Synthesis and In Silico Evaluation of GABA, Pregabalin and Baclofen N-Heterocyclic Analogues as GABA_B Receptor Agonists

Abstract

γ-amino butyric acid (GABA) is an inhibitory neurotransmitter whose deficiency has been associated with various neurological disorders. However, its low liposolubility limits its use as a supplement. Thus, multiple investigations have focused on searching for lipophilic GABA analogs that can modulate the activity of the GABA_B receptor, which could be associated with the etiology of some central nervous system disorders. The GABA analogs available on the market are Vigabatrin, Gabapentin as well as Pregabalin and Baclofen. In this work, we report on the synthesis of GABA analogs, taking the scaffold of GABA, Pregabalin, and Baclofen as a starting point. The analogs include structural features that could favor the affinity of the molecules for the GABA_B receptor, such as heterocyclic rings in the γ-position and alkyl or p-Cl-phenyl substituents (in analogy to Pregabalin and Baclofen, respectively). These analogs were synthesized by a sequence of reactions involving an N-alkylation, a 1,4-conjugated addition of dialkyl and diarylcuprates and a basic hydrolysis. Furthermore, a computational molecular docking over the GABA_B receptor was performed to evaluate the interaction of each compound in the Baclofen binding site. With this information, we evaluated our compounds as GABA_B agonists through a QSAR analysis. Finally, by means of molecular similarity analysis, and in silico ADME prediction, we support our three best compounds (8a–b, 8d) as potential GABA_B receptor agonists.

1. Introduction

γ-aminobutyric acid (GABA) 1 (Figure 1), is the most important inhibitory neurotransmitter of the central nervous system (CNS), where its main function is to reduce neuronal excitability [1]. This inhibitory activity is expressed through the interaction with the GABA_A, GABA_B, and GABA_C receptor subtypes [2]. These interactions cause the opening of specific ion channels that allow for the entry of chloride ions and the exit of potassium ions, which causes a change in the transmembrane potential, inducing hyperpolarization [3,4].

GABA_B receptors belong to the family of G-protein-coupled receptors [5]. These are functional as heterodimers composed of two subunits, GABA_B1 and GABA_B2; each GABA_B receptor subunit contains two different domains, called LB1 and LB2. The GABA_B1 subunit contains the γ-aminobutyric acid (GABA) binding site, while the GABA_B2 subunit, located inside the cell surface, is responsible for the G-protein activation. In response to GABA binding, GABA_B receptors regulate slow and prolonged synaptic transmission [6]. Currently, no compounds have been found that interact directly with the GABA_B2 subunit; however, its domain interacts directly with the GABA_B1 subunit domain to increase the affinity of agonist ligands [7,8,9].

The only FDA-approved agonist ligand of GABA_B receptor is Baclofen (2), a GABA analog substituted in β-position with a p-chlorophenyl ring, which is indicated to treat muscle spasticity in patients with multiple sclerosis and spinal cord injury [10,11]. Baclofen is the prototype drug in the design of potential agonists of the GABA_B receptor; this drug presents diverse limitations since it cannot penetrate the blood–brain barrier (BBB) in a passive way, and due to its short half-life, a larger number of doses are required [12]. Currently, there are some structural analogs of Baclofen such as phenibut, a neuropsychotropic drug that acts as GABA-mimetic, primarily at the GABA_B receptor [13], Arbaclofen placarbil, a prodrug of active (R)-Baclofen, which overcomes the pharmacokinetic limitations of racemic Baclofen [14], and 2-hydroxysaclofen, which can abolish nicotine-induced hypolocomotor effects and increase the antinociceptive effects [15]. Other agonists of the GABA_B receptor are 3-APPA (CGP27492), 3-APMPA (SKF97541), CGP44532, CGP47656 [16], and lesogaberan (AZD3355), which has been evaluated in the treatment of gastroesophageal reflux disease [17]. Additionally, there are some drugs reported as low-affinity antagonists of the GABA_B receptor, such as CGP54626 and CGP56999A, whereas GABA phosphinic acids analogs have presented high-affinity antagonist profile [12,18]. Those mentioned above have shown that the modulation of GABA_B receptors could be involved in the treatment of these disorders, such as epilepsy, Huntington’s disease, and Parkinson, among others [19].

Given the low lipophilicity of GABA, most of the analogs that have been synthesized and reported in the literature include structural modifications with functional groups whose purpose is to increase its lipophilicity, such is the case of Pregabalin (3) [20], Vigabatrin (4) [21], and Gabapentin (5) [22] (Figure 1). In fact, Pregabalin and Baclofen include an isobutyl and phenyl group in the β-position, respectively, which increase their lipophilicity [20,23,24]. Because of this, we included these modifications in the proposed analogues in this work, as well as heterocyclic ring systems in the γ-position. Since nitrogen heterocycles represent a highly important class of compounds in medicinal chemistry [25,26], in recent years, our research group has explored the synthesis of GABA analogs, in which the nitrogen atom in the γ position is part of different heterocycle systems [27,28,29,30]. The use of heterocycles in the design of drugs to improve their potency and selectivity, in some cases, can be explained by their ability to participate in hydrogen bonding with the therapeutic target, either as hydrogen bound-acceptors, (HBAs), or donors, (HBDs) [31], and to help in the modulation of some pharmacokinetic properties such as their lipophilicity, polarity, and solubility, among others [32,33,34].

In silico approaches to the drug design of agonists and antagonists of GABA_B receptor include ligand-based and structure-based methods, such as a combination of docking and QSAR studies, and quantum simulations [12,35,36,37].

In this work, we report the synthesis and in silico evaluation of the new heterocyclic GABA 8a–d, and the Pregabalin 13a–d and Baclofen 14a–d-related analogs, which include heterocyclic ring systems in the γ-position. These compounds were obtained following a previously reported protocol [25,26], which was tested in a computational model to predict their activity as GABA_B receptor agonists.

2. Materials and Methods

2.1. General Information

All reagents were of analytical grade (Aldrich Chemical Company, Milwaukee, WI, USA) and were used as received. Diethyl ether (Et₂O) and tetrahydrofuran (THF) were dried and purified by distillation from a Na/benzophenone system. Hexane, dichloromethane (CH₂Cl₂), and ethyl acetate (AcOEt) were distilled before use. Methanol (MeOH) and acetonitrile (CH₃CN) were of reagent grade and used without prior treatment. Column chromatography on silica gel was carried out on Merck Kieselgel 60 (230–400 mesh). ¹H and ¹³C{1H} NMR spectra were recorded on a Bruker AVANCE III HD 500 MHz (11.74 T), Varian Innova 400 MHz (9.4 T), Varian Mercury 200 (4.7 T), Varian Mercury 400 MHz (9.4 T), or Varian VNMRS 700 MHz (16.45 T) spectrometer in CDCl₃, DMSO-d₆, or CD₃OD (with TMS as the internal standard) at room temperature. Used abbreviations: s = singlet, d = doublet, t = triplet, q = quartet, p = pentet, dt = doublet of triplets, dq = doublet of quartets, m = multiplet, bs = broad signal. High-resolution mass spectra (HMRS), using FAB+ or EI+, were measured using a JEOL MS-700 spectrometer with m/z values accurate to four decimal places.

2.2. General Method for the N-Alkylation Reaction. Synthesis of Compounds 7a–d

A solution of 1.0 equivalent 6 in 7.0 mL of CH₃CN was treated with DIPEA (1.5 equivalents). The reaction mixture was stirred at room temperature for 5 min and then the corresponding heterocyclic system (a–d) 1.1 equivalents were added. The mixture was stirred at room temperature for 24–72 h. The solvents were evaporated under reduced pressure and the residue obtained was dissolved in 10 mL of water and then extracted with CH₂Cl₂ (3 × 20 mL). The organic layer was dried over anhydrous Na₂SO₄, filtered, and concentrated under reduced pressure. The crude product was purified by flash chromatography, giving the corresponding ester 7a–d.

