Discovery of a Novel Class of Heteroaryl-Pyrrolidinones as Positive Allosteric Modulators of the Muscarinic Acetylcholine Receptor M₁

Abstract

This Letter describes the synthesis and optimization of a series of heteroaryl-pyrrolidinone positive allosteric modulators (PAMs) of the muscarinic acetylcholine receptor M₁ (mAChR M₁). Through the continued optimization of M₁ PAM tool compound VU0453595, with a focus on replacement of the 6,7-dihydro-5H-pyrrolo[3,4-b]pyridin-5-one with a wide variety of alternative 4,5-dihydropyrrolo-fused heteroaromatics, the generation of M₁ PAMs with structurally novel chemotypes is disclosed. Two compounds from these subseries, 8b (VU6005610) and 20a (VU6005852), show robust selectivity for the M₁ mAChR, and no M₁ agonism. Both compounds have favorable preliminary PK profiles in vitro; 8b additionally demonstrates high brain exposure in a rodent IV PK cassette model.

Graphical Abstract

The muscarinic acetylcholine receptors (mAChRs, subtypes M₁-M₅) are Class A G protein-coupled receptors (GPCRs) with important roles in a variety of biological processes. The mAChRs are expressed in both peripheral and central nervous system (CNS) tissues, and all 5 subtypes have emerged as attractive targets for drug discovery.^1,2 In particular, the M₁ mAChR has seen a steady increase in interest from the scientific community. A growing body of evidence suggests that M₁ plays a crucial role in the modulation of biological pathways relating to cognition, learning and memory.^3-5 Specifically, reduction in M₁ expression in the prefrontal cortex has been noted in populations of schizophrenic patients⁴, and allosteric M₁ modulators such as benzylquinolone carboxylic acid (BQCA) have been shown to restore discrimination reversal learning in rodent models of Alzheimer’s disease.⁵ As such, the need for novel (and selective) tool compounds to further probe M₁ biology has never been more important.

Xanomeline is an extensively characterized muscarinic agonist that showed promise in treating the cognitive deficits related to Alzheimer’s disease (AD) and schizophrenia.⁶ The compound is often referred to in the literature as “M₁/M₄ preferring”, a fitting description given the reports highlighting the roles of these two subtypes in xanomeline’s pro-cognitive efficacy.^6-8 Although xanomeline ultimately proved intolerable to patients due to severe gastrointestinal side effects, the drug’s promise as a CNS therapeutic continues to this day. Recently, promising phase 2 clinical data with a combination of xanomeline and trospium (a peripheral muscarinic antagonist) have been disclosed.⁹

Inspired by the efficacy of xanomeline and other early muscarinic agonists, both industrial^10-13 and academic^14-20 groups have disclosed a variety of novel and selective M₁ positive allosteric modulator (PAM) chemotypes. Although the structure activity relationships (SAR) necessary for receptor activity are relatively well understood for M₁ PAMs, novel chemotypes continue to emerge. In general, M₁ PAM compounds are comprised of a flexible tail pendant and a substituted amide or lactam (often substituted with a trans-1,2-aminohydroxycyclohexane or tetrahydropyran), linked by a heteroaryl core (Figure 1).

Figure 1.

A common chemotype for M₁ PAMs, and structures of selected compounds.

In SAR campaigns around such compounds, it has proven crucial to monitor the agonist and/or “ago-PAM” activity of the given chemotypes; high M₁ agonist activity has been linked to severe side effects including seizures.²¹ Indeed, compounds that were initially reported to have negligible M₁ agonist activity have been shown to act as robust allosteric agonists in native systems,²² highlighting the necessity of detailed characterization of these ligands.

As part of our ongoing medicinal chemistry efforts toward selective M₁ PAMs devoid of agonist activity, we have previously disclosed VU0453595 (3, Figure 1), a highly subtype-selective M₁ PAM with favorable properties both in vitro and in vivo, and further showed that the compound reverses both the negative and cognitive symptoms in a rodent model of schizophrenia.²³ Importantly, the pro-cognitive effects of 3 are not mediated through direct agonism at M₁; the compound functions as a “pure PAM”.²⁴ VU0453595 features a chemically distinct aza-isoindolinone scaffold, a template that has proven fruitful for our group and others in follow up medicinal chemistry efforts on M₁ PAMs (Figure 1).^10,12,25,26 This “reversed lactam” chemotype was originally derived from a class of M_1,3,5-preferring isatin compounds,²⁸ and an exploration of positional lactam isomers has allowed for further elaboration (Figure 2).^25,26 Molecules derived from the VU0453595 chemotype are reported to have minimal ago-PAM activity, a profile that has crippled the development of compounds in the alternate series.^12,21,26

Figure 2.

