D₂ Dopamine Receptor G Protein-Biased Partial Agonists Based on Cariprazine

Abstract

Functionally selective G protein-coupled receptor ligands are valuable tools for deciphering the roles of downstream signaling pathways that potentially contribute to therapeutic effects versus side effects. Recently, we discovered both G_i/o-biased and β-arrestin2-biased D₂ receptor agonists based on the Food and Drug Administration (FDA)-approved drug aripiprazole. In this work, based on another FDA-approved drug, cariprazine, we conducted a structure–functional selectivity relationship study and discovered compound 38 (MS1768) as a potent partial agonist that selectively activates the G_i/o pathway over β-arrestin2. Unlike the dual D₂R/D₃R partial agonist cariprazine, compound 38 showed selective agonist activity for D₂R over D₃R. In fact, compound 38 exhibited potent antagonism of dopamine-stimulated β-arrestin2 recruitment. In our docking studies, compound 38 directly interacts with S193^5.42 on TM5 but has no interactions with extracellular loop 2, which appears to be in contrast to the binding poses of D₂R β-arrestin2-biased ligands. In in vivo studies, compound 38 showed high D₂R receptor occupancy in mice and effectively inhibited phencyclidine-induced hyperlocomotion.

Graphical Abstract

INTRODUCTION

G protein-coupled receptors (GPCRs) are seven-transmembrane domain receptors¹ that have been implicated in the pathophysiology of numerous human diseases including schizophrenia,² Parkinson’s disease,³ attention-deficit disorders,⁴ and obesity.⁵ More than 30% of current Food and Drug Administration (FDA) approved drugs target GPCRs. ^1,6,7 Historically, GPCR drug discovery has postulated that ligand engagement with a receptor could only lead to a single unique outcome.¹ Under this simplified paradigm, partial and full agonists were classified as ligands that stimulated the recognized downstream signaling pathway, whereas antagonists were ligands that simply blocked the actions of agonists.⁸ Current research, however, suggests that GPCR ligands can induce conformational changes that result in selective activation of distinct downstream signaling pathways.^{1, 7} This concept has been widely accepted by the academic and pharmaceutical communities and is termed biased signaling or functional selectivity.^9–14

Functional selectivity in GPCR ligands has important potential clinical implications as it has been suggested that different signaling pathways contribute to various drug pharmacological profiles including both therapeutic actions and adverse effects.^15–17 For example, carvedilol and metoprolol tartrate are both inverse agonists of cyclic adenosine monophosphate (cAMP) production at β₂AR, but carvedilol has also been shown to possess phospho-ERK1/2 activity, likely leading to the reduction in mortality observed in a clinical trial on patients with chronic heart failure.^18,19 Another example includes PZM21, a potent Gi-biased agonist of the μ-opioud-receptor (μOR) that was shown to efficiently reduce pain without causing lethal side effects, such as fatal respiratory depression and morphine-like reinforcing activity. In contrast, both of these side effects are observed alongside therapeutic analgesia when the unbiased μOR agonist morphine is administered.²⁰ These findings have fueled great interest in the discovery of new functionally selective ligands for various GPCRs.^13,21–36

Dopaminergic receptors are a subfamily of aminergic GPCRs that are highly expressed in the brain. Dopaminergic receptors are generally classified into subtypes as D₁-like (D₁ and D₅) and D₂-like (D₂, D₃, and D₄). D₂R is the most highly studied dopaminergic receptor due to its implication in neuropsychiatric diseases.^37,38 Previously, we discovered both β-arrestin2-biased and G protein-biased D₂R agonists through investigation of the structure–functional selectivity relationship (SFSR) of the FDA-approved drug aripirazole.^35,39–41 Other groups have also reported multiple functionally selective D₂ receptor agonists.^{27–30,42–45} Among them, cariprazine and its 1,2,3,4-tetrahydroisoquinoline (THIQ) isosteres were reported to selectively activate the D₂R G_i/o-mediated cAMP inhibition pathway over the phospho-ERK1/2 pathway.²⁷ However, for these cariprazine analogues, there are no reports on ligand bias with regard to β-arrestin recruitment.

Cariprazine is an FDA-approved drug for the treatment of schizophrenia and bipolar mania in adults and is currently also under clinical investigation for the treatment of bipolar depression and major depressive disorder.⁴⁶ Cariprazine, however, is also known to carry several side effects including extrapyramidal symptoms, indigestion, nausea, headache, and weight gain.⁴⁷ A functionally selective cariprazine-biased D₂R agonist has the potential to serve as a tool for dissecting the therapeutic and adverse effects of different downstream pathways, which may ultimately lead to improved therapeutics. In the present study, we conducted an SFSR study starting from the cariprazine scaffold and report the discovery of compound 38, a THIQ isostere of cariprazine, as a G protein-biased D₂R partial agonist. Here, we present the design, synthesis, and biological characterization of compound 38 and its analogues.

RESULTS AND DISCUSSION

Discovery of Compound 16 as a G Protein-Biased Agonist.

We conducted an SFSR study on three regions of the cariprazine scaffold: (1) the left-hand side (LHS) 2,3-dichlorophenylpiperazinyl moiety, (2) the middle linker cyclohexylene ring moiety, and (3) the right-hand side (RHS) urea moiety, including a few known compounds in ref 26. The syntheses of 9 (cariprazine) and the designed analogues, 10–12, 15 and 16 are outlined in Schemes 1 and 2. The Boc-protected ethyl 2-(trans-4-aminocyclohexyl)acetate 1 was prepared by esterification of commercially available 2-(trans-4-aminocyclohexyl)acetic acid, followed by protection of the amino group with di-tert-butyl dicarbonate. Lithium borohydride-mediated reduction of the methyl ester provided the alcohol 2. After activation of the hydroxyl group, the resulting reactive methanesulfonate ester intermediate was converted into the cyano compound 3. Diisobutylaluminium hydride (DIBAL-H)-mediated reduction of the cyano group afforded the trans-aldehyde 4. The cis-aldehyde 8 was synthesized from 2-(cis-4-aminocyclohexyl)acetic acid using the same procedures described above for the preparation of compound 4. Compound 10 was obtained via reductive amination of 4 and 1-(2,3-dichlorophenyl)piperazine in the presence of sodium triacetoxyborohydride. Boc-group deprotection under acidic conditions afforded compound 11, which was subsequently converted to 9 by dimethylurea formation with dimethylcarbamoyl chloride. Compound 12 was prepared from the cis-aldehyde 8 using the same procedures as compound 9. Compound 15 was synthesized using a 5-carbon linker moiety. Boc protection of commercially available 5-aminopentanol gave 13. Activation of the free hydroxyl group followed by substitution with 2,3-dichlorophenylpiperazine yielded 14. Boc deprotection followed by dimethylurea formation afforded compound 15 (Scheme 1). Compound 16 was prepared from the cis-aldehyde 4 and 1,2,3,4-tetrahydroisoquinoline using the same procedures as compound 9 (Scheme 2).

Scheme 1. Synthesis of Cariprazine Analogues^a.

^aReagents and conditions: (a) SOCl₂ reflux; (b) Boc₂O, triethylamine (TEA), dichloromethane (DCM), 85–95% over two steps; (c) LiBH₄, Et₂O, MeOH, 94–95%; (d) MsCl, TEA, DCM; (e) NaCN, dimethyl sulfoxide (DMSO), 90 °C, 90–96% over two steps; (f) DIBAL-H, DCM, −78 °C, 76–80%; (g) NaBH(OAc)₃, DCM, 75%; (h) trifluoroacetic acid (TFA), DCM, 58%; (i) dimethylcarbamoyl chloride, TEA, DCM, 33–47%; (j) 1-(2,3-dichlorophenyl)piperazine, K₂CO₃, CH₃CN, 12%.

Scheme 2. Synthesis of Compound 6 and Its Analogues 16–25^a.

^a(a) NaBH(OAc)₃, DCM, 73%; (b) TFA, DCM; (c) TEA, DCM, 41–74% over two steps.

All of these synthesized compounds were then evaluated for their effects on D₂R G_i/o-mediated cAMP inhibition and β-arrestin2 recruitment. The D₂R G_i/o-mediated cAMP inhibition assay measures inhibition of isoproterenol-stimulated cAMP production via the G_i/o-coupled signaling pathway, whereas the D₂R-mediated β-arrestin2 translocation Tango assay measured the recruitment of β-arrestin2 to D₂R.^41,48 Quinpirole, a full D₂R agonist, was used as a balanced positive control in both assays (G_i/o: EC₅₀ = 1.8 nM, pEC₅₀ = 8.74 ± 0.08; β-arrestin2: EC₅₀ = 2.7 nM, pEC₅₀ = 8.57 ± 0.10). The potency (EC₅₀) and efficacy (E_max) in both pathways for each compound in this series are displayed in Table 1.

Table 1.

SFSR Study on Cariprazine^a

CPD	Structure	cAMP Inhibition			β-arrestin			Bias for Gi/o
		EC50 (nM)	pEC50	Emax (%)	EC50 (nM)	pEC50	Emax (%)
9		0.4	9.35 ± 0.07	70 ± 1	0.6	9.20 ± 0.09	66 ± 2	1
10		14	7.84 ± 0.07	77 ± 2	9	8.04 ± 0.08	75 ± 2	0.4
11		0.7	9.19 ± 0.12	62 ± 2	2	8.71 ± 0.14	64 ± 3	2
12		3	8.49 ± 0.11	48 ± 2	12	7.93 ± 0.08	37 ± 1	3
15		7	8.17 ± 0.24	29 ± 3	10	7.98 ± 0.11	24 ± 1	1
16		12	7.92 ± 0.11	60 ± 3	69	7.16 ± 0.10	22 ± 1	11

^aEC₅₀, pEC₅₀, and E_max values are the average of at least three independent experiments performed in triplicate with standard error of the mean (SEM). Bias factors were calculated as described in the methods, with respect to quinpirole as a positive control (G_i/o: EC₅₀ = 1.8 nM, pEC₅₀ = 8.74 ± 0.08; β-arrestin2: EC₅₀ = 2.7 nM, pEC₅₀ = 8.57 ± 0.10).

Although compound 9 was reported to be functionally selective for the cAMP (G_i/o-protein) pathway over the ERK1/2 phosphorylation pathway,²⁷ this compound did not show significant bias for the G_i/o pathway (EC₅₀ = 0.4 nM; E_max = 70%) over the β-arrestin2 pathway (EC₅₀ = 0.6 nM; E_max = 66%). Modification of the dimethylurea group to a tert-butyl carbamate (10) drastically reduced the potencies for both the G_i/o (35-fold) and β-arrestin2 (15-fold) pathways, while slightly increasing the efficacies to approximately 75% of the maximal responsiveness in both pathways. Truncation of the dimethylurea group to the primary amino group (11) did not obviously change either potency or efficacy for the G_i/o pathway. Compared to 9, compound 11 displayed decreased potency (3-fold) and retained almost equal efficacy in the β-arrestin2 recruitment assay, suggesting that a positively charged amino group in this region is detrimental to potency in β-arrestin2 recruitment. Switching the trans-1,4-cyclohexylene of 9 to a cis-1,4-cyclohexylene moiety (12) resulted in a significant decrease in the potency and efficacy in both the G_i/o and β-arrestin2 pathways (G_i/o: EC₅₀ = 3 nM; E_max = 48%; β-arrestin2: EC₅₀ = 12 nM; E_max = 37%). Replacement of the trans-1,4-cyclohexylene group-containing linker to a more flexible 5-carbon chain led to a significant decrease in potency (17-fold) and efficacy (E_max = 24%) for β-arrestin2 recruitment and also an obvious decrease in potency (18-fold) and efficacy (E_max = 29%) for the G_i/o pathway. Replacement of the 2,3-dichlorophenylpiperazine to its isostere 1,2,3,4-tetrahydroisoquinoline (THIQ) has been explored in D₂R and D₃R studies.^27,49 The THIQ compound 16 displayed significantly reduced potency and efficacy in both pathways (G_i/o: EC₅₀ = 12 nM; E_max = 60%; β-arrestin2: EC₅₀ = 69 nM; E_max = 22%). However, 16 also displayed bias toward G_i/o signaling over β-arrestin2 recruitment with a bias factor of 11 relative to quinpirole.

