Discovery of the first M₅-selective and CNS penetrant negative allosteric modulator (NAM) of a muscarinic acetylcholine receptor: (S)-9b-(4-chlorophenyl)-1-(3,4-difluorobenzoyl)-2,3-dihydro-1H-imidazo[2,1-a]isoindol-5(9bH)-one (ML375)

Abstract

A functional high throughput screen and subsequent multi-dimensional, iterative parallel synthesis effort identified the first muscarinic acetylcholine receptor (mAChR) negative allosteric modulator (NAM) selective for the M₅ subtype. ML375 is a highly selective M₅ NAM with sub-micromolar potency (human M₅ IC₅₀ = 300 nM, rat M₅ IC₅₀ = 790 nM, M_1–4 IC₅₀ >30 μM), excellent multi-species PK, high CNS penetration, and enantiospecific inhibition.

INTRODUCTION

The five G protein-coupled muscarinic acetylcholine receptors (mAChRs or M₁-M₅) utilize acetylcholine as their endogenous agonist and are broadly distributed throughout the periphery and central nervous system (CNS) where they regulate a diverse array of physiological processes. ^1–4 M₁ and M₄ are predominantly expressed within the CNS and have the highest expression levels, whereas M₂ and M₃ are expressed in both the periphery and moderately within the CNS, while M₅ expression is low (<2% of the total CNS mAChR population).^2,5 Localization studies have found low levels of M₅ expression in multiple brain regions, but M₅ mRNA is the only mAChR transcript identified in dopaminergic neurons of the substantia nigra pars compacta (SNc) and ventral tegmental area (VTA).^4–7 Here, M₅ is co-expressed with D₂ dopamine mRNA, which has led to the hypothesis that M₅ might modulate dopaminergic neurotransmission and function in addiction/reward mechanisms.⁶ Subsequent studies in M₅ ^−/− mice confirmed this hypothesis, with M₅ ^−/− mice showing reduced morphine and cocaine-conditioned place preference and self-administration, with no effect on food intake, suggesting preferential abuse-related effects. ^8–10 Thus, much of our current understanding of the function of M₅ has come from M₅ receptor localization, M₅ ^−/− mice and experiments conducted with non-selective, orthosteric muscarinic ligands, as no M₅-selective antagonists or negative allosteric modulators (NAMs) have been reported.⁴ Recently, targeting allosteric sites on mAChRs has led to the discovery of highly selective positive allosteric modulators (PAMs) of M₁, M₄ and M₅;^11–16 however, highly selective NAMs for individual subtypes have not yet been identified for any of the five mAChRs - and only one M₁-selective orthosteric antagonist chemotype has been reported.¹⁷ Therefore, to address this limitation in small molecule tools to study M₅ function, we elected to pursue the discovery and development of selective M₅ NAMs to enable the dissection of the physiological role and therapeutic potential of M₅ inhibition.

RESULTS AND DISCUSSION

High-Throughput Screen

We performed a triple-add, functional high-throughput screen to identify M₅ modulator leads.^16,18 For this effort, we screened the MLPCN₁₉ collection (360,000 compounds) in Chinese hamster ovary (CHO) cells stably expressing human M₅ (hM5) and measuring intracellular calcium mobilization. This effort identified 3,920 M₅ primary hits (1.07% hit rate). Counter-screening against the parental CHO cell line as well as CHO cells expressing human M₁ and human M₄ and reconfirmation of powders in 10-point concentration-response curves (CRCs) resulted in 9 confirmed, selective antagonists of hM₅.²⁰ At this point, it was not clear if these confirmed hits were selective M₅ orthosteric antagonists or negative allosteric modulators (NAMs).

