Synthesis, Pharmacological Evaluation and Molecular Modeling Studies of Triazole Containing Dopamine D₃ Receptor Ligands

Abstract

A series of 2-methoxyphenyl piperazine analogues containing a triazole ring were synthesized and their in vitro binding affinities at human dopamine D₂ and D₃ receptors were evaluated. Compounds 5b, 5c, 5d, and 4g, demonstrate high affinity for dopamine D₃ receptors and moderate selectivity for the dopamine D₃ versus D₂ receptor subtypes. To further examine their potential as therapeutic agents, their intrinsic efficacy at both D₂ and D₃ receptors was determined using a forskolin-dependent adenylyl cyclase inhibition assay. Affinity at dopamine D₄ and serotonin 5-HT_1A receptors was also determined. In addition, information from previous molecular modeling studies of the binding of a panel of 163 structurally-related benzamide analogues at dopamine D₂ and D₃ receptors was applied to this series of compounds. The results of the modeling studies were consistent with our previous experimental data. More importantly, the modeling study results explained why the replacement of the amide linkage with the hetero-aromatic ring leads to a reduction in the affinity of these compounds at D₃ receptors.

Dopamine is a neurotransmitter that plays a number of important roles in the central nervous system (CNS) by activating both presynaptic and postsynaptic dopamine receptors. Dopamine receptors belong to the family of seven transmembrane domain G-protein coupled receptors (GPCRs) and are classified into two types: D₁-like and D₂-like receptors. D₁-like receptors consist of D₁ and D₅ receptors, and D₂-like receptors consist of D₂, D₃ and D₄ receptors. Agonist activation of D₁-like receptors leads to an increase of adenylyl cyclase activity, whereas activation of D₂-like receptors leads to adenylyl cyclase inhibition.¹

Experimental evidence has suggested that changes in the expression of D₃ receptors may be associated with several CNS diseases and behavioral disorders, including psychostimulant abuse. For instance, increased CNS D₃ receptor expression was found in human cocaine overdose victims when compared with age-matched controls.² In addition, several reports suggest that D₃ receptor antagonists and partial agonists attenuate cocaine's rewarding effects.^3-5 An up-regulation of D₃ receptors has been found in animal models of dyskinesia^6-8 and D₃ receptor selective compounds were found to attenuate L-DOPA-induced dyskinetic-like involuntary movements in hemi-parkinsonian rats, suggesting that D₃ receptors could be therapeutic targets for L-DOPA-induced dyskinesia.⁹ Therefore, highly selective D₃ receptor ligands represent potential pharmacotherapeutic agents for the treatment of D₃-related CNS disorders.

There is substantial amino acid sequence homology between the transmembrane helices of the D₂ and D₃ receptors (78% homology)¹⁰, which makes it difficult to develop highly selective D₃ receptor ligands. Our group has developed several benzamide analogues as D₃-selective ligands as candidates including (Figure 1): WC 10 and WC 44.¹¹ Tritium-labeled WC 10 has been used for quantitative autoradiography studies.¹²

Figure 1.

Structure and D₂/D₃ Receptor Binding Data for Compounds WC 10 and WC 44.

In this work, we explored new synthons as replacements for the amide linkage in benzamide based D₃ receptor ligands. 5-Membered heterocycle ring, such as triazole, has been reported to represent an isosteric substitution of an amide linkage.^{13, 14} In addition, the hetero atoms in the triazole ring may participate in hydrogen bonding and dipole-dipole interactions with receptors,^{14, 15} which could result in a more favorable receptor binding affinity. Gmeiner et al first synthesized a series of analogues with 1,2,3-triazole linkers as ligands for GPCR receptors.¹⁶ Micheli et al reported 1,2,4-Triazol-3-yl-thiopropyl-tetrahydrobenzazepines as D₃ antagonists.¹⁷ Carro et al has explored 1,4-disubstituted triazole analogues as D₃ ligands.¹⁸ We have synthesized a series of 1,2,3-triazole based D₃ receptor ligands with reversed substitution pattern from Carro et al's report. The aryl substitutions are on N1 position of the triazole in our series of ligands. Moreover, we explored varied carbon linkages in conjunction with triazole moiety. This letter reports the synthesis, evaluation, and SAR studies of 1,2,3-triazole based D₃ receptor ligands and molecular modeling studies with these ligands.

