Discovery of a potent M₅ antagonist with improved clearance profile. Part 2: Pyrrolidine amide-based antagonists

Abstract

This Letter describes our ongoing effort to improve the clearance of selective M₅ antagonists. Herein, we report the replacement of the previously disclosed piperidine amide (4, disclosed in Part 1) with a pyrrolidine amide core. Several compounds within this series provided good potency, subtype selectivity, and low to moderate clearance profiles. Interestingly, the left-hand side SAR for this series diverged from our earlier efforts.

More than 30 years after identification of the M₅ receptor,^1,2 the field is still struggling to discover potent and subtype-selective tool compounds with pharmaceutically suitable ADME profiles.³ Encouragingly, recent biological data discovered with early generation tool compounds (Fig. 1, both orthosteric inhibitors 1–4 especially VU6019650 (3)⁴ and allosteric inhibitors (or NAMs) such as ML375 (5)⁵ and VU6008667 (6)⁶) indicates the importance of M₅ as a pharmaceutically valuable target for neurological disorders such as substance abuse disorder, depression, and anxiety.^3,7-11 Thus, improved tool compounds are critically important to better understand M₅ biology and validate M₅ as a drug target. Fig. 2 illustrates the SAR development of a piperidine sulfonamide-based series of M₅ antagonists, wherein a highly potent and subtype-selective tool compound VU6019650 (3) was discovered starting from two structurally related HTS hits (8 and 9).⁴ The thioether linkage of 3, a putative metabolic soft spot, was then replaced with an amide linker to further improve the clearance profile (disclosed in Part 1).²⁰ A parallel investigation of the piperidine core moiety is disclosed herein. Comparing 11 with first-generation M₅ antagonist ML381 (1),¹² we hypothesized that replacing the piperidine core with a pyrrolidine might be tolerated. This modification brings chirality and may potentially lead to different metabolism and clearance profiles.¹³ This core replacement idea was further supported by a close alignment of the low-energy conformers of 11 and 12a (Fig. 2).

Fig. 1.

Structures of selected orthosteric M₅ antagonist (1–4) as well as M₅ negative allosteric modulator (NAM) tool compounds (5–7).^4-6,11-15,20

Fig. 2.

Next-generation M₅ antagonist design: ^a The low-energy conformers of 11 and 12a were aligned using MOE software (Rigid Body Alignment mode; Version 2020.09; Chemical Computing Group ULC)16 and visualized with Maestro software (Version 12.4.072; Schrödinger).¹⁷

We conducted an iterative parallel synthesis approach to explore SAR around pyrrolidine amide-based M₅ antagonists (Scheme 1). Desired products were quickly assembled in 3 steps. Commercially available and suitably substituted carboxylic acids 13 were coupled with readily available anilines to provide intermediates 14. Deprotection 15, followed by sulfonamide formation under basic conditions, gave desired products 16 in poor to moderate yield. Various isolated yields were observed due to the reagent condition of sulfonyl chlorides and low compound solubility (in some cases).

Scheme 1.

General synthesis of pyrrolidine amides: (a) anilines, HATU, DIPEA, DMF, rt, 0.5–3 h, 33–99 %; (b) TFA or HCl, DCM, 1–6 h, 57–99 %; (c) sulfonyl chlorides, DIPEA, DCM, 0 °C to rt, 1–8 h, 3–53 %.

To test whether a pyrrolidine core would maintain activity, we synthesized 12a based on analogy to 11 (Table 1). Despite 12a being racemic, 12a was 2 fold more potent (hM₅ IC₅₀ = 47 nM) than 11 (hM₅ IC₅₀ = 111 nM). Encouraged by this result, we independently synthesized enantiopure 12c and 12d using commercially available chiral starting materials following Scheme 1. As predicted by the low energy alignment (Fig. 2), (R)-12a (12d, hM₅ IC₅₀ = 21 nM) was 21-fold more potent than (S)-12a (12c, hM₅ IC₅₀ = 440 nM). Substituted pyrrolidines and other small heterocyclic amine-containing cores were also examined. Azetidine 12b showed a significant reduction in potency (hM₅ IC₅₀ = 1,800 nM), indicating further core size reduction is not be ideal. However, larger 4-methyl substituted pyrrolidines (12e and 12f) were well tolerated (hM₅ IC₅₀ = 35 and 87 nM respectively). While trans-methyl analog 12f (hM₅ IC₅₀ = 87 nM) was slightly less potent than cis-methyl analog 12e (hM₅ IC₅₀ = 35 nM), trans-methyl 12f significantly improved the in vitro clearance profile (Predicted CL_hep = 4 (h) and 29 (r) mL/min/kg). Therefore, trans substitutions were investigated further. We hypothesized that more lipophilic CF₃ may improve potency while maintaining improved clearance. However, 12g containing 4R-trifluoromethyl was not active, presumably due to insufficient space. Lastly, we tested 3-azabicyclo[3.1.0] hexane 12h as well. While this core was also tolerated (12h, hM₅ IC₅₀ = 270 nM), the clearance profile was not improved (Predicted CL_hep = 18 (h) and 60 (r) mL/min/kg).