Given the promising functional selectivity of 16, we opted to focus our SFSR campaign on the THIQ series by exploring three regions of this scaffold: the LHS 1,2,3,4-tetrahydroisoquinoline moiety, the middle linker cyclohexylene ring moiety, and the RHS urea moiety.

SFSR of the RHS of the THIQ Scaffold.

We explored a series of compounds (17–25) with different sized urea, carbamate, and amide moieties at the RHS. These compounds were synthesized according to the procedures for the preparation of 16, as outlined in Scheme 2. The results from both cAMP and β-arrestin2 assays are summarized in Table 2.

Table 2.

SFSR Study on the Right-Hand Side Moiety of Compound 16^a

CPD	RHS moiety	cAMP Inhibition			β-arrestin			Bias for Gi/o
		EC50 (nM)	pEC50	Emax (%)	EC50 (nM)	pEC50	Emax (%)
17		47	7.33 ± 0.27	36 ± 4	312	6.51 ± 0.14	16 ± 1	10
18		71	7.15 ± 0.17	52 ± 4	183	6.74 ± 0.11	14 ± 1	6
19		14	7.84 ± 0.14	66 ± 3	88	7.05 ± 0.08	42 ± 1	6
20		33	7.48 ± 0.08	59 ± 2	195	6.71 ± 0.08	52 ± 3	4
21		74	7.13 ± 0.21	49 ± 5	238	6.62 ± 0.04	26 ± 1	4
22		144	6.84 ± 0.15	53 ± 4	636	6.20 ± 0.11	22 ± 1	7
23		8	8.12 ± 0.13	62 ± 3	65	7.19 ± 0.07	50 ± 2	7
24		132	6.88 ± 0.28	33 ± 6	N.C.	N.C.	<10	N.C.
25		54	7.26 ± 0.36	22 ± 4	N.C.	N.C.	<10	N.C.

^aEC₅₀, pEC₅₀, and E_max values are the average of at least three independent experiments performed in triplicate with standard error of the mean (SEM). Bias factors were calculated as described in the methods, with respect to quinpirole as a positive control (G_i/o: EC₅₀ = 1.8 nM, pEC₅₀ = 8.74 ± 0.08; β-arrestin2: EC₅₀ = 2.7 nM, pEC₅₀ = 8.57 ± 0.10). N.C., not calculated.

Increasing the bulkiness of the dimethylurea moiety (16) to diethylurea (17) or diisopropyl urea (18) significantly reduced the potencies by 3–5-fold, while maintaining the maximal responsiveness for both the G_i/o and β-arrestin2 recruitment pathways. Compared to 16, a pyrrolidine carboxamide group (19) retained similar potency and increased the efficacy (G_i/o: EC₅₀ = 14 nM; E_max = 66%; β-arrestin2: EC₅₀ = 88 nM; E_max = 42%) in both pathways. This modification, however, did not improve the bias profile toward G_i/o protein activity. For the carbamate analogues, the tert-butyl group (20) showed better G_i/o pathway activation (EC₅₀ = 33 nM; E_max = 59%) than isopropyl (21) (EC₅₀ = 74 nM; E_max = 49%) and ethyl groups (EC₅₀ = 144 nM; E_max = 53%). Similarly, the tert-butyl group (20) also demonstrated better activity in the β-arrestin2 recruitment assay (EC₅₀ = 195 nM; E_max = 52%) than isopropyl (21) (EC₅₀ = 238 nM; E_max = 26%) and ethyl (22) groups (EC₅₀ = 636 nM; E_max = 22%). For the amide RHS, benzamide (23) was more potent in inducing both G_i/o-mediated cAMP inhibition (EC₅₀ = 8 nM; E_max = 62%) and β-arrestin2 recruitment (EC₅₀ = 65 nM; E_max = 50%) than propionamide (24) and acetamide (25). Albeit rather weak agonists for the G_i/o pathway (for 24: EC₅₀ = 132 nM; E_max = 33%; for 25: EC₅₀ = 54 nM; E_max = 22%), both propionamide (24) and acetamide (25) showed diminished efficacy in β-arrestin2 recruitment (EC₅₀ not calculated; E_max < 10%). Consequently, these smaller amide moieties (24 and 25) confer apparent functional selectivity toward the G_i/o pathway.

Overall, this SFSR study with the urea, carbamate, and amide moieties of the THIQ scaffold did not provide a significantly superior functionally selective D₂R agonist compared to compound 16. Modifications that lead to improved activities in the G_i/o pathway were always accompanied with improved β-arrestin2 recruitment. To improve the functional selectivity and efficacy of this series in the G_i/o pathway, we retained the dimethylurea moiety of compound 16 for the following studies on the middle linkers and LHS moieties.

SFSR of the Middle Linker of the THIQ Scaffold.

To determine effects of the middle linker on D₂R functional selectivity, we explored compounds 26, 28, 30, and 32. The synthetic routes are summarized in Scheme 3. Their biological results are outlined in Table 3.

Scheme 3. Synthesis of Compounds for Exploring the Middle Linker^a.

^a(a) NaBH(OAc)₃, DCM; (b) TFA, DCM; (c) dimethylcarbamoyl chloride, TEA, DCM, 64% over three steps; (d) MsCl, TEA, DCM; (e) 1,2,3,4-tetrahydroisoquinoline, K₂CO₃, CH₃CN, 95 °C, 15–43%; (f) Boc₂O, NaHCO₃, tetrahydrofuran (THF), 80%; (g) dimethylcarbamoyl chloride, K₂CO₃, THF, 23% over four steps.

Table 3.

SFSR Study of the Middle Linker of Compound 16^a

CPD	Middle linker	cAMP Inhibition			β-arrestin			Bias for Gi/o
		EC50 (nM)	pEC50	Emax (%)	EC50 (nM)	pEC50	Emax(%)
26		259	6.59 ± 0.15	58 ± 5	>1000	<6.00	N.C.	N.C.
28		N.A	N.A	N.A	N.A	N.A	N.A	N.C.
30		N.A	N.A	N.A	N.A	N.A	N.A	N.C.
32		N.A	N.A	N.A	N.A	N.A	N.A	N.C.

^aEC₅₀, pEC₅₀, and E_max values are the average of at least three independent experiments performed in triplicate with standard error of the mean (SEM). Bias factors were calculated as described in the methods, with respect to quinpirole as a positive control (G_i/o: EC₅₀ = 1.8 nM, pEC₅₀ = 8.74 ± 0.08; β-arrestin2: EC₅₀ = 2.7 nM, pEC₅₀ = 8.57 ± 0.10). N.C., not calculated. N.A., no activity.

Compound 26 was synthesized according to the same procedures for the preparation of 16 from the cis- aldehyde 8. The preparation of compound 28 started from compound 2. The activation of the hydroxyl group of 2 with methanesulfonyl chloride, followed by substitution with 1,2,3,4-tetrahydroisoquinoline, provided the Boc-protected intermediate 27. Boc deprotection and subsequent urea formation yielded 28. Compound 30 was synthesized following the protocols for the preparation of 28 from the alcohol intermediate 29, which was prepared by selective Boc protection of the commercially available 2-(4-aminophenyl)-ethan-1-ol. Similarly, compound 32 was synthesized using the same route as the preparation of 28 from the alcohol 13 (Scheme 3).

Replacement of the trans-1,4-cyclohexylene ring of 16 with a cis-1,4-cyclohexylene group (26) significantly decreased potency (22-fold), while maintaining efficacy (E_max = 58%) for the G_i/o pathway. Interestingly, compound 26 showed very low activity for β-arrestin2 recruitment (EC₅₀ > 1000 nM), resulting in our inability to calculate a bias factor for this compound. However, 26 still showed low potency for the G_i/o pathway (EC₅₀ = 259 nM), indicating weak G protein-bias. Incorporation of a short linker (28), a rigid aromatic linker (30), or a flexible 5-C chain linker (32) resulted in compounds that were all inactive in both the cAMP inhibition and β-arrestin2 recruitment assays. Taken together, these findings suggest that the central linker plays an important role in the functional potency and efficacy of this scaffold in both the G_i/o and β-arrestin2 signaling pathways.

SFSR of the LHS of the THIQ Scaffold.

We next explored compounds 33–43 to study the effects of phenyl substituents on D₂R functional selectivity. Compounds 33–43 were prepared using the same synthetic routes for 16, starting from the appropriately substituted 1,2,3,4-tetrahydroisoquinoline precursor (Scheme 4). The biological assay data are summarized in Table 4.

Scheme 4. Synthesis of Compounds with Various Substituents on the LHS Phenyl Ring^a.

^a(a) NaBH(OAc)₃, DCM; (b) TFA, DCM; (c) dimethylcarbamoyl chloride, TEA, DCM, 29–53% over three steps.

Table 4.

SFSR Study of Left-Hand Side Moiety of Compound 16^a

CPD	LHS moiety	cAMP Inhibition			P-arrestin			Bias for Gi/o
		EC50 (nM)	pEC50	Emax (%)	EC50 (nM)	pEC50	Emax (%)
33		3	8.58 ± 0.05	116 ± 2	2	8.62 ± 0.04	90 ± 1	0.8
34		10	8.01 ± 0.08	71 ± 2	26	7.59 ± 0.05	64 ± 1	2
35		3	8.55 ± 0.05	95 ± 1	4	8.39 ± 0.07	92 ± 2	1
36		5	8.26 ± 0.16	44 ± 2	17	7.78 ± 0.11	16 ± 1	6
37		8	8.10 ± 0.16	66 ± 3	5	8.32 ± 0.09	24 ± 1	1
38		11	7.97 ± 0.13	62 ± 3	40	7.39 ± 0.14	11 ± 1	14
39		67	7.17 ± 0.09	83 ± 3	N.C.	N.C.	<10	N.C.
40		129	6.89 ± 0.09	68 ± 2	N.C.	N.C.	<10	N.C.
41		27	7.57 ± 0.00	68 ± 0	60	7.22 ± 0.00	49 ± 1	2
42		86	7.07 ± 0.16	49 ± 3	202	6.69 ± 0.08	16 ± 1	5
43		19	7.71 ± 0.07	89 ± 2	54	7.27 ± 0.07	76 ± 2	2

^aEC₅₀, pEC₅₀, and E_max values are the average of at least three independent experiments performed in triplicate with standard error of the mean (SEM). Bias factors were calculated as described in the methods, with respect to quinpirole as a positive control (G_i/o: EC₅₀ = 1.8 nM, pEC₅₀ = 8.74 ± 0.08; β-arrestin2: EC₅₀ = 2.7 nM, pEC₅₀ = 8.57 ± 0.10). N.C., not calculated.