The VU0453595 aza-isoindolinone chemotype allows for extensive SAR exploration.

The present work describes a novel class of 5-substituted, fused heteroaryl-pyrrolidinones based on the VU0453595 aza-isoindolinone structure (Figure 2). In holding the γ-lactam moiety constant, along with historically M₁-preferring pendant tail pieces, we have developed a series of novel 5,5-bicyclic lactams, featuring a wide variety of 5-membered heteroaryls. Importantly, compounds in this series are devoid of agonist activity and display robust selectivity for the M₁ mAChR.

Compounds were synthesized as shown in Schemes 1-5.²⁹ For bicyclic imidazoles, thiazoles and thiophenes 8, commercially available aldehydes 6 were condensed with substituted benzylamines²⁹ to give intermediates 7, which were then hydrolyzed and cyclized under standard amide forming conditions to give final analogs 8 (Scheme 1). The synthesis of substituted oxazoles and furans is shown in Scheme 2. Commercially available esters of type 9 were brominated under radical conditions to give intermediates 10 (oxazoles; methyl 2-(bromomethyl)furan-3-carboxylate is available commercially). Bromides 10 were then subjected to reductive amination, hydrolysis, and lactam-formation conditions analogously to Scheme 1 to give final analogs 12. Substituted 4,5-dihydropyrrolo[3,4-b]pyrrol-6(1H)-ones and substituted 4,5-dihydropyrrolo[3,4-c]pyrazol-6(1H)-ones were synthesized in a similar fashion (Schemes 3 and 4), with N-alkylation giving final products 16 and 20, respectively. The pyrazolo analogs (20a-j) were generally alkylated at the beginning of the sequence, as the resulting alkyl isomers were readily separable and identifiable by NOESY at this stage (Scheme 4) (alkylations can be also conducted and separated at the end of the sequence after deprotection of a para-methoxybenzyl intermediate, as in the case of analog 20a (see the experimental section for further synthetic details)). Lastly, methyl pyrazole 25a (Scheme 5) was synthesized by route of a silyl protection-Vilsmeier-Haack reaction sequence,³⁰ and carried through to give final analog 25a in a similar fashion. This chemistry generally proved quite facile, although certain ester hydrolyses required elevated temperatures and/or microwave irradiation to proceed. These routes were also found to be amenable to re-ordering – substituted tail pieces could be generated in a late-stage library format through substitution of fluorinated (4-bromophenyl)methanamines under standard Suzuki-Miyaura conditions.²⁶

Scheme 1. Reagents and conditions:

(a) R¹ benzylamine, NaBH(OAc)₃, AcOH, r.t. or 80 °C; (b) NaOH, MeOH, THF, H₂O, r.t. or 80 °C; (c) HATU, DIPEA, DMF, r.t.

Scheme 5. Reagents and conditions:

(a) TIPS-Cl, NaH, THF, 0 °C to r.t.; (b) DMF, oxalyl chloride, DMF, 0 °C to r.t.; (c) R¹ benzylamine, NaBH(OAc)₃, DCE, r.t.; (d) NaOH, THF, H₂O, 70 °C; (e) HATU, DIPEA, DMF, r.t.

Scheme 2. Reagents and conditions:

(a) NBS, benzoyl peroxide, DCE, 80 °C (oxazoles); (b) R¹ benzylamine, DIPEA, MeCN, r.t.; (c) LiOH or NaOH, MeOH, THF, H₂O, r.t. or 80 °C; (d) HATU, DIPEA, DMF, r.t.

Scheme 3. Reagents and conditions:

(a) R¹ benzylamine, NaBH(OAc)₃, DCE, r.t.; (b) LiOH, THF, H₂O, 120 °C, microwave; (c) HATU, DIPEA, DMF, r.t.; (d) R² alkyl chloride or bromide, Cs₂CO₃, DMF, 70 °C.

Scheme 4. Reagents and conditions:

(a) R² alkyl chloride or bromide, K₂CO₃, DMF, 50 °C, chromatographic separation of isomers; (b) R¹ benzylamine, NaBH(OAc)₃, DCE, r.t.; (c) LiOH, THF, H₂O, r.t.; (d) HATU, DIPEA, DMF, r.t.

hM₁ potency data for all analogs is listed in Table 1. Imidazole 8a was found to be inactive at hM₁ (Table 1), while thiazoles 8b and 8c displayed activities in the micromolar range. N-methylindazole compound 8b was found to have superior potency within the thiazole series, a trend that has been previously observed for similar indazole tail pieces in related M₁ reports.²⁶ Both thiophene regioisomers (8d-k) demonstrated similar micromolar hM₁ potency to 8b, and N-methylimidazole compound 8j displayed remarkably high potency (EC₅₀ 246 nM, although ACh Max was relatively low (40%)).

