Pharmacology and Therapeutic Potential of Benzothiazole Analogues for Cocaine Use Disorder

Pharmacological targeting of the dopamine D4 receptor (D4R)—expressed in brain regions that control cognition, attention, and decision-making—could be useful for several neuropsychiatric disorders including substance use disorders (SUDs). This study focused on the synthesis and evaluation of a novel series of benzothiazole analogues designed to target D4R. We identified several compounds with high D4R binding affinity (Ki ≤ 6.9 nM) and >91-fold selectivity over other D2-like receptors (D2R, D3R) with diverse partial agonist and antagonist profiles. Novel analogue 16f is a potent low-efficacy D4R partial agonist, metabolically stable in rat and human liver microsomes, and has excellent brain penetration in rats (AUCbrain/plasma > 3). 16f (5–30 mg/kg, i.p.) dose-dependently decreased iv cocaine self-administration in rats, consistent with previous results produced by D4R-selective antagonists. Off-target antagonism of 5-HT2A or 5-HT2B may also contribute to these effects. Results with 16f support further efforts to target D4R in SUD treatment.


■ INTRODUCTION
D 1 -like (D 1 R, D 5 R) and D 2 -like (D 2 R, D 3 R, D 4 R) dopamine receptors are G protein-coupled receptors that regulate physiological functions such as movement, emotion, and cognition. 1,2Compared to D 2 R and D 3 R, D 4 Rs have the lowest level of expression in the brain and are uniquely distributed primarily in the prefrontal cortex and hippocampus.In the prefrontal cortex, D 4 R plays important roles in cognition, attention, decision making, and executive function.−5 A better understanding of D 4 Rmediated signaling is essential to developing new pharmacotherapeutic treatments.
Studies with D 4 R-selective ligands demonstrate that D 4 R is a promising therapeutic target for the treatment of several neuropsychiatric conditions, notably ADHD and SUDs. 6,7 4 R-selective agonists alter cognition in animal models. 8,9In rats, D 4 R-selective agonists A-412997 (1) and PD168077 (2)  improved performance in the novel object recognition task 10,11 and the five-trial inhibitory avoidance task. 124 R antagonism may be useful to treat L-DOPA-induced dyskinesias and SUDs, particularly for psychostimulants like cocaine.7,13−17 D 4 R antagonist NGD-94−1 (3) 18 reduced cocaine-maintained behavior in nonhuman primates, 15 and antagonist L-745,870 (4; 3-(4-[4-chlorophenyl]piperazin-1yl)-methyl-1H-pyrrolo [2,3-b]pyridine; Figure 1) reduced alcohol-taking and alcohol-seeking behaviors in rats.19 One of the most well-studied D 4 R-selective compounds, L-745,870, is >100-fold selective for D 4 R over other dopamine receptors (D 1 R, D 2 R, D 3 R, and D 5 R) with subnanomolar affinity.20,21 However, the compound acts as a partial agonist, binds to several nondopaminergic receptors, 22 and failed to reduce psychotic symptoms in a phase IIa clinical study evaluating its utility as an antipsychotic.23,24 These issues are emblematic of many of the research ligands that have been used to investigate D 4 R physiology to date, which stem from unsuccessful attempts in the 1990s to develop new antipsychotics based on the idea that clozapine's unusually high D 4 R affinity may have been a key driver in its clinical success.The recent resurgence in D 4 R medication development, by our group and many others, has been reviewed ably by others. 6,7 5−29 Despite the clinical implications of cocaine use disorder (CUD), there are no FDA-approved medications for CUD treatment or approved drugs that selectively target D 4 R.The goal of this study was the development and characterization of novel ligands with high D 4 R affinity and selectivity for investigation in rodent models of CUD. Co32 We used 16a 31 as the chemical template for two rounds of structure−activity relationship (SAR)-guided synthesis with the goal of identifying novel druggable D 4 R-selective lead compounds.
We synthesized an initial library of analogues (16b−16f) featuring modifications to the pyrimidinylpiperazine region of 16a and variations in the linker chain length.Following extensive in vitro analyses, including binding and functional studies, we determined that these modifications resulted in several novel analogues that retained high D 4 R binding affinity, improved D 2 -like subtype selectivity, and showed low-efficacy partial agonist or antagonist profiles at D 4 R. Based on its initial profile, we expanded the characterization of 16f to include diverse receptor screening, pharmacokinetic and metabolic stability studies, including rat and human liver microsomes, and in vivo behavioral analysis in rats trained to self-administer cocaine.In parallel with the in vivo characterizations of 16f, we pursued a second round of SAR studies exploring substitutions on both the terminal benzo [d]thiazole and the pyrimidine/ pyridine scaffolds as well as piperazine versus piperidine moieties, as identified in an in silico screen using the D 4 R crystal structure (crystallized with the antipsychotic nemonapride (6; Figure 1) modified via molecular docking. 28hese diverse libraries provide valuable insight into the SAR driving the D 4 R ligand affinity, efficacy, and target selectivity.(5) and nemonapride (6).Following the discovery of the relatively high affinity of 5 for D 4 R, 30 several pharmaceutical companies developed novel ligands targeting D 4 R, including 1−4.A recent resurgence in D 4 R ligand discovery has identified a diverse array of novel agonists (7), partial agonists (8), and antagonists (9−13) including lead compound 16a from this study.