These analogues were evaluated using in vitro receptor binding assays to determine the binding affinity and selectivity at D₂ and D₃ receptors. Additional assays, such as dopamine D₄, 5-HT_1A and the intrinsic efficacy of selected ligands at D₂ and D₃ receptors were also conducted at D3-selective compounds.

Our group had previously generated comprehensive three dimensional quantitative structure–activity relationship (3D-QSAR) models for a set of 163 D₂/D₃ benzamide compounds based on homology models of the D₂ and D₃ receptors.¹⁹ The homology model of the D₃ receptor is consistent with the D₃ X-ray crystal structure²⁰ (PDB code: 3PBL) (RMSD of the model to the X-ray structure is 2.88 Å, which is lower than the resolution of the X-ray structure). The 3DQSAR models for the compound library also showed excellent correlation and predictive power. Thus, we used the new series of triazole phenylpiperazine analogues as another external test set to evaluate the accuracy of our QSAR models. Moreover, QSAR has provided insights on the different binding mode of this series of ligands versus our previous benzamide ligands.

The target compounds were synthesized using click chemistry as depicted in Schemes 1. Anilines were first converted to the corresponding diazoniums using 1:1 HCl/H₂O as the solvent, in the presence of sodium nitrite. The transformation of diazoniums to azides was then achieved by treating with sodium azide, using sodium acetate to neutralize the excess HCl.

Scheme 1.

Synthesis of Triazole Analogues

The substituted 4-phenypiperazinyl aryl-1-ynes with different carbon spacers were synthesized by N-alkylation of 1-(2-methoxylphenyl)-piperazine and corresponding tosylates, using acetonitrile as the solvent and sodium carbonate as the base. Tosylates were converted from corresponding substituted alcohols by stirring with p-TsCl and triethylamine in dichloromethane.

The target triazole analogues were synthesized by click reactions using classic Cu(I)-catalyzed conditions reported by Fokin et al.²¹

The triazole analogues were evaluated for affinity at human D₂ and D₃ dopamine receptors expressed in stably transfected HEK cells (Table 1) using a competitive inhibition binding assay, which was previously reported.²² The ligand binding selectivity is calculated as K_i (D₂)/ K_i (D₃). Analogues that exhibited >10-fold binding selectivity for dopamine D₃ receptors compared to the D₂ receptor subtype were further evaluated for a) affinity at D₄ dopamine receptors and 5-HT_1A serotonin receptors; b) intrinsic efficacy at D₂ and D₃ receptors using a forskolin-dependent adenylyl cyclase inhibition assay.

Table 1.

D₂, D₃, D₃, 5-HT_1A Affinities and Log D Values of Triazole Analogues

Ar	Carbon Spacer	Compound number	D2 (Ki nM)	D3 (Ki nM)	D2/D3 Ratio	D4 (Ki nM)	D4/D3 Ratio	5-HT1A (Ki nM)	Log Db
		WC 10	34.4 ± 3.3	0.8 ± 0.1	43	896 ± 193	1120	7.5	3.1
		WC 44	54.5 ± 4.4	2.4 ± 0.4	23	804 ± 33	335		2.9
	3	3a	64.0 ±11.4	22.2 ± 3.8	2.9				3.8
	4	4a	30.9 ± 3.9	4.7 ± 0.4	6.6				3.9
	5	5a	19.4 ± 2.5	3.1 ± 0.2	6.3				4.7
	3	3b	79.3 ± 5.3	36.3 ± 1.3	2.2				3.5
	4	4b	42.3 ± 5.2	3.7 ± 0.5	11.5	875 ± 117	237	6.0 ± 1.3	3.7
	5	5b	17.7 ± 0.7	1.8 ± 0.3	9.8	186 ± 23	103	5.8 ± 1.5	4.4
	3	3c	158 ± 15	8.9 ± 2.1	17.7	90± 19	10	7.4 ± 0.5	5.2
	4	4c	94.3 ± 12.7	8.2 ± 1.8	11.6	395 ± 13	48	1.3 ± 0.2	5.4
	5	5c	109 ±19	6.3 ±1.2	17.3	163 ± 10	26	153 ±12	6.1
	3	3d	73.5 ± 9	26.5 ± 1.1	2.8				3.8
	4	4d	46.7 ± 1.2	6.5 ± 0.7	7.2				4.0
	5	5d	23.7 ± 4.2	3.2 ± 0.5	7.4				4.8
	3	3e	272 ± 42	93.1 ± 8.8	2.9				3.7
	4	4e	139 ± 3	25.4 ± 2.3	5.5				3.8
	5	5e	62.7 ± 7.6	21.1 ± 1.9	3.0				4.6
	3	3f	162 ± 13	29.6 ±5.2	5.5				3.9
	4	4f	171 ± 13	13.9 ±1.9	12.4	758 ± 23	55	9.4 ± 1.1	4.0
	5	5f	62.6 ± 6.1	6.7 ± 0.6	9.3				4.8
	3	3g	44.1 ± 2.0	9.6 ± 2.3	4.6				4.2
	4	4g	37.7 ± 4.2	2.1 ± 0.3	18.1	200 ± 26	95	5.3 ± 0.6	4.4
	5	5g	26.6 ± 3.6	2.3 ± 0.3	11.6	227 ± 14	99	3.9 ± 0.9	5.2
	3	3h	12.8 ± 1.7	6.8 ± 0.84	1.9				3.0
	4	4h	9.4 ± 1.1	2.9 ± 0.6	3.2				3.1