Methyl 4-(3,4-dihydroisoquinolin-2(1H)-yl)butanoate (7a): Compound 7a was obtained as a yellow oil (0.39 g, 75%). ¹H NMR (CDCl₃, 200 MHz): δ 7.16–6.97 (m, 4H), 3.64 (s, 3H), 3.61 (s, 1H), 2.89 (t, 2H, J = 5.8 Hz), 2.71 (t, 2H, J = 5.8 Hz), 2.53 (t, 2H, J = 7.1 Hz), 2.40 (t, 2H, J = 7.3 Hz), 1.92 (q, 2H, J = 7.19 Hz). ¹³C{¹H} NMR (CDCl₃, 50 MHz): δ 174.2, 134.9, 134.4, 128.7, 126.6, 126.1, 125.6, 57.5, 56.1, 51.6, 50.9, 32.1, 29.2, 22.5. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₄H₂₀NO₂: 234.1494; found: 234.1483.

Methyl 4-(1H-benzo[d]imidazol-1-yl)butanoate (7b): Compound 7b was obtained as a yellow oil (1.38 g, 50%). ¹H NMR (CDCl₃, 200 MHz): δ 7.8 (s, 1H), 7.76–7.71 (m, 1H), 7.39–7.15 (m, 3H), 4.17 (t, 2H, J = 6.8 Hz), 3.59 (s, 3H), 2.31–2.19 (m, 2H), 2.18–2.03 (m, 2H). ¹³C{¹H} NMR (CDCl₃, 50 MHz): δ 172.8, 143.8, 142.9, 122.9, 122.1, 120.4, 109.6, 51.8, 43.8, 30.4, 24.9. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₂H₁₅N₂O₂: 219.1134; found: 219.1143.

Methyl 4-(6,7-dihydrothieno [3,2-c]pyridin-5(4H)-yl)butanoate (7c): Compound 7c was obtained as a yellow oil (1.72 g, 67%). ¹H NMR (CDCl₃, 200 MHz): δ 7.07 (d, 1H, J = 5.1 Hz), 6.72 (d, 1H, J = 5.1 Hz), 3.64 (s, 3H), 3.55 (s, 2H), 2.91–2.82 (m, 2H) 2.81–2.73 (m, 2H), 2.57 (t, 2H, J = 7.1 Hz), 2.4 (t, 2H, J = 7.3 Hz), 1.99–1.85 (m, 2H). ¹³C{¹H} NMR (CDCl₃, 50 MHz): δ 174.1, 133.9, 133.5, 125.3, 122.7, 56.9, 53.1, 51.6, 50.9, 32.0, 25.5, 22.7. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₂H₁₈NO₂S: 240.1058; found: 240.1102.

Methyl 4-(2,3-dihydro-1H-pyrrolo[2,3-b]pyridin-1-yl)butanoate(7d): Compound 7d was obtained as a yellow oil (0.124 g, 20%). ¹H NMR (CDCl₃, 200 MHz): δ 7.86–7.79 (m, 1H), 7.19–7.10 (m, 1H), 6.40 (dd, 1H, J = 6.95, 5.37 Hz), 3.64 (s, 3H), 3.48 (t, 2H, J = 8.3 Hz), 3.38 (t, 2H, J = 7.1 Hz), 2.94 (t, 2H, J = 8.4 Hz), 2.41 (t, 2H, J = 7.52 Hz), 1.95 (q, 2H, J = 7.49 Hz). ¹³C{¹H} NMR (CDCl₃, 50 MHz): δ 173.9, 145.7, 142.7, 130.8, 123.0, 112.1, 51.5, 49.6, 44.9, 31.6, 25.9, 23.0. HRMS (EI⁺): m/z [M]⁺ calcd. for C₁₂H₁₆N₂O₂: 221.1212; found: 220.1219.

2.3. General Method for the Synthesis of Compounds 8a–d

To a solution of the corresponding ester 7a–d (1.0 equivalent) in 8.0 mL methanol/THF mixture (1:1), 1.1 equivalents of lithium hydroxide dissolved in 2 mL of water were added and the mixture was stirred at 50 °C for 24 h. Solvents were evaporated under reduced pressure and the residue was dissolved in water and extracted with CH₂Cl₂ (3 × 10 mL). The aqueous layer acidified with concentrated HCl at pH of 3. The product was isolated by filtration.

4-(3,4-dihydroisoquinolin-2(1H)-yl)butanoic acid (8a): Compound 8a was obtained as a yellow oil (0.94 g, 87%). ¹H NMR (400 MHz, DMSO-d₆): δ 7.32–7.15 (m, 4H), 4.38 (bs, 2H), 3.24–3.15 (m, 3H), 3.14–3.03 (bs, 2H) 2.37 (t, 2H, J = 7.28 Hz), 2.07–1.98 (m, 2H). ¹³C{¹H} NMR (100 MHz, DMSO-d₆): δ 173.6, 131.5, 128.6, 128.5, 127.7, 126.73, 126.7, 54.5, 51.7, 48.7, 30.8, 24.8, 19.1. HRMS (EI⁺): m/z [M]⁺ calcd. for C₁₃H₁₇NO₂: 219.1259; found: 219.1256.

4-(1H-benzo[d]imidazol-1-yl)butanoic acid (8b): Compound 8b was obtained as a yellow oil (0.25 g, 70%). ¹H NMR (400 MHz, CD₃OD): δ 9.54 (s, 1H), 8.06–8.02 (m, 1H), 7.91–7.87 (m, 1H), 7.71–7.67 (m, 2H), 4.64 (t, 2H, J = 7.4 Hz), 2.5 (t, 2H, J = 6.92 Hz), 2.29 (p, 2H, J = 6.94 Hz). ¹³C{¹H} NMR (100 MHz, CD₃OD): δ 175.7, 142.0, 132.5, 132.3, 128.2, 127.9, 115.8, 114.1, 47.5, 31.2, 25.6. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₁H₁₃N₂O₂: 205.0977; found: 205.1087.

4-(6,7-dihydrothieno[3,2-c]pyridin-5(4H)-yl)butanoic acid (8c): Compound 8c was obtained as a yellow oil (0.58 g, 75%). ¹H NMR (400 MHz, CDCl₃): δ 7.4 (d, 1H, J = 5.23 Hz), 6.9 (d, 1H, J = 5.24 Hz), 4.43 (s, 2H), 3.74–3.61 (m, 2H), 3.39–3.33 (m, 2H), 3.24 (t, 2H, J = 6.31 Hz), 2.5 (t, 2H, J = 6.93 Hz), 2.15–2.07 (m, 2H). ¹³C{¹H} NMR (100 MHz, CDCl₃): δ 175.8, 132.6, 128.8, 126.6, 125.9, 56.5, 52.7, 51.5, 31.4, 23.1, 20.7. HRMS (EI⁺): m/z [M]⁺ calcd. for C₁₁H₁₅NO₂S: 225.0823; found: 225.0829.

4-(2,3-dihydro-1H-pyrrolo[2,3-b]pyridin-1-yl)butanoic acid (8d): Compound 8d was obtained as a yellow oil (0.15 g, 65%). ¹H NMR (CD₃OD, 400 MHz) δ 7.67–7.63 (m, 1H), 7.22–7.18 (m, 1H), 6.4 (dd, 1H, J = 6.98, 5.49 Hz), 3.63–3.3 (m, 6H), 2.96 (t, 2H, J = 8.22 Hz), 2.27–2.22 (m, 2H). ¹³C{¹H} NMR (CD₃OD, 100 MHz): δ 182.2, 164.2, 145.3, 132.2, 125.5, 112.9, 46.36, 36.4, 26.5, 25.4, 24.2. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₁H₁₅N₂O₂: 207.1134; found: 207.1115.

2.4. General Procedure for the Synthesis of Compounds 10a–d

A solution of (E)-ethyl 4-bromocrotonate 9 (1.0 equivalents) in 10.0 mL of CH₃CN was treated with DIPEA (1.5 equivalents). The reaction mixture was stirred at room temperature for 5 min and then the corresponding heterocyclic system a–d (1.0 equivalent) was added. The mixture was stirred at room temperature for 24–72 h. The solvents were evaporated under reduced pressure and the residue obtained was dissolved in 10 mL of water and then extracted with CH₂Cl₂ (3 × 20 mL). The organic layer was dried over anhydrous Na₂SO₄, filtered, and concentrated under reduced pressure. The crude product was purified by flash chromatography, giving the corresponding ester 10a–d.