■ CHEMISTRY
Our first and second series of ligands were synthesized as outlined in Schemes 1 and 2, respectively, using routine Nalkylation reactions previously reported. 2In Schemes 1 and 2, substituted or unsubstituted 2-aminobenzenethiol compound 14 or 17 was reacted with 4-chlorobutanoyl chloride or 5chloropentanoyl chloride to give intermediate compound 15 or 18. Alkylation of the unsubstituted arylpiperazine moiety with intermediate compound 15 delivered target compound 16.The same procedure was employed to make target compounds 19 and 20 by alkylation of the substituted or unsubstituted arylpiperazine or arylpiperidine moiety with intermediate compounds 15b and 18.The requisite substituted or unsubstituted aminothiophenols and substituted or unsubstituted arylpiperazines or arylpiperidines are commercially available.
In a small initial SAR series, we employed two classes of modifications to parent compound 16a, varying the alkyl linker chain length and substitutions on the arylpiperazine, producing analogues 16b−16f.In this initial series, the 2-pyrimidine moiety of 16a was replaced with 2-pyridine in 16b, 5methylpyridin-2-yl in 16c, and 5-chloropyridin-2-yl in 16d.To evaluate the contribution of the alkyl chain to binding affinity and subtype selectivity, we synthesized alkyl chain length analogues of compounds 16a and 16b, removing one methylene from the linker chain in compounds 16e and 16f, respectively.
Based on our pharmacological evaluation of this initial series, one compound (16f) was chosen for further investigation.Simultaneously, the overall library was later extended with a second series (19a−f, 20a−e) designed after in silico docking studies, as shown in Scheme 2.
This extended library featured additional arylpiperazine/ arylpiperidine substitutions as well as modifications to the benzo [d]thiazole.In our second series, the pyridine of compound 16f was substituted with 4-methylpyridin-2-yl to form 19a. The piperazine attached to the three-C-linker chain on compound 19b with a 3-methoxy-2-pyridinyl moiety was replaced with a piperidine to form 20a with a 3-methyl-2pyridinyl moiety.The 3-methyl-2-pyridine of 20a was substituted with 5-methoxy-2-pyridine to form 20b. Finally, we probed the contribution of the benzo [d]thiazole moiety via substitution on the phenyl ring with electron-donating (methyl) and electron-withdrawing (chloro) groups (compound 19c compared to compound 19d, respectively) with a propyl linker attached to the pyridin-2-yl-piperazin-1-yl moiety.Additional substitutions on the benzo[d]thiazole moiety produced compounds 19e,19f.We also probed combinations of benzo[d]thiazole moiety substitutions with pyridin-2-

Journal of Medicinal Chemistry
ylpiperidin-1-yl moieties attached to the propyl linker to obtain compounds 20c−20e.

