SAR inspired by aldehyde oxidase (AO) metabolism: Discovery of novel, CNS penetrant tricyclic M₄ PAMs

Abstract

This letter describes progress towards an M₄ PAM preclinical candidate driven by an unexpected aldehyde oxidase (AO) metabolite of a novel, CNS penetrant thieno[2,3-c]pyridine core to an equipotent, non-CNS penetrant thieno[2,3-c]pyrdin-7(6H)-one core. Medicinal chemistry design efforts yielded two novel tricyclic cores that enhanced M₄ PAM potency, regained CNS penetration, displayed favorable DMPK properties and afforded robust in vivo efficacy in reversing amphetamine-induced hyperlocomotion in rats.

Graphical Abstract

Positive allosteric modulators (PAMs) that act on the muscarinic acetylcholine receptor subtype 4 (M₄) continue to be the focus of programs developing new antipsychotic therapeutics, and other indications including Huntington’s disease (HD) and Parkinson’s disease (PD).^1-2,5-7 In many of these studies, structure-activity relationships (SAR) of the thieno[2,3-b]pyridine 2-carboxamides and the thieno[2,3-c]pyridazine 6-carboxamides were derived from varied functionalization of these cores and driven by potency results from cell-based calcium efflux assays.^3-4 Recently, we have reported on our efforts to identify novel M₄ PAM chemotypes through scaffold hopping and a reexamination of hydrogenbonding motifs contained within these privileged scaffolds.^8-23Concurrent efforts also identified a series of ligands where an aldehyde oxidase (AO) metabolite of the thieno[2,3-c]pyridine 2-carboxamides drove the SAR towards new tricyclic analogues, culminating in the identification of a thieno[3,2-e][1,2,4]triazolo-[1,5-a]pyridine-7-carboxamide system. In this Letter, we report on the syntheses of these tricyclic ligands and their varied SAR and DMPK results that generated this novel class^25-27 of selective M₄ PAMs with excellent in vitro potency, good ADME properties, and one of the most pronounced reversals of amphetamine-induced hyperlocomotion (AHL) in an in vivo model reported for this class of compounds.

While the dominant M₄ PAM chemotype is exemplified by either a thieno[2,3-b]pyridine 2-carboxamide or a thieno[2,3-c]pyridazine 6-carboxamide, these cores typically suffer from poor physicochemical properties and variable species muscarinic potency. To overcome solubility limited absorption, we explored other, related congeners with the potential for salt formation via a more basic nitrogen atom. Despite steep SAR that precluded the majority of targeted cores, a thieno[2,3-c]pyridine 2-carboxamide chemotype, exemplified by 5 (VU0467206), was identified as a novel M₄ PAM harboring a more basic nitrogen, with the potential to modulate physicochemical and DMPK properties.

The synthesis of 5 is illustrated in Scheme 1. Condensation of methyl 2-mercaptoacetate with isonicotinitrile 6 proceeds to provide 7 in 83% yield. Suzuki cross coupling reaction provides the 4-methylthieno[2,3-c]pyridine scaffold 8 in 56% yield. Saponification followed by HATU mediated coupling with the privileged trifluoromethylsulfone amine 9, delivers compound 5 in 35% yield after reverse phase HPLC.

Scheme 1.

Synthesis of thieno[2,3-c]pyridine M₄ PAM 5.^a

^aReagents and conditions: (a) methyl 2-mercaptoacetate, K₂CO₃, IPA, 65 °C, 83%; (b) CH₃-BF₃K, Pd(dppf)Cl₂·DCM, Cs₂CO₃, THF/H₂O (10:1), MW 145 °C, 56%; (c) aq. KOH, MeOH/H₂O (3:2), 50 °C, >99%; (d) R-NH2, HATU, DIEA, DMF, rt, 35%.