Given that cariprazine featured a dichloro substitution pattern on its LHS, we designed and synthesized compound 33 bearing a 5,6-dicholoro-substituted THIQ. Similar to cariprazine, compound 33 was not functionally selective in the two signaling pathways. Notably, however, this compound exhibited higher efficacy than cariprazine for both the G_i/o (E_max = 116%) and β-arrestin2 (E_max = 90%) pathways. In fact, its G_i/o signaling efficacy was even higher than that of the control compound quinpirole (E_max = 100%), indicating superagonism. A 6,7-dioxole on the THIQ (34) did not show significant G_i/o pathway selectivity. Because of the enhanced maximal responsiveness of both signaling pathways to the 5,6-dichloro substituent (33), we explored the effects of monochloro substituents at the 5-, 6-, 7-, and 8-positions on the THIQ. Similar to 33, the 5-chloro compound 35 was a balanced near-full agonist for both the G_i/o signaling pathway (E_max = 95%) and β-arrestin2 recruitment (E_max = 92%). A monochloro substituent at the 7-position (37) also resulted in a balanced effect in both pathways, however, a 6- or 8-chloro substituent on the THIQ moiety led to biased partial agonists for activating G_i/o signaling. Although the functional selectivity of 36 (6-chloro) was moderate, compound 38 (8-chloro) displayed enhanced bias toward activating the G_i/o signaling pathway (EC₅₀ = 11 nM; E_max = 62%) over β-arrestin2 recruitment (EC₅₀ = 40 nM; E_max = 11%) with a bias factor of 14. We next explored the methyl (39), bromo (40), fluoro (41), trifluoromethyl (42), and methoxy (43) substituents at the 8-position. Albeit with lower functional potencies (27–86 nM) for the G_i/o signaling pathway than 38, all of these compounds appeared to be partial agonists for stimulating G_i/o signaling. Taken together, our results suggest that the substituents of the LHS 1,2,3,4-tetrahydroisoquinoline moiety play a critical role in modulating functional selectivity of D₂R. The 8-chloro substituent on the THIQ moiety provided a potent and G_i/o-biased compound, 38.

Evaluation of G_i/o Protein-Biased D₂R Agonist 38 in Orthogonal Assays.

To further confirm the observed signaling bias, which has been shown to depend on cell background and readout.^14,50 we tested 38 in a bioluminescence resonance energy transfer (BRET)-based assay⁵¹ using quinpirole as the control. As demonstrated in Figure 1A, in HEK 293T cells co-expressing D₂R C-terminal tagged renilla luciferase (Rluc), a Venus-tagged β-arrestin2, and G protein-coupled receptor kinase 2 (GRK2), compound 38 displayed no activity for D₂R-mediated β-arrestin2 recruitment compared to quinpirole (EC₅₀ = 115 nM; E_max = 100%). Next, to confirm compound 38 is an agonist in the D₂R G protein pathway, we tested compound 38 in a Gαi1–Gγ2 dissociation BRET-based assay⁵² using quinpirole as a control. In this assay, compound 38 showed potent partial agonist activity (EC₅₀ = 0.40 nM; E_max = 47%) compared to quinpirole (EC₅₀ = 2.4 nM; E_max = 100%) (Figure 1B). Taken together, BRET-based orthologous assay platforms for either G protein activation or β-arrestin2 recruitment confirmed that compound 38 is a G protein-biased D₂R partial agonist.

Figure 1.

Further confirmation that compound 38 is a G protein-biased D₂R partial agonist. (A) D₂R-mediated BRET β-arrestin recruitment activity of compounds 38 and quinpirole (EC₅₀ = 115 nM). (B) D₂R-mediated Gi1 BRET activity of compounds 38 (EC5₀ = 0.37 nM, E_max = 47%) and quinpirole (EC₅₀ = 2.4 nM). Data are the average of at least three independent experiments performed in duplicate.

Evaluation of G_i/o Protein-Biased D₂R Agonist 38 in D₂R β-Arrestin Antagonist Mode and in D₃R Functional Assays.

Considering that compound 38 is a G_i/o protein-biased partial agonist showing no activity for D₂R β-arrestin2 activity, we examined the antagonist activity of 38 for dopamine (DA)-induced D₂R β-arrestin2 recruitment. Different from dopamine, compound 38 did not show agonist activity toward β-arrestin2 recruitment (Figure 2A). However, similar to clozapine, 38 exhibited potent antagonism in blocking dopamine-stimulated β-arrestin2 recruitment (Figure 2A).

Figure 2.

(A) Antagonist activity of compound 38 (IC₅₀ = 71 nM) on dopamine (DA)-stimulated D₂R β-arrestin recruitment is similar to clozapine (IC₅₀ = 140 nM) as measured by Tango. Lack of D₃R agonist activity of compound 38 compared to positive control quinpirole in the G_i/o-mediated cAMP inhibition GloSensor assay (B) and β-arrestin recruitment Tango assay (C). Data are the average of three independent experiments performed in triplicate.

Because cariprazine is a D₂R/D₃R dual partial agonist, we also tested whether the functionally selective D₂R partial agonist, compound 38, is also a D₃R agonist. As demonstrated in Figure 2B,C, compound 38 acts as a selective agonist at D₂R over D₃R, showing no agonist activity in D₃R G_i/o-mediated cAMP inhibition assay (Figure 2B) or β-arrestin2 recruitment assay (Figure 2C).

In Silico Studies.

We next sought to explore how these findings from our SFSR campaign fit into the existing paradigms for biased signaling at D₂R. To do this, we docked cariprazine, 16, and 38 to a model based on the co-crystal structure of D₂R in the complex with risperidone (PDB: 6CM4).³⁸ Because antagonists, such as risperidone, may stabilize different receptor conformations compared to agonists, we first compared this model to that of a previous D₂R model from the literature that successfully modeled functional selectivity in D₂R agonists.⁴¹ The all-atom binding site root-mean-square deviation (RMSD) between our present model and this previously reported D₂R model was 0.74 Å,⁴¹ indicating a high degree of similarity between the atom arrangements lining the binding pockets of both structures (Figure 3A).

Figure 3.

In silico structure-based SFSR studies of D₂R agonists. (A) When the D₂R crystal structure utilized for this study (gold) is overlayed with a D₂R homology model from a previous study that examined the structural determinants of D₂R agonist functional selectivity (blue), a high degree of similarity was noted between the two structures (all-atom binding site RMSD = 0.74 Å).⁴¹ (B) The docked pose of cariprazine suggests that together with I184^EL2, F189^5.38 may function as a “lid” over the binding pocket leading to increase ligand rESIdence times and β-arrestin recruitment. (C) Compound 16 engages D₂R from deeper within the orthosteric site, and the steric constraints it imposes on TM5 may partially underlie its G_i/o bias. (D) Compound 38’s chloro substituent forms a hydrogen bond with conserved TM5 serines and angles with the rest of the scaffold away from extracellular loop 2 (EL2), which would be predicted to enhance its G_i bias compared to 16.

Cariprazine docked to D₂R revealed key binding pocket interactions including a salt bridge interaction between cariprazine’s protonated piperazinyl nitrogen and the conserved D114^3.32 residue (2.52 Å) and a close edge-to-face π interaction between cariprazine’s LHS dichlorophenyl entity and F390^6.52 (2.89 Å, Figure 3B). Notably in our pose, the LHS dichlorophenyl entity is oriented upward and toward TM5, such that the 3-chloro group can form a hydrogen bond with S193^5.42 (2.52 Å), which is a residue found to be important for D₂R G_i/o-dependent signaling.⁴¹ Furthermore, the 3-chloro appears to form an n–π interaction with F189^5.38, and the 2-chloro group is pointed toward I184^EL2 (4.70 Å), which is a residue previously implicated in D₂R β-arrestin2 recruitment efficacy.^41,45,53 Taken together, cariprazine’s binding pose appears to be consistent with its balanced agonist activity (Figure 3B).

By contrast, D₂R G protein-biased compounds 16 (Figure 3C) and 38 (Figure 3D) appear to dock deeper in the orthosteric binding site, lacking clear interactions with I184^EL2 and engaging TM5 further via strict steric constraints, which has previously been shown to play an important role in receptor activation and G protein signaling.⁵⁴ Specifically, the chloro substitution of 38 is able to directly interact with S193^5.42 on TM5 (2.07 Å). Given previous findings that ligand engagement with TM5 serines is critical for G protein signaling at D₂R,⁴¹ this binding pose suggests that the TM5–LHS chloro interaction may contribute to 38’s dramatic increase in G_i/o bias compared to 16, which by comparison has a weaker interaction with TM5. These docking poses suggest that the selective engagement with TM5 over EL2 may be important for the D₂R G_i/o-bias of compounds 16 and 38, which is consistent with mechanisms previously proposed in the literature.⁴¹

Receptor Occupancy and Behavioral Studies in Mice.

To evaluate the in vivo D₂R target engagement of compound 38, we implemented a radioactive competitive assay to assess the displacement of [³H]-raclopride in mouse striatum and cerebellum.⁵⁵ Compound 38 dose-dependently bound to the D₂R in mouse striatum and cerebellum, with full receptor occupancy observed at 32 mg/kg intravenous dose (Figure 4). The high D₂R target engagement of 38 encouraged us to examine the in vivo pharmacological effects of this ligand on supressing the NMDA receptor antagonist phencyclidine (PCP) in a PCP-stimulated hyperlocomotion open field test. Compound 38 significantly reduced PCP-induced hyperlocomotion at 1 mg/kg. At an elevated dose of 3.5 mg/kg, 38 almost completely suppressed the induced hyperlocomotion (Figure 5). It remains unclear whether the suppression of hyperlocomotor activity is due to compound 38’s G protein-biased agonism or antagonism in this model of high dopamine efflux. Taken together, compound 38 was efficacious in vivo and may serve as a potential in vivo tool compound for elucidating the role of G protein-mediated D₂R signaling.

Figure 4.

Compound 38 exhibits dose dependence to binding to D₂ receptor in male c57 mice. [³H]Raclopride in vivo binding assay was used.

Figure 5.

Compound 38 displays a potent antipsychotic-like activity in a hyperlocomotion study. C57Bl/6 mice were given the vehicle (Veh), 1, or 3.5 mg/kg 38 (intraperitoneally (i.p.)), followed 30 min later with the Veh or 6 mg/kg phencyclidine (PCP, i.p.). (A) Locomotor activities are shown as 5 min binned intervals. The repeated measures analysis of variance (RMANOVA) are provided for baseline [time: F(5, 145) = 12.731, p < 0.001; treatment: F(3, 29) = 2.722, p = 0.063; time by treatment: F(15, 145) = 1.878, p = 0.099] and for PCP-stimulated activity [time: F(17, 493) = 22.056, P < 0.001; treatment: F(3, 29) = 16.992, p < 0.001; time by treatment: F(51, 493) = 6.159, p = < 0.001]. (B) Locomotor activities presented as cumulative activity (0–30 min) and PCP-stimulated activity (31–90 min). A RMANOVA found the following: [pre–post: F(1, 29) = 55.625, p < 0.001; treatment: F(3, 29) = 20.004, p < 0.001; pre–post by treatment: F(3, 29) = 17.651, p < 0.001]. N = 8–9 mice/group; Bonferroni-corrected pair-wise comparisons—*p < 0.05, vs baseline within treatment; +p < 0.05, vs the Veh-PCP group within the post-stimulation interval; #p < 0.05, 1 mg/kg compound vs 3.5 mg/kg compound 38 within the post-stimulation interval.