■ PHARMACOLOGICAL RESULTS AND DISCUSSION
SARs at Dopamine D 2 -Like Receptors: Initial Library.Compound 16a and the structures of all synthesized analogues across both series are shown in Table 1.Overall, all compounds exhibited cLogP values of less than 5 and library members consistently demonstrated higher binding affinity for D 4 R over D 2 R and D 3 R.The binding affinities of all compounds were evaluated via radioligand competition binding studies using [ 3 H]N-methylspiperone and membranes prepared from HEK293 cells stably expressing human dopamine D 2 -like receptors (D 2 R, D 3 R, or D 4 R).Binding data for all of the ligands are shown in Table 1.In addition, cLogP and polar surface area (PSA) values were calculated to provide measures of polarity (Table 1).Functional analyses of each compound were completed using the LANCE assay for cAMP (Table 2) and the DiscoverX β-arrestin recruitment assay (Table 3), in both agonist and antagonist modes, using Chinese hamster ovary (CHO) cells stably expressing D 2 R, D 3 R, or D 4 R.In agonist mode, E max values for each compound are in comparison to dopamine and EC 50 values represent agonist potency.In antagonist mode, I max values for each compound represent inhibition of dopamine-induced signaling and IC 50 values represent antagonist potency.
In our initial series, modification of the pyrimidine ring of 16a to a pyridine or substituted pyridine (16b−16d) modestly decreased D 4 R binding affinity in competition binding assays and decreased fold selectivity over the structurally related D 3 R. Compounds 16b and 16c had decreased but still highly potent D 4 R binding affinity at 9.9 ± 2.0 and 21.2 nM ± 1.37, respectively.The D 2 R/D 4 R fold selectivity was similar to compound 16a at ∼40-fold, but the D 3 R/D 4 R fold selectivity decreased to 6-fold (16b) and 10-fold (16c).In cAMP antagonism assays, 16b and 16c were full antagonists (∼100% inhibition) at D 4 R but lost potency (IC 50 = 123 nM [95.9− 157] and 600 nM [467−771], respectively) compared to 16a (IC 50  = 31.8nM [24.7−40.9]).Furthermore, 16b inhibited cAMP production via activation of D 2 R (E max = 59.4 ± 3.0%, EC 50 = 124 nM [68.9−222]) while the methyl substitution in 16c remained inactive in agonist mode but was a full D 2 R antagonist with low potency (I max = 107 ± 5, IC 50  The chloro-substituted pyridine 16d maintained a high D 4 R binding affinity (K i = 4.9 ± 0.6 nM) with decreased D 2 R affinity (K i = 830 ± 160 nM) compared to 16a (K i = 127 ± 10 nM), which improved D 4 R selectivity over D 2 R 171-fold.16d maintained a D 3 R binding affinity similar to that of 16a and thus had no improvement in D 4 R selectivity over D 3 R.However, cAMP and β-arrestin recruitment were greatly diminished across all receptors.Starting with D 4 R, 16d lost potency in cAMP assays with the estimated IC 50 = 10,800 nM  [7520−15,500].In the β-arrestin assay, 16d was a full antagonist (I max = 103 ± 4%, IC 50 = 414 nM [274−621]).At D 2 R, 16d had greatly reduced potency at both cAMP (I max and IC 50 = not determined) and β-arrestin recruitment (I max = 108 ± 19%, IC 50  = 23,000 nM [8960−67,700]).16d did not recruit β-arrestin to D 3 R, indicating no detectable agonist activity, and very low potency antagonism was suggested but accurate potency was not determined due to incomplete (unsaturated) curves.Although the D 4 R binding affinity was not affected by the chloro substituent, functional activity was greatly diminished for all D 2 -like receptors.
Removing one methylene unit from the linker chain� decreasing the alkyl linker from four carbons to three� markedly improved selectivity for D 4 R by reducing D 2 R and D 3 R binding affinities.Compared to similar butyl linker compounds 16a−16c, the propyl linker chain in 16e,fand 19a maintained the D 4 R binding affinity and greatly reduced the D 2 R and D 3 R binding affinity.For example, 16e had a D 4 R binding affinity (K i = 6.5 ± 0.6 nM) similar to that of 16a (K i = 3.1 ± 0.2 nM) but a dramatically decreased affinity for D 2 R (16e: K i = 6370 ± 1020 nM; 16a: K i = 127 ± 10 nM) and D 3 R (16e: K i = 1650 ± 120 nM; 16a: K i = 93.2± 8.3 nM).Exchanging the pyrimidine in 16e for a pyridine in 16f improved D 4 R binding affinity (K i = 2.2 ± 0.1 nM) and further increased D 4 R selectivity over both D 2 R (1326-fold) and D 3 R (520-fold).In cAMP functional assays, 16e (E max = 14.0 ± 0.8%, EC 50  = 4.3   (chemical structure shown in Figure 1) was prepared and preprocessed using Maestro's Protein Preparation Wizard. 33he preprocessed protein's charge state was optimized at pH 7.4.Then, a restrained minimization was performed to relax the protein structure using the OPLS3 force field. 34Using Maestro Elements allowed for the preparation of the 3D structures of nemonapride 2 and the 4400 16f analogues.The 3D structure of nemonapride was extracted from the crystal structure (PDB ID: 5WIU), and the initial structure of the 16f analogues was from the Combinatorial Library Enumeration tool.In order to generate each ligand's ionization/tautomeric states at pH 7.4, Maestro's Epik tool was used based on the more accurate Hammett and Taft methodologies. 33During this step, the lowest ionization/tautomeric state was chosen.Afterward, the geometry was minimized to the most energetically favorable structure to relax the ligand's structure.
Glide XP Docking and Compound Selection.The receptor grid files were generated from the prepared receptor complex in which the centroid of the crystal ligand, nemonapride, was used to specify the active site.The prepared ligands were docked into their corresponding generated grids using Glide XP scoring with default procedures and parameters. 34In detail, the receptor grid required for the docking process was generated using a van der Waals scaling factor of 1 and a partial charge cutoff of 0.25.Docking was performed by using a ligand-centered grid and an OPLS3 force field.Glide XP Dock was used to perform a comprehensive systematic search for the best receptor conformations and orientations to fit the ligand.The docked pose of the crystal ligand was confirmed with its crystal pose, thus validating the docking protocol.Figure S2 shows the interactions between 16f and residues within the D 4 R binding pocket.
Following Glide XP docking, we selected several compounds per library, focusing on compounds with the simplest substitutions while still maintaining improved docking scores relative to those of the lead scaffold.Table S2 represents the 10 identified ligands with desirable docking scores and with the greatest synthetic feasibility.Both interaction diagrams (left) and 3D representations with interacting residues (right) are provided for the best selected candidate in each of our list.
SARs at Dopamine D 2 -Like Receptors: Expanded Library.Adding a 4-methyl substituent onto the pyridine ring (19a) maintained high D 4 R affinity (K i = 2.9 ± 1.0 nM) and subtype selectivity of >450-fold over both D 2 R and D 3 R, with no agonist activity detected for any receptor in either cAMP or β-arrestin recruitment assays.At D 4 R, 19a potently antagonized cAMP inhibition (I max = 88.0 ± 4.4%, IC 50  Overall, the initial and expanded library included four key classes of modifications with distinct effects on binding and efficacy profiles across the D 2 -like receptors.We find the following SARs: (1) reducing the linker chain length from a butyl linker to a propyl linker dramatically improved D 4 R binding selectivity over D 2 R and D 3 R.This is consistent with prior literature 25,35,36 that supports alkyl linker length substantially driving D 2 -like subtype selectivity, with a propyl linker providing optimal separation of the lipophilic and basic moieties to improve D 4 R affinity and selectivity.(2) Substitution of the pyrimidine ring in initial lead 16a with a pyridinyl moiety further improved D 4 R binding affinity and selectivity over D 2 R and D 3 R. (3) Piperazine and piperidine ring moieties produce differential effects on cAMP and β-arrestin signaling at each receptor.(4) Substitutions at different positions on the benzo[d]thiazole moiety substantially altered binding and functional profiles and warranted more detailed follow-up studies.We also note that 5substituted pyridine rings (16c, 16d, 20b) were full antagonists, consistent with prior published reports. 2,37,386f was chosen for further analysis based on its pharmacological profile: high D 4 R binding affinity with excellent selectivity over D 2 R and D 3 R (1326-and 520-fold, respectively), as measured by [ 3 H]N-methylspiperone com-petition (Table 1), and excellent D 4 R selectivity in both cAMP and β-arrestin recruitment antagonism (Tables 2 and 3).It was also one of our first compounds to complete in vitro characterization.16f is a low-efficacy D 4 R partial agonist, as measured in cAMP inhibition assays (Figure 2A and Table 2), and a full antagonist in β-arrestin recruitment assays (Figure 2B and Table 3), maintaining 97-fold D 4 R selectivity over D 2 R in cAMP antagonist assays, 391-fold D 4 R selectivity over D 2 R in β-arrestin recruitment antagonist assays, and 859-fold D 4 R selectivity over D 3 R in β-arrestin recruitment antagonist assays, indicating it is highly subtype-selective.We conducted Schildtype analysis of 16f using a β-arrestin recruitment assay to determine whether 16f was a competitive orthosteric antagonist without any allosteric activity.Dopamine concentration−response curves were conducted in the presence of DMSO and with increasing concentrations of 16f (Figure 2C).The dopamine curves shifted to the right without decreasing dopamine efficacy, indicating that 16f is a competitive antagonist.Schild-type analysis revealed the slope-approached unity (slope = 1.09), and its affinity was 11.0 nM (Figure 2C inset).Together, these results indicated that 16f is a potent and selective D 4 R antagonist suitable for further analyses.
Compound 16f Selectivity at an Array of CNS GPCRs and Monoamine Transporters.Our results indicated that 16f is highly D 4 R-selective compared to the other D 2 -like receptors (D 2 R and D 3 R).In order to determine the selectivity of 16f at other biogenic amine receptors, 16f was tested at the Psychoactive Drug Screening Program (PDSP), which tests compounds at an array of GPCRs and monoamine transporters. 39In an initial high-concentration (10 μM) screen, 16f displayed greater than 50% radioligand inhibition at 23 GPCRs/transporters.These were further tested in full concentration−response analyses to determine the affinity of 16f for each receptor/transporter (Table 4).Only six GPCRs showed an affinity higher than 100 nM: σ 2 , 5-HT 1A , 5-HT 2A , 5-HT 2B , α 2C , and D 4 R (Table 4).The PDSP-determined affinity of 16f for D 4 R was 9.48 nM, consistent with the results we obtained (K i = 2.21 nM).16f had comparable affinity for 5-HT 1A (5.8 nM) and 5-HT 2B (13 nM) and lower affinity for 5-HT 2A (46 nM), α 2C (73 nM), and σ 2 (73 nM; Table 4).Given the important roles that serotonin receptors can play in SUDs, we further characterized the signaling effects of 16f at the three highest affinity secondary targets, 5-HT 1A , 5-HT 2A , and 5-HT 2B .
To determine its functional activity at 5-HT 1A , 5-HT 2A , and 5-HT 2B , 16f was tested in Gα i or Gα q calcium flux assays by Eurofins Discovery (Eurofins Cerep SA, Celle l'Evescault, France; Eurofins DiscoverX Corporation, Fremont, California).The results of these assays are listed in Table 5.When tested at 5-HT 1A in the Gα i calcium flux assay, 16f showed agonist activity with an estimated potency of 4.6 μM.The curve did not saturate at the highest tested concentration (10 μM), so the estimated E max (58.8% of the serotonin response) may be an underestimate due to the low apparent potency of 16f at 5-HT 1A for Gα i -coupled responses.Because agonist responses interfere with analysis of the antagonist mode of this assay, the IC 50 of 16f as >370 nM reflects the highest drug concentration not excluded from analysis.In Gα q calcium flux assays, 16f had no agonist activity at 5-HT 2A or 5-HT 2B (EC 50 > 10 μM) and was a full antagonist at both receptors (5-HT 2A I max = 106%, IC 50 = 532 nM; 5-HT 2B I max = 111%, IC 50 = 770 nM).In comparison, 16f had higher potency at D 4 R, with an IC 50 of 69.3 nM in cAMP assays that is also responsive to Gα i/o -mediated signaling (Table 2).While one should be cautious about overinterpreting relative potencies collected across different assay conditions, these results suggest that 16f is modestly D 4 R-selective over 5-HT 2A and 5-HT 2B with potential low potency agonism at 5-HT 1A .
In Silico and In Vitro Pharmacokinetics Studies of 16f.The potential for brain penetrance of 16f was evaluated in silico by using central nervous system multiparameter optimization (CNS MPO) tools.16f, and the brain-penetrant CNS ligand buspirone as a comparator, had calculated CNS MPO scores of 4.5 and 5.8, respectively, shown in Supplemental Table S3; scores >4 correlate with CNS drug-like properties. 4016f was also tested in Caco-2 membrane permeability assays (Eurofins Panlabs, St. Charles, Missouri) and the apical-to-basolateral (A-B) permeability of 16f was 27 × 10 −6 cm/s, comparable to the assay control compounds propranolol (24 × 10 −6 cm/s)
We then evaluated the phase I metabolic stability of 16f using rat and human liver microsomes, as previously described. 41Incubation of 16f with rat liver microsomes in the presence of NADPH resulted in time-dependent degradation, with ∼33% remaining after 1 h (Figure 3A).In human liver microsomes, 16f showed greater stability, with ∼60% remaining after 1 h incubation (Figure 3B).These results predict that 16f has modest liver metabolic stability in humans and relatively lower stability in the rat liver.HPLC traces of 16f and the major metabolite of 16f�a dealkylated 1-(pyridin-2-yl)piperazine product�are shown in Figure S2.
Pharmacokinetic Assessment of 16f in Rats.Given its adequate stability profile, we next evaluated the in vivo pharmacokinetic profile of 16f in rats.Sprague−Dawley rats were dosed with 16f (10 mg/kg, i.p.), and plasma and brain levels were measured 0−6 h postdose.The results from the pharmacokinetic analysis are listed in Figure 4A,B.16f demonstrated good exposure in both plasma and brain, with AUC 0−t values of 1.05 and 3.67 nmol•h/g, respectively.Compound 16f was observed to have a brain penetration index (AUC brain/plasma ratio) of 3.5 with an apparent half-life of ∼1 h (t 1/2 ).The detailed pharmacokinetic parameters of 16f are provided in Figure 4B.
Behavioral Effects of 16f in Rats Trained to Self-Administer Food and Cocaine.In order to test our hypothesis that D 4 R antagonism is a viable route for CUD