^bCalculated using ACD log D software, Advanced Chemistry Development, Toronto, Canada. All Ki values for the affinity at D2 and D3 receptors are reported as nanomolar values and are the mean value ± S.E.M. for n > 3.

The radioligand binding analysis identified several triazole analogues with ≥10-fold selectivity for D₃ versus D₂ receptors including compounds 4b, 5b, 3c, 4c, 5c, 4f, 4g, and 5g. Among these compounds, 4b, 5b, 4g, and 5g have the highest affinity, with K_i values for the D₃ receptor ≤ 4 nM.

Compound 4a, which is the reversed substituted analogue of compound 15 in Carro et al's paper,¹⁸ shows a much higher affinity at D₃ receptor and higher D₂/D₃ selectivity compared to 15 (Figure 2).

Figure 2.

Comparison between Compound 4a and 15

Another observation from the experimental data is that varying the length of the carbon spacer group while maintaining the same aromatic substituents changes both D₂ and D₃ receptor affinity. It was found that triazole analogues with a three-carbon spacer generally exhibits lower D₂ and D₃ affinity than their four or five-carbon-spacer analogues. When the carbon spacer group increases from four to five, D₂ receptor affinities generally increase to greater extent than D₃ receptor affinities. It leads to the fact that although most of the five-carbon spacer compounds have the highest D₃ receptor affinity, they exhibit lower D₂ versus D₃ selectivity.

The structure-activity relationship was further explored by changing the substituents on the aromatic ring. It was found that compounds with highest affinity and/or selectivity are para-substituted. For example: 3c, 4c, 5c, 5d, 5f, 4g, and 5g.

Unfortunately, when amide bonds were replaced with triazoles, both D₃ receptor affinity and selectivity decreased. For example, WC 10 (K_i D₃= 0.8 nM, D₂/D₃ = 43) versus the triazole analogue 4f (K_i D₃ = 13.9 nM, D₂/D₃ =12).

Compounds 4b, 5b, 3c, 4c, 5c, 4f, 4g, and 5g were selected for further binding assays because their selectivity for the D₃ versus the D₂ receptor • 10-fold. The affinity of these compounds was found to be low at the D₄ dopamine receptor subtype. Additionally, as we previously reported, D₃-selective benzamide analogues can exhibit high binding affinity at 5-HT_1A receptors.¹¹Unfortunately, the triazole analogues did not improve this property. All of the eight compounds we tested except 3c bind to 5-HT_1A receptors with high affinity. It is necessary to point out that the replacement of amide with triazoles increases the ligands’ logD value.

An adenylyl cyclase inhibition assay was used to determine the intrinsic activity of compounds 4b, 5b, 3c, 4c, 5c, 4f, 4g, and 5g. All test compounds were used at a concentration equal to 10 × the calculated K_i value; therefore 90% of the D₂ or D₃ binding sites were occupied (Table 2). For each compound, the inhibition was compared to the intrinsic activity of quinpirole (full agonist) and haloperidol (antagonist). All of the compounds were found to be partial agonists at D₂ dopamine receptors, with the intrinsic activity ranging from approximately 31-75% of the full agonist. However, for D₃ receptor a broader range of intrinsic activity was observed (from 13% efficacy to a 107% full agonist). Interestingly, compound 4f is a full agonist at D₃ receptors, while its benzamide analogue, WC 10 is a D₃ receptor antagonist/weak partial agonist (18.7%). ¹¹

Table 2.