Table 1.

Structures and Activities of Cyclic Pyrrolidinone M₁ PAMs^a

Compound	Structure	hM1 EC50 (nM)	hM1 ACh Max
8a		>10,000	-
8b		2000b	60
8c		8000	30
8d		1920b	75
8e		1020	59
8f		2870	60
8g		3970	67
8h		2780	51
8i		2164b	66
8j		246	40
8k		4900	67
12a		>10,000	-
12b		>10,000	-
12c		992	45
16a		6450	58
16b		5580	67
16c		2920	56
20a		1120c	85
20b		3540b	61
20c		3930	84
20d		739b	87
20e		5180	57
20f		1510c	84
20g		1660b	79
20h		3220	77
20i		6890	66
20j		3580	65
25a		>10,000	-

^aCalcium mobilization assays with hM₁-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine for PAM, no exogenous acetylcholine (ACh) for the agonist EC₅₀; values represent means from one (n = 1)

^btwo (n = 2)

^cthree (n = 3) independent experiments performed in triplicate.

Interestingly, both oxazole regioisomers (12a and 12b) were found to be inactive, while furan (12c) was approximately equipotent to its thiophene congener 8e. Substituted pyrroles (16a-c) and pyrazoles (20a-j) proved even more fruitful, and compounds in these series consistently displayed activities in the low micromolar range for hM₁. Within the pyrazole series, 1-substituted pyrazoles (i.e. 20a) were generally more potent than the 2-substituted analogs (20b). Methyl pyrrolo compound 25a was found to be inactive, further demonstrating that subtle changes in ring substituents and/or electronics can have notable effects on potency for M₁ PAMs; this kind of SAR has been previously established in the field.²⁶

Because of their acceptable potencies, distinct chemotypes, and relatively high ACh max values, thiazole 8b (VU6005610) and pyrazole 20a (VU6005852) were selected for further profiling. Both compounds displayed robust selectivity for hM₁ over the other 4 mAChR subtypes (>10,000 nM hM₂-hM₅) at the concentrations tested (Table 2). Both compounds were also devoid of any hM₁ agonist activity, an encouraging profile for the further development of this series. Additionally, 8b displayed no species disconnect between human and rat M₁ (rM₁ EC₅₀ = 2080 nM). Both compounds also displayed good free fraction and low to moderate predicted hepatic clearance across species (human and rat, see Table 3).²⁰ After IV administration in a rat cassette model (0.20 mg/kg, brain sampling 0.25 h post dose), 8b (VU6005610) was also found to be highly brain penetrant (K_p = 2.1, K_p,uu = 2.2 (average of 3 experiments)). In contrast, 20a was found to have much lower brain exposure in this model (K_p = 0.12).

Table 2.

Muscarinic Selectivity Data for Compounds 8b and 20a.^a

Compound	hM1 EC50 (nM)	hM2 EC50 (nM)	hM3 EC50 (nM)	hM4 EC50 (nM)	hM5 EC50 (nM)
8b	2000	>10,000	>10,000	>10,000	>10,000
20a	1120	>10,000	>10,000	>10,000	>10,000

^aCalcium mobilization assays with hM₁- or hM₅-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine for PAM, no exogenous ACh for the agonist EC₅₀; values represent means from one (n = 1) independent experiment performed in triplicate

Table 3.

Physicochemical Properties; Human and Rat Hepatic Clearance and Free Fraction Data for Compounds 8b and 20a.^a

Compound	MW	tSPA	hClhep(mL/min/kg)	rClhep(mL/min/kg)	fu (plasma, rat)	fu (brain, rat)	fu (plasma, human)
8b	378.4	48.3	10.4	43	0.040	0.042	0.050
20a	463.5	60.7	18.6	58.7	0.027	0.019	0.025

^aSee ref. 20 for experimental details.

In summary, we have developed a series of structurally novel and highly mAChR subtype-selective M₁ PAMs devoid of agonist activity. Preliminary PK data for VU6005610 (8b) was particularly encouraging. The favorable muscarinic selectivity, lack of species potency disconnect, and high brain exposure should facilitate the use of this compound as a tool to further probe M₁ biology in rodent models. Additional synthetic efforts within these series are underway in our laboratory and will be reported in due course.