Chemistry

Of the confirmed hits (Figure 1), our attention focused on 1, a unique 2,3-dihydro-1H-imidazo[2,1-a] isoindol-5(9bH)-one-based scaffold, which was inactive on M₁ and M₄, but displayed weak inhibition of M₅ (IC₅₀ >10 μM, 41% ACh Max). Upon a simple two-step resynthesis involving condensation of ethylene diamine and 2-benzoylbenzoic acid 2 to provide 3 and subsequent acylation (Scheme 1), we were pleased to observe enhanced activity of fresh powder of 1 at M₅ (hM₅ IC₅₀ = 3.5 μM, rM₅ IC₅₀ = 5.7 μM) and no activity at the other four mAChRs (hM₁-M₄ IC₅₀ > 30 μM).²⁰ Figure 1 also highlights the chemical optimization strategy for 1, evaluating multiple dimensions simultaneously through iterative parallel synthesis, necessitated by the often shallow nature of allosteric SAR.¹¹

Figure 1.

Structure, pharmacology (hM₅ IC₅₀ = 3.49 μM, rM₅ IC₅₀ = 5.67 μM, M₁-M₄ IC₅₀>30 μM) and chemical optimization plan for 1.

Scheme 1.

Synthesis of 1 and route for analog synthesis^a

^aReagents and conditions: (a) ethylene diamine, p-TSA, toluene, reflux, Dean-Stark trap, 80%; (b) 4-fluorobenzoyl chloride, DCM, DIPEA, 93%.

The first round of library synthesis evaluated alternate amides; thus, 3 was subsequently acylated with an array of 40 acid chlorides (aryl, heteroaryl and aliphatic) according to scheme 1 to provide analogs 4. The library was efficiently triaged in a 10 μM single point assay against an EC₈₀ of ACh on human M₅.¹⁵ SAR was shallow, with only seven analogs significantly reducing the EC₈₀, and only benzamide congeners were active. As shown in Table 1, a 3,4-difluro analog, 4g, proved the most active (hM₅ IC₅₀ = 1.0 μM, rM₅ IC₅₀ = 2.1 μM), and it also maintained excellent mAChR selectivity (M₁-M₄ IC₅₀s >30 μM).

Table 1.

Structures and activities of analogs 4.

Entry	R	hM5 pIC50*	hM5 IC50 (μM)	ACh Min* (%)
4a	4-OCF3	5.35±0.03	4.47	0.2±2.4
4b	4-SCF3	5.86±0.02	1.38	0.5±1.5
4c	4-CF3	5.71±0.03	1.95	−2.4±1.4
4d	3-CF3	5.52±0.06	3.02	−0.7±5.2
4e	3,5-diCl	5.24±0.17	5.75	0.0±18.9
4f	3,5-diF	5.42±0.06	3.80	−2.2±5.6
4g	3,4-diF	5.98±0.02	1.05	0.0±1.3

^*hM₅ pIC₅₀ and ACh Min data reported as averages±SEM from our calcium mobilization assay; n = 3

Other library efforts led exclusively to inactive analogs. ²⁰ Expansion of the aminal ring to a six-membered congener was inactive, conversion of the isoindolinone core to an azaindolinone proved inactive, as did urea, sulfonamide and tertiary amine derivatives as replacements for the amide linkage. These data suggested that both the 2,3-dihydro-1H-imidazo[2,1-a] isoindol-5(9bH)-one core and the benzamide moiety were critical for M₅ activity.

Our attention then focused on introducing substituents into the 9b phenyl ring of 1, and assessing the impact on M₅ activity. Once again, analogs 5–7 were prepared following Scheme 1, but employing functionalized congeners of 2 in a matrix library (3 × 10). As Shown in Table 2, SAR was shallow and unpredictable, with the data suggesting a cooperative relationship between benzamide and 9b phenyl substituents. However, one analog, 5g, possessing the 3,4-difluorobenzamide and a 9b 4-chlorophenyl moiety, afforded sub-micromolar potency at M₅ (hM₅ IC₅₀ = 0.48 μM, rM₅ IC₅₀ = 1.1 μM) and no activity at the other four mAChRs (hM₁-M₄ IC₅₀ > 30 μM). This was an exciting result, as this was the racemic form of the analog with potential for enantioselective inhibition of M₅.²⁰

Table 2.

Structures and activities of analogs 5–7.

