Synthesis of novel 5-HT1A arylpiperazine ligands: Binding data and computer-aided analysis of pharmacological potency

2.1 Experimental

The reagents and solvents used in this work were obtained from Alfa Aesar or Sigma-Aldrich and used without further purification. Solvents were routinely dried over anhydrous Na₂SO₄ before evaporation. ¹H- and ¹³C NMR spectra were recorded on: ¹H NMR (200 MHz) and ¹³C NMR (50 MHz), Gemini 2000 spectrometer; ¹H NMR (400 MHz) and ¹³C NMR (101 MHz), Varian/Agilent spectrometer; and ¹H NMR (500 MHz) and ¹³C NMR (126 MHz), Bruker Avance III spectrometer. Chemical shifts (δ) are reported in parts per million (ppm) from tetramethylsilane (TMS) as an internal standard in deuterated chloroform (CDCl₃); all coupling constants (J values) are reported in Hz; high-resolution mass spectra (HRMS) were obtained with a heated ESI (HESI)-LTQ Orbitrap XL spectrometer. Melting points were obtained by a Boetius PHMK apparatus (VEB Analytic, Dresden, Germany) at a heating rate of 4 °C/min and are uncorrected. IR spectra were recorded using a Thermo Scientific Nicolet 6700 Fourier-transform spectrometer operated in the ATR mode. For analytical thin-layer chromatography (TLC), Polygram SIL G/UV254 plastic-backed thin-layer silica gel plates were used (Macherey-Nagel, Germany).

2.2 Binding assays

The affinity of the synthesized compounds towards 5-HT1A was determined in radioligand competition binding assays. The assays were performed using crude cell membrane homogenates obtained from HEK cells stably transfected with human 5-HT1A receptor and the 5-HT1A specific radioligand [³H]–OH-DPAT (obtained from PerkinElmer, NET929250UC). Membrane suspension was incubated with 1 nM [³H]–OH-DPAT and various concentrations of the test compound in 25 mM Tris-HCl, pH 7.4 buffer containing 120 mM NaCl, 5 mM KCl at room temperature for 60 min. Non-specific binding was determined in the presence of 10 µM of serotonin. The reaction was terminated by rapid filtration using Whatman GF/B glass-fibre filters, pre-soaked in 0.3 % polyethyleneimine, and a 48-channel harvester (Biomedical Research and Development Laboratories, Gaithersburg, MD, USA) followed by washing four times with ice-cold TRIS-HCl buffer. Filter-bound radioactivity was quantified by liquid scintillation counting. At least three separate experiments in triplicate were performed to determine inhibitory constant (Ki) values. The data were analyzed by GraphPad Prism, Version 4.1 (GraphPad Inc., La Jolla, CA). The Ki was calculated according to the Cheng-Prusoff equation:

Ki = IC₅₀/(1 + ([L]/K_D)) where Ki is the inhibition constant, IC₅₀ is the concentration of competitive inhibitor that displaces 50 % of the specifically bound labelled ligand, [L] is the concentration of the labelled ligand, and K_D = the affinity constant of the labelled ligand.

2.3 Absorption, distribution, metabolism, excretion and toxicity (ADMET) analysis

To predict absorption, distribution, metabolism, and excretion (ADME) qualities of tested ligands, we used the SwissADME webserver (www.swissadme.ch) (Daina et al., 2017). Toxicology prediction was done through Pro Tox-II virtual lab server for the prediction of toxicities of small molecules (https://tox-new.charite.de/protox_II/index.php?site=home) (Drwal et al., 2014). Ligand structures were provided as SMILES using ChemDraw.

2.4 Docking simulations

Docking simulations were done in Maestro Suite software (Schrödinger, 2018). 3D model of the 5-HT1A receptor with bound aripiprazole (PDB Code 7E2Z) was obtained from the GPCR database (Xu et al., 2021). 2D structures of ligands were drawn in ChemDraw software, and prepared for docking in Maestro software using default LigPrep procedures.

Induced fit docking (IFD) simulation using standard sampling protocol and default values, was carried out for prepared receptor model and selected ligands (Sherman et al., 2006). Bind site was defined, based on bound aripiprazole and centered on ASP 116 residue. The inner grid box was set to 10x10x10 Å and the outer box size according to the size of each tested ligand. Grid spacing was set to 1 Å. Obtained docking structures were examined and selected for further analysis based on the number of receptor-ligand interactions and calculated post docking MM-GBSA energy.