Ethyl (E)-4-(3,4-dihydroisoquinolin-2(1H)-yl)but-2-enoate (10a): Compound 10a was obtained as a yellow oil (0.53 g, 65%). ¹H NMR (CDCl₃, 200 MHz): δ 7.14–6.85 (m, 5H), 5.98 (dt, 1H, J = 15.7, 1.6 Hz), 4.13 (q, 2H, J = 7.15 Hz), 3.59 (s, 2H), 3.26 (dd, 2H, J = 6.14, 1.64 Hz), 2.89–2.79 (m, 2H), 2.74–2.65 (m, 2H), 1.22 (t, 3H, J = 7.14 Hz). ¹³C{¹H} NMR (CDCl₃, 50 MHz): δ166.3, 145.0, 134.2, 134.0, 128.8, 126.6, 126.4, 125.8, 123.7, 60.5, 59.0, 56.1, 50.9, 28.9, 14.3. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₅H₂₀NO₂: 246.1494; found: 246.1467

Ethyl (E)-4-(1H-benzimidazole-1-yl)ethyl but-2-enoate (10b): Compound 10b was obtained as a yellow oil (0.82 g, 70%). ¹H NMR (CDCl₃, 400 MHz): δ 7.91 (s, 1H), 7.87–7.8 (m, 1H), 7.34–7.25 (m, 3H), 7.06 (dt, J = 15.67, 4.85 Hz), 5.69 (dt, 1H, J = 15.64, 1.94 Hz), 4.95 (dd, 2H, J = 4.83, 1.95 Hz), 4.16 (q, 2H, J = 7.14 Hz), 1.24 (t, 3H, J = 7.13 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 165.4, 143.8, 142.9, 140.9, 133.6, 123.7, 123.5, 122.6, 120.7, 109.7, 60.9, 45.5, 14.2. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₃H₁₅N₂O₂: 231.1134; found: 231.1147.

Ethyl (E)-4-(4,7-dihydrothienopyridin-6(5H)-yl)but-2-enoate (10c): Compound 10c was obtained as a yellow oil (1.51 g, 40%). ¹H NMR (CDCl₃, 200 MHz): δ 7.12–6.93 (m, 2H), 6.72 (d, 1H, J = 5.13 Hz), 6.05 (dt, 1H, J = 15.69, 1.67 Hz), 4.21 (q, 2H, J = 7.1 Hz), 3.61–3.57 (m, 2H), 3.35 (dd, 2H, J = 6.15, 1.64 Hz), 2.95–2.86 (m, 2H), 2.85–2.76 (m, 2H), 1.29 (t, 3H, J = 7.2 Hz). ¹³C{¹H} NMR (CDCl₃, 50 MHz): 166.2, 145.1, 133.4, 133.3, 125.2, 123.6, 122.9, 60.5, 58.5, 53.2, 50.8, 25.4, 14.3. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₃H₁₈NO₂S: 252.1058; found: 252.1042.

Ethyl (E)-4-(2,3-dihydro-1H-pyrrolopyridin-1-yl)but-2-enoate (10d): Compound 10d was obtained as a yellow oil (0.41 g, 30%). ¹H NMR (CDCl₃, 200 MHz): δ 7.88–7.83 (m, 1H), 7.21 (dq, 1H, J = 7.01, 1.4 Hz), 6.98 (dt, 1H, J = 15.64, 5.21 Hz), 6.47 (dd, 1H, J = 7.02, 5.31 Hz), 6.01 (dt, 1H, J = 15.67, 1.79 Hz), 4.24–4.16 (m, 2H), 4.15–4.11 (m, 2H), 3.48 (t, 2H, J = 8.28 Hz), 3.0 (t, 2H, J = 8.37 Hz), 1.28 (t, 3H, J = 7.13 Hz). ¹³C{¹H} NMR (CDCl₃, 50 MHz): δ 166.3, 145.9, 145.8, 144.3, 131.2, 122.8, 122.54, 112.9, 60.5, 49.8, 46.7, 26.0, 14.3. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₃H₁₇N₂O₂: 233.1290; found: 233.1262.

2.5. General Procedure for the 1,4-Addition Reactions

In a round-bottom flask provided with magnetic stirring, was placed copper iodine (2.0 equivalents) in 10.0 mL of anhydrous diethyl ether under nitrogen atmosphere. The mixture was stirred at 0 °C. In another round-bottom flask, magnesium turnings and a catalytic amount of iodine in anhydrous diethyl ether were placed under nitrogen atmosphere at room temperature Then, 1-bromo-2-methylpropane (4.0 equivalents) or 1-bromo-4-chlorobenzene (4.0 equivalents) was added to the mixture dropwise. The resulting mixture was stirred until the color of the solution changed from amber to gray and the disappearance of the metallic magnesium was observed. The Grignard reagent was added to the copper iodide suspension via cannula, and the resulting mixture was stirred for 30 min at 0 °C to generate the corresponding diorganocopper compound. Then, the corresponding compound 10a–d (1.0 equivalent) dissolved in anhydrous diethyl ether was added to the mixture dropwise at 0 °C for 30 min. The reaction was allowed to proceed for 24 h before being quenched with a saturated aqueous solution of NH₄Cl. The mixture was extracted with ether (3 × 25 mL) dried over Na₂SO₄ and the solvent removed in vacuo to afford the crude products, which were purified by flash chromatography eluting with Hexane–EtOAc (95:5). Some modifications made to this procedure are mentioned in the specific examples below.

Ethyl 3-((3,4-dihydroisoquinolin-2(1H)-yl)-methyl-5-methylhexanoate (11a): Compound 11a was obtained as a yellow oil (0.41 g, 40%). ¹H NMR (CDCl₃, 400 MHz): δ 7.12–7.0 (m, 4H), 3.95 (dq, 2H, J = 7.13, 2.86 Hz), 3.67 (d, 1H, J = 14.75 Hz), 3.5 (d, 1H, J = 14.8 Hz), 2.87–2.82 (m, 2H), 2.80–2.73 (m, 1H), 2.64–2.57 (m, 1H), 2.42–2.19 (m, 5H), 1.74–1.63 (m, 1H), 1.31–1.16 (m, 2H), 1.13 (t, 3H, J = 7.14 Hz), 0.92 (d, 3H, J = 6.15 Hz), 0.90 (d, 3H, J = 6.15 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 173.8, 135. 4, 134.7, 128.7, 126.6, 126.0, 125.5, 63.5, 60.0, 56.6, 51.3, 42.8, 39.1, 31.2, 29.3, 25.4, 23.0, 22.9, 14.2. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₉H₃₀NO₂: 304.2277; found: 304.2247.

Ethyl 3-((1H-benzo[d]imidazol-1-yl)methyl)-5-methylhexanoate (11b): Compound 11b was obtained as a yellow oil (0.42 g, 45%). ¹H NMR (CDCl₃, 400 MHz): δ 8.43 (s, 1H), 8.41–8.34 (m, 1H), 7.51–7.43 (m, 1H), 7.35–7.27 (m, 2H), 4.26–4.27 (m, 1H), 4.15–4.02 (m, 3H), 2.58–2.48 (m, 1H), 2.35–2.15 (m, 2H), 1.71–1.58 (m, 1H), 1.32–1.16 (m, 5H), 0.91 (d, 3H, J = 6.53 Hz), 0.82 (d, 3H, J = 6.51 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 172.1, 145.4, 141.7, 123.8, 123.0, 120.9, 110.2, 60.8, 49.3, 41.1, 36.1, 33.3, 25.3, 23.1, 22.2, 14.4. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₇H₂₅N₂O₂: 289.1916; found: 289.1946.

Ethyl 3-((6,7-dihydrothieno[3,2-c]pyridin-5(4H)-yl)methyl)-5-methylhexanoate (11c): Compound 11c was obtained as a yellow oil (0.27 g, 30%). ¹H NMR (CDCl₃, 400 MHz): δ 7.05 (d, 1H, J = 5.11 Hz), 6.72 (d, 1H, J = 5.13 Hz), 3.96 (dq, 2H, J = 7.15, 2.52 Hz), 3.62–3.56 (m, 1H), 3.45–3.39 (m, 1H), 2.87–2.80 (m, 3H), 2.71–2.61 (m, 1H), 2.5–2.17 (m, 5H), 1.63–1.61 (m, 1H), 1.21–1.12 (m, 5H), 0.91 (d, 3H, J = 5.89 Hz), 0.9 (d, 3H, J = 5.88 Hz).¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 173.7, 134.3, 133.6, 125.3, 122.4, 62.9, 60.0, 53.6, 51.3, 42.8, 39.1, 31.5, 25.6, 25.4, 23.0, 22.8, 14.2. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₇H₂₈NO₂S: 310.1841; found: 310.1875.