Compound 5 proved to be a potent rat M₄ PAM (EC₅₀ = 290 nM), but with a moderate to high predicted hepatic clearance (rat CL_hep = 51 mL/min/kg; rat microsomal CL_int = 200 mL/min/kg), which was confirmed in vivo (rat CL_p = 56 mL/min/kg), and thus a good in vitro/in vivo correlation (IVIVC). In our standard rat IV PBL cassette, 5 showed excellent brain penetration (brain:plasma K_p = 2.5, K_p,uu = 1.0; 0.2 mg/kg, 10% EtOH, 40% PEG 400, 50% DMSO 2 mg/mL). Next, we dosed 5 in a discrete rat 10 mg/kg IP PBL study (10% Tween 80 in water, 4 mg/ mL) and observed similarly high CNS penetration (brain:plasma K_p = 1.8, K_p,uu = 0.76). Encouraged by the potential of the thieno[2,3-c]pyridine core, we performed a metabolic soft-spot experiment in rat hepatic microsomes (+/− NADPH) to inform the rational design of analogs possessing increased metabolic stability. As shown in Figure 2, interestingly, we observed only minor

Figure 2.

Soft-spot analysis of 5. Confirmation of aldehyde oxidase-dependent (AO-dependent) metabolism was confirmed through AO inhibition in vitro suggesting either 5a or 5b as the core of the major AO metabolite.

NADPH-dependent metabolism (hydroxylation of the benzylic carbon), but noted extensive NADPH-independent oxidation of the western pyridine ring, later confirmed to be via aldehyde oxidase (AO). Based on previous experience with AO oxidation of heteroarenes,²⁴ we suspected that the major metabolite was either the thieno[2,3-c]pyridine-7(6H)-one 5a or the thieno[2,3-c]pyridine-5(6H)-one 5b.

After exhaustive 2D NMR experiments (¹H-¹H COSY, HSQC, HMBC), we identified the putative AO-mediated metabolite of M₄ PAM 5, as the thieno[2,3-c]pyridine-7(6H)-one 5a core, and specifically, compound 15 (VU6016365). We initiated the synthesis of compound 15 in order to independently confirm its structure. In Scheme 2, pyridone 10 was bis-brominated using N-bromosuccinimide to afford 11 and it was subsequently condensed with methyl thioglycolate to provide 12. Protection of the penultimate aminothiophene with methoxymethyl chloride (MOMCl) proceeded in an efficient 56% yield over three steps, providing compound 13. Saponification of the ester followed by standard amide coupling with our preferred amine provided compound 14 in good yield. Synthesis of the AO metabolite 15 was completed via palladium-mediated installation of the methyl group followed by removal of the hemi-aminal ether under aqueous acidic conditions.

Scheme 2.

Synthesis of AO metabolite 15 (VU6016365).^a

^a Reagents and conditions:(a) NBS, ACN, 65 °C; (b) methylthioglycolate, K₂CO₃, IPA, 65 °C; (c) MOMCl, DIEA, DCM, 50 °C, 56% over 3 steps; (d) i. LiOH, THF/H₂O (3:2) 50 °C; ii. (4-((trifluoromethyl)sulfonyl)phenyl)methanamine, HATU, DIEA, DMF, rt, 31% over 2 steps; (e) CH₃BF₃K, Pd(dppf)Cl₂, Cs₂CO₃, 1,4-dioxane/H₂O (10:1) 100 °C; (f) 4M HCl in 1,4-dioxane/H₂O, 85 °C, 3% over 2 steps.

With an authentic sample of 15 in hand, we compared the HPLC-UV chromatograms of the parent compound 5 to that of the AO metabolite 15 (Figure 3), and to compound 5 subjected to rat hepatic S9 incubation (without NADPH). There was excellent retention time correlation of the standards to the observed peaks from the hepatic S9 incubation. Confirmation of the metabolite identity was also provided by simultaneous mass spectrometric (MSⁿ) analysis, which revealed matching fragmentation patterns and ion spectra for the respective standards and S9 peaks. Further validation was then obtained by analysis of the sample from the S9 incubation after addition of the authentic standard of 15 (10 μM final concentration), which produced an increase in the corresponding peak area with the same MSⁿ spectra observed prior to addition. In light of these findings and the NMR data collected, we concluded that we had correctly identified the structure of 15. Evaluation of the in vitro pharmacology of 15 showed it was similarly potent to 5 (rat M₄ EC₅₀ = 260 nM) but had an increased fraction unbound in both rat and human plasma relative to 5 (15: human f_{u plasma} = 0.024, rat f_{u plasma} = 0.051), but was not centrally penetrant (brain:plasma K_p < 0.13, K_p,uu < 0.03), likely as a result of an additional hydrogen bond donor conferring activity as a substrate for efflux transporter(s). While not a productive scaffold for CNS indications, we envisioned constraining the amide in 15 into five-membered heterocycles might maintain favorable properties while restoring CNS.

Figure 3.

HPLC-UV (reverse phase) chromatograms of parent standard alone (5, retention = 14.7-14.8 min), metabolite standard alone (15, retention = 13.2-13.3 min), and parent (5) incubated (25 μM, 1 hr, 37 °C) in rat hepatic S9 (5 mg/mL, absence of NADPH).

All of the potential tricyclic scaffolds required de novo syntheses. The first tricyclic series that we investigated mimicked the halogenation pattern of 1 (LY2033298, Figure 1). Starting with dichloropyridine (R₁ = H) or trichloropyridine (R₁ = Cl) 16, S_NAr with sodium methanethiolate afforded compound 17 in good yield. Condensation with methyl thioglycolate, followed by saponification and amide coupling under standard conditions yielded a series of compounds with the generic structure of 19. Oxidation of the sulfide to the sulfoxide 20 proceeded in moderate yield using hydrogen peroxide under acidic conditions. Treatment of 20 with hydrazine and triethylorthoacetate (R₂= Me) or triethylorthoformate (R₂ = H) in one pot under microwave conditions produced a compound library of the structure 21, Scheme 3.

Figure 1.

Historical M₄ PAMs showing a conserved ×-amino carboxamide (in red) with poor physicochemical properties. Compound 5 (VU0467206, in blue) has potential to improve solubility via salt formation.

Scheme 3.

Synthesis of 1,3,4-triazole tricycles of 21.^a

^aReagents and conditions:(a) NaSMe, MeOH, 91%; (b) methylthioglycolate, K₂CO₃, IPA, 50 °C, 65%; (c) KOH, H₂O, 100 °C, 98%; (d) R₃NH₂, HATU, DIEA, DMF, rt, 22-54%; (e)30% H₂O₂, AcOH, 41%; (f) i. hydrazine hydrate, DMSO, MW 140 °C; ii. Triethylorthoformate or triethylorthoacetate, NMP, MW 180 °C, 21-53% over 2 steps.

The 1,3,4-triazolo tricyclic series of compounds (Table 1) was was separated into two classes where R₁ = H or R₁ = Cl. For compounds 21a-e, we expected that R₁= Cl would be very potent M₄ PAMs as chloro-substituted methylpyridine scaffolds^1-23 were typically more potent than their proteo-counterparts: 21a (EC₅₀ = 16 nM), 21b (EC₅₀ = 12 nM), 21c (EC₅₀ = 20 nM), 21d (EC₅₀ = 22 nM), and 21e (EC₅₀ = 8 nM). However, despite their exquisite potency for the human M₄ receptor, these compounds exhibited low fraction unbound in brain homogenate (21b, rat f_{u bram} = 0.002; 21c, rat f_{u brain} = 0.011) and many were substrates for P-gp efflux (MDCK-MDR1 cell line, efflux ratios from 5 to 73 for 21a-e). Similarly, 21f-m showed undesirable profiles with a 2-10 fold decrease in potency and high efflux ratios (MDCK-MDR1 cell line, efflux ratios > 15 for 21f-m), which precluded their advancement.

Table 1.

Structures and activities for M₄ PAM 1,3,4-triazole tricyclic analogs 21.^a

Cmpd	R1	R2	R3	hM4 EC50 (μM)a [% ACh Max]
15	-	-		0.26 [84]
21a	Cl	H		0.016 [90]
21b	Cl	H		0.012 [82]
21c	Cl	H		0.020 [73]
21d	Cl	CH3		0.022 [88]
21e	Cl	CH3		0.008 [94]
21f	H	H		0.11 [79]
21g	H	H		0.017 [87]
21h	H	H		0.22 [84]
21i	H	H		0.12 [71]
21j	H	H		0.089 [78]
21k	H	H		0.99 [60]
21l	H	H		0.012 [84]
21m	H	H		0.026 [87]

^aFor SAR determination, calcium mobilization human M₄/G_qi5 assays were performed n = 1 independent times in triplicate with an EC₂₀ fixed concentration of acetylcholine.

Undaunted, other tricyclic scaffolds were investigated, including the imidazo- and the 1,3,5-triazolo tricyclic variants. The synthesis of either scaffold began with dichloropyridine 22 (Scheme 4, panel A). Nucleophilic aromatic substitution with ammonia in methanol provided aminopyridine 23 with high regioselctivity in excellent yield. Condensation with methylthioglycolate at high temperature afforded diamino-4-mcthylthieno[2.3-b] pyridinc 24. In Scheme 4, panel B, the imidazo-tricyclic scaffold was constructed via condensation with chloroacetaldehyde or 1-chloropropan-2-one to provide either the unsubstituted hydrogen version or the methyl substituted congener 25, (R₁ = Me or H). Saponification followed by amide coupling using HATU conditions provided a small library of compounds 26. Similarly from compound 24, 1,3,5-triazolo tricycles were synthesized via condensation with N,N-dimethylformamide dimethyl acetal or triethylorthoacetate in the presence of hydroxylamine. Cyclization/dehydration was accomplished using trifluoroacetic anhydride to afford 27. Saponification with sodium hydroxide followed by amide coupling under standard HATU conditions provided a library of 1,3,5-triazolo tricycles 28.

Scheme 4.

Synthesis of tricyclic analogs 26 and 28.^a

^aReagents and conditions: Panel A. Synthesis of diamino-4-methylthieno[2,3-b]pyridine 24. Reagents and conditions: (a) 7 N NH₃ in MeOH, MW 150 °C, 90%; (b) methylthioglycolate, K₂CO₃, IPA, MW 120 °C, 51%. Panel B. Synthesis of imidazo-tricycles 26 and 1,3,5-triazolo tricycles 28: (c) chloroacetaldehyde (R₁ = H) or 1-chloropropan-2-one (R₁ = Me), NaHCO₃, EtOH, 80 °C, 98%; (d) 6M NaOH, THF/MeOH (3:2), 53%; (e) R₂NH₂, HATU, DIEA, DMF, 18-43%; (f) DMF-DMA (R₁ = H) or triethylorthoacetate (R₁ = Me), HONH₂, IPA, 99%; (g) TFAA, THF, 0 °C, 90%; (h) NaOH, EtOH, 63%; (i) R₂NH₂, HATU, DIEA, DMF, 11-28%.

As shown in Table 2, imidazo-tricyclic analogs 26 proved to be less than optimal with a significant loss of potency compared to previous triazolo tricycles. While small alkyl amines (26a, EC₅₀ = 130 nM) were still preferred, providing a potent compound, there were no compounds which were improved over previous library iterations with respect to potency (26b-h, EC₅₀ s range from 240 nM to 2.1 μM).

Table 2.

Structures and activities for M₄ PAM 1,3,4-triazolo tricyclic analogs 26.^a

Cmpd	R1	R2	hM4 EC50 (μM)a [% ACh Max]
26a	H		0.13 [77]
26b	H		0.27 [80]
26c	H		0.24 [66]
26d	H		2.1 [63]
26e	H		0.31 [65]
26f	CH3		1.2 [76]
26g	CH3		0.30 [71]
26h	CH3		2.0 [47]

^aFor SAR determination, calcium mobilization human M₄/G_qi5 assays were performed n = 1 independent times in triplicate with an EC₂₀ fixed concentration of acetylcholine.

Table 3 highlights the final tricyclic series investigated, the 1,3,5-triazolo tricycles. A limited set of unsubstituted triazoles (R₁ = H) was synthesized, 28a-28c, with amides that possess good potency and were representative of well tolerated analogs across multiple scaffolds. These same amides were utilized in the substituted case (R₁ = Me) for compounds 28d-r. Many compounds containing privileged M₄ PAM amides (28d, EC₅₀ = 46 nM; 28e, EC₅₀ = 60 nM; 28g, EC₅₀ = 84 nM) had desirable potencies, but once again possessed poor physicochemical properties (xLogPs > 3.5). N-aryl-azetidine amides were previously reported to be potent analogs with the potential to be centrally penetrant. Here, compounds 28j-r were synthesized and showed comparable potencies to previously described M₄ PAMs. 28j is a moderately potent M₄PAM (EC₅₀ = 140 nM) with low in vivo clearance in rat (rat CL_p = 13 mL/min/kg); however, it suffered from low brain distribution (rat brain:plasma K_p,uu = 0.02) and off-target activity at human muscarinic M₂ receptor (hM₂