloride in a mixture of 50% w/v aqueous NaOH solution and dichloromethane (Scheme 1).⁴⁸

Scheme 1.

Synthesis of 7-azaindolequinuclidinone analogs (7-AIQD; 7a, 7b and 8a-8g, 9a, 9b). Reagents and conditions: (a) Ac₂O, Et₃N, DCM, reflux; (b) quiniclidin-3-one hydrochloride (3), THF, LDA, −78°C; (c) 4-(2-chloroethyl)morpholine (5a) and 1-(2-chloroethyl)pyrrolidine (5b), 50% NaOH (aq), benzyltriethylammonium chloride, DCM, rt; (d) 50% NaOH (aq), benzyltriethylammonium chloride, DCM, rt; (e) NaBH₄, methanol, rt.

Reduction of 8a and 8b utilizing sodium borohydride in methanol at room temperature afforded the racemic quinuclidinol analogs 9a and 9b in 88% and 86% yield, respectively (Scheme 1). All the above novel 7-AIQD derivatives 7a and 7b, 8a-8g, and 9a and 9b were characterized by ¹H, ¹³C NMR and HRMS spectroscopic analysis (see Supporting Information).

Initial receptor binding screens for the above 7-AIQD analogs at both CB1 and CB2 receptors were conducted and are presented in Table 1 (See Supporting Information for binding assay conditions). Initial binding screens examining the ability of a single analog concentration of 1 μM to compete for receptor binding with the high affinity CB1/CB2R agonist [³H]CP-55,940 (Table 1), allowed rapid determination of an approximate affinity of all compounds tested at the two cannabinoid receptors. If the 1 μM concentration of an analog tested produced greater than 50% displacement of a 0.1 nM concentration of [³H]CP- 55,940, this indicated that that compound exhibited a relatively high sub-micromolar affinity for CB1 and/or CB2 receptors. For example, employing the conditions used for this screen, it would be predicted by the Cheng-Prusoff equation that the concentration of a compound producing 50% displacement of [³H]CP-55,940 from a receptor will estimate the compounds affinity for that receptor.

Table 1.

Binding screen of 7-AIQD analogs at CB1 and CB2 receptors

Compound	% Displacement of [3H]CP-55,940 (0.1 nM)
	CB1Rs	CB2Rs
7a	0.00 ± 8.14 (3)	7.92 ± 1.01 (3)
7b	39.4 ± 7.29 (3)	56.6 ± 2.55 (3)
8a	15.1 ± 3.13 (4)	79.9 ±1.87 (4)
8b	68.9 ± 3.45 (4)	74.8 ± 0.92 (4)
8c	22.4 ± 2.81 (4)	27.5 ± 2.39 (4)
8d	55.7 ± 2.47 (4)	64.1 ± 3.86 (4)
8e	42.5 ± 1.56 (4)	56.7 ± 2.91 (4)
8f	74.2 ± 1.39 (4)	67.6 ± 2.05 (4)
8g	87.6 ± 4.80 (3)	100.0 ± 2.10 (3)
9a	0.00 ± 4.02 (4)	0.00 ± 3.55 (4)
9b	58.2 ± 0.87 (4)	58.5 ± 0.81 (4)

^aK_i values of shaded analogs for CB1 and CB2 receptors were determined and are presented in Table 2.

Data presented in Table 1 show that a 1μM concentration of a 7-AIQD analog containing a N-substituted 2-morpholinoethyl moiety (7a) produce little to no [³H]CP-55,940 displacement, predicting this compound exhibits no measurable affinity for either CB1 or CB2 receptors. Another analog, containing a N-substituted 2-pyrrolidinoethyl moiety (7b) exhibited only a small amount of [³H]CP-55,940 displacement from CB1 or CB2 receptors when compared to the N-benzyl-7-AIQD compounds 8a-8g. For example, a 1μM concentration 7-AIQD analogs 8b, 8d, 8f, 8g and 9b produced more that 50% [³H]CP-55,940 displacement from both CB1 and CB2 receptors, predicting that these compounds should exhibit sub-micromolar affinity for cannabinoid receptors.