Ethyl 3-((2,3-dihydro-1H-pyrrolo[2,3-b]pyridin-1-yl)methyl)-5-methyl hexanoate (11d): Compound 11d was obtained as a yellow oil (0.074 g, 22%). ¹H NMR (CDCl₃, 400 MHz): δ 7.85 (dd, 1H, J = 5.2, 1.6 Hz), 7.16–7.12 (m, 1H), 6.4 (dd, 1H, J = 6.94, 5.34 Hz), 4.08–3.94 (m, 2H), 3.62–3.55 (m, 1H), 3.43–3.32 (m, 2H), 3.10 (dd, 1H, J = 13.5, 5.25 Hz), 2.97–2.90 (m, 2H), 2.40–2.32 (m, 1H), 2.28–2.21 (m, 1H), 1.73–1.64 (m, 1H), 1.28–1.18 (m, 2H), 1.14 (t, 3H, J = 8.38 Hz), 0.92 (d, 3H, J = 6.6 Hz), 0.9 (d, 3H, J = 6.55 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 173.5, 163.7, 145.8, 130.7, 122.79, 112.1, 60.1, 50.6, 50.2, 42.2, 38.2, 32.2, 25.9, 25.4, 22.9, 22.8, 14.1. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₇H₂₇N₂O₂: 291.2073; found: 291.2083.

Ethyl 3-(4-chlorophenyl)-4-(3,4-dihydroisoquinolin-2(1H)-yl) butanoate (12a): Compound 12a was obtained as a yellow oil (0.20 g, 30%). ¹H NMR (CDCl₃, 400 MHz): δ 7.28–7.24 (m, 2H), 7.19–7.15 (m, 2H), 7.14–7.05 (m, 4H), 3.92 (q, 2H, J = 7.13 Hz), 3.73 (d, 1H, J = 14.74 Hz), 3.57 (d, 1H, J = 14.75 Hz), 3.51–3.42 (m, 1H), 2.94–2.86 (m, 2H) 2.85–2.79 (m, 2H), 2.7–2.62 (m, 2H), 2.59–2.47 (m, 2H), 1.07 (t, 3H, J = 7.13 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 172.6, 141.5, 135.0, 134.5, 132.4, 129.0, 128.7, 126.6, 126.2, 125.6, 116.8, 64.2, 60.3, 56.3, 51.3, 39.7, 39.4, 29.2, 14.1. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₂₁H₂₅NO₂Cl: 358.1574; found: 358.1566.

Ethyl 4-(1H-benzimidazole-1-yl)-3-(4-chlorophenyl) butanoate (12b): Compound 12b was obtained as a brown solid (0.54 g, 50%). ¹H NMR (CDCl₃, 400 MHz): δ 8.23 (d, 1H, J = 7.13 Hz), 8.01 (s, 1H), 7.31–7.17 (m, 3H), 7.12 (d, 2H, J = 8.39 Hz) 6.88 (d, 2H, J = 8.42 Hz) 4.33 (dd, 1H, J = 14.21, 6.34 Hz), 4.09 (dd, 1H, J = 14.24, 8.13 Hz), 3.98 (q, 2H, J = 7.07 Hz), 3.63–3.51 (m, 1H), 2.61 (d, 2H, J = 7.39 Hz), 1.08 (t, 3H, J = 7.13 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 171.1, 145.1, 141.4, 138.2, 133.7, 133.4, 129.4, 128.8, 123.9, 123.1, 120.8, 110.0, 60.1, 50.4, 41.5, 37.5, 14.2. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₉H₂₀N₂O₂Cl: 343.1213; found: 343.1214.

Ethyl 3-(4-chlorophenyl)-4-(6,7-dihydrothieno[3,2-c]pyridin-5(4H)-yl) butanoate (12c): Compound 12c was obtained as a yellow oil (0.41 g, 41%). ¹H NMR(CDCl₃, 700 MHz): δ 7.29–7.24 (m, 2H), 7.19–7.14 (m, 2H), 7.06 (d, 1H, J = 5.11 Hz), 6.71 (d, 1H, J = 5.11 Hz), 3.93 (q, 2H, J = 7.13 Hz), 3.65 (d, 1H, J = 14.2 Hz), 3.5 (d, 1H, J = 14.22 Hz), 3.48–3.42 (m, 1H), 2.9–2.85 (m, 2H), 2.85–2.81 (m, 2H), 2.74–2.69 (m, 1H), 2.69–2.63 (m, 2H), 2.51 (dd, 1H, J = 15.47, 8.06 Hz), 1.09 (t, 3H, J = 7.14 Hz). ¹³C{¹H} NMR (CDCl₃, 175 MHz): δ 172.6, 141.4, 133.9, 133.5, 132.5, 129.0, 128.7, 125.3, 122.6, 63.6, 60.4, 53.4, 51.2, 39.9, 39.4, 25.4, 14.1. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₉H₂₃NO₂SCl: 364.1138; found: 364.1122.

Ethyl 3-(4-chlorophenyl)-4-(2,3-dihydro-1H-pyrrolo[2,3-b]pyridin-1-yl) butanoate (12d): Compound 12d was obtained as a yellow oil (0.13 g, 30%). ¹H NMR (CDCl₃, 400 MHz): δ 7.79–7.74 (m, 1H), 7.24–7.11 (m, 4H), 7.08–7.03 (m, 1H), 6.34 (dd, 1H, J = 6.98, 5.32 Hz), 3.88 (q, 2H, J = 7.16 Hz), 3.67 (dd, 1H, J = 13.53, 8.71 Hz), 3.52–3.42 (m, 1H), 3.41–3.3 (m, 1H), 3.29–3.15 (m, 2H), 2.81 (t, 2H, J = 8.17 Hz), 2.72 (dd, 2H, J = 15.68, 5.98 Hz), 2.52 (dd, 1H, J = 15.69, 8.83 Hz), 1.0 (t, 3H, J = 7.18 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 172.2, 163.2, 145.8, 140.7, 132.6, 130.9, 129.2, 128.7, 122.6, 112.5, 60.4, 51.5, 50.2, 40.6, 38.9, 25.9, 14.1. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₉H₂₂N₂O₂Cl: 345.1370; found: 345.1404.

2.6. General Procedure for the Synthesis of the Pregabalin Analogs 13a–d

To a solution of the corresponding ester 11a–d (1.0 equivalent) in 8.0 mL methanol/THF mixture (1:1), 1.1 equivalents of lithium hydroxide dissolved in 2 mL of water were added and the mixture was stirred at 50 °C for 24 h. Solvents were evaporated under reduced pressure and the residue was dissolved in water and extracted with CH₂Cl₂ (3 × 10 mL). The aqueous layer acidified with concentrated HCl at pH of 3. The product was isolated by filtration.

3-((3,4-dihydroisoquinolin-2(1H)-yl)methyl)-5-methylhexanoic acid (13a): Compound 13a was obtained as a yellow oil (0.3 g, 68%). ¹H NMR (400 MHz, CDCl₃): δ 7.29–7.0 (m, 4H), 4.63–3.97 (bs, 2H), 3.74–3.34 (bs, 2H), 3.32–3.21 (m, 1H), 3.20–2.97 (bs, 2H), 2.96–2.84 (m, 1H), 2.72–2.52 (m, 2H), 2.36–2.25 (bs, 1H), 1.54–1.41 (m, 1H), 1.26–1.17 (m, 2H), 0.82 (d, 3H, J = 4.54 Hz), 0.8 (d, 3H, J = 4.62 Hz). ¹³C{¹H} NMR (100 MHz, CDCl₃): δ 174.9, 130.5, 128.9, 128.4, 127.4, 127.1, 126.7, 58.6, 53.2, 49.5, 41.8, 37.4, 28.7, 25.1, 23.9, 22.8, 22.4. HRMS (EI⁺): m/z [M]⁺ calcd. for C₁₇H₂₅NO₂: 275.1885; found: 275.1885.