CONCLUSIONS

In summary, we designed and synthesized a series of cariprazine analogues and evaluated them in cAMP accumulation and β-arrestin2 recruitment assays. Our initial SFSR study revealed compound 16, bearing a THIQ moiety, which exhibited bias for activating G_i/o signaling over β-arrestin2 pathway with a bias factor of 11. Further optimization of 16 led to the discovery of compound 38, which displayed an enhanced functional selectivity for the G_i/o signaling pathway β-arrestin2 recruitment with a bias factor of 14. Orthologous BRET assays were used to confirm that 38 was a G protein-biased partial agonist. Unlike cariprazine, which was a dual partial agonist of D₂R/D₃R, compound 38 showed no agonist activity at D₃R but demonstrated partial G_i/o protein agonist activity at D₂R. Compound 38 also exhibited potent D₂R antagonism of dopamine-stimulated β-arrestin2 recruitment. Docking studies suggested that both a TM5–LHS chloro interaction and the resulting scaffold separation from EL2 could contribute to 38’s high functional selectivity in the G_i/o pathway. Finally, compound 38 was assessed in an in vivo receptor occupancy assay and a psychotomimetic-induced hyperlocomotion assay in rodents. Compound 38 showed high in vivo D₂R target engagement as well as dose-dependent inhibition of PCP-induced hyperlocomotion in mice. Taken together, compound 38 is a highly G_i/o-biased D₂R partial agonist with the potential to serve as a tool compound for elucidating the role of G protein-mediated D₂R signaling in pathophysiological systems.

EXPERIMENTAL SECTION

Chemistry General Procedures.

High-performance liquid chromatography (HPLC) spectra for all compounds were acquired using an Agilent 1200 series system with a diode array detector. Chromatography was performed on a 2.1 × 150 mm² Zorbax 300SB-C18 5 μm column with water containing 0.1% formic acid as solvent A and acetonitrile containing 0.1% formic acid as solvent B at a flow rate of 0.4 mL/min. The gradient program was as follows: 1% B (0–1 min), 1–99% B (1–4 min), and 99% B (4–8 min). High-resolution mass spectra (HRMS) data were acquired in a positive ion mode using an Agilent G1969A API-TOF with an electrospray ionization (ESI) source. Nuclear magnetic resonance (NMR) spectra were acquired on either a Bruker DRX-600 spectrometer (600 MHz ¹H) or a Bruker Avance-III 800 MHz spectrometer (201 MHz ¹³C). Chemical shifts are reported in ppm (δ). Preparative HPLC was performed on Agilent Prep 1200 series with UV detector set to 254 nm. Samples were injected into a Phenomenex Luna 250 × 30 mm², 5 μM C₁₈ column at room temperature (rt). The flow rate was 40 mL/min. A linear gradient was used with 10% (or 50%) of MeOH (A) in H2O (with 0.1% TFA) (B) to 100% of MeOH (A). HPLC was used to establish the purity of target compounds. All final compounds had >95% purity using the HPLC methods described above. All final compounds are characterized as trifluoroacetic acid salt form, except compound 38, of which form is free base.

Methyl trans-4-((tert-Butoxycarbonyl)amino)cyclohexane-1-carboxylate (1).

To a solution of 4-aminocyclohexane-1-carboxylic acid HCl salt (5 g, 27.9 mmol) in methanol (100 mL) was added thionyl chloride (7.5 mL, 103.5 mmol) at 0 °C. The resulting solution was warmed to rt and stirred for 18 h. Solvent removal under reduced pressure yielded the crude product as an off-white solid, which was used in the next step without further purification.

To a solution of the crude intermediate in dichloromethane (100 mL) was added triethylamine (6 mL, 43 mmol) at 0 °C, followed by di-tert-butyl dicarbonate (8.6 g, 39.5 mmol). The reaction was stirred overnight before being quenched with saturated aqueous sodium bicarbonate. The mixture was extracted with dichloromethane (3 × 50 mL). Combined organic phase was dried over anhydrous sodium sulfate and concentrated under reduced pressure. The resulting residue was purified by silica gel flash chromatography (elution with hexane/EtOAc = 1:1) to give compound 1 as white powder (6.84 g, 95%). ¹H NMR (600 MHz, CD₃OD) δ 3.67 (s, 3H), 3.33–3.25 (m, 1H), 2.36–2.16 (m, 1H), 2.03–1.88 (m, 4H), 1.45–1.47 (m, 11H), 1.31–1.16 (m, 2H).

tert-Butyl (trans-4-(Hydroxymethyl)cyclohexyl)carbamate (2).

To a solution of methyl trans-4-((tert-butoxycarbonyl)amino)cyclohexane-1-carboxylate (6.84, 26.5 mmol) in diethyl ether (120 mL) and methanol (0.6 mL) was added lithium borohydride (1.83 g, 84 mmol) in portions, followed by dropwise addition of methanol (3 mL). The resulting mixture was stirred for 45 min before being quenched with methanol and concentrated under reduced pressure. The resulting residue was treated with 1% aqueous sodium hydroxide and extracted with ethyl acetate (3 × 60 mL). The combined organic phase was dried over anhydrous sodium sulfate and concentrated under reduced pressure. The resulting residue was purified with silica gel flash chromatography (elution with MeOH/DCM = 0–10%) to give the title compound as white solid (5.8 g, 95%). ¹H NMR (600 MHz, CD₃OD) δ 3.37 (dd, J = 6.4, 1.3 Hz, 2H), 3.31–3.20 (m, 1H), 1.98–1.90 (m, 2H), 1.89–1.81 (m, 2H), 1.52–1.36 (m, 10H), 1.23–1.16 (m, 2H), 1.09–0.97 (m, 2H).

tert-Butyl (trans-4-(Cyanomethyl)cyclohexyl)carbamate (3).

To the solution of tert-butyl (trans-4-(hydroxymethyl)cyclohexyl)carbamate (4.34 g, 18.9 mmol) and triethylamine (5.3 mL, 37.8 mmol) in dichloromethane (150 mL) was added methanesulfonyl chloride (2.2 mL, 28.3 mmol) dropwise at 0 °C. The resulting mixture was stirred for 2 h before being quenched with water. The mixture was extracted with dichloromethane (3 × 60 mL). Combined organic phase was dried over sodium sulfate. After filtration, the solvent was removed under reduced pressure to provide the crude intermediate, which was used in next step without purification.

The crude material and potassium cyanide (3.6 g, 55.2 mmol) were mixed in dimethyl sulfoxide (60 mL). The mixture was stirred for 5 h at 100 °C, before being cooled to room temperature. After being cooled to rt, the reaction was diluted with water and extracted with ethyl acetate (3 × 100 mL). Combined organic phase was dried over sodium sulfate. The concentrated residue was purified by silica gel column (elution with hexane/EtOAc = 1:1) to yield the title compound as gray solid (4.3 g, 96%). ¹H NMR (600 MHz, CD₃OD) δ 3.32–3.25 (m, 1H), 2.40 (d, J = 6.5, 1.4 Hz, 2H), 1.96 (d, J = 10.4 Hz, 2H), 1.89 (d, J = 11.1 Hz, 2H), 1.69–1.58 (m, 1H), 1.45 (s, 9H), 1.28–1.09 (m, 4H).

tert-Butyl (trans-4-(2-Oxoethyl)cyclohexyl)carbamate (4).

To a solution of tert-butyl (trans-4-(hydroxymethyl)cyclohexyl)carbamate (1.2 g, 5 mmol) in dichloromethane (20 mL), was added diisobutylaluminium hydride (1 M in hexane, 15 mL) at −78 °C. The resulting solution was stirred for 2 h before being quenched with saturated aqueous solution of Rochelle’s salt. The mixture was warmed to rt slowly and stirred vigorously until two phases were visualized. The organic phase was concentrated under reduced pressure. The resulting residue was purified by silica gel column to give the title compound as pale yellow solid (960 mg, 76%). ¹H NMR (600 MHz, CDCl₃) δ 9.75 (d, J = 2.3 Hz, 1H), 4.38 (s, 1H), 3.47–3.30 (m, 1H), 2.32 (dd, J = 6.8, 2.2 Hz, 2H), 2.00 (d, J = 10.7 Hz, 2H), 1.90–1.74 (m, 3H), 1.43 (s, 9H), 1.15–1.08 (m, 4H).

tert-Butyl (cis-4-(Hydroxymethyl)cyclohexyl)carbamate (6).

Compound 6 was prepared according to the same procedures as preparing compound 2 (yield 95%). ¹H NMR (600 MHz, CD₃OD) δ 3.56 (m, 1H), 3.43 (d, J = 6.4 Hz, 2H), 1.65–1.53 (m, 8H), 1.43 (s, 9H), 1.41–1.34 (m, 1H).

tert-Butyl (cis-4-(Cyanomethyl)cyclohexyl)carbamate (7).

Compound 7 was prepared according to the same procedures as preparing compound 3 (yield 96%). ¹H NMR (600 MHz, CD₃OD) δ 3.68–3.54 (m, 1H), 2.41 (dd, J = 7.2, 3.6 Hz, 2H), 1.83–1.55 (m, 9H), 1.44 (s, 9H).

tert-Butyl ((1s,4s)-4-(2-Oxoethyl)cyclohexyl)carbamate (8).

Compound 8 was prepared according to the same procedures as preparing compound 4 (yield 80%). ¹H NMR (600 MHz, CDCl₃) δ 9.75 (d, J = 2.2 Hz, 1H), 4.61 (s, 1H), 3.77–3.65 (m, 1H), 2.37 (dd, J = 6.9, 2.1 Hz, 2H), 2.12–1.97 (m, 1H), 1.66–1.56 (m, 6H), 1.44 (s, 9H), 1.29–1.18 (m, 2H).

3-(trans-4-(2-(4-(2,3-Dichlorophenyl)piperazin-1-yl)ethyl)cyclohexyl)-1,1-dimethylurea (9).

To the solution of Compound 11 (20 mg, 0.03 mmol) and triethylamine (12 mg, 0.12 mmol) in dichloromethane (2 mL) was added dimethylcarbamoyl chloride (7 mg, 0.06 mmol). The mixture was stirred for few minutes before being condensed. The resulting residue was purified by preparative HPLC (10–100% methanol/0.1% TFA in H₂O) to give the title compound as white powder (8 mg, 43%). ¹H NMR (600 MHz, CD₃OD) δ 7.37–7.29 (m, 2H), 7.24–7.18 (m, 1H), 3.73–3.67 (m, 2H), 3.603.46 (m, 3H), 3.36–3.24 (m, 4H), 3.18–3.04 (m, 2H), 2.90 (s, 6H), 1.97–1.91 (m, 2H), 1.92–1.84 (m, 2H), 1.77–1.66 (m, 2H), 1.43–1.25 (m, 3H), 1.17 (dd, J = 12.4, 3.5 Hz, 2H). MS (ESI) m/z 427.4 [M + H]⁺.

tert-Butyl (trans-4-(2-(4-(2,3-Dichlorophenyl)piperazin-1-yl)ethyl)cyclohexyl)carbamate (10).