Intrinsic Activity of Selected Analogues at Dopamine D₂ and D₃ Receptors^c

Compound	hD2 HEK	hD3 HEK
Haloperidol	−0.6 ± 1.6	4.0 ± 5.5
WC 10	33.5 ± 3.1	18.7 ± 2.2
WC 44	35.3 ± 1.0	96.2 ± 4.2
3c	53.8 ± 6.1	50.5 ± 9.9
4b	49.0 ± 3.2	48.2 ± 5.4
4c	67.8 ± 5.3	72.4 ± 4.4
4f	38.8 ± 7.6	107.0 ± 19.3
4g	41.1 ± 8.8	39.4 ± 5.0
5b	43.1 ± 5.4	25.9 ± 6.3
5c	31.4 ± 5.0	67.8 ± 15.6
5g	75.4 ± 5.5	82.2 ± 10.3
Quinpirole	100	100

^cThe intrinsic activity of the test compounds was evaluated by determining the percent inhibition of a forskolin-dependent whole cell adenylyl cyclase assay. The results were normalized to the percent inhibition obtained using the full agonist quinpirole at human D₂ (1 μM) and D₃ (100 nM) receptors expressed in stably transfected HEK 293 cells. For D₂ receptors the maximum inhibition was ±90% and for D₃ receptors the maximum inhibition ranged from 38 to 53%. The test drug was used at a concentration equal to approximately 10× the K_i value that was determined from the radioligand binding analysis. The mean ± the S.E.M. values are reported for n ≥ 3.

Computational molecular modeling studies were initiated to begin to define at a molecular level how triazole phenylpiperazines bind to the D₂ or D₃ receptors. The Comparative Molecular Similarity Index Analysis (CoMSIA1) for both D₂ and D₃ provided results in good agreement with the experimental values for ligand-receptor affinity (Supplementary Table 2). Because their values only span one to two orders of magnitude, no statistical analysis was performed. However, a) the residuals are all reasonably small and b) overall trends for the compounds’ predicted affinities agree with the experimental binding data on both D₂ and D₃. These modeling studies indicate that the compounds have similar binding modes at the D₂ receptor site as their benzamide counterparts (WC series), while losing about 10-fold affinity at D₃ site.

In the molecular modeling study, both CoMSIA1 models utilized a combination of steric, electrostatic, hydrophobic and hydrogen bond donor and acceptor fields. An analysis of the contribution of each of these components suggests that hydrophobic interactions and hydrogen binding are the major determinant factors for the compounds’ binding affinities.

The replacement of the amide bond in WC series with triazole was found to reduce both the affinity and selectivity to D₃ receptors compared to the WC series (WC 10 vs. 4f). A closer examination of our D₃ receptor-ligand binding model reveals that, besides the salt bridge between the highly conserved residue Asp3.32 in the transmembrane (TM) 3 helix of the receptor and the protonated nitrogen on the piperidine scaffold, the benzamide group on WC 10 potentially forms a critical hydrogen bond with the D₃ receptor (Figure 3). This hydrogen bond is between the amine on the benzamide group and the carbonyl of the backbone from C181 on extracellular loop (EL) 2 of D₃. C181 is a highly conserved cysteine residue, which is involved in the formation of a disulfide bond between EL2 and the third transmembrane (TM 3) spanning helix. This disulfide bond plays an important role in the stabilization of the GPCR structure. Consequently, the ligand is securely anchored at the binding site, with its benzamide group extending toward the extracellular opening of the binding pocket. The extension of this amine-containing scaffold into a second binding pocket is crucial for high D₃ receptor binding affinity.²⁰

Figure 3. Binding of WC 10 to the Human D₃ Dopamine Receptor.

The binding of WC 10 and the human D₃ receptor model is shown. A stick model is shown for the backbone atoms of the D₃ receptor. Oxygen is depicted in red and the nitrogen atoms are depicted in blue. The hydrogen bond between the C181 carbonyl and the amide nitrogen of WC 10 is shown as a dashed green line. This hydrogen bond appears to be a pivotal interaction for the binding of arylamide phenylpiperazines. The triazole and isoxazole analogues _are unable to make this pivotal hydrogen bond. Therefore an interaction that is essential for guiding the aryl moiety of the benzamide towards a second binding pocket has been eliminated.

This second binding pocket is delineated by the residues at the junction of EL1, EL2 and the interfaces of TM helices 1, 2 and 7 and opens to the extracellular regions. This region was described in the dopamine D3 crystal structure paper as a secondary binding pocket,²⁰ and recently, as a allosteric site for dopamine D3 receptors.²³ The four-carbon linker on the ligand scaffold is conformationally flexible. The benzamide hydrogen bond anchors the ligand and helps to orient it toward the secondary binding pocket. Therefore, interactions between the substituents on the benzene ring with amino acid residues in the secondary binding site become possible.