3-((1H-benzo[d]imidazol-1-yl)methyl)-5-methylhexanoic acid (13b): Compound 13b was obtained as a yellow oil (0.35 g, 80%). ¹H NMR (CD₃OD, 400 MHz): δ 8.17 (s, 1H), 7.81–7.58 (m, 2H) 7.43–7.30 (m, 2H), 4.17 (dd, 1H, J = 14.23, 6.08 Hz), 4.08 (dd, 1H, J = 14.23, 8.05 Hz), 2.49–2.38 (m, 1H) 2.22 (dd, 1H, J = 14.28, 6.68 Hz), 2.03 (dd, 1H, J = 14.29, 7.54 Hz), 1.52–1.39 (m, 1H), 1.23–1.12 (m, 1H), 1.04–0.94 (m, 1H) 0.76 (d, 3H, J = 6.55 Hz), 0.65 (d, 3H, J = 6.51 Hz). ¹³C{¹H} NMR (CD₃OD, 100 MHz): δ 181.5, 123.3, 122.5, 118.7, 111.2, 49.3, 41.2, 40.9, 34.0, 24.6, 22.2, 21.5. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₅H₂₁N₂O₂: 261.1603; found: 261.1696.

3-((6,7-dihydrothieno[3,2-c]pyridin-5(4H)-yl)methyl)-5-methylhexanoic acid (13c): Compound 13c was obtained as a yellow oil (0.34 g, 60%). ¹H NMR (400 MHz, CDCl₃): δ 7.14 (d, 1H, J = 5.16 Hz), 6.74 (d, 1H, J = 5.17 Hz), 3.74–3.55 (m, 2H), 2.97–2.81 (m, 3H), 2.51 (d, 1H), 2.35–2.21 (m, 2H), 2.20–2.08 (m, 1H), 1.78–1.65 (m, 1H), 1.34–1.26 (m, 1H), 1.14–1.13 (m, 1H), 0.99–0.77 (m, 8H). ¹³C{¹H} NMR (100 MHz, CDCl₃): δ 182.1, 134.4, 134.1, 126.1, 123.8, 64.1, 54.3, 52.2, 44.3, 43.8, 32.4, 26.4, 25.6, 23.5, 23.2. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₅H₂₄NO₂S: 282.1528; found: 282.1516.

3-((2,3-dihydro-1H-pyrrolo[2,3-b]pyridin-1-yl)methyl)-5-methylhexanoic acid (13d): Compound 13d was obtained as a yellow oil (0.29 g, 76%). ¹H NMR (CD₃OD, 400 MHz): δ 7.65–7.6 (m, 1H), 7.23–7.17 (m, 1H) 6.39 (dd, 1H, J = 6.95, 5.49 Hz), 3.67–3.56 (m, 1H), 3.55–3.45 (m, 1H), 3.23 (d, 2H, J = 7.21 Hz), 2.96 (t, 2H, J = 8.5 Hz), 2.4–2.27 (m, 1H), 2.25–2.07 (m, 2H), 1.76 (m, 1H), 1.25 (t, 2H, J = 6.9 Hz), 0.92 (d, 3H, J = 2.5 Hz), 0.9 (d, 3H, J = 2.4 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 182.0, 164.7, 145.3, 132.2, 125.3, 112.8, 51.3, 50.8, 43.8, 43.1, 33.8, 26.57, 26.5, 23.5, 23.3. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₅H₂₃N₂O₂: 263.1760; found: 263.1798.

2.7. General Procedure for the Synthesis of the Baclofen Analogs 14a, 14b and 14d

To a solution of the corresponding ester 12a, 12b or 12d (1.0 equivalent) in 8.0 mL methanol/THF mixture (1:1), 1.1 equivalents of lithium hydroxide dissolved in 2 mL of water were added and the mixture was stirred at 50 °C for 24 h. Solvents were evaporated under reduced pressure and the residue was dissolved in water and extracted with CH₂Cl₂ (3 × 10 mL). The aqueous layer acidified with concentrated HCl at pH of 3. The product was isolated by filtration.

3-(4-chlorophenyl)-4-(3,4-dihydroisoquinolin-2(1H)-yl)butanoic acid (14a): Compound 14a was obtained as a yellow solid (0.20 g, 30%). ¹H NMR (CDCl₃, 400 MHz): δ 7.26–7.12 (m, 4H), 7.11–6.83 (m, 4H) 4.33–3.97 (m, 2H), 3.90–3.52 (m, 2H), 3.47–3.15 (m, 3H), 3.12–2.81 (m, 3H), 2.6 (dd, 1H, J = 16.76, 5.65 Hz). ¹³C{¹H} NMR (CDCl₃, 100 MHz): δ 173.9, 139.4, 133.6, 130.5, 129.5, 129.3, 128.8, 128.4, 127.3, 127.1, 126.8, 59.1, 53.4, 50.2, 40.4, 37.3, 24.3. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₂₁H₂₅NO₂Cl: 330.1261; found: 330.1263.

4-(1H-benzimidazol-1-yl)-3-(4-chlorophenyl)butanoic acid (14b): Compound 14b was obtained as a yellow solid (0.65 g, 70%). ¹H NMR (DMSO_-d₆, 400 MHz): δ 9.44 (s, 1H), 8.0 (d, 1H, J = 7.7 Hz), 7.81 (d, 1H, J = 7.7 Hz), 7.62–7.49 (m, 2H), 7.25 (s, 3H), 4.89–4.63 (m, 2H), 3.79–3.66 (m, 1H), 2.91–2.65 (m, 2H). ¹³C{¹H} NMR (DMSO_-d₆, 100 MHz): δ 172.2, 141.6, 138.9, 131.8, 131.5, 131.2, 129.7, 128.4, 126.0, 125.8, 115.3, 113.1, 50.5, 41.0, 37.2. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₇H₁₆N₂O₂Cl: 315.0900; found: 315.0878.

3-(4-chlorophenyl)-4-(2,3-dihydro-1H-pyrrolo[2,3-b]pyridin-1-yl)butanoic acid (14d): Compound 14d was obtained as a yellow solid (0.75 g, 80%). ¹H NMR (CD₃OD, 400 MHz): δ 7.54–7.5 (m, 1H), 7.49–7.45 (m, 1H), 7.43–7.37 (m, 2H), 7.36–7.30 (m, 2H), 6.67 (t, 1H, J = 6.8 Hz), 3.94–3.82 (m, 2H), 3.78–3.58 (m, 3H), 3.18–3.10 (m, 2H), 2.84 (dd, 1H, J = 16.11, 6.46 Hz), 2.74 (dd, 1H, J = 16.07, 7.6 Hz). ¹³C{¹H} NMR (CD₃OD, 100 MHz): δ 173.4, 139.27, 134.9, 132.9, 131.1, 130.4, 129.2, 128.5, 111.97, 50.8, 50.6, 39.0, 37.0, 24.6. HRMS (FAB⁺): m/z [M+H]⁺ calcd. for C₁₇H₁₈N₂O₂Cl: 317.1057; found: 317.1067.

2.8. Computational Details

2.8.1. Chemical-Structure Optimization

Three-dimensional structures of molecules 8a–d, 13a–d, 14a–d, GABA, and Baclofen were constructed in Spartan’24 [38], starting from the structure of Baclofen (PDB: 4MS4) [39], with the respective substitutions, considering the protonation state of the compounds at a physiological pH of 7.4. Molecules were submitted to a geometry optimization at the density-functional-theory level (DFT) using the M06 functional [40] and the 6-31+G* basis set [41]. These calculations considered water as a solvent with the SM8 model [42]. A vibrational analysis was carried out to ensure that all the structures were at a minimum of the potential energy surface. All calculations were performed in Spartan’24.

2.8.2. Molecular Docking

GABA_B receptor crystal structure (PDB: 4MS4) [39] was corrected, by adding missing hydrogen atoms and bonds, with tools implemented in Molegro Virtual Docker (MVD) [43]. Additionally, all water molecules and the rest of the ligands (N-acetyl-D-glucosamine) were removed from the GABA_B receptor crystal structure.

The Baclofen binding site was settled as the cavity (525 ų, Figure 2) where all the molecular docking calculations would be performed. Furthermore, all Baclofen analogs were docked (rigid approximation) using the Mulliken partial charges. This docking was achieved using default parameters: MolDock Score [GRID] as the scoring function with 0.3 Å of grid resolution. The evaluation of ligand internal energy considered internal ES (steric), internal HBond, and sp²-sp² torsions. The MolDock SE (Simplex Evolution) search algorithm was used with the following parameters: 10 runs with a maximum of 1500 iterations using a population of 50 individuals. Structural docking validation was performed by reproducing the crystal conformation of Baclofen (based on its RMSD = 0.12 value). GABA scaffold interactions are shown in Supplementary Figure S56.