Journal of Medicinal Chemistry
pharmacotherapy, we evaluated whether 16f altered cocaine self-administration, using food self-administration as a natural reward comparator.Separate groups of male Fischer 344 rats were trained to respond on a lever to receive food pellets or iv cocaine in multicomponent procedures.Both procedures included three components (60 min each for cocaine, 30 min each for food) in each test wherein the reinforcer was reduced across components (food: four, two, and one food pellets across components 1, 2, and 3, respectively; iv cocaine: 166, 83, and 41.5 mg/infusion across components 1, 2, and 3, respectively).After successful training, the saline vehicle and 16f (5, 15, and 30 mg/kg, i.p.) were tested.16f pretreatment produced a significant decrease in the number of infusions for each cocaine dose, an effect that was dependent upon the dose of the compound (5, 15, and 30 mg/ kg, i.p.) (Figure 5A).Intake following saline pretreatment was not significantly different from baseline [F 2, 16 = 0.2935, P = 0.75].A significant main effect of compound 16f on cocaine self-administration was observed [F 1.239, 34.69 = 57.79,P < 0.0001] and a significant interaction of component and 16f on cocaine intake [F 6, 56 = 3.181, P = 0.0093].The number of infusions obtained for each cocaine dose was significantly different after 16f treatment, and the magnitude of effect was dependent on the dose of 16f as well as the dose of cocaine self-administered.
Overall, these results indicate that 16f is centrally active and reduces cocaine-and food-maintained responding.The effects of 16f are most pronounced at lower unit doses of cocaine but at higher unit doses of food, suggesting some differentiation of these effects that will be more fully evaluated in follow-up studies.Future testing will also determine whether 16f affects relapse-like responses and other behaviors relevant to CUD.