To a solution of 1-(2,3-dichlorophenyl)piperzine (46.2 mg, 0.2 mmol) and aldehyde 4 (48.2 mg, 0.2 mmol) in chloromethane (4 mL) was added sodium triacetoxyborohydride (84.8 mg, 0.4 mmol) in portions. The resulting mixture was stirred overnight before being quenched with saturated aqueous sodium bicarbonate. The resulting mixture was extracted with dichloromethane (3 × 10 mL). The combined organic phase was dried over sodium sulfate and concentrated under reduced pressure. The resulting residue was purified by preparative HPLC (10–100% methanol/0.1% TFA in H₂O) and lyophilized to give the title compounds as white powder (85.5 mg, yield 75%). ¹H NMR (600 MHz, CD₃OD) δ 7.36–7.30 (m, 2H), 7.20–7.16 (m, 1H), 3.73–3.68 (m, 2H), 3.59–3.52 (m, 2H), 3.36–3.25 (m, 5H), 3.21–3.09 (m, 2H), 1.99–1.91 (m, 2H), 1.89–1.82 (m, 2H), 1.77–1.67 (m, 2H), 1.46 (s, 9H), 1.39–1.32 (m, 1H), 1.27–1.09 (m, 4H). MS (ESI) m/z 456.7 [M + H]⁺.

trans-4-(2-(4-(2,3-Dichlorophenyl)piperazin-1-yl)ethyl)-cyclohexan-1-amine (11).

Compound 10 (60 mg, 0.1 mmol) was treated with dichloromethane (1 mL) and trifluoroacetic acid (1 mL) for 1 h. After removal of the solvents, the resulting residue was purified by preparative HPLC (10–100% methanol/0.1% TFA in H₂O) to give the title compound as white powder (35.5 mg, 58%). ¹H NMR (600 MHz, CD₃OD) δ 7.37–7.29 (m, 2H), 7.18 (dd, J = 7.5, 2.1 Hz, 1H), 3.71 (d, J = 12.1 Hz, 2H), 3.56 (d, J = 13.2 Hz, 2H), 3.38–3.23 (m, 5H), 3.17–3.04 (m, 2H), 2.12–2.04 (m, 2H), 2.00–1.92 (m, 2H), 1.81–1.70 (m, 2H), 1.44 (qd, J = 12.5, 3.5 Hz, 3H), 1.25–1.18 (m, 2H). MS (ESI) m/z 356.2 [M + H]⁺.

3-(cis-4-(2-(4-(2,3-Dichlorophenyl)piperazin-1-yl)ethyl)cyclohexyl)-1,1-dimethylurea (12).

To a solution of 1-(2,3-dichlorophenyl)piperzine (23.1 mg, 0.1 mmol) and compound 8 (24.1 mg, 0.1 mmol) in dichloromethane (2 mL) was added sodium triacetoxyborohydride (42.4 mg, 0.2 mmol) in portions. The resulting mixture was stirred overnight before being quenched with saturated aqueous sodium bicarbonate. The mixture was extracted with dichloromethane (3 × 5 mL). The combined organic phase was dried over sodium sulfate and concentrated under reduced pressure. The residue was treated with dichloromethane (1 mL) and trifluoroacetic acid (1 mL) for 1 h. After the solvent was removed under reduced pressure, the resulting residue was used in the next step without further purification. The crude material was dissolved in dichloromethane (2 mL). To the resulting solution was added triethylamine (50 mg, 0.5 mmol). After the solution was stirred for a few minutes, dimethylcarbamoyl chloride (22 mg, 0.2 mmol) in dichloromethane (1 mL) was added dropwise. After the resulting solution was stirred for 1 h, the mixture was concentrated under reduced pressure. The resulting residue was purified by preparative HPLC (10–100% methanol/0.1% TFA in H₂O) to give the title compound as white powder (25.3 mg, yield 47%). ¹H NMR (600 MHz, CD₃OD) δ 7.39–7.26 (m, 2H), 7.22–7.16 (m, 1H), 3.77–3.65 (m, 2H), 3.61–3.51 (m, 2H), 3.39–3.24 (m, 5H), 3.20–3.10 (m, 2H), 2.93 (s, 6H), 1.89–1.80 (m, 2H), 1.73–1.60 (m, 7H), 1.58–1.46 (m, 2H). MS (ESI) m/z 427.5 [M + H]⁺.

3-(5-(4-(2,3-Dichlorophenyl)piperazin-1-yl)pentyl)-1,1-dimethylurea (15).

To a solution of 13 (203 mg, 1 mmol) and triethylamine (200 mg, 2 mmol) in dichloromethane (10 mL) was added methanesulfonyl chloride (137.4 mg, 1.2 mmol) dropwise at 0 °C. The resulting solution was stirred for 2 h at rt before being quenched with 1 mL of saturated aqueous sodium bicarbonate. The mixture was extracted with dichloromethane and washed with water. The organic phase was dried over anhydrous sodium sulfate, filtered, and concentrated under reduced pressure. The resulting residue was dissolved in acetonitrile (10 mL) before potassium carbonate (414.6 mg, 3 mmol) and 1-(2,3-dichlorophenyl)piperzine (231 mg, 1 mmol) were added successively. The mixture was refluxed overnight before being diluted with 10 mL of water. After being cooled to rt, the mixture was extracted with ethyl acetate (3 × 10 mL). The combined organic phase was dried over anhydrous sodium sulfate and concentrated under reduced pressure. The resulting residue was purified by silica gel column (CH₃OH/DCM, 0–10%) to give compound 14 (50 mg, yield 12%). Compound 14 was treated with trifluoroacetic acid (1 mL) and dichloromethane (2 mL) for 3 h at rt. The reaction was condensed under reduced pressure. The resulting residue was used for next step without further purification. To a solution of the residue above and triethylamine (40 mg, 0.4 mmol) in dichloromethane (2 mL), was added dimethylcarbamoyl chloride (16.1 mg, 0.15 mmol). The mixture was stirred for 1 h before being concentrated under reduced pressure. The resulting residue was purified by preparative HPLC (10–100% methanol/0.1% TFA in H₂O) to give the title compound as a white powder (20 mg, yield 33%). ¹H NMR (600 MHz, CD₃OD) δ 7.40–7.25 (m, 2H), 7.24–7.17 (m, 1H), 3.71 (d, J = 12.1 Hz, 2H), 3.55 (d, J = 13.1 Hz, 2H), 3.36–3.28 (m, 2H), 3.27–3.23 (m, 2H), 3.22–3.13 (m, 4H), 2.91 (s, 6H), 1.89–1.80 (m, 2H), 1.65–1.53 (m, 2H), 1.45 (t, J = 7.3 Hz, 2H). MS (ESI) m/z 387.4 [M + H]⁺.

General Procedure for Preparing 16–19, 21–25.

Compound 20 (212 mg, 0.45 mmol) was treated with trifluoroacetic acid (10 mL) and dichloromethane (10 mL) for 3 h at rt. The solution was concentrated under reduced pressure. The resulting residue was dissolved in dichloromethane (9 mL). To the solution was added triethylamine (404 mg, 4.0 mmol). The mixture was equally divided into nine portions. To each portion was added the corresponding acyl chloride or anhydride (0.1 mmol, 2 equiv). The resulting mixture was stirred overnight, followed by concentration under reduced pressure. The resulting residue was purified by preparative HPLC (10–100% methanol/0.1% TFA in H₂O) to give the title compound as a white powder.

3-(trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (16).

Dimethylcarbamoyl chloride was used to afford a powder (15 mg, yield 68%). ¹H NMR (600 MHz, CD₃OD) δ 7.38–7.26 (m, 3H), 7.23 (d, J = 7.5 Hz, 1H), 4.62 (d, J = 15.3 Hz, 1H), 4.34 (d, J = 15.3 Hz, 1H), 3.89–3.79 (m, 1H), 3.56–3.47 (m, 1H), 3.44–3.23 (m, 4H), 3.19–3.12 (m, 1H), 2.92–2.84 (m, 6H), 1.96–1.91 (m, 2H), 1.90–1.83 (m, 2H), 1.81–1.73 (m, 2H), 1.43–1.38 (m, 1H), 1.36–1.27 (m, 2H), 1.22–1.12 (m, 2H). MS (ESI) m/z 330.8 [M + H]⁺.

3-(trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-diethylurea (17).

Diethylcarbamoyl chloride was used to afford a powder (yield 54%). ¹H NMR (600 MHz, CD₃OD) δ 7.35–7.26 (m, 3H), 7.23–7.19 (m, 1H), 4.69–4.27 (m, 2H), 3.88–3.48 (m, 3H), 3.38–3.31 (m, 8H), 1.96–1.89 (m, 2H), 1.85 (d, J = 12.5 Hz, 2H), 1.79–1.70 (m, 2H), 1.42–1.25 (m, 3H), 1.22–1.05 (m, 8H). MS (ESI) m/z 358.4 [M + H]⁺.

3-(trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-diisopropylurea (18).

Diisopropylcarbamoyl chloride was used to afford a powder (yield 60%). ¹H NMR (600 MHz, CD₃OD) δ 7.34–7.25 (m, 3H), 7.21 (d, J = 7.4 Hz, 1H), 4.65–4.51 (m, 2H), 3.81–3.73 (m, 2H), 3.57–3.42 (m, 3H), 3.41–3.26 (m, 2H), 3.24–3.10 (m, 2H), 1.93 (d, J = 12.2 Hz, 2H), 1.88–1.81 (m, 2H), 1.80–1.68 (m, 2H), 1.43–1.34 (m, 1H), 1.34–1.06 (m, 16H). MS (ESI) m/z 386.8 [M + H]⁺.

N-(trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)pyrrolidine-1-carboxamide (19).

Pyrrolidine-1-carbonyl chloride was used to afford a powder (yield 39%). ¹H NMR (600 MHz, CD₃OD) δ 7.38–7.24 (m, 3H), 7.24–7.19 (m, 1H), 4.70–4.27 (m, 2H), 3.90–3.49 (m, 3H), 3.46–3.39 (m, 4H), 3.25–3.06 (m, 4H), 2.01–1.81 (m, 8H), 1.78–1.72 (m, 2H), 1.45–1.35 (m, 1H), 1.33–1.26 (m, 2H), 1.24–1.13 (m, 2H). MS (ESI) m/z [M + H]⁺.

tert-Butyl (trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)carbamate (20).

Compound 20 was prepared by the same procedures as preparing compound 10. 1,2,3,4-Tetrahydroisoquinoline (266 mg, 2 mmol), compound 4 (482 mg, 2 mmol) and sodium triacetoxyborohydride (848 mg, 4 mmol) were used to give the title compound as an oil (460 mg, yield 49%). ¹H NMR (600 MHz, CD₃OD) δ 7.40–7.14 (m, 4H), 4.62 (d, J = 15.0 Hz, 1H), 4.34 (d, J = 15.1 Hz, 1H), 3.94–3.62 (m, 3H), 3.43–2.98 (m, 4H), 2.13–1.63 (m, 6H), 1.48–1.28 (m, 10H), 1.27–1.06 (m, 4H). MS (ESI) m/z 359.9 [M + H]⁺.