Entry	R1		R2	hM5 pIC50*	hM5 IC50 (μM)	ACh Min* (%)
5a	Cl		4-OCF3	5.63±0.06	2.3	−5.3±4.4
5b			4-SCF3	5.52±0.13	3.0	−9.2±8.7
5c			4-CF3	5.69±0.04	2.0	0.7±2.3
5d			3-CF3	5.31±0.11	4.9	−7.2±8.0
5e			4-Me	5.58±0.04	2.6	0.1±2.8
5f			3,5-diF	6.00±0.04	1.0	2.8±2.0
5g			3,4-diF	6.32±0.02	4.8	0.1±1.0
6a	F		4-OCF3	---	>10	---
6b			4-SCF3	5.89±0.03	1.3	−0.2±1.7
6c			4-CF3	---	>10	---
6d			3-CF3	5.68±0.03	2.1	−2.4±1.8
6e			4-Me	---	>10	---
6f			3,5-diF	---	>10	---
6g			3,4-diF	---	>10	---
7a	Me		4-OCF3	5.46±0.06	3.5	−0.2±4.3
7b			4-SCF3	5.39±0.08	4.1	4.1±5.5
7c			4-CF3	---	>10	---
7d			3-CF3	---	>10	---
7e			4-Me	---	>10	---
7f			3,5-diF	5.63±0.05	2.3	−1.1±3.3
7g			3,4-diF	5.58±0.06	2.6	−6.8±4.3

^*hM₅ pIC₅₀ and ACh mMin data reported as averages±SEM from our calcium mobilization assay; n = 3; ---, not determined

We were able to quickly develop SFC conditions to separate the pure enantiomers of 5g to provide 8 and 9, both in >99% ee.²⁰ Optical rotations were recorded, and the (−)-enantiomer 8 proved to be active (hM5 IC₅₀ = 300 nM), where the (+)-enantiomer 9 was devoid of M₅ activity (hM₅ IC₅₀ >30 μM) as shown in Figure 2. However, the absolute stereochemistry was unknown. Ultimately, single X-ray crystallography indicated that the active (−)-enantiomer 8 possessed the (S)-stereochemistry. Thus, this core showed enantiospecific activity for the inhibition of M₅.

Figure 2.

Structures and activities of the (S)-and (R)-enantiomers, 8 (hM₅ IC₅₀ = 300 nM) and 9 (hM₅ IC₅₀ >30 μM) respectively of 5g (hM₅ IC₅₀ = 480 nM). A) M₅ CRC for racemic 5g, 8 and 9. B) X-ray crystal structure of 8 (CCDC 953105); C) structures of 5g, 8 and 9.

Molecular Pharmacology

The SAR was driven using a human M₅ functional assay, and, since we desired an in vivo tool compound, we also evaluated 8 against rat M₅. There was a slight species difference, with 8 displaying a >2-fold loss in activity at the rat M₅ receptor (rM₅ IC₅₀ = 790 nM). However, 8 was inactive at human M₁-M₄ (Figure 3A) as well as rat M₁-M₄ (Figure 3B), representing the first M₅ selective small molecule inhibitor. To determine the mechanism of action of 8, whether orthosteric or allosteric, we first performed competition binding experiments with the orthosteric mAChR antagonist [³H]-NMS and compared this to atropine. Compound 8 displayed no competition [³H]-NMS for binding to hM₅, suggesting an allosteric mode of receptor inhibition (Figure 3C).^13–15,20 To further investigate an allosteric mechanism, we also performed [³H]-NMS dissociation kinetic experiments (Figure 3D), with hM₅ cell membranes, which revealed that 8 decreased the dissociation rate of [³H-NMS], further confirming an allosteric effect of 8 on the orthosteric site.^13–15,19 Thus, 8 is the first M₅-selective negative allosteric modulator (NAM).

Figure 3.

Molecular pharmacology profile of 8. A) Human mAChR selectivity. 8 is selective for hM₅ (hM5 IC50 = 300 nM, hM₁-M₄ IC_5os >30 μM); B) rat mAChR selectivity. 8 is selective for rM₅ (rM5 IC50 = 790 nM rM₁-M₄ IC_5os >30 μM); C) [³H]-NMS competition binding [N=3] in membranes prepared from hM5 cells. This stongly suggests that 8 does not bind at the M₅ orthosteric site; D) [³H]-NMS dissociative kinetics [N=3}. Atropine (alone) t_1/2 = 49.9 min, atropine (+8) t_1/2 = 68.8 min. 8 exerts an allosteric effect on the orthosteric site.