The loss of this hydrogen bonding in the triazole analogues eliminates an important interaction that is essential for guiding the aryl moiety (of the arylamide) towards a second binding pocket (Figure 4). The importance of the carboxamide linker for achieving D₃ vs. D₂ receptor binding affinity and selectivity has been previously reported. ²⁴ The current work further defines the role that this pivotal pharmacophore plays in the development of dopamine D₃ receptor selective ligands.

Figure 4. Binding Modes for Triazole Analogues of WC 10 with Varying Carbon Spacer Lengths in the Human Dopamine D₃ Receptor Binding Site.

A comparison of the alignments of the triazole analogues (3f, 4f, 5f), with WC 10 in the D₃ dopamine receptor binding site is shown. Oxygen is depicted in red and the nitrogen atoms are depicted in blue. The hydrogen bond between the carbonyl group of C181 and the amide nitrogen of WC 10 is shown as a dashed green line. A composite comparison of the binding 3f, 4f and 5f in the D₃ receptor binding site is shown.

Alignments for the analogues WC 10, 3f, 4f and 5f on the molecular field maps of CoMSIA1 derived for D₃ shows that longer carbon spacer compounds (4f and 5f) have the aromatic ring sticking into the hydrophobic favorable regions within the second binding pocket (Figure 5B). Since hydrophobic interactions are a major contributor to the ligand affinity, we would expect the compounds with the longer spacer chain to bind with higher affinity. This also partially explains why a substituent in the para position of the aromatic ring enhances activity. However, in our D₂ receptor model, this class of compounds binds to the D₂ receptor with an L-shaped carbon spacer chain. As a result, the alignment of the triazole analogues overlaps quite well with WC 10 (Figure 5A). Therefore, these analogues have comparable binding affinities with the WC series at the D₂ receptor. These modeling studies suggest that the origins of D₂ vs. D₃ receptor subtype selectivity for the ligands arise primarily from contour differences of the two binding sites. Therefore ligands must exploit subtle topographical differences within the two ligand binding sites to achieve high selectivity between the D₂ and D₃ dopamine receptor subtypes.

Figure 5. Structural Alignment of WC and the Triazole Analogues as They Occupy the D₂ and D₃ Dopamine Receptor Binding Sites.

A structural alignment of WC 10 with triazole compounds 3f, 4f and 5f is shown as they occupy the D₂ and D₃ dopamine receptor binding sites. Oxygen is depicted in red and nitrogen atoms are depicted in blue. Carbon atoms are colored in white for WC 10, in cyan for 3f, in yellow for 4f, and in green for 5f. Hydrogen atoms are undisplayed for clarity. The contours are displayed at the 80% favored and 20% disfavored level. A. The triazole analogues 3f, 4f, and 5f aligned together with WC 10 at the D₂ receptor binding pocket. Each of the ligands assumes a bent configuration. B. The triazole analogues (3f, 4f, and 5f) together with WC 10 are illustrated by a stick model in the hydrophobic molecular field map based on our previously generated CoMSIA1 model for D₃ receptor. When bound to the D₃ receptor, each of the ligands assume a more linear configuration than that observed for D₂ receptor binding. Increased binding affinity is correlated with more hydrophobic near orange, less hydrophobic near violet.

In summary, we have synthesized and evaluated a series of triazole analogues which showed moderate to high affinity for D₃ receptors and variable selectivity for D₃ versus D₂ receptors. For this series of compounds we identified several compounds that exhibited high D₃ binding affinity (<4.0 nM) with moderate D₃ versus D₂ receptor binding selectivity. The SAR could be partially explained by our previously generated 3D-QSAR models for selective dopamine receptor ligands. This study expands the SAR of the D₃-selective ligands and our knowledge on the binding modes of dopamine receptors. The development of new class of compounds for dopamine D3 receptors is ongoing.

Abbreviations

3D-QSAR models

three dimensional quantitative structure–activity relationship models

CoMSIA1

Comparative Molecular Similarity Index Analysis

EL1

Extracellular Loop 1

EL2

Extracellular Loop 2

GPCR

G-protein coupled receptor

RMSD

Root Mean-Square Deviation

SAR

Structure–Activity Relationship

TM

Transmembrane