As a second step, flexible docking over the same GABA_B receptor crystal structure (PDB: 4MS4) was performed. For this, five amino acids were set as flexible (Trp65, His170, Ile276, Trp278 and Glu349) to evaluate the inductive effect caused by the binding of each ligand; 2000 minimization steps for each flexible residue, and 2000 steps of global minimization per run. The scoring function and the parameters used for the rigid docking calculation were also employed for the flexible approximation. The MolDock Optimizer search algorithm was used with the following parameters: 10 runs with a maximum of 2000 iterations using a population of 50 individuals.

2.8.3. QSAR Analysis

In our previous work, Equation (1) was reported [35], where Y_pred represents the predicted agonist activity of the molecules.

Y_pred = 0.244E_TRP278 + 0.005E_LUMO + (–137.01)PW5 + 10.24PIJ2 + 0.06T(N..O) + 13.24

(1)

The conformation obtained from the flexible docking calculations of the molecules was used to calculate the quantum molecular descriptor related to the Energy of the Lowest Unoccupied Molecular Orbital (E_LUMO). For this purpose, a single-point energy calculation (in vacuum) with the M06 functional and the 6-31+G* basis set in Spartan’24 was carried out. Then, the LUMO graphics and E_LUMO values were obtained for all compounds. Also, from the results of the flexible docking, the interaction energy of the amino acid Trp278 with each ligand (E_TRP278) was obtained. On the other hand, the topological descriptors PW5 and PJI2 related to the shape, as well as the T(N..O) descriptor, were calculated with the AlvaDesc package [44]. PJI2 is the Petitjean index calculated as the difference between the topological diameter and the radius divided by the radius [45]. This descriptor has a positive coefficient, which indicates that the larger the value of this descriptor, the higher the probability of obtaining an increase in Y (agonic activity). PW5 is obtained from the Pi/Wi matrix constructed by the paths (Pi) and walks (Wi) matrixes [46]. PW5 has a negative coefficient, which limits the structural modifications (ramifications) of the GABA_B receptor agonists. Finally, the T(N..O) descriptor represents the topological distance between the nitrogen and oxygen atoms.

It is observed that with a decrease in the values (higher negative value) of the molecular descriptors E_TRP278 and E_LUMO, a reduction in the predicted biological activity value is obtained. On the other hand, an increase in PJI2 and T(N..O) values counteract the negative values from the sum of the E_TRP278, E_LUMO, and PW5 descriptors, favoring the compound’s activity as GABA_B receptor agonists. The T(N..O) descriptor indicates that increasing the number of nitrogen and oxygen atoms, and the distance between them, could increase the probability that Baclofen analogs demonstrate the same activity as GABA_B receptor agonists. The statistical validation of the QSAR model is reported in a previous paper [35].

2.8.4. Molecular Docking Re-Evaluation

The 8a–d binding mode was also evaluated in the GABA_B crystal co-crystallized with GABA (PDB:4MS3) [39], since the 4MS4 crystal, co-crystallized with Baclofen, has a much larger cavity that could interfere with the optimal accommodation of GABA derivatives.

For this analysis, we repeated the docking process in 4MS4 described above. First, a rigid docking was carried out over the cavity of the GABA binding site (Figure 3), which turned out to be much smaller (65.563 ų) than the cavity of the 4MS4 crystal. The docking parameters used were the same as for 4M4 rigid docking. The structural validation was performed by reproducing the crystal conformation of GABA (based on its RMSD = 0.15 value). Additionally, flexible docking over the 4MS3 crystal was performed; Glu349, Met312, Trp278, and His170 were settled as flexible. These residues are the closest ones to the GABA amino group, where the structural modifications of 8a–d were made. We used the best poses obtained from the rigid docking to start the flexible docking. We used the same parameters for the search algorithm (MolDock Optimizer) employed in 4MS4 for this crystal.

2.8.5. Molecular Similarity Analysis

For this similarity analysis, the LUMO graphics of our best candidates, in (R)-enantiomer, according to the QSAR and the (R)-Baclofen and GABA, were obtained; the LUMO graphics of the (S)-enantiomer are shown in Figure S57. Additionally, molecular electrostatic potential (MEP) mapped onto an iso-density surface (0.002 e⁻/ų) for each candidate was obtained and compared to determine the electronic zones with higher and lower electronic density [47]. In MEP graphics, regions with higher electronic density are in red, and those with less density are in blue [30]. Moreover, to evaluate the force of the binding regions where intermolecular interactions with GABA_B receptor will take place, molecular electrostatic potential surfaces of iso-values (±320 kJ/mol) were obtained. All the molecular graphics were performed in Spartan’24.

2.8.6. Drug-Likeness Prediction

The pharmacokinetics of the best candidates were predicted using the ADMETlab2 server [48]. ADMElab2 allows us to predict how a candidate drug will behave or interact in the body once ingested, associated with absorption, distribution, excretion, metabolism, toxicity, and physicochemical properties, such as molecular weight (MW), the number of hydrogen bond donors (NHD), the number of rotary bonds (NRB), the topological polar surface area (TPSA), the number of hydrogen bond acceptors (NHA), the logarithm of the n-octanol/water distribution coefficients (LogP), among others.

3. Results and Discussion

3.1. Synthesis of GABA Analogs

GABA analogs were synthetized from 4-bromobutyric acid, which was previously esterified to protect the carboxylic acid (

Table 1. Reaction conditions for the synthesis of compounds 7a–d.

Compound	Base	Time (h)	Temperature (°C)	Solvent	Yield (%)
7a	-	8	r.t.	-	72
7b	DIPEA	72	60	CH3CN	65
7c	DIPEA	72	60	CH3CN	70
7d	DIPEA	72	60	CH3CN	30

). Subsequently, N-alkylation of ester 6 with the corresponding heterocycles (a–c) was carried out. Due to the nucleophilic character of the nitrogen in the tetrahydroisoquinoline (a), the reaction proceeded without the use of base, providing compound 7a with 75% of chemical yield. In the case of benzimidazole (b), we attempted to carry out the N-alkylation under the same reaction conditions; however, the reaction did not proceed as expected due to the lower nucleophilic character of the nitrogen atom and the low solubility of heterocycle (b) in ester 6. To find the optimal reaction conditions for the N-alkylation of ester 6, we carried out experiments with different bases such as N,N-diisopropylethylamine (DIPEA), K₂CO₃, and Et₃N, but the yields were very low; however, good yields with N,N-diisopropylethylamine (DIPEA) and the use of CH₃CN as a solvent were obtained (Table 1). After some experimentation, compound 7b was isolated in 65% yield when the reaction was carried out at 60 °C with 1.5 eq of DIPEA as a base in CH₃CN. In a similar manner, we were also able to perform the N-alkylation substitutions with 4,5,6,7-tetrahydrothieno[2,3-c]pyridine (c) and 2,3-dihydro-1H-pyrrolo[2,3-b]pyridine (d), using CH₃CN as a solvent and DIPEA as the base, to obtain products 7c and 7d with 70% and 30%, respectively.

All compounds were purified by column chromatography and, once characterized, they were hydrolyzed with lithium hydroxide to obtain the corresponding GABA analogs 8a–d (Scheme 1).

3.2. Synthesis of Pregabalin and Baclofen Analogs

The synthesis of the Pregabalin and Baclofen analogs starts with the N-alkylation reaction of ethyl 4-bromocrotonate 9, with heterocycles (a–c) using the same reaction conditions used previously (

Table 2. Reaction conditions for the synthesis of compounds 10a–d.

Compound	Base (1.1 eq)	Time (h)	Temperature (°C)	Solvent	Yield (%)
10a	-	24	r.t.	-	65
10b	DIPEA	48	r.t.	CH3CN	70
10c	DIPEA	72	r.t.	CH3CN	65
10d	DIPEA	72	r.t.	CH3CN	40

). Once again, in the case of compound 10a, the reaction proceeds without the use of solvent and base; in the case of compounds 10b–d, the use of DIPEA as the base and CH₃CN as the solvent was necessary (Table 2).

Successive treatment of the α,β-unsaturated systems 10a–d with the organocopper reagents (CH₃)₂CHCH₂)₂CuLi and (p-Cl-C₆H₄)₂CuLi led to the Pregabalin and Baclofen precursors 11a–d and 12a–d, respectively, with moderate to good yields (Scheme 2). Several reaction conditions were tested to increase the chemical yields of these conjugate additions; unfortunately, an increase in the equivalents of the copper reagent and reaction times had no effect on the chemical yield, probably due to the labile nature of these reagents, which easily decompose under the reaction conditions used (Scheme 2).