Table 1.

SAR exploration of core.

Cmpd	Core	${\mathbf{IC}}_{50}^{\mathbf{a}}$ [ACh Mina (%)] ELogD7.4	CLint (mL/min/kg) Predicted CLhep (mL/min/kg)
12a		47 [2]	72 (h), 252 (r)
		–	16 (h), 55 (r)
12b b		1,800 [2]	61 (h), 301 (r)
		–	16 (h), 57 (r)
12c		440 [3]	64 (h), 201 (r)
		2.51	16 (h), 52 (r)
12d		21 [2]	50.6 (h), 356 (r)
		2.53	15 (h), 59 (r)
12e		35 [3]	202 (h), 769 (r)
		2.88	19 (h), 64 (r)
12f		87 [2]	5 (h), 51 (r)
		2.94	4 (h), 29 (r)
12g		inactive	–
		3.65	–
12h		270 [2]	115 (h), 396 (r)
		2.88	18 (h), 60 (r)

^aCalcium mobilization assay with hM₅-CHO cells was performed in the presence of an EC₈₀ fixed concentration of acetylcholine. One experiment was performed in triplicate.

^b1-Boc-azetidine-3-carboxylic acid was used as a starting material.

As shown in Fig. 3, we previously reported that modifying the left-hand ring improves clearance (see Part 1 for details),²⁰ and substituted pyrazoles were well tolerated (e.g. 7 in Fig. 1). Therefore, we incorporated disubstituted pyrazoles in the current series (17a and b), which were inactive (Table 2), revealing an SAR mismatch between the piperidine amide-based series and the pyrrolidine amide-based series. This trend persisted in substituted 3-pyridines as well (17c-e). We hypothesized that, with the smaller pyrrolidine-based core, a larger left-hand heterocycle is required to properly position the H-bond acceptor. Therefore, we focused on bicyclic heterocycles.

Fig. 3.

Previously discovered SAR knowledge.

Table 2.

SAR exploration of western heteroaromatic ring.

Cmpd	het	IC50a (nM) [ACh Mina (%)] ELogD7.4	CLint(mL/min/kg) Predicted CLhep (mL/min/kg)
17a		>10,000 [6] 1.62	–
17b		>10,000 [59] 2.62	–
17c		inactive 1.65	–
17d		inactive 2.41	–
17e		>10,000 [41] 1.85	–
17f		35 [2] 2.45	50 (h), 168 (r) 15 (h), 49 (r)
17g		350 [3] 2.43	30 (h), 267 (r) 12 (h), 55 (r)
17h		>10,000 [31] 3.54	–
17i		350 [3] 1.99	42 (h), 191 (r) 14 (h), 51 (r)
17j		290 [2] 2.04	33 (h), 51 (r) 13 (h), 29 (r)

- Means not tested.

^aCalcium mobilization assay with hM₅-CHO cells was performed in the presence of an EC₈₀ fixed concentration of acetylcholine. One experiment was performed in triplicate.

As the methylene moiety of 2,3-dihydrobenzofuran is a potential metabolic soft spot, we generated per-deuterated analog 17f, anticipating a potential kinetic isotope effect. 17f was synthesized by following Scheme 1 using 2,3-dihydrobenzofuran-5-sulfonyl chloride-2,2,3,3-d₄ (21), which was synthesized according to Scheme 2. 21 was synthesized in 3 steps starting from 2-bromophenol 18. 18 underwent double alkylation under basic conditions to provide deuterated intermediate 20. SO₃·DMF and SOCl₂ mediated sulfonyl chloride formation reaction provided the desired sulfonyl chloride 21 in good yield.