The 7-AIQD analogs 8a, 8b, 8f and 8g were selected for additional dose-response experiments employing full competition binding curves to determine K_i values of these compounds at CB1 and CB2 receptors (Table 2; Fig. 4). The affinity of these 7-AIQDs for CB1 and CB2 receptors was derived by employing the Cheng-Prusoff equation which converts the observed IC₅₀ values into K_i values from 3-4 separate competition receptor binding curves for each ligand (Fig. 4). These studies showed that while analog 8f displayed relatively equivalent affinity for CB1 (1107 nM) and CB2 (625 nM) receptors, analogs 8a, 8b and 8g exhibited high nanomolar affinity (110, 115 and 23.7 nM, respectively) and selectivity (26.3, 6.1 and 9.2-fold, respectively) for binding to CB2 relative to CB1 receptors, respectively (Table 2). The reduced 7-AIQD analogs 9a and 9b, exhibited decreased binding affinity for both CB1 and CB2 receptors compared to their parent compounds.

Table 2.

Affinity (K_i values)^a of selected 7-AIQD analogs for CB1 and CB2 receptors

Compound	Ki Values (nM)		CB2-Selectivity
	CB1Rs	CB2Rs	(CB1-Ki / CB2-Ki)
8a	2898 ± 370 (4)	110 ± 6.8 (4)	26.3-fold
8b	704 ± 139 (4)	115 ±41.9 (4)	6.1-fold
8f	1107 ± 229 (4)	625 ± 111 (4)	1.8-fold
8g	218 ± 37.0 (3)	23.7 ± 4.5 (4)	9.2-fold

^aKi values for CB1 and CB2 receptors were determined for shaded analogs in Table 1 exhibiting favorable initial single dose receptor binding results.

Figure 4.

Affinity (Ki values) of selected 7-AIQD analogs (8a, 8b, 8f and 8g) for CB1 (open symbols) and CB2 (filled symbols) receptors.

Finally, to determine whether the 7-AIQD analogs 8a, 8b, 8f and 8g acted as agonists, antagonists or inverse agonists at CB1 and CB2 receptors, the ability of each analog to modulate G-protein activity was examined (Fig. 5). Both CB1 and CB2 receptors are G-protein coupled receptors that produce intracellular effects via interaction with the Gi/Go-subtype of G-proteins. Upon binding to CB1/CB2 receptors, agonists produce activation of G-proteins that can be quantified in membrane preparations by measuring increases in agonistinduced binding of [³⁵S]GTPγS, a nonhydrolyzable GTP analogue. Agonists increase, while inverse agonists decrease and neutral antagonists have no effect on [³⁵S]GTPγS binding. In these experiments, the ability of a single high (near receptor saturating) 10 μM concentration of each compound to modulate [³⁵S]GTPγS binding in CHO-hCB₁ (for hCB1Rs; Fig. 5, upper panel) or CHO-hCB₂ (for hCB2Rs; Figure 5, lower panel) cell membranes was examined.

Figure 5.

Modulation of G-protein activity via CB1 (open bars) and CB2 (filled bars) receptors by selected 7-AIQD analogs 8a, 8b, 8f and 8g.

*Significantly different from basal binding (p<0.05; One-sample, t-test)

As anticipated, the CB1/CB2 agonist CP-55,940 significantly increased [³⁵S]GTPγS binding approximately 40% in both hCB1 and hCB2-expressing membranes. In addition, the CB1 receptor inverse agonist AM-251 (in CHO-hCB1 membranes) and the CB2 receptor inverse agonist AM-630 (in CHO-hCB2 membranes) both significantly reduced [³⁵S]GTPγS binding below basal levels, indicative of inverse agonist activity. Incubation of 8a with either CB1 or CB2 receptor expressing membranes produced non-significant effects on basal [³⁵S]GTPγS binding. These results indicate that analog 8a likely lacks significant agonist or inverse agonist activity and may be acting as a neutral antagonist at both hCB1 and hCB2 receptors. In contrast, the 7-AIQD analog 8b significantly reduces [³⁵S]GTPγS binding below basal levels in both CB1 and CB2 receptor expressing membranes, indicative of inverse agonist activity at both receptors. Analog 8f appears to act as an inverse agonist at hCB1 receptors, while exhibiting neutral antagonist properties at hCB2 receptors. Finally, analog 8g would be predicted to exhibit neutral antagonist and partial agonist properties at hCB1 and hCB2 receptors, respectively.

To understand the CB2 receptor-selective mechanism at the atomic level, we performed molecular docking and binding free energy calculations for compounds 8a, 8b, 8f, and 8g. The crystal structure of human CB1 receptor complexed with MDMB-Fubinaca (PDBID: 6n4b), which was obtained from the RCSB Protein Databank, was used to perform molecular simulation.⁴⁹ The structure of human CB2 receptor was constructed by using the CB1 receptor structure as a template for homology modeling with the Modeler software.⁵⁰ Software AutoDock v4.2 was applied to dock these compounds into the receptors. ⁵¹ The molecular mechanics-generalized Born surface area (MM/GBSA) method was used to estimate the binding free energies.⁵² According to the calculated binding free energies summarized in Table 3, compound 8g had the highest binding affinity with both CB1 and CB2 receptors, and all of these compounds had a higher binding affinity with the CB2 receptor compared to the corresponding binding affinity with the CB1 receptor, which is qualitatively consistent with the experimental data listed in Table 2. The consistency between the computational and experimental data suggests that the computationally obtained binding structures were reasonable.