■ CONCLUSIONS
Evidence from human genetic studies and animal models suggests D 4 R signaling modulates drug-taking and -seeking behaviors.Newer highly selective D 4 R antagonists will be useful to better characterize the role of D 4 R signaling in vivo, particularly in behavioral models of CUD.This study provided a detailed SAR analysis of a novel series of D 4 R partial agonists and antagonists.We identified several compounds with high D 4 R binding affinity and selectivity over other D 2 -like receptors (D 2 R, D 3 R) with diverse partial agonist and antagonist profiles.The low-efficacy D 4 R partial agonist 16f was chosen as a lead compound suitable for pharmacokinetic and behavioral testing on the basis of its high selectivity over D 2 R and D 3 R. 16f displayed acceptable in vitro metabolic stability in rat and human liver microsomes and good in vivo half-life and brain penetration parameters.
In behavioral testing, 16f dose-dependently decreased cocaine-and food-maintained operant responses, with diverging effects on the reinforcer unit dose.These results suggest that D 4 R antagonism reduces the rewarding effects of cocaine and is a plausible route for CUD pharmacotherapy development.We cannot rule out the importance of off-target effects in the behavioral response to 16f�the compound is only modestly D 4 R-selective over its antagonistic effects at 5-HT 2A and 5-HT 2B receptors.−46 Our results also suggest the possibility of 16f producing low-potency agonism at 5-HT 1A , which has been previously reported to increase the reinforcing strength of a low cocaine dose. 47This is at odds with our behavioral results, which seem to indicate a stronger effect of 16f at lower unit doses of cocaine.
The extended analogue library created while analyzing 16f in vitro and in vivo identified several additional modifications that improved D 4 R affinity and selectivity over D 2 R and D 3 R; future analyses will determine whether these modifications alter activity at 5-HT 2A , 5-HT 2B , and 5-HT 1A .We are optimistic that these analogues will be useful as novel in vivo research tools, and we plan to examine additional ADME characteristics of selected library members.It is interesting to speculate that a collection of ligands with varying efficacies may allow for the fine-tuning of D 4 R inhibition, potentially leading to a fuller understanding of functional consequences of D 4 R signaling levels in the development of therapeutics for SUDs and other neuropsychiatric disorders.