Metabolism and Disposition

Evaluation of the in vitro and in vivo DMPK profile^20,21 of 8 (Table 3) revealed the compound to possess high metabolic stability with low hepatic microsomal intrinsic clearance (CL_int;; human 2.6 mL/min/kg, cynomolgus monkey (cyno), 20 mL/min/kg, rat, 24 mL/min/kg) and a corresponding low predicted hepatic clearance in multiple species (CL_hep; human, 2.3 mL/min/kg, cyno, 14 mL/min/kg rat, 18 mL/min/kg).

Table 3.

DMPK profile of 8.

Parameter/Species	Rat (male, Sprague-Dawley)	NHP (male, cynomolgus)	Human
Hepatic microsome CLint(mL/min/kg)	24	20	2.6
Predicted CLhep*	18	14	2.3
fu plasma, fu brain	0.029, 0.003	0.001, -	0.013, -
CYP inhibition (P450, IC50)	-	-	3A4, 2D6, 1A2: >15; 2C9: 7.4
CLp (mL/min/kg), Elimination t1/2 (hr)	2.5, 80	3.0,10	-
Vdss(IV)	16 L/kg	1.9 L/kg	-
%F (PO)	80	-	-
Brain-plasma Kp, Kp,uu (1.0hr, PO)	1.8, 0.2	-	-

^*determined using CL_int values in well-stirred model of organ clearance uncorrected for fraction unbound in plasma

Correspondingly, 8 exhibited low clearance (CL_p, 2.5 mL/min/kg) and a long elimination half-life (t_1/2, 80hr) in rodents (male, Sprague-Dawley rat, 1 mg/kg IV, n = 2) and nonhuman primates (male, cynomolgus monkey, 1 mg/kg, CL_p, 3.0 mL/min/kg, t_1/2, 10 hr, n = 3). Consistent with a low clearance compound, 8 also demonstrated high oral bioavailability (%F, 80) following administration of a suspension-dose to male SD rats (n=2) with a maximal plasma concentration (C_max) of 1.4 μM and a corresponding time to reach C_max (T_max) of 7 hours. The distribution of compound 8 was characterized by a low fraction unbound in plasma (f_u,p; human: 0.013, cyno: 0.001, rat: 0.029) and a high nonspecific binding in brain homogenate (f_u,br; rat: 0.003). Following an oral CNS distribution study in rat (male, Sprague-Dawley, n = 2; 10 mg/kg) we observed total and unbound brain-plasma partition coefficients of 1.8 and 0.2 (K_p, K_p,uu, respectively) one hour post-administration.

Compound 8 displayed an acceptable human cytochrome P450 inhibition profile producing acceptable IC₅₀ values for 3A4 (16 μM), 1A2 (25 μM, 2C9 (7.4 μM) and 2D6 (26 μM). Moreover, in a Eurofins radioligand binding panel of 68 GPCRs, ion channels and transporter,²⁰ compound 8 displayed significant binding (>50% inhibition @10 μM) at only 1 target (CB₁, 66%), but no functional activity at this target in a subsequent assay.

Conclusion

In summary, we have developed 8 (also referred to as ML375 or VU0483253), the first mAChR NAM that selectively targets M₅ (hM₅ IC₅₀ = 300 nM, hM₁-M₄ IC50 >30 μM), with a favorable DMPK profile and CNS penetration. Enantiospecific M₅ activity was noted, with all activity residing in the (S)-enantiomer, 8. Due to the unexpected human-rodent species difference in regards to M₅ potency and brain homogenate binding, 8 is not suitable for in vivo work in rodents, but may achieve sufficient exposure in non-human primate. Current efforts are focused on a new optimization program, driving the SAR on rat M₅ to deliver an in vivo tool for rodent addiction studies, and progress will be reported in due course.