2.5 Cytochrome P450 site of metabolism analysis

In addition to basic sites of metabolism analysis (SoMs) prepared with NERDD (https://nerdd.univie.ac.at/) FAME3 (Šícho et al., 2019) / GLORY (de Bruyn Kops et al., 2019) modules, Maestro Suite (Schrödinger, 2018) P450 Site of Metabolism workflow was used to predict the metabolic stability of tested ligands. All three cytochrome available isoforms (PDB code: 2C9, 2D6 and 3A4) were used for the analysis. All calculation parameters were left at default values.

3.1 Chemistry

The general synthetic route and chemical structures of the compounds synthesized in the present study are shown in Scheme 1.

Close

Scheme 1

Synthetic pathways for N-{3-[2-(4-aryl-piperazin-1-yl)ethyl]phenyl}arylamides and tabelar representation of aryl part of new and parent compounds. Reagents and conditions for the synthesis of 4a,b-10a,b: (i) Arylpiperazine, Et₃N, 10 °C, propylphosphonic acid anhydride (PPAA), N,N-dimethylformamide (DMF), r.t.; (ii) B₂H₆, THF, 0 °C for 6 h, r.t. for 1 h, then reflux for 2 h; (iii) Ra/Ni, NH₂NH₂, EtOH, 1,2-dichloroethane; (iv) ArCO₂H, Et₃N, PPAA, DMF, r.t. over night. Parent ligands 4c^a-10c^a (R = R¹ = H) were previously synthesized (Sukalovic et al., 2013).

Export to PPT

2-Methoxy and 2,3-dichloro substituents were introduced into the phenylpiperazine part of the already active parent compounds (4c-10c, R = R¹ = H, Scheme 1, Fig. 2) (Sukalovic et al., 2013) to improve affinity toward 5-HT1A receptors. In brief, N-{3-[2-(4-aryl-piperazin-1-yl)ethyl]phenyl}arylamides were obtained: by acylation of N-arylpiperazine using 3-nitrophenylacetic acid and gave rise to 2-(3-nitrophenyl)-1-(4-arylpiperazin-1-yl)ethan-1-ones (1a,b), followed by reduction of amides 1a,b using diborane in tetrahydrofuran (THF). 1-(3-nitrophenethyl)-4-arylpiperazines (2a,b) were reduced by Raney/Ni to provide 3-[2-(4-arylpiperazin-1-yl)ethyl]anilines (3a,b). Target arylmides 4a,b-10a,b were obtained by condensation of anilines 3a,b with corresponding aryl acid in the presence of propylphosphonic acid anhydride (PPAA) in N,N-dimethylformamide (DMF). All the compounds were characterized by ¹H and ¹³C NMR spectroscopy and mass spectroscopy (for HRMS spectra of new ligands 4a,b-10a,b please, see Suppl. S1).

3.2 5-HT1A receptor affinity

As presented in Table 1, eight (4a-b; 5a-b; 6b; 7a; 8a-b) of the 14 newly synthesized ligands had an enhancement in affinity for 5-HT1A receptors compared to parent compounds (4c-8c). Analysis of the obtained experimental affinity values of the newly synthesized ligands and the affinities of the parent compounds shows that the conspicuous increase was observed with ligands 5a, followed by 8a, 8b, and 5b. The decreasing order of increase of other ligands is 4a > 4b > 6b > 7a (from moderate to mild). In the case of the other newly synthesized ligands, the desired, pronounced increase in affinity was absent (for a diagram processing the results, please see Suppl. S2). Also, in the analysis of the relationship between the experimental affinity values of the newly synthesized ligands and aripiprazole, the most pronounced increase was observed for ligands 8a and 8b (2.1 and 1.8-fold, respectively). Obtained experimental data showed that in the case of starting structures 9c and 10c, neither methoxy nor 2,3-dichloro substitution did not lead to an affinity increment towards the 5-HT1A receptors. The same is true for resulted compounds 6a and 7b. The proposed modification did not result in a significant increase in affinity.

Two compounds stood out as representatives of the successful approach, N-(3-{2-[4-(2-methoxyphenyl)piperazin-1-yl]ethyl}phenyl)isonicotinamide (8a), and N-(3-{2-[4-(2,3-dichlorophenyl)piperazin-1-yl]ethyl}phenyl)isonicotinamide (8b), with Ki values of 0.8 nM and 0.9 nM, respectively. Presented Ki values correspond to commercial, structurally resembling drugs.

3.3 ADME prediction and toxicity

Determining a compound’s toxicity and absorption, distribution, metabolism, and excretion characteristics is a material step in the drug development process.