■ EXPERIMENTAL METHODS
Reaction conditions and yields were not optimized.Anhydrous solvents were purchased from Aldrich and used without further purification.All other chemicals and reagents were purchased from Sigma-Aldrich Co., LLC, Aurora Fine Chemicals LLC, VWR Chemicals, Enamine, Acros Organics, and Alfa Aesar.All amine final products were converted to either the oxalate or hydrochloride salt.Spectroscopic data and yields refer to the free base forms of compounds.Flash chromatography was performed using silica gel (EMD Chemicals, Inc.; 230−400 mesh, 60 Å) by using a Teledyne Isco CombiFlash RF system. 1 H and 13 C spectra were acquired by using a JEOL ECZ-400S NMR spectrometer. 1 H chemical shifts are reported as parts per million (δ ppm) relative to tetramethylsilane (0.00 ppm).All of the coupling constants are measured in Hz.Chemical shifts for 13 C NMR spectra are reported as parts per million (δ ppm) and referenced according to deuterated solvent for 1 H spectra (CDCl 3 , 7.26, or CD 3 OD, 3.31) and 13 C spectra (CDCl 3 , 77.1, or CD 3 OD, 49.0).Chemical shifts, multiplicities, and coupling constants (J) were reported and calculated using Mnova 64.Combustion analysis was performed by Atlantic Microlab, Inc. (Norcross, Georgia), and the results agree within ±0.4% of calculated values (Table S1).cLogP and PSA values were calculated using ChemDraw version 20.0.Melting point determination was conducted using an SRS OptiMelt MPA100-Automated melting point apparatus and are uncorrected.Based on NMR and combustion analysis data, all final compounds are ≥95% pure.All compounds within this series are covered under an existing patent, 32 but only 16a and 16e 31 have been previously described in the peer-reviewed literature.
General Method A. 31 4-Chlorobutanoyl chloride or 5-chloropentanoyl chloride (1.24 equiv) was added dropwise to a solution of substituted or unsubstituted 2-aminobenzenethiol (1.00 equiv) in toluene at 0 °C over 15 min, and an off-white precipitate was formed.The reaction mixture was stirred at room temperature for 48 h, under a N 2 atmosphere.After the reaction was complete, the solvent was removed in vacuo.The crude mixture was diluted with aqueous NaHCO 3 (100 mL) and EtOAc (100 mL), and the two layers were separated and then extracted with EtOAc (2 × 100 mL) and washed with brine (100 mL).The combined organic layer was dried over Na 2 SO 4 , filtered, and concentrated.The product was purified by flash column chromatography (5−95% EtOAc:hexanes) gradient to give the desired substituted or unsubstituted 2-(3-chloropropyl)benzo[d]thiazole or 2-(4-chlorobutyl)benzo[d]thiazole compounds.
Radioligand competition binding experiments were conducted using thawed membranes on test day, and each test compound was diluted into 10 half-log serial dilutions using 30% DMSO vehicle, starting from 1 mM or 100 μM concentration.Previously frozen membranes were diluted in fresh EBSS binding buffer to 200 μg/mL (for hD 2L R or hD 3 R) or 400 μg/mL (for hD 4 R) for binding.Radioligand competition experiments were conducted in 96-well plates containing 300 μL of fresh EBSS binding buffer, 50 μL of diluted test compound, 100 μL of membranes (20 μg/well total protein for hD 2L R and hD 3 R, and 50 μL of [ 3 H]N-methylspiperone radioligand diluted in binding buffer (0.4 nM final concentration; PerkinElmer).Nonspecific binding was determined using 10 μM (+)-butaclamol (Sigma-Aldrich, St. Louis, Missouri), and total binding was determined with 30% DMSO vehicle.All compound dilutions were tested in triplicate and the reaction incubated for 1 h at RT.The reaction was terminated by filtration through PerkinElmer Uni-Filter-96 GF/C plates, presoaked for 1 h in 0.5% polyethylenimine, using a Brandel 96-Well Plates Harvester Manifold (Brandel Instruments, Gaithersburg, Maryland).The filters were washed (3 × 1 mL/well) in ice-cold binding buffer.PerkinElmer MicroScint 20 Scintillation Cocktail (65 μL) was added to each well, and filters were counted using a PerkinElmer MicroBeta Microplate Counter.IC 50 values for each compound were determined from dose−response curves, and K i values were calculated using the Cheng−Prusoff equation. 47When a complete inhibition could not be achieved at the highest tested concentrations, K i values were extrapolated by constraining the bottom of the dose−response curves (= 0% residual specific binding) in the nonlinear regression analysis.These analyses were performed using GraphPad Prism versions 6.00−8.00(Graph-Pad Software, San Diego, California).All results were rounded to three significant figures.K i values were determined from at least three independent experiments and are reported as means ± SEM.
Functional Assays.cAMP Inhibition Assay.D 4 R-and D 2 R -mediated inhibition of forskolin-stimulated cAMP production was assayed using the PerkinElmer LANCE UltracAMP assay kit (PerkinElmer, Inc., Waltham, Massachusetts).CHO-K1 cells stably expressing the human D 2 R long isoform or D 4 R were maintained in Ham's F12 supplemented with 10% fetal bovine serum, 100 U/mL penicillin, 100 μg/mL streptomycin, 800 μg/mL G418, and 300 μg/ mL hygromycin at 37 °C, 5% CO 2 , and 90% humidity.Cells were seeded in 5 μL of Hank's Balanced Salt Solution (with CaCl and MgCl 2 ) with 5 mM HEPES buffer and 0.2 μM sodium metabisulfite at a density of 5000 cells/well in 384-well white plates.Compounds and forskolin were made in the same buffer.Immediately after plating, cells were treated with 2.5 μL of compound (at various concentrations) and 2.5 μL of forskolin and incubated at room temperature for 30 min.The final concentration of forskolin was 10 μM.When running the assay in antagonist mode, the EC 80 of dopamine (10 nM) was added to the forskolin solution.Eu-cAMP tracer and ULight-anti-cAMP solutions were added as directed by the manufacturer, and cells were incubated for 2 h in the dark at room temperature, after which a TR-FRET signal was measured using a BMG Labtech PHERAstar FS (BMG Labtech, North Carolina).Values were normalized to a percentage of the control TR-FRET signal seen with a maximum concentration of dopamine for agonist mode assays and the EC 80 of dopamine for antagonist mode assays.Data were collected in triplicate from at least three independent experiments.Data analysis and normalization were performed in GraphPad Prism 9 (GraphPad Software, California).First, raw data was fit using a log(agonist/antagonist) vs response − variable slope (four parameters) curve fit.The data were normalized to the percent maximum dopamine response (agonist mode) or the EC 80 of dopamine (antagonist mode).The Hill coefficients of the concentration−response curves did not significantly differ from unity with the data fitting to a single-site model.Graphs are meaned concentration response curves from at least three independent experiments.Data in Table 2 was extracted from the meaned curves where E max /I max are expressed as mean ± SEM and the potencies are expressed as mean [95% confidence interval].Fold selectivities for D 4 R over the were also calculated and are presented in Table 2.
β-Arrestin Recruitment Assay.Assays were conducted with minor modifications, as previously published by our laboratory 2,19−23 using the DiscoverX PathHunter technology (Eurofins DiscoverX, Fremont, California).Briefly, CHO-K1 cells stably expressing the human D 2 R long isoform, D 3 R, or D 4 R (Eurofins DiscoverX) were maintained in Ham's F12 media supplemented with 10% fetal bovine serum, 100 U/ mL penicillin, 100 μg/mL streptomycin, 800 μg/mL G418, and 300 μg/mL hygromycin at 37 °C, 5% CO 2 , and 90% humidity.The cells were seeded in 7.5 μL media at a density of 2625 cells/well in 384well black, clear-bottom plates.The following day, the compounds were diluted in PBS with 0.2 μM sodium metabisulfite.The cells were treated with 16 concentrations of a compound in triplicate and incubated at 37 °C for 90 min.Tropix Gal-Screen Substrate (Applied Biosystems, Massachusetts) was diluted in Gal-Screen buffer A (Applied Biosystems) 1:25 and added to cells according to the manufacturer's recommendations followed by a 30−45 min incubation at room temperature in the dark.Luminescence was measured on a Hamamatsu FDSS μCELL reader.Data was collected in triplicate and transferred to GraphPad Prism 9 where it was fit with a log(agonist/antagonist) vs response − variable slope (four parameters) curve fit.The data were normalized to the percent maximum dopamine response (agonist mode) or the EC 80 of dopamine (antagonist mode).The Hill coefficients of the concentration−response curves did not significantly differ from unity, with the data fitting to a single-site model.Graphs are mean concentration response curves from at least three independent experiments.Data in Table 3 was extracted from the meaned curves where E max /I max are expressed as mean ± SEM and the potencies are expressed as mean [95% confidence interval].Fold selectivities for D 4 R over D 2 R and D 3 R were also calculated and are presented in Table 3.
Schild-Type Analysis − β-Arrestin Recruitment Assay.Schild-type analysis using the β-arrestin recruitment assay was conducted similarly except for compound preparation.Compounds were diluted in PBS with 0.2 μM sodium metabisulfite at eight concentrations ranging from 10 μM to 10 nM (final in assay concentrations) and a DMSO control.The compounds were added to the cells followed immediately by a dopamine concentration response curve and returned to the incubator at 37 °C for 90 min.The Tropic Gal-Screen substrate and buffer were prepared and added as previously described.All other aspects of the Schild-type analysis were identical to those of the β-arrestin recruitment assay procedure.Data was collected in triplicate and transferred to GraphPad Prism 9 (GraphPad Software, California) where it was fit with a log(agonist) vs response-variable slope (four parameters) curve fit.The data were normalized to the maximum dopamine/DMSO response.Graphs are meaned concentration response curves from at least three independent experiments.Schild-type plots were generated by plotting the log-scale compound concentration (x-axis) versus the log((A′/A) − 1) where A′ is the EC 50 of the dopamine curve obtained for each concentration of antagonist and A is the EC 50 of dopamine in the DMSO control.Simple linear regression was performed in GraphPad Prism 9 where the slope and x-intercept indicate competitiveness and the affinity of the compound, respectively.