The 1,4-addition reactions generate the intermediates 11 and 12 as racemic mixtures, since there is no chirality inducer in 10 that favors one enantiomer over the other. Therefore, in the next basic hydrolysis step, the final products (13 and 14) are obtained as racemic mixtures. The reaction was carried out using an equimolar amount of LiOH to form the corresponding carboxylates and then protonated with aqueous hydrochloric acid. Analogs 13a–d were obtained with good yields (Scheme 3).

On the other hand, compounds 12a–d were subjected to basic hydrolysis. In the first step, molecules 12a–d were treated with lithium hydroxide, and then the corresponding carboxylates were treated with aqueous HCl to form the corresponding acid derivatives 14a, 14b, and 14d in good yields; however, it was not possible to isolate compound 14c (Scheme 4) due to its oily nature.

For the computational evaluation as GABA_B agonists, we used all the compounds of these series, including 14c, so that we had the entire series of compounds.

3.3. Molecular Docking Results

As observed in Figure S55, the active conformation of Baclofen of the crystal was reproduced (RMSD 0.8563), thus achieving one of the docking validation criteria. The other one was related to the binding mode of GABA and Pregabalin, which were oriented in an adequate way, maintaining the carboxylic group interactions with Cys129, Ser153 and Ser130. At the same time, the amino fragment keeps the interactions with Trp278, Glu349, and His170.

In Figure 4, the binding mode of Baclofen at the 4MS4 crystal is displayed, as well as the residues established as flexible for docking (within pink circles). The best poses of each of the molecules 8a–d, (R)-13a–d, and (R)-14a–d can be appreciated according to their scaffold (Figure 4). In all three types of derivatives, it is observed that the interactions formed between the carboxylic group of Baclofen with Ser130 and Ser152 (H-bond interactions) are preserved. GABA derivatives (8a–d) preserve the interactions of the p-chlorophenyl ring of Baclofen and create new interactions within the flexible residues at the binding site. These derivatives, however, have weaker interactions with Trp278 compared to Baclofen.

Pregabalin (R)-13a–d analogs also bind within the defined cavity of the protein. However, these molecules can acquire more conformations within the cavity formed by the flexible residues, but due to the incorporation of the nitrogenous heterocycles, they maintain and enhance the interactions observed between Baclofen and the GABA_B receptor.

Baclofen analogs, (R)-14a–d, are the molecules that dock better inside the cavity of the protein, maximizing the interactions with the flexible residues and possessing the best shape of the three series of derivatives. These molecules interact mainly with Trp278, improving the binding energies more than those of the co-crystallized Baclofen due to the orientation of both rings in the lipophilic cavity generated by the p-chlorophenyl ring. The results for the GABA and (S)-compounds dockings are shown in Figure S55.

3.4. QSAR Study

LUMO graphics of GABA and (R)-Baclofen and their analogs 8a–d, (R)-13a–d, and (R)-14a–d were obtained from the best pose of the molecular docking and are shown in Figure 5. LUMO is observed at the amine group of the aromatic bicycle, but with specific patterns depending on the scaffold (GABA or (R)-Baclofen) and the structural modification. The LUMO of 8a, (R)-13a, and (R)-14a are distributed over the two six-membered heterocycles. It is observed that the size of these orbitals decreases as follows: 8a > 13a > 14a. In the last compound, 14a, LUMO is delocalized mainly over the phenyl ring. In 8b, (R)-13b, and (R)-14b, LUMO is delocalized over the bicycle, but in a bigger size for (R)-13b. On the other hand, a substituent with sulfur destabilizes the location of the LUMO, as it is observed in 8c, (R)-13c, and (R)-14c. In these molecules, LUMO is mostly found over the five-membered sulfur heterocycle. LUMO of 8d, (R)-13d, and (R)-14d are delocalized over the carbon atoms on opposite sides to nitrogen atoms. The LUMO graphics in the (S)-configuration of the sets of 13 and 14 are shown in Figure S57; some compounds present a different delocalization of the LUMO in some isomers.

Specific molecular descriptors were obtained for 8a–d, 13a–d, 14a–d, GABA, and Baclofen, to obtain the Y_pred value of each molecule by using the QSAR model described above.

Table 3. Molecular descriptors present in the QSAR model and Y_pred of all the designed analogs, including GABA and (R)-Baclofen.

Compound	ETRP278	PIJ2	PW5	T(N..O)	ELUMO (kJ/mol)	Ypred
(R)-Baclofen	−19.699	1.0000	0.0920	10	−213.07	5.601
GABA	−4.919	0.6667	0.0438	10	−275.50	12.084
8a	−16.374	1.0000	0.0869	10	−212.47	7.119
8b	−19.910	0.8000	0.0832	24	126.85	7.243
8c	−20.023	0.8000	0.0761	10	−190.38	5.766
8d	−20.702	0.8000	0.0832	24	122.32	7.027
(R)-13a	−25.100	1.0000	0.0916	10	−173.77	4.533
(R)-13b	−23.836	0.8000	0.0900	24	132.32	5.387
(R)-13c	−21.550	0.8000	0.0834	10	−187.47	4.412
(R)-13d	−26.530	0.8000	0.0900	24	108.77	4.612
(R)-14a	−26.520	1.0000	0.1036	10	−182.49	2.501
(R)-14b	−28.810	0.8333	0.1019	24	116.00	2.806
(R)-14c	−22.383	0.8333	0.0970	10	−188.74	2.673
(R)-14d	−28.000	0.8333	0.1019	24	104.93	2.948

shows the values of each molecular descriptor and the predicted biological activity. Although all molecules tested showed possible bioactivity, only four molecules (8a–d) show a higher predicted activity than Baclofen, but lower than GABA. Molecular descriptors and the Y_pred values for the (S)-derivates are shown in Table S1.

GABA derivatives had the highest Y_pred when compared to the rest of our molecules evaluated in this study. This performance may be due to a good ovality value, indicated by the PJI2 descriptor. PW5 and E_TRP278 descriptors of 8a to 8d shows less branching and a higher interaction with the Trp278 residue, indicating a relationship between the branching of the molecules and their interaction with Trp278.

It is also seen that, even though 8a–d have the highest Y_pred, 8c has the lowest Y_pred. The molecular interactions for compounds 8a, 8b, and 8d are shown in Figure 6. This behavior is constant in all the molecules evaluated with sulfur ((R)-13c and (R)-14c), although (R)-13c does not show a low predicted activity compared to the rest of its derivatives. (R)-13c, as observed in the docking, does not maintain the conformation as expected, thus not allowing trust in its prediction. As observed in the LUMO analysis, the presence of the sulfur atom in all cases could be responsible of the low predicted activity; however, this will be discussed in the next sections.

In Figure 6, we can appreciate that the aromatic nature of 8a, 8b, and 8d increases the intermolecular interactions with aromatic residues (Tyr250 and Trp65). Molecular interactions like alkyl–Pi, Pi–Pi, and cation–Pi are displayed. Additionally, we can see that the molecular interactions of the carboxylate group are conserved in all the compounds, with the same interactions (strong HB) as GABA and Baclofen with Ser130 and Ser153. Furthermore, we performed molecular docking calculations and Y_pred determination using another GABA_B structure (PDB:4MS3), the one co-crystallized with GABA, to evaluate the potential activity of our best candidates as GABA_B agonists in a more robust way.

In Figure 7, GABA and our best candidates’ Y_pred values are displayed. Also, these candidates preserve the same interactions of GABA with the GABA_B receptor and present a new interaction.

Unlike GABA, molecules 8a–d generate a new interaction with the Met312 residue, this interaction is of Van der Waals type in the case of molecule 8a, Pi–sulfur in the case of 8b, Pi–alkyl in the case of 8c and alkyl–alkyl in the case of 8d. This new interaction is due to the modifications in the amino group and to the movement of the side chains in the flexible docking.

In the case of Trp278, 8a–d molecules considerably improve the interaction compared with GABA, which has an energy of −12.73 kcal/mol, while for GABA derivatives the interaction energies range from −20.7 to −24.2 kcal/mol. These molecules maintain the Pi–cation type interactions, and this change in energy is because a better shape complementarity is generated with the receptor, by shortening the distances between the amino group substituents and Trp278, maximizing the contact and interaction energy.