EXPERIMENTAL SECTION

Chemistry

The general chemistry, experimental information, and syntheses of all other compounds are supplied in the Supporting Information.

(S)-9b-(4-chlorophenyl)-1-(3,4-difluorobenzoyl)-2,3-dihydro-1H-imidazo[2,1-a]isoindol-5(9bH)-one (8)

To a mixture of 2-(4-chlorobenzoyl)benzoic acid (5.21 g, 20.0 mmol, 1 eq.) and ethylenediamine (2.67 mL, 40.0 mmol, 2 eq.) in toluene (30 mL, 0.67 M) was added p-toluenesulfonic acid monohydrate (~0.1 g, 3 mol%). A Dean-Stark trap was used to remove water while the mixture was allowed to stir at reflux for 4 hours. After cooling to ambient temperature, the reaction mixture was dissolved in dichloromethane. The organic layer was washed with a saturated aqueous solution of sodium bicarbonate and then with brine. Solvent was removed under reduced pressure and the crude product was recrystallized from ethanol to give 2.86 g of pure 9b-(4-chlorophenyl)-2,3-dihydro-1H-imidazo[2,1-a]isoindol-5(9bH)-one (50.2% yield). To a solution of 9b-(4-chlorophenyl)-2,3-dihydro-1H-imidazo[2,1-a]isoindol-5(9bH)-one (15 mg, 0.053 mmol, 1.0 eq.) and DIPEA (18 μL, 0.105 mmol, 2.0 eq.) in DCM (0.53 mL, 0.1 M) was added 3,4-difluorobenzoyl chloride (9.9 μL, 0.079 mmol, 1.5 eq.). The reaction was allowed to stir at ambient temperature for 2 hours. The reaction was quenched with methanol and the organics were concentrated on a heated air-drying block. Crude product was purified via Gilson preparative LC to obtain 12.0 mg of 5g (53.3 % yield). The second eluting pure enantiomer of 5g was separated via CO₂ supercritical fluid chromatography (Lux cellulose-3 10 × 250 mm column at 40 °C, backpressure regulated at 100 bar, MeOH co-solvent, 10% isocratic prep over 7 minutes at 15 mL/min) and was determined to have an ee of >98% by chiral HPLC analysis (Lux cellulose-3 4.6 × 250 mm column at 40 °C, backpressure regulated at 100 bar, MeOH co-solvent, 5–50% over 7 minutes at 3.5 mL/min). ¹H NMR (400.1 MHz, CDCl₃) δ (ppm): 8.04-7.99 (m, 1H); 7.90-7.85 (m, 1H); 7.65-7.56 (m, 2H); 7.38-7.30 (m, 3H); 7.25-7.19 (m, 2H); 7.18-7.14 (m, 2H); 4.38-4.30 (m, 1H); 4.01-3.93 (m, 1H); 3.82-3.75 (m, 1H); 3.34-3.25 (m, 1H). ¹³C NMR (100.6 MHz, CDCl₃) δ (ppm): 172.07, 166.84, 151.81 (dd, J_C=F = 254 Hz, 12.7 Hz), 150.33 (dd, J_C–F = 252 Hz, 13 Hz) 145.77, 136.65, 134.94, 133.55, 132.91 (t, J = 4.8 Hz), 131.88, 130.61, 129.06, 128.97, 127.53, 124.03, 123.62 (dd, J = 6.8 Hz, 4 Hz), 117.94 (d, J = 17 Hz), 116.83 (d, J = 18 Hz), 87.37, 52.24, 39.70. SFC (214 nM) R_T = 3.591 min (>98%). HRMS (TOF, ES+) C₂₃H₁₆N₂O₂F₂Cl [M+H]⁺ calc. mass 425.0868, found 425.0872. Specific rotation $[\alpha ]\frac{23}{D}=-168.6°$ (c = 0.75, CHCl₃).

ABBREVIATIONS USED

M₅

muscarinic acetylcholine receptor subtype 5

CRC

concentration-response-curve

NAM

negative allosteric modulator

MLPCN

Molecular Libraries Probe Production Centers Network