Scheme 2.

Synthesis of 2,3-dihydrobenzofuran-5-sulfonyl chloride-2,2,3,3-d₄: (a) 1,2-dibromoethane-1,1,2,2-d₄, K₂CO₃, acetone, 60 °C, overnight, 85 %; (b) n-BuLi, THF, −78 °C, 79 %; (c) i) SO₃.DMF under N₂, 85 °C, overnight, ii) SOCl₂, rt to 75 °C over the course of 1 h, assumed quantitative yield and used for the next step without purification.

Unfortunately, 17f did not exhibit reduced metabolic clearance (Predicted CL_hep = 15 (h) and 49 (r) mL/min/kg). We therefore tried to reduce the lipophilicity (ElogD_7.4). However, a slightly more polar analog 17g (ELogD_7.4 = 2.43 compared to 12d, ELogD_7.4 = 2.53), while maintaining potency showed no notable improvement in metabolic clearance, while maintaining potency. Additional fused heterocycles were explored (such as 17i and 17j), though only the less-potent benzothiazole 17j slightly improved metabolic clearance (hM₅ IC₅₀ = 290 nM, Predicted CL_hep = 13 (h) and 29 (r) mL/min/kg). As expected, compounds lacking an H-bond acceptor (such as 17h) were significantly less active.

Although the piperidine amide-based series discussed in Part 1 showed promising in vitro profiles (potency, subtype selectivity, and clearance), the series remained sub-optimal due to low brain exposure and a P-gp efflux liability.²⁰ Therefore, we explored the P-gp efflux ratio of 12f and 17j. As shown in Table 3, the efflux ratio of 12f was slightly improved (ER = 32.3) compared to 4 (ER = 110). However, compounds within this series generally showed high ER (data not shown) indicating further optimization is required to deliver pharmaceutically favored tool compounds.

Table 3.

Tier 1 DMPK profile data for selected compounds.

Property	12f, VU6029051	17j, VU6029919
hM5 (nM) [ACh Min (%)]a	87 [2]	290 [2]
rM5 (nM) [ACh Min (%)]a	190 [2]	550 [4]
hM1 (nM) [ACh Min (%)]a	>10,000 [8]	–
hM4 (nM) [ACh Min (%)]a	>10,000	–
MW	443.54	444.55
cLogP	4.07	3.26
TPSA	88.6	92.3
Fu, plasma (rat)	0.01	0.01
Fu, plasma (human)	0.03	0.12
Fu, brain (rat)	0.05	0.08
Kp	0.11	0.1
Kp,uu	0.56	0.67
Clint (mL/min/kg)	5 (h), 51 (r)	33 (h), 51 (r)
Predicted Clhep (mL/min/kg)	4 (h), 29 (r)	13 (h), 29 (r)
MDCK-MDR1 P-gp ER	32.3	–
MDCK-MDR1 Papp (10−6 cm/s)	1.75	–

- Means not tested.

^aCalcium mobilization assay with M₅-CHO cells (h or r), M₁-CHO cells (h or r), or hM₄-Gqi5-CHO cells was performed in the presence of an EC₈₀ fixed concentration of acetylcholine. One experiment was performed in triplicate.

In summary, we further improved the in vitro profile of our selective M₅ antagonists. Compounds bearing a pyrrolidine core in place of a piperidine core showed good potency and subtype selectivity. Interestingly, we observed a different left-hand side SAR trend between the piperidine and pyrrolidine amide-based series. Unfortunately, both series possess high P-gp mediated efflux liability. As compounds with fewer H-bond donors have a lower probability of P-gp mediated efflux,^18,19 our current focus on removing the amide bond N─H will be reported in due course.

Abbreviations:

ADME

Absorption, Distribution, Metabolism, and Excretion

DIPEA

N,N-Diisopropylethylamine

DMPK

Drug metabolism and pharmacokinetics

ER

Efflux ratio

HATU

1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate

HTS

High throughput screening

SAR

Structure-activity relationship

Data availability

The data that has been used is confidential.