For Glu349, the interaction was retained; however, there was no energetic contribution to the overall interaction energy. In the case of His170, a small improvement in terms of interaction energies was obtained; however, better interaction energies were obtained with respect to GABA, as well as a change in the interactions performed, which change from an H-bond typical of GABA to Van der Waals interactions in the case of molecules 8a, 8b and 8d, and to a Pi–sulfur interaction for 8c.

Considering that docking program detects favorable interactions of Met312 and His170 with a favorable positioning of both residues in molecules 8a, 8b, and 8d, a shape complementarity can be perceived with these molecules with respect to the cavity so that Van der Waals-type interactions are favored.

The LUMO of 8a, 8b, and 8d in their active conformation (according to the docking analysis) are presented in Figure 8. Although these proposed compounds are GABA derivatives, they present a particular LUMO localization, depending on the substituent. In 8a, LUMO is placed over the two rings, but with a higher size on the nitrogen, while in 8b it is located on the bicycle due to the aromaticity of the two rings. In 8d, LUMO is mostly observed on the nitrogen of each ring. 8c LUMO is located over the five-membered sulfur heterocycle.

3.5. Molecular Similarity Analysis Results

To analyze the electronic properties of our best candidates, we studied the MEP of these compounds. In Figure 9, the MEPs of 8a, 8b, and 8d, as well as GABA and Baclofen are presented. Of the three best candidates, only 8a is a zwitterionic molecule that presents a dipole as observed in GABA and (R)-Baclofen. A negative-basic region is observed on the carboxylate anion (red color), while a positive-acidic region is present on the amine substitution region. In the case of 8b and 8d, as they are anionic molecules, the MEP’s surface is colored red with some yellow regions where acidic hydrogen atoms are located. 8c is also a zwitterionic molecule that concentrates the acidic region on the bicyclic substituent, but presents less acidic protons in compared to 8a, 8b, and 8d.

A molecular electrostatic potential surface of ±320 kJ/mol was calculated for our three best derivatives to evaluate their interaction elements (Figure 10). The regions where the GABA analogs could interact with positively charged aminoacidic residues at the GABA_B receptor with an electrostatic potential value of −320 kJ/mol are represented as purple mesh surfaces. In all the cases, these regions are located on the carboxylate anion. In 8b, there is an extra basic region on the nitrogen sp²; in 8d, this basic region is extended to both nitrogen atoms (sp² and sp³) located in the bicycle and is a bigger surface than in 8a and 8b. These findings correlate with the MEP maps presented in Figure 9. The acidic regions, with a positive electrostatic potential value, are over all the molecular skeleton, with the highest extensions on the most acid protons; these protons are mostly found in the nitrogen substitution and this finding is better observed in Figure S58.

3.6. Drug-Likeness Prediction Results

ADME properties for all compounds were obtained and are listed in

Table 4. ADME properties values of GABA, Baclofen, Vigabatrin, Gabapentin, valproic acid, SCH–50911, and our candidates.

Molecule	LogS	LogD	LogP	MW	nHA	nHD	TPSA	nROT	PAINS
GABA	0.639	−2.143	−2.598	103.06	3	3	67.77	3	0
Baclofen	−0.674	0.276	−1.127	213.06	3	3	67.77	4	0
Pregabalin	0.379	−0.748	−1.022	159.13	3	3	67.77	5	0
Vigabatrin	0.716	−1.649	−2.280	129.08	3	3	67.77	4	0
Gabapentin	0.402	−1.145	−0.845	171.13	3	3	67.77	3	0
Valproic acid	−0.572	−0.054	1.095	143.11	2	0	40.13	5	0
SCH–50911	0.572	−1.650	−2.550	187.12	4	2	65.97	2	0
8a	0.221	−0.561	−2.364	219.13	3	1	44.57	4	0
8b	0.204	−1.071	−1.063	203.08	4	0	57.95	4	0
8c	0.120	−0.673	−2.431	225.08	3	1	44.57	4	0
8d	0.237	−0.991	−0.727	205.1	4	0	56.26	4	0
(R)-13a	−0.268	1.496	−0.940	275.19	3	1	44.57	6	0
(R)-13b	−3.147	2.513	3.094	260.15	4	1	55.12	6	0
(R)-13c	−0.350	1.298	−1.057	281.14	3	1	44.57	6	0
(R)-13d	−2.666	0.932	1.133	261.16	4	0	56.26	6	0
(R)-14a	−0.69	1.793	−0.591	329.12	3	1	44.57	5	0
(R)-14b	−1.933	0.663	0.916	313.07	4	0	57.95	5	0
(R)-14c	−0.482	1.440	−0.846	335.07	3	1	44.57	5	0
(R)-14d	−3.161	1.079	1.805	315.09	4	0	56.26	5	0

. From the Lipinski parameters [49,50], all molecules are good candidates, as they possess less than 500 umas, LogP values below 5, and less than 5 and 10 HBD and HBA, respectively. TPSA values below ≤140 Ų [51] are considered good values for oral bioavailability, as proposed by Veber et al. [52], although other effects may be considered. In this sense, all molecules analyzed in Table 4 have TPSA below 70 Ų, where the reference compounds have the highest values (~67.77 Ų), while valproic acid has the smallest one (40.13 Ų).

SCH-50911, an antagonist of the GABA_B receptor, also has a similar TPSA value (65.97 Ų), which is very close to other approved drugs. Other drug-likeness parameters, for example the ones proposed by Mozziconacci [53], state that molecules should have less than seven halogens per molecule, as well as oxygen and nitrogen atoms (nO and nN), above one for all compounds, which are seen for all Baclofen analogues. This analysis also includes the total number of rings and rotatable bonds (nRings and nRot) which are below 6 and 15, respectively, consistent with the structure of all molecules that contain heterocycles. Finally, the parameters stated by Ghose et al. [54] are more rigorous than others. For example, molecular weight is considered optimal when molecules are in the range of 160 to 480 g/mol, while for the LogP values are in a range of −0.4 to 5.6. For this analysis, GABA, Pregabalin, Vigabatrin, and valproic acid have less than 160 g/mol, while the rest of the molecules have values below 340 g/mol, where 14c has the highest value (335.07 g/mol). On the other hand, the only acceptable LogP values are encountered for molecules 13b, 13d, 14b, and 14d, in addition to valproic acid.

From the ADME properties of the GABA analogs, it is observed that 8a has a LogP value of −2.364, suggesting better solubility in water. 8b (LogP = −1.063) and 8d (LogP = −0.727) have a decreased solubility in water, which could be beneficial for their activity in contrast to 8a. On the other hand, it can be seen in Figure 11 that molecules 8b and 8d have a greater TPSA value (57.95 Ų and 56.26 Ų) than 8a (44.57 Ų), but low values when compared to GABA or Baclofen (67.77 Ų for both molecules). The ADME properties of Gabapentin and Pregabalin are shown in Figure S59.

8a, 8b, and 8d have a good predicted oral bioavailability. 13a–d and 14a–d present nROT values above 5, which suggest high degrees of freedom in the molecules. Only GABA, Gabapentin, and SCH–50911 have a low number of rotatable bonds; the first two with nROT = 3, while for SCH–50911 only two rotatable bonds are present. 8a–d have four rotatable bonds, which are the main backbone of the molecules, the same number of nROT as Baclofen.

Finally, our QSAR model predicted higher potency for most of the analyzed molecules than Baclofen and, according to our molecular similarity analysis, these compounds possess good chemical properties related to a good ADME profile. Following the molecular docking analysis on both enantiomers (for the (S)-compounds, the information can be viewed in the Supplementary Material), molecules are stated as bioactive, with preference for the (R)-enantiomers. Nevertheless, a biological evaluation of these compounds as their enantiomerically pure form [56] is needed to confirm our results.

4. Conclusions

N-alkylation reactions were obtained in good yields without the use of base, if the electron pair on the nitrogen atom is sufficiently nucleophilic. In the case of heterocycles b and d, the electron pair in the nitrogen atom may delocalize in the aromatic ring, which decreases its nucleophilic character and favors lower yields. On the other hand, the low yields in the 1,4-addition reactions with dialkyl cuprate reagents are attributed to two main factors: the possible incomplete synthesis of the organometallic reagent and the lability of the organometallic reagent.

We propose 8a, 8b, and 8d as potential GABAB receptor agonists because they showed the same intermolecular interactions as GABA and Baclofen in the receptor binding site from the molecular docking analysis.