Investigating ADME and toxicology through animal experiments is expensive and time-consuming. On the other hand, the prediction of ADMET in silico is a fast and affordable alternative (Banerjee et al., 2018a; Daina et al., 2017). The method relies on existing databases to construct a capable model for the ADMET prediction of novel compounds (Table 1), where results of Lipinski, Muegge and Ghose rules are presented as the total number of violations (for the detailed ADMET profile of all tested ligands, and criteria for violation of Lipinski, Muegge and Ghose rules, please see Suppl. S3 and S4) (Banerjee et al., 2018b; Drwal et al., 2014). Results of toxicology prediction suggested that substances 5a and 7a have the potential to be immunotoxic. Performed comparative study of two different formulas of aripiprazole (free base crystal or cocrystal formula) suggested that aripiprazole, in prescribed doses, is not cytotoxic or immunotoxic. The same is true in the case of cariprazine and brilaroxazine (Baek et al., 2015; Citrome, 2013; Bhat et al., 2018).

ADME analysis pointed out that several newly synthesized compounds suffer from high molecular weight and low solubility (4b, 5b, and 6b). Both parameters are substantial for bio-oral availability (substance ability to be administrated and transported through the body). High molecular weight, by itself, doesn’t have to be eliminating factor. Some commercially available substances fall into this category but, combined with low solubility, rule out ligand as a desirable drug candidate. Betterment of solubility was not the subject of this research. Therefore substances 4b, 5b, and 6b were excluded from a further survey (Table 1). After ADMET analysis, ligands 4a, 8a, and 8b emerged as possible candidates for the additional analysis (for the ADMET profile of all tested ligands, please see Suppl. S3 and S4).

3.4 Docking simulations, molecular dynamics, and electrostatic potential calculations

Ligands 4a, 8a, and 8b were subject to docking simulations, molecular dynamics (MD), and electrostatic potential calculations (ESP) (Schrödinger, 2018; Bowers et al., 2006). A potential drug candidate should make a stable receptor-ligand complex, forming interactions comparable to commercial substances with a similar structure (Table 1).

Since the 5-HT1A receptor model with bound aripiprazole is a readily available crystal structure (PDB code 7E2Z) (Xu et al., 2021), all tested substances were docked into the receptor using the Induced Fit docking protocol (IFD) (Sherman et al., 2006). IFD protocol can take into account both flexible ligand and flexible binding site to closely simulate natural conditions.

All docked substances share the same bind cavity and number of key interactions with aripiprazole (and other selected commercial substances). The superposition of docked ligand 8a and aripiprazole inside the receptor bind site was shown in Fig. 3 (for the interactions of all superposed ligands, please see Suppl. S5).

Close

Fig. 3

Superposition of docked ligand 8a (red) and aripiprazole (green) inside the receptor bind site with interacting residues forming OBS (yellow) and EBS (blue). Only key amino acid residues are shown for clarity.

Export to PPT

The formation of a short salt bridge with Asp 116, the first requirement for high ligand affinity, was observed in all docking runs. Arylpiperazine part of the ligand has correct orientation inside the orthosteric bind site (OBS), formed by TM 3, TM 5, and TM 6 helices. The remaining part of the ligand binds into the extended bind site (EBS), made in part by TM 2, TM 7 and extracellular loop 3 (ECL3) (Fig. 3).

ESP calculations show compatible positive and negative areas on ligands that correspond to key amino acid residues on the receptor. The positive area around protonated nitrogen atom orients ligand toward negatively charged Asp 116. The negative area inside the aryl ring of the ligands arylpiperazine part forms edge-to-face interactions with Phe 361 and 362. The introduction of the methoxy group in position 2 of the aryl part enhances mentioned negative area and facilitates hydrogen bonding with Thr 121 (4a and 8a). Much the same is achieved with the introduction of 2,3-dichloro substitution in compound 8b, but this time hydrogen bond is formed with Ser 199 (Table 2).

Additional affinity gains come from ligand interactions with EBS. Here, the optimal orientation of corresponding ligand functional groups is of the utmost importance. Hydrogen bond with Gln 97 and aromatic interactions with Tyr 96 and Trp 387 are needed to further enhance affinity (Table 2). MD simulations, carried over 100 ns, showed that all obtained docking poses are stable and that crucial interactions hold more than 20 % of the overall interaction time (Fig. 4).