Table 3.

Binding free energies (kcal/mol) obtained from the MM/GBSA calculations for compounds 8a, 8b, 8f and 8g

Compound	ΔGcal(CB1R)	ΔGcal(CB2R)	ΔΔGcala
8a	−27.9	−35.4	−7.5
8b	−32.1	−34.6	−2.5
8f	−32.0	−32.3	−0.3
8g	−32.2	−36.9	−4.7

^aΔΔG_cal = ΔG_cal(CB2R) – ΔG_cal(CB1R)

The structures of CB1 and CB2 receptor binding with 8a were superimposed and are shown in Fig. 6. According to the binding mode, there is a conserved hydrogen bond between the carbonyl of the quinuclidin-3-one moiety and the Nε atom of His178/95^2.65 in the (CB1/CB2) receptor. Also, 8a forms T-π interactions with Phe200/117^3.36and Phe268/183^ECL2, and π-π interaction with Trp279/194^5.43 in the CB1/CB2 receptor. However, there was a deflection of 8a in the CB1 receptor reducing the binding affinity because of the steric effect of Leu193^3.29 and Leu359^6.51. The longer side chain of Leu359^6.51 makes the ligand move close to Phe200^3.36. Furthermore, the phenyl ring flips under the influence of Leu193^3.29. Thus, the para-substituent of the phenyl ring near to helix 5 significantly affects activity and selectivity. For the CB2 receptor, the distance between the phenyl ring and helix 5 is 2.8 Å. With increasing size of the phenyl substituent, the potential steric clash between the ligand and helix becomes stronger, decreasing the binding affinity. For the CB1 receptor, the distance between the para-substituent on the phenyl ring and helix 5 is longer than that observed with the CB2 receptor. Therefore, as the size of the phenyl substituent increases, the binding affinity becomes stronger and then weaker, indicating that this is an interesting site for improving selectivity. In other words, the L359/V261^6.51 and L193/I110^3.29 residues of the CB1/CB2 receptor determine the binding affinity and selectivity of the various 7-AIQD analogs.

Figure 6.

The binding modes of compounds A) 8a and B) 8g in the superimposed pockets of CB1 and CB2. The cartoons of CB1 and CB2 were colored in light pink and light blue. The important amino acid residues in the superimposed pockets of CB1 and CB2 were colored in hot pink and blue. Each ligand was illustrated in stick colored format colored with magenta and cyan at CB1 and CB2.

In summary, a series of N-benzyl-7-azaindolequinuclidinone (7-AIQD) analogs are identified as a novel class of CB1/CB2 receptor ligands. 7-AIQD analogs are dual CB1/CB2 receptor ligands exhibiting rather greater selectivity towards CB2 receptors compared to the previously reported indolequinuclidinone (IQD) analogs. 7-AIQD analogs 8b, 8d, 8f, 8g and 9b at 1 μM concentration produced more that 50% displacement of the CB1/CB2 non-selective agonist CP-55,940 (0.1 nM) in binding assays. Compounds 8a, 8b and 8g exhibited the highest affinity for CB2 receptors with IQ values of 110, 115 and 23.7 nM, respectively. Compounds 8a and 8g were identified as more selective compounds than other analogs in the 7-AIQD series, exhibiting 26.3 and 9.2 fold selectivity, respectively, for CB2 over CB1 receptors. Modulation of G-protein activity confirmed that compound 8a acts as a neutral antagonist, and 8b exhibits inverse agonist activity at CB1 and CB2 receptors. Compounds 8f and 8g exhibit different intrinsic activities, depending on the receptor examined. For the most active compounds 8a, 8b, 8f, and 8g binding affinity was also studied utilizing molecular docking and binding free energy calculations to better understand the CB2 receptor-selective nature of the compounds. These four analogs afforded higher predicted binding affinities with the CB2 receptor compared to their corresponding binding affinities with the CB1 receptor, which is consistent with the experimental data. Collectively, these initial studies suggest that further structure optimization of 7-AIQD analogs may lead to the identification of cannabinoid agents with clinical potential.