Rat and Human Microsomal Stability Assays.Phase I metabolic stability assays were conducted using rat and human liver microsomes as previously described 41,49 with minor modifications.In brief, the reactions were carried out with 100 mM potassium phosphate buffer, pH 7.4, in the presence of a NADPH regenerating system (1.3 mM NADPH, 3.3 mM glucose 6-phosphate, 3.3 mM MgCl 2 , 0.4 U/mL glucose-6-phosphate dehydrogenase, 50 μM sodium citrate).Negative controls without cofactors were assessed to determine the non-CYP-mediated metabolism.Positive controls for phase I metabolism (buprenorphine) were also evaluated.Compound disappearance was monitored over time using a liquid chromatography and tandem mass spectrometry (LC/MS) method.All reactions were performed in triplicate.
Chromatographic analysis was performed on a Dionex ultra highperformance LC system coupled with a Q Exactive Focus orbitrap mass spectrometer (Thermo Fisher Scientific Inc., Waltham, Massachusetts).Separation was achieved using an Agilent Eclipse Plus column (100 × 2.1 mm i.d; maintained at 35 °C) packed with a 1.8 μm C18 stationary phase.The mobile phase used was composed of 0.1% formic acid in acetonitrile and 0.1% formic acid in water with gradient elution, starting with 2.5% organic phase (from 0 to 2 min) linearly increasing to 99% (from 2 to 5.5 min) and re-equilibrating to 2.5% by 6.5 min.The total run time for each analyte was 6.5 min.Pumps were operated at a flow rate of 0.3 mL/min.The mass spectrometer controlled by Xcalibur software v.4.0.27.13 (Thermo Scientific) was operated with a HESI ion source in positive ionization mode.Compounds were identified in the full-scan mode (from m/z 50 to 750) by comparing t = 0 samples with t = 30 min and t = 60 min samples.
Pharmacokinetics Study in Rats.Pharmacokinetic studies in Sprague−Dawley (SD) rats were conducted according to protocols approved by the Animal Care and Use Committee at Johns Hopkins University.SD rats obtained from Harlan were maintained on a 12 h light−dark cycle with ad libitum access to food and water.A test compound was administered via i.p. injection at a dose of 10 mg/kg (100% saline vehicle, 10 mL/kg volume).The rats were sacrificed at specified time points (0.25, 0.5, 1, 2, 4, and 6 h) post drug administration.For the collection of plasma and brain tissue, animals were euthanized with CO 2 and blood samples were collected in heparinized microtubes by cardiac puncture.Brains were dissected and immediately flash-frozen (−80 °C).Blood samples were spun at 2000g for 15 min, and the plasma was removed and stored at −80 °C until analysis (as described below).
Bioanalysis.Quantitation of 16f was performed using liquid chromatography with tandem mass spectrometry (LC/MS-MS) methods.Briefly, calibration standards were prepared using respective tissue (nai ̈ve plasma and brain) with additions of the test compound.For quantifying the test compound in the pharmacokinetic samples, plasma samples (20 μL) were processed using a single liquid extraction method by addition of 100 μL of acetonitrile containing internal standard (losartan: 0.5 μM), followed by vortex-mixing for 30 s and then centrifugation at 10,000 × g for 10 min at 4 °C.Brain tissues were diluted 1:5 w/v with acetonitrile containing losartan (0.5 μm) and homogenized, followed by vortex-mixing and centrifugation at 10,000 × g for 10 min at 4 °C.A 50 μL aliquot of the supernatant was diluted with 50 μL of water and transferred to 250 μL polypropylene autosampler vials sealed with Teflon caps. 2 μL of the sample was injected into the LC/MS/MS system for analysis.Chromatographic analysis was performed using an Accela ultra highperformance system consisting of an analytical pump and an autosampler coupled with a TSQ Vantage mass spectrometer.Separation of the analyte was achieved at ambient temperature using an Agilent Eclipse Plus column (100 × 2.1 mm inner diameter) packed with a 1.8 μm C18 stationary phase.The mobile phase consisted of 0.1% formic acid in acetonitrile and 0.1% formic acid in water with gradient elution, starting with 10% organic phase (from 0 to 1 min) linearly increasing to 95% (from 1 to 2 min) and reequilibrating to 10% by 3 min.The total run time for each analyte was 3.5 min.Pumps were operated at a flow rate of 0. Food-Maintained Responding.For experiments, rats were transferred to operant conditioning chambers (ENV-008CT; Med Associates, St. Albans, Vermont) enclosed in sound-attenuating cubicles (ENV-018; Med Associates).The front panel of the operant chambers contained two response levers (4 cm above the floor and 3 cm from each side wall), a cue light (3 cm above each of the two levers), and a food chute centered on the front wall (2 cm above the floor) that was connected to a food pellet dispenser (ENV-023; Med Associates) located behind the front wall and a tone generator to mask extraneous noise.Food-maintained responding was assessed using a multicomponent procedure consisting of three 30 min components separated by 4 min blackout periods between components.Responding was engendered and maintained by delivery of food pellets (45 mg; Noyes, Lancaster, New Hampshire; four, two, and one pellets for components 1, 2 and 3, respectively) under an FR3 schedule of reinforcement.Completion of the response requirement on the active lever extinguished lights, retracted both levers, and delivered food and was followed by a 20 s time-out (TO) period.After the TO, the lights were illuminated, the levers were extended, and the FR schedule was in effect.The presentation of 16f doses (5, 15, 20, and 30 mg/kg, i.p.) and saline were randomly assigned and administered 15 min before the start of the session.The criterion for stable response was two consecutive sessions in which the total number of reinforcers did not vary by more than 10% from baseline levels.
Cocaine Self-Administration.The operant apparatus was described above.For self-administration studies, a counterbalance arm was connected at the rear corner of the operant chamber, onto which a single channel swivel was mounted.The rat's leash was attached to the swivel, and the catheter tubing was connected to the bottom port of the swivel.A motor-driven 20 mL syringe pump (PHM-100; Med Associates) was attached outside of the soundattenuating chamber, and polyethylene tubing connected the needle on the syringe to the entry port of the swivel.A PC was used for session programming and data collection (Med Associates, Inc., East Fairfield, Vermont).For lever training, subjects were transferred to the operant chambers for daily experimental sessions and responding was engendered and maintained by delivery of food pellets (45 mg pellets; Noyes, Lancaster, NH) under an FR1 schedule of reinforcement that was gradually increased to FR3.The lever light was illuminated when the schedule was in effect.Completion of the response requirement on the active lever extinguished lights, retracted both levers, delivered food, and was followed by a 20 s time-out (TO) period during which all lights were off.After the TO, the lights were illuminated and the FR schedule was in effect.Sessions lasted 30 min or until 50 food pellets were delivered.The criterion for stable response was five consecutive sessions in which the total number of reinforcers did not vary by more than 20% from control levels.Responses on the inactive lever were recorded but had no scheduled consequences.
Intravenous Jugular Surgery.−52 Catheters were anchored to the muscle near the point of entry into the vein.The distal end of the catheter was guided subcutaneously to exit above the scapulae through a Teflon shoulder harness.The harness provided a point of attachment for a spring leash connected to a single-channel fluid swivel at the opposing end.The catheter was threaded through the leash and attached to the swivel.The other end of the swivel was connected to a syringe (for saline and drug delivery) mounted on a syringe pump.Rats were administered penicillin G procaine (75,000 units in 0.25 mL, i.m.) and allowed a minimum of 5 days to recover before self-administration studies were initiated.Following surgery, rats received hourly infusions of heparinized 0.9% bacteriostatic saline (1.7 U/mL; 200 μL/h) using a computer-controlled motor-driven syringe pump in the home cage vivarium.The health of the rats was monitored daily by the experimenters and weekly by an institutional veterinarian per the guidelines issued by the High Point University Institutional Animal Care and Use Committee and the National Institutes of Health.Infusions of propofol (6 mg/kg; (iv) were administered to assess the catheter patency, as needed.
Responding was maintained under an FR3:20 s TO of three 1 h components.Subjects were allowed to self-administer cocaine iv (166, 83, and 41.5 mg/infusion).Each dose was available during a different component, and the doses were presented in descending order.The infusion volume for the first component was 400 μL infused over 12 s, and the volumes for the successive components were 200 μL for component 2 (infused over 6 s) and 100 μL for component 3 (infused over 3 s).Before each component, a 10 min blackout was followed by a priming infusion of the dose to be administered in the succeeding component.After an additional 10 min blackout period, the lever was activated and the cue light above the lever was illuminated.The start of each session was indicated by the illumination of the house light, stimulus light above the active lever, and the extension of both levers.Upon completion of the response requirement, a drug infusion was delivered, the lever light was extinguished, a tone was generated, and the house light was illuminated.During the 20 s TO after the infusion, responses on the lever were recorded but had no scheduled consequence.A minimum of 3 days of stable responding (less than 10% variation in the number of infusions) at FR3 in all components was required before administration of compounds was initiated.
Effects of 16f on Cocaine Self-Administration. Rats were transferred to the operant chambers for self-administration sessions.Before each session, the swivel and catheter were flushed with 500 μL of heparinized saline before connecting the catheter to the syringe via a 20 ga Leur hub and 28 ga male connector.Completion of the response requirement on the active lever extinguished lights, retracted both levers, and delivered food,and was followed by 20 s TO.−54 After a minimum of 5 days of stable responding (defined as consecutive sessions in which the total number of infusions did not vary by more than 20% from the mean of previous sessions), saline vehicle and 16f (5, 15, 20, and 30 mg/kg, i.p.) were tested.Dose order was randomly assigned for each subject.16f and saline were administered 15 min before the first component.