Close

Fig. 4

2D Diagrams of selected ligands key receptor-ligand interactions with 3D ESP map (a, c, e) and RMSD profiles throughout the 100 ns MD simulation (b, d, f): for ligand 4a (a and b); for ligand 8a (c and d); for ligand 8b (e and f). Only key amino acid residues are shown for clarity. Aromatic interactions are shown as green and electrostatic interactions are pink lines. Numbers in square denote the percentage of time interactions are observed during the 100 ns MD run. Negative ESP is red, while positive is blue.

Export to PPT

3.5 Prediction of drug metabolism

In vivo interaction of drug molecules with Cytochromes P450 superfamily of metabolic enzymes can cause the breakdown or alteration of their structure (Tyzack et al., 2017; Tyzack and Kirchmair, 2019). The change in character or composition, typically in a comparatively small but significant way, happens at the sites of metabolism (SoMs) and results in a modified, usually more polar structure. Consequently, formed metabolites have different pharmacological and toxicological properties regarding the parent compound (Kirchmair et al., 2013; Tyzack and Glen, 2014).

Prediction of SoMs and metabolites was completed using the NERDD server and FAME3 (Šícho et al., 2019) / GLORY (de Bruyn Kops et al., 2019) protocol. Due to the chemical nature of compounds 4a, 8a and 8b, the location of their major SOM is in the arylpiperzine part of the molecule and foreseen metabolites correspond to metabolic modifications in this part of the molecule. Although another SOM is predicted in the aryl amide part, this metabolic modification is minor and less likely (for the complete list of ranked metabolites, please see Suppl. S6). Predicted metabolites are summarized in Table 3 and their modifications are comparable to known metabolites of commercial substances, namely aripiprazole. Aripiprazole is metabolized in the liver by the cytochrome P450 isoenzymes CYP3A4 and CYP2D6. Dehydroaripiprazole is a predominant metabolite, along with products of N-dealkylation, dehydrogenation, and hydroxylation (in the same manner as shown for 8b, Table 3) (Kirschbaum et al., 2008).

Table 3 Predicted metabolites that can be formed in humans by enzymes belonging to the cytochrome P450 (CYP) enzyme family of selected ligands, their binding energies, and ADME characteristics.

Ligand No	Metabolite structure	Predict. Score	Tox.	ADME						MM-GBSA energy (kcal/mol)
				L.	M.	G.	C*logP	M.W. (g/mo)	ESOL Class
4a		4.12	inactive	0	0	0	3.54	401.50	Moderately soluble	−30.15
		4.12	inactive	0	0	1	3.34	431.53	Moderately soluble	−31.21
8a		4.24	inactive	0	0	0	2.80	402.49	Moderately soluble	−36.54
		4.24	inactive	0	0	0	2.66	418.49	Moderately soluble	−37.60
8b		2.77	inacitve	0	0	1	3.84	471.38	Moderately soluble	−34.16

Designation of: Predict. Score-Prediction Score; Tox.-Toxicity; Total number of violations (L.-Lipinski; M.- Muegge; G.- Ghose); C*logP-Consensus logP. Only the most probable rank 1 metabolites are included, as well as the most probable products according to GLORY prediction. For a complete predicted metabolite list, please see Suppl. S6 and S7.

Structurally, formed metabolites include the addition of –OH group in position 4 on arylpiperazine part in case of ligand 8b, and for ligands, 8a and 4a, methoxy group transformation into –OH or -O-CH₂-OH group. Their ADMET profile is very much the same compared to parent compounds. To determine the mode of metabolite interaction with 5-HT1A, we performed the IFD docking of listed metabolites (Table 3) and compared them with the initial docking results. All tested metabolites can bind to the 5-HT1A receptor. Interaction with the 5-HT1A receptor takes place in the same or altered way, compared to the parent substances.

In the case of metabolites of ligands 8a and 4a, all key interactions formed by the parent substances and metabolites are identical (Fig. 5).

Close

Fig. 5

2D docking schematics of selected metabolite products of a) 8a and b) 8b, with the key receptor-ligand interactions (salt bridge–red, hydrogen bonds–blue, and aromatic interactions–green).

Export to PPT

Ligand 8b metabolite with –OH group at position 4 of the arylpiperazine ring can form a stronger hydrogen bond with Ser 199 compared to the parent compound. Thus, 8b metabolite may benefit from this interaction. Predicted MM-GBSA energies, put all tested metabolites affinities in the similar range as parent compounds (for the ADME profile and 2D diagrams of all tested metabolites, please see Suppl. S7 and S8).