Isopropyl (trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)carbamate (21).

Isopropyl chloroformate was used to afford a powder (yield 74%). ¹H NMR (600 MHz, CD₃OD) δ 7.34–7.24 (m, 3H), 7.21 (d, J = 7.6 Hz, 1H), 4.57–4.32 (m, 2H), 3.74–3.40 (m, 3H), 3.37–3.28 (m, 3H), 3.23–3.12 (m, 2H), 1.94 (d, J = 12.5 Hz, 2H), 1.85 (d, J = 12.8 Hz, 2H), 1.79–1.71 (m, 2H), 1.42–1.33 (m, 1H), 1.31–1.07 (m, 10H). MS (ESI) m/z 345.5 [M + H]⁺.

Ethyl (trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)carbamate (22).

Ethyl chloroformate was used to afford a powder (yield 41%). ¹H NMR (600 MHz, CD₃OD) δ 7.35–7.23 (m, 3H), 7.21 (d, J = 7.6 Hz, 1H), 4.67–4.24 (m, 2H), 4.10–3.98 (m, 2H), 3.93–3.55 (m, 3H), 3.48–3.09 (m, 4H), 1.94 (d, J = 12.3 Hz, 2H), 1.87 (dd, J = 27.4, 13.1 Hz, 2H), 1.80–1.70 (m, 2H), 1.43–1.36 (m, 2H), 1.29–1.07 (m, 6H). MS (ESI) m/z 331.7 [M + H]⁺.

N-(trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)benzamide (23).

Benzoyl chloride was used to afford a powder (yield 59%). ¹H NMR (600 MHz, CD₃OD) δ 7.82–7.77 (m, 2H), 7.54–7.49 (m, 1H), 7.44 (dd, J = 8.4, 7.0 Hz, 2H), 7.35–7.26 (m, 3H), 7.22 (d, J = 7.2 Hz, 1H), 4.64–4.58 (m, 1H), 4.36–4.30 (m, 1H), 3.90–3.77 (m, 1H), 3.43–3.21 (m, 5H), 3.17 (s, 1H), 2.06–1.99 (m, 2H), 1.91 (d, J = 13.1 Hz, 2H), 1.84–1.74 (m, 2H), 1.47–1.40 (m, 3H), 1.28–1.18 (m, 2H). MS (ESI) m/z 363.5 [M + H]⁺.

N-(trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)propionamide (24).

Propionyl chloride was used to afford a powder (yield 74%). ¹H NMR (600 MHz, CD₃OD) δ 7.34–7.26 (m, 3H), 7.23–7.17 (m, 1H), 4.78–4.42 (m, 2H), 3.70–3.54 (m, 3H), 3.42–3.30 (m, 2H), 3.24–3.14 (m, 2H), 2.16 (q, J = 7.5 Hz, 2H), 1.92 (d, J = 12.2 Hz, 2H), 1.86 (d, J = 12.6 Hz, 2H), 1.78–1.72 (m, 2H), 1.41–1.35 (m, 1H), 1.34–1.15 (m, 4H), 1.11 (t, J = 7.7 Hz, 3H). MS (ESI) m/z 315.3 [M + H]⁺.

N-(trans-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)acetamide (25).

Acetic anhydride was used to afford a powder (yield 58%). ¹H NMR (600 MHz, CD₃OD) δ 7.35–7.23 (m, 3H), 7.21 (d, J = 7.5 Hz, 1H), 4.65–4.26 (m, 2H), 3.84–3.43 (m, 3H), 3.36–3.29 (m, 2H), 3.24–3.13 (m, 2H), 1.96–1.89 (m, 5H), 1.88–1.81 (m, 2H), 1.79–1.69 (m, 2H), 1.43–1.33 (m, 1H), 1.28–1.12 (m, 4H). MS (ESI) m/z 301.4 [M + H]⁺.

3-(cis-4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (26).

Compound 26 (21 mg, yield 64%) was prepared using the same procedures as preparing compound 16–19 and 21–25 from 1,2,3,4-tetrahydroisoquinoline (13.3 mg, 0.1 mmol), compound 4 (24.1 mg, 0.1 mmol) and sodium triacetoxyborohydride (42.4 mg, 0.2 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.37–7.27 (m, 3H), 7.26–7.20 (m, 1H), 4.64 (d, J = 15.1 Hz, 1H), 4.36 (d, J = 15.1 Hz, 1H), 3.87–3.83 (m, 1H), 3.75–3.67 (m, 1H), 3.46–3.39 (m, 1H), 3.36–3.26 (m, 3H), 3.21–3.12 (m, 1H), 2.93 (s, 6H), 1.89 (t, J = 7.1 Hz, 2H), 1.74–1.59 (m, 7H), 1.59–1.45 (m, 2H). MS (ESI) m/z 330.8 [M + H]⁺.

tert-Butyl (trans-4-((3,4-Dihydroisoquinolin-2(1H)-yl)methyl)cyclohexyl)carbamate (27).

Compound 27 (31 mg, yield 15%) was prepared using the same procedures as preparing compound 14 from 1,2,3,4-tetrahydroisoquinoline (133 mg, 1 mmol) and compound 6 (100 mg, 0.44 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.34–7.22 (m, 3H), 7.21 (dd, J = 7.5, 1.4 Hz, 1H), 4.59 (d, J = 15.4 Hz, 1H), 4.32 (d, J = 15.3 Hz, 1H), 3.93–3.69 (m, 1H), 3.43–3.34 (m, 1H), 3.30 (d, J = 1.6 Hz, 3H), 3.19–3.08 (m, 2H), 2.01–1.82 (m, 5H), 1.43 (s, 9H), 1.34–1.25 (m, 2H), 1.24–1.12 (m, 2H). MS (ESI) m/z 345.6 [M + H]⁺.

3-(trans-4-((3,4-Dihydroisoquinolin-2(1H)-yl)methyl)cyclohexyl)-1,1-dimethylurea (28).

Compound 28 (15 mg, yield 68%) was prepared using the same procedures as preparing compound 15 from compound 27 (31 mg, 0.07 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.33–7.25 (m, 3H), 7.21 (d, J = 7.5 Hz, 1H), 4.64–4.54 (m, 1H), 4.37–4.29 (m, 1H), 3.87–3.75 (m, 1H), 3.56–3.48 (m, 1H), 3.44–3.36 (m, 1H), 3.32–3.24 (m, 2H), 3.19–3.10 (m, 2H), 2.88 (s, 6H), 2.01–1.84 (m, 5H), 1.43–1.33 (m, 2H), 1.26–1.14 (m, 2H). MS (ESI) m/z 316.6 [M + H]⁺.

3-(4-(2-(3,4-Dihydroisoquinolin-2(1H)-yl)ethyl)phenyl)-1,1-dimethylurea (30).

To a solution of compound 29 (40 mg, 0.16 mmol) and triethylamine (32 mg, 0.32 mmol) in dichloromethane (5 mL) was added methanesulfonyl chloride (22.8 mg, 0.2 mmol) dropwise at 0 °C. The solution was stirred for 2 h at rt before being quenched with saturated aqueous sodium bicarbonate. The mixture was diluted with water and extracted with dichloromethane. The organic phase was dried over anhydrous sodium sulfate and concentrated under reduced pressure. The resulting rESIdue was dissolved in acetonitrile (5 mL). To the resulting solution were added potassium carbonate (69 mg, 0.5 mmol) and 1,2,3,4-tetrahydroisoquinoline (65 mg, 0.5 mmol). The mixture was refluxed overnight before being diluted with water (10 mL) and extracted with ethyl acetate (3 × 10 mL). The organic phase was dried over anhydrous sodium sulfate and concentrated under reduce pressure. The resulting rESIdue was purified by silica gel column (CH3OH/DCM, 0–10%) to give an oil intermediate (30 mg). The intermediate was treated with trifluoroacetic acid (1 mL) and dichloromethane (2 mL) for 3 h at rt before being condensed under reduced pressure. To a mixture of the resulting rESIdue and potassium carbonate (40 mg, 0.3 mmol) in tetrahydrofuran (2 mL) was added dimethylcarbamoyl chloride (16.1 mg, 0.2 mmol). The mixture was stirred for 1 h before being condensed under reduced pressure. The resulting rESIdue was purified by preparative HPLC (10–100% methanol/0.1% TFA in H₂O) to give the title compound as white powder (17 mg, yield 23%). ¹H NMR (600 MHz, CD₃OD) δ 7.37 (d, J = 8.2 Hz, 2H), 7.35–7.26 (m, 3H), 7.23 (dd, J = 12.3, 7.8 Hz, 3H), 4.51 (d, J = 105.7 Hz, 2H), 3.99–3.64 (m, 1H), 3.54–3.44 (m, 3H), 3.27–3.17 (m, 2H), 3.12 (dd, J = 10.4, 6.3 Hz, 2H), 3.01 (s, 6H). MS (ESI) m/z 324.3 [M + H]⁺.

3-(5-(3,4-Dihydroisoquinolin-2(1H)-yl)pentyl)-1,1-dimethylurea (32).

Compound 32 (40 mg, yield 10%) was prepared using the same procedures as preparing compound 15 from 1,2,3,4-tetrahydroisoquinoline (133 mg, 1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.34–7.24 (m, 3H), 7.21 (d, J = 7.5 Hz, 1H), 4.44 (s, 2H), 3.61–3.55 (m, 2H), 3.28–3.24 (m, 2H), 3.23–3.17 (m, 4H), 2.87 (s, 6H), 1.89–1.83 (m, 2H), 1.63–1.55 (m, 2H), 1.47–1.42 (m, 2H). MS (ESI) m/z 290.3 [M + H]⁺.

3-(trans-4-(2-(5,6-Dichloro-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (33).

Compound 33 (27 mg, 53%) was prepared using the same procedures as preparing compound 12 from 5,6-dichloro-1,2,3,4-tetrahydroisoquinoline (20.2 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.51 (d, J = 8.3 Hz, 1H), 7.21 (d, J = 8.4 Hz, 1H), 4.64–4.30 (m, 2H), 3.83–3.45 (m, 3H), 3.37–3.32 (m, 2H), 3.26–3.17 (m, 2H), 2.87 (s, 6H), 1.92 (dd, J = 13.4, 3.7 Hz, 2H), 1.85 (d, J = 12.5 Hz, 2H), 1.78–1.71 (m, 2H), 1.42–1.35 (m, 1H), 1.34–1.29 (m, 2H), 1.19–1.12 (m, 2H). MS (ESI) m/z 398.1 [M + H]⁺.

3-(trans-4-(2-(7,8-Dihydro-[1,3]dioxolo[4,5-g]isoquinolin-6(5H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (34).

Compound 34 (24 mg, 49%) was prepared using the same procedures as preparing compound 12 from 5,6,7,8-tetrahydro-[1,3]dioxolo[4,5-g]isoquinoline (17.7 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 6.72 (s, 1H), 6.67 (s, 1H), 5.95 (s, 2H), 4.52–4.39 (m, 1H), 4.25–4.12 (m, 1H), 3.82–3.67 (m, 1H), 3.51–3.45 (m, 1H), 3.35–3.26 (m, 3H), 3.17–2.98 (m, 2H), 2.87 (s, 6H), 1.92 (dd, J = 13.5, 3.8 Hz, 2H), 1.87–1.81 (m, 2H), 1.76–1.71 (m, 2H), 1.39–1.33 (m, 1H), 1.34–1.26 (m, 2H), 1.18–1.09 (m, 2H). MS (ESI) m/z 374.2 [M + H]⁺.

3-(trans-4-(2-(5-Chloro-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (35).

Compound 35 (23 mg, 48%) was prepared using the same procedures as preparing compound 12 from 5-chloro-1,2,3,4-tetrahydroisoquinoline (13.3 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.43 (d, J = 7.9 Hz, 1H), 7.34–7.29 (m, 1H), 7.20 (d, J = 7.7 Hz, 1H), 4.64–4.28 (m, 2H), 3.87–3.45 (m, 3H), 3.35–3.29 (m, 2H), 3.20–3.13 (m, 2H), 2.87 (s, 6H), 1.94–1.89 (m, 2H), 1.85 (d, J = 13.1 Hz, 2H), 1.78–1.72 (m, 2H), 1.43–1.35 (m, 1H), 1.33–1.27 (m, 2H), 1.19–1.12 (m, 2H). MS (ESI) m/z 364.2 [M + H]⁺.

3-(trans-4-(2-(6-Chloro-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (36).

Compound 36 (19 mg, 40%) was prepared using the same procedures as preparing compound 12 from 6-chloro-1,2,3,4-tetrahydroisoquinoline (13.3 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.32 (dd, J = 8.3, 2.2 Hz, 1H), 7.29–7.25 (m, 2H), 4.71–4.13 (m, 2H), 3.92–3.41 (m, 3H), 3.34–3.30 (m, 2H), 3.24–3.10 (m, 2H), 2.87 (s, 6H), 1.94–1.89 (m, 2H), 1.85 (d, J = 12.5 Hz, 2H), 1.77–1.72 (m, 2H), 1.42–1.35 (m, 1H), 1.34–1.26 (m, 2H), 1.17–1.12 (m, 2H). MS (ESI) m/z 364.2 [M + H]⁺.

3-(trans-4-(2-(7-Chloro-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (37).

Compound 37 (19 mg, 42%) was prepared using the same procedures as preparing compound 12 from 7-chloro-1,2,3,4-tetrahydroisoquinoline (13.3 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.35–7.29 (m, 2H), 7.21 (dd, J = 8.4, 1.8 Hz, 1H), 4.56–4.31 (m, 2H), 3.79–3.44 (m, 3H), 3.34–3.27 (m, 2H), 3.23–3.11 (m, 2H), 2.87 (s, 6H), 1.92–1.93 (m, 2H), 1.85 (d, J = 13.1 Hz, 2H), 1.77–1.69 (m, 2H), 1.42–1.35 (m, 1H), 1.34–1.25 (m, 2H), 1.18–1.12 (m, 2H). MS (ESI) m/z 364.1 [M + H]⁺.

3-(trans-4-(2-(8-Chloro-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (38).

Compound 38 (25 mg, 52%) was prepared using the same procedures as preparing compound 12 from 8-chloro-1,2,3,4-tetrahydroisoquinoline (13.3 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). A portion of compound 38 was converted to free base form with aqueous sodium bicarbonate for ¹H NMR and ¹³C NMR. ¹H NMR (800 MHz, CD₃OD) δ 7.22 (d, J = 7.9 Hz, 1H), 7.19–7.15 (m, 1H), 7.10 (d, J = 7.6 Hz, 1H), 3.66 (s, 2H), 3.56–3.49 (m, 1H), 2.95 (t, J = 6.1 Hz, 2H), 2.89 (s, 6H), 2.76 (t, J = 6.0 Hz, 2H), 2.66–2.61 (m, 2H), 1.93 (dd, J = 13.1, 4.0 Hz, 2H), 1.89–1.84 (m, 2H), 1.59–1.53 (m, 2H), 1.36–1.29 (m, 3H), 1.12 (dd, J = 12.3, 3.3 Hz, 2H). ¹³C NMR (201 MHz, CD₃OD) δ 159.1, 136.7, 131.9, 131.8, 127.1, 127.1, 126.2, 56.0, 53.6, 50.0, 49.7, 35.4, 35.1, 33.4, 33.0, 32.1, 28.5. MS (ESI) m/z 364.2 [M + H]⁺. HRMS m/z [M + H]⁺ calcd for C₂₀H₃₁ClN₃O₊ 364.2150, found 364.2165.

1,1-Dimethyl-3-(trans-4-(2-(8-methyl-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)urea (39).

Compound 39 (19 mg, 42%) was prepared using the same procedures as preparing compound 12 from 8-methyl-1,2,3,4-tetrahydroisoquinoline (14.7 mg, 0.1 mmol) and compound 8 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.22–7.16 (m, 1H), 7.10 (dd, J = 17.3, 7.5 Hz, 2H), 4.55 (d, J = 15.5 Hz, 1H), 4.20 (d, J = 15.5 Hz, 1H), 3.80 (d, J = 11.1 Hz, 1H), 3.54–3.46 (m, 1H), 3.44–3.20 (m, 4H), 3.10 (d, J = 16.5 Hz, 1H), 2.87 (s, 6H), 2.27 (s, 3H), 1.98–1.89 (m, 2H), 1.86 (d, J = 13.1 Hz, 2H), 1.81–1.75 (m, 2H), 1.43–1.35 (m, 1H), 1.34–1.26 (m, 2H), 1.21–1.09 (m, 2H). MS (ESI) m/z 344.2 [M + H]⁺.

3-(trans-4-(2-(8-Bromo-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (40).

Compound 40 (24 mg, 46%) was prepared using the same procedures as preparing compound 12 from 8-bromo-1,2,3,4-tetrahydroisoquinoline (21.2 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.56 (d, J = 7.8 Hz, 1H), 7.30 (d, J = 7.7 Hz, 1H), 7.25 (t, J = 7.8 Hz, 1H), 4.64–4.18 (m, 2H), 3.85–3.44 (m, 3H), 3.42–3.37 (m, 2H), 3.26–3.20 (m, 2H), 2.87 (s, 6H), 1.92 (dd, J = 13.4, 3.7 Hz, 2H), 1.86 (d, J = 12.3 Hz, 2H), 1.77 (q, J = 7.4 Hz, 2H), 1.43–1.36 (m, 1H), 1.35–1.28 (m, 2H), 1.19–1.13 (m, 2H). MS (ESI) m/z 408.2 [M + H]⁺.

3-(trans-4-(2-(8-Fluoro-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (41).

Compound 41 (20 mg, 43%) was prepared using the same procedures as preparing compound 12 from 8-fluoro-1,2,3,4-tetrahydroisoquinoline (15.1 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.39–7.34 (m, 1H), 7.11 (d, J = 7.8 Hz, 1H), 7.06 (dd, J = 9.9, 8.3 Hz, 1H), 4.33 (s, 2H), 3.85–3.44 (m, 3H), 3.40–3.35 (m, 2H), 3.27–3.19 (m, 2H), 2.87 (s, 6H), 1.93–1.89 (m, 2H), 1.88–1.82 (m, 2H), 1.80–1.73 (m, 2H), 1.43–1.35 (m, 1H), 1.34–1.26 (m, 2H), 1.19–1.12 (m, 2H). MS (ESI) m/z 348.2 [M +H]⁺.

1,1-Dimethyl-3-(trans-4-(2-(8-(trifluoromethyl)-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)urea (42).

Compound 42 (15 mg, 29%) was prepared using the same procedures as preparing compound 12 from 8-(trifluoromethyl)-1,2,3,4-tetrahydroisoquinoline (15.1 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹H NMR (600 MHz, CD₃OD) δ 7.68 (d, J = 7.6 Hz, 1H), 7.57 (d, J = 7.7 Hz, 1H), 7.54–7.49 (m, 1H), 4.78–4.42 (m, 2H), 4.03–3.69 (m, 1H), 3.53–3.46 (m, 1H), 3.45–3.28 (m, 5H), 2.87 (s, 6H), 1.95–1.89 (m, 2H), 1.88–1.83 (m, 2H), 1.81–1.69 (m, 2H), 1.44–1.35 (m, 1H), 1.34–1.24 (m, 2H), 1.19–1.10 (m, 2H). MS (ESI) m/z 398.2 [M + H]⁺.

3-(trans-4-(2-(8-Methoxy-3,4-dihydroisoquinolin-2(1H)-yl)ethyl)cyclohexyl)-1,1-dimethylurea (43).

Compound 43 (19 mg, 40%) was prepared using the same procedures as preparing compound 12 from 8-methoxy-1,2,3,4-tetrahydroisoquinoline (16.3 mg, 0.1 mmol) and compound 4 (24.1 mg, 0.1 mmol). ¹²H NMR (600 MHz, CD₃OD) δ 7.12 (d, J = 8.5 Hz, 1H), 6.88–6.82 (m, 1H), 6.81 (d, J = 2.6 Hz, 1H), 4.59–4.17 (m, 2H), 3.78 (s, 3H), 3.53–3.34 (m, 1H), 3.33–3.24 (m, 4H), 3.23–3.11 (m, 2H), 2.87 (d, J = 1.3 Hz, 6h), 1.92 (dd, J = 12.9, 3.7 Hz, 2H), 1.87–1.82 (m, 2H), 1.76–1.70 (m, 2H), 1.42–1.34 (m, 1H), 1.34–1.23 (m, 2H), 1.19–1.06 (m, 2H). MS (ESI) m/z 360.4 [M + H]⁺.

Experimental Procedures for in Vitro Biochemical Assays.

D₂R G_i/O-Mediated cAMP Inhibition Assay.

D₂R or D³R G_i/o-mediated cAMP inhibition assays were performed in parallel with D₂R or D₃R β-Arrestin recruitment Tango assays. Human D₂R or D₃R transfected HEK293T cells co-expressing the cAMP biosensor GloSensor-22F (Promega) were seeded (15 000 cells/40 μL/well) into white 384 clear-bottom, tissue culture plates in Dulbecco’s modified Eagle’s medium (DMEM) containing 1% dialyzed fetal bovine serum (FBS). Next day, drug dilutions were made in drug buffer (Hank’s balanced salt solution (HBSS), 20 mM N-(2-hydroxyethyl)piperazine-N'-ethanesulfonic acid (HEPES), 0.1% bovine serum albumin (BSA), 0.01% ascorbic acid, pH 7.4), and the same drug dilutions used for the G_i/o-mediated cAMP inhibition assays were also used for β-arrestin recruitment Tango assay. Media were removed and 20 μL of drug buffer was added per well and allowed to equilibrate for at least 15 min room temperature. To start the assay, cells were treated with 5 μL/well of 5× drug using a FLIPR (Molecular Devices). After 15 min, cAMP accumulation was initiated by the addition of 5 μL/well of either 0.3 μM isoproterenol (final concentration) for D₂R or 1 μM of Forskolin for D₃R diluted in GloSensor reagent. Luminescence per well per second was read on a MicroBeta plate counter (PerkinElmer). Data were normalized to maximum cAMP inhibition by quinpirole (100%) and basal cAMP accumulation by isoproterenol (0%). Data were analyzed using the sigmoidal dose–response function built into GraphPad Prism 5.0. Notably, HEK293T cells expressing the GloSensor-22F alone (no hD₂R) were assayed in parallel and displayed no inhibition of isoproterenol-stimulated cAMP, either by quinpirole or by the test compounds, suggesting that the effect observed in hD_2L-expressing cells was due to compound acting via the recombinant receptor.