* sı Supporting Information
The Supporting Information is available free of charge on the ACS Web site.The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jmedchem.3c00734.
Elemental analysis for all final compounds results; library of 16f analogues for virtual docking; computational Analysis; molecular docking analysis, CNS-MPO value and method of calculation for 16f; and HPLC and MS traces of 16f (PDF) SMILES data (CSV)

Figure 1 .
Figure1.Selected structures of notable D 4 R ligands, including the atypical antipsychotics clozapine (5) and nemonapride(6).Following the discovery of the relatively high affinity of 5 for D 4 R,30 several pharmaceutical companies developed novel ligands targeting D 4 R, including 1−4.A recent resurgence in D 4 R ligand discovery has identified a diverse array of novel agonists(7), partial agonists(8), and antagonists (9−13) including lead compound 16a from this study.

Figure 2 .
Figure 2. Lead compound 16f (CAB-01−019) demonstrated excellent D 4 R selectivity in functional assays and is a competitive antagonist at D 4 R. (A) 16f is a potent full D 4 R antagonist for β-arrestin recruitment with no D 4 R agonist activity detected.At D 2 R and D 3 R, 16f is a low-potency antagonist that shows 391-and 859-fold selectivity for the D 4 R, respectively (Table3).D 3 R exhibits partial agonist activity with 16f, while D 2 R has very low partial agonist activity.(B) 16f potently antagonizes D 4 R-mediated cAMP inhibition and is 97-fold more potent at D 4 R than at D 2 R (Table2).Furthermore, 16f has low-efficacy D 4 R agonism and is a low-potency partial agonist at D 2 R. (C) With increasing concentrations of 16f, dopamine concentration−response curves are shifted to the right with no decrease in E max , indicating that 16f is a competitive orthosteric ligand.Furthermore, the Schild plot (inset) of these data had a slope of 1.09 and K b = 11.0 nM.All data are presented as means ± SEM from at least three independent experiments run in triplicate.

Figure 3 .
Figure 3. Phase I metabolic stability of 16f in rat (A) and human (B) liver microsomes.16f shows time-dependent degradation in human and rat liver microsomes.16f was modestly stable in human liver microsomes.Data are expressed as mean ± SEM, n = 3.As a positive control for phase I metabolism, metabolic stability of buprenorphine in rat (C) and human (D) liver microsomes is also presented.

Table 1 .
Human Dopamine D 2 -Like Receptor Competition Binding in HEK293 Cells for Benzothiazole Analogues with Varying Three-or Four-Carbon Linker Chains

Table 2
a Compounds were tested alone (agonist mode) and with an EC 80 concentration of dopamine (antagonist mode) for their ability to alter cAMP production mediated by D 2 R and D 4 R signaling.Dopamine was used as a control in all agonist mode assays.Spiperone was included in all antagonist mode assays for the D 2 R and D 4 R. ND, not determined due to an incomplete curve.Inactive, no measurable activity.b Efficacy/antagonist % (ant.%) values obtained from nonlinear regression of meaned data obtained from at least three independent experiments with triplicate measures.Values are presented as means ± SEM. c Potency values obtained from nonlinear regression of meaned data obtained from at least three independent experiments with triplicate measures.Values are presented as mean [95% confidence interval].

Table 4 .
Psychoactive Drug Screening Program (PDSP) Results from Primary and Secondary Assays on an Array of Receptors and Monoamine Transporters a Receptors and transporters were initially tested with 10 μM 16f and % inhibition measured compared to a known reference compound.Receptors and transporters with greater than 50% inhibition were selected for full assays to determine the affinity of 16f for the receptor/transporter. Receptors with <100 nM affinity for 16f in bold.
a b NT − Not Tested due to <50% inhibition in primary assessment.

Table 5 .
Functional Assessment of 16f and Control Compounds at 5-HT 1A , 5-HT 2A , and 5-HT 2B Receptors in Gα i and Gα q Calcium Flux Assays a 370 nM was the highest concentration of 16f that was not excluded by Eurofins in the Gα i assay as higher concentrations had agonist affects that interfere with interpretation of the antagonist assay.