D₂R β-Arrestin Recruitment Tango Assay.

Recruitment of β-arrestin to agonist-stimulated D_2L or D₃ receptors was performed using Tango.^41,56 Briefly, HTLA cells stably expressing β-arrestin-TEV protease and a tetracycline transactivator-driven luciferase were plated in 15 cm dishes in DMEM containing 10% FBS and transfected (via calcium phosphate) with 10 μg of a D₂V₂-TCS-tTA or D₃V₂-TCS-tTA construct. The next day, cells were plated in white, clear-bottom, 384-well plates (Greiner; 15 000 cells/well, 40 μL/well) in DMEM containing 1% dialyzed FBS. The following day, media were decanted and exchanged for fresh DMEM media containing 1% dialyzed FBS. Importantly, the same drug dilutions used for the G_i/o-mediated cAMP inhibition assay were used for the Tango assay to prevent compound solubility variability between assays. Cells were treated with 10 μL/well of drug using a FLIPR and for D₂R antagonist assays, antagonists were diluted in 10 nM dopamine. After at least 20–22 h, the medium were removed and replaced with 1:20 diluted BrightGlo reagent (Promega), and luminescence per well was read using a MicroBeta plate counter (PerkinElmer). Data were normalized to vehicle (0%) and quinpirole or dopamine-stimulated controls (100%) and analyzed using the sigmoidal dose–response function built into GraphPad Prism 5.0.

D₂R β-Arrestin Recruitment BRET Assay.

To measure D₂R-mediated β-arrestin recruitment, HEK293T cells were co-transfected in a 1:1:15 ratio with human D_2long containing a C-terminal renilla luciferase (RLuc8), GRK2, and Venus-tagged N-terminal β-arrestin2. Next day, transfected cells were plated in polylysine coated 96-well white clear-bottom cell culture plates in plating media (DMEM +1% dialyzed FBS) at a density of 30 000–40 000 cells in 200 μL/well and incubated overnight. Next day, media was decanted and cells were washed twice with 60 μL of drug buffer (1× HBSS, 20 mM HEPES, 0.1% BSA, 0.01% ascorbic acid, pH 7.4). Then 60 μL of drug buffer was added per well and drug stimulation was initiated with addition of 30 μL of drug (3×) per well. Then 10 μL/well of RLuc substrate, coelenterazine h (Promega, 5 μM final concentration) was added and read 15 min post drug addition, which is the same time point for D₂R Gαi1-γ2 dissociation BRET assay. Plates were read for both luminescence at 485 nm and fluorescent eYFP emission at 530 nm for 1 s/well using a Mithras LB940. The ratio of eYFP/RLuc was calculated per well and the net BRET ratio was calculated by subtracting the eYFP/RLuc per well from the eYFP/RLuc ratio in wells without Venus-β-arrestin2 present. The net BRET ratio was plotted as a function of drug concentration using GraphPad Prism 5 (GraphPad Software Inc., San Diego, CA). Data were normalized to % quinpirole or dopamine stimulation and analyzed using nonlinear regression log(agonist) versus response.

D₂R Gαi1-γ2 Dissociation BRET Assay.

To measure D₂R-mediated Gαi1-γ2 dissociation, HEK293T cells were co-transfected in a 1:1:1:1 ratio of Gαi1-RLuc, Gβ1, GFP₂-Gγ2, and human D_2long, respectively. Gαi1-RLuc, Gβ1, and Gγ2-GFP₂ constructs were generously provided by Dr. Michel Bouvier. Cells were plated and assays were performed exactly similar to the BRET arrestin assay, except the substrate used was Coelenterazine 400a (NanoLight, 5 μM final concentration). Plates were read after 15 min drug incubation, which is the same time point for D₂R β-arrestin recruitment BRET assay, measuring luminescence at 400 nm and fluorescent GFP₂ emission at 515 nm for 1 s/well using a Mithras LB940. The ratio of GFP₂/RLuc was calculated per well and plotted as a function of drug concentration using GraphPad Prism 5 (GraphPad Software Inc., San Diego, CA). Data were normalized to % quinpirole stimulation and analyzed using nonlinear regression log(agonist) versus response.

Bias Calculation.

Transduction coefficients (log(τ/K_A)) were calculated using the Black and Leff operational model in GraphPad Prism 5.0, where τ is agonist efficacy and K_A is the equilibrium dissociation constant and were calculated with respect to quinpirole, the D2 full agonist reference. Transduction coefficients were calculated for both pathways and averaged across experiments. Calculation of bias factors utilized the method by Kenakin et al.,⁵⁷ where the Δlog/τK_A) was calculated relative to the reference and the ΔΔlog(τ/K_A.) was calculated by subtracting the β-arrestin2 from the G_i/o transduction coefficient to yield the antilog transformed bias factors with respect to G_i/o activity.

Molecular Docking of Compounds 1, 16, and 38.

The crystal structure of the dopamine D₂ receptor (D₂R) in complex with risperidone was retrieved from the Protein Data Bank (www.rcsb.org) by its identifying code (PDB: 6CM4). The structure was then manually edited to remove the T4 lysozyme, oleic acid, and the di(hydroxyethyl)ether. The structure also contained three thermo-stabilizing mutations: I122^3.40A, L375^6.37A, and L379^6.41A, which were retained, as these rESIdue positions are not near the ligand binding site and were not predicted to impact binding pose. Risperidone was retained for purposes of maintaining an orthosteric site that was conducive to ligand binding during the modeling process. Because antagonists, such as risperidone, are known to stabilize different receptor conformations compared to agonists, we sought to compare this structure to that of a previous D₂R model that was successfully implemented to model functionally selective D₂ agonists.⁴¹ The two structures were aligned and superimposed in PyMol 2.2.0. The allatom binding site root-mean-square deviation (RMSD) was computed in PyMol after overlaying the two D₂R structures and selecting the atoms in the transmembrane helices and extracellular domains that border the orthosteric binding pocket of the receptor. The RMSD was calculated to ensure sufficient similarity between the atom positions of the rESIdues important for binding. The all-atom binding site RMSD between our model and this previous D₂R model was 0.74 Å,⁴¹ which indicates a high degree of similarity between the arrangements of the atoms lining the binding pockets of both structures (Figure 3A). Compounds 1, 16, and 38 were docked to the orthosteric binding site of our D₂R model using DOCK3.7.2.⁵⁸ As previously described,⁵⁹ the Dock 3.7.2 program superimposes atoms of each molecule onto matching spheres to dock flexible ligands into a binding site. Matching spheres represent favorable positions for individual ligand atoms, and in this study, 45 was used, based on the atom positions of the original risperidone complex. For docking, the receptor structure was prepared, and AMBER united-atom charges were assigned. Reduce⁶⁰ was utilized to protonate receptor structures, whereas partial charges from the united-atom AMBER⁶¹ force field were used for all receptor atoms. Energy grids were used to assess the various energy terms of the DOCK scoring function including AMBER van der Waals potential, Poisson–Boltzmann electrostatic potentials using QNIFFT, and ligand desolvation from the occluded volume of the target for different ligand orientations. Prior to docking, ligand conformations and protonation states were calculated, and ligands were prepared in an energy-minimized state. Marvin (version 18.12, ChemAxon, 2018; https://www.chemaxon.com) was used to protonate each ligand at pH 7.4. Corina⁶² (Molecular Networks GmbH) was then used to create a three-dimensional rendering of each protomer before conformational sampling via Omega (OpenEye Scientific Software).⁶³ Finally, AMSOL⁶⁴ was used to compute the charges and initial solvation energies of the ligands. Following docking to the matching spheres, the lowest energy poses for each ligand were visualized and compared with existing crystal structures of ligands bound to aminergic GPCRs.^38,65,66

Experimental Procedures for in Vivo Studies in Mice.

Receptor Occupancy Assay.

In vivo receptor occupancy was assessed in male c57 mice. All procedures performed on animals were in accordance with regulations and established guidelines and were reviewed and approved by an Institutional Animal Care and Use Committee.

Mice were dosed with vehicle (5% DMSO: 5% Cremophor: 90% saline plus 1 mequiv of HCl) or 38 (0.32, 1, 3.2, 10, or 32 mg/kg) subcutaneously at a dose volume of 10 mL/kg. After 20 min of the subcutaneous dose, the mice were dosed retro-orbitally with 100 μci [³H]raclopride at a dose volume of 5 mL/kg. After a 30 min drug pretreatment period, the animals were euthanized by live decapitation. Trunk blood was collected and centrifuged at 7500 rpm for 10 min, and the plasma was collected and stored at −80 °C until exposure level analysis. The striatum and the cerebellum (used to define nonspecific binding) were immediately dissected and homogenized in 20 mg/mL of assay buffer (20 mM HEPES, 4.16 mM NaHCO₃, 0.44 mM KH₂PO₄, 0.63 mM NaH₂PO₄ 127 mM NaCl, 5.36 mM KCl, 1.26 mM CaCl₂, 0.98 mM MgCl₂) with a polytron for 10 s. The rest of the brain was immediately frozen on dry ice and stored at −80 °C until exposure level analysis. Striatum or cerebellum (400 μL) was filtered through presoaked Whatman GF/B filter circles followed by 2 × 5 mL of assay buffer. Filters were transferred to scintillation vials containing Ultima Gold MV scintillation cocktail. Radioactivity was measured using the PerkinElmer TriCarb liquid scintillation counter.

Locomotor Activity Assay.

Adult male and female C57BL/6 mice were housed 3–5/cage on a 14:10 h light/dark cycle (lights on 0700 h) in a humidity- and temperature-controlled room with food and water provided ad libitum. All experiments were conducted with an approved protocol from the Duke University Institutional Animal Care and Use Committee. Motor activity was assessed in a 21 × 21 × 30 cm³ open field (Omnitech Inc., Columbus, OH) under 340 lx illumination. C57BL/6 mice were housed in the test room 24 h prior to testing. Animals were injected (i.p.) with vehicle or different doses of 38 and immediately placed into the open field. After 30 min, mice were removed from the test arena and given (i.p.) the vehicle or 6 mg/kg PCP (Sigma-Aldrich, St. Louis, MO) and returned immediately to the open field for an additional 90 min. Horizontal distance traveled (cm) was quantitated with Fusion software (Omnitech) and scored in 5 min bins across testing. The behavioral data are presented as means and standard errors of the mean and analyzed by repeated measures ANOVA followed by Bonferroni-corrected pair-wise comparisons where a p < 0.05 was considered significant.

ABBREVIATIONS

GPCR

G protein-coupled receptor

SFSR

structure—functional selectivity relationships

D₂R

dopamine D₂ receptor

cAMP

cyclic adenosine monophosphate

LHS

left-hand side

RHS

right-hand side

BRET

bioluminescence resonance energy transfer

GRK2

G protein-coupled receptor kinase 2

D₁R

dopamine D₁ receptor

D₃R

dopamine D₃ receptor

D₄R

dopamine D₄ receptor

D₄R

dopamine D₅ receptor

TFA

trifluoroacetic acid

DCM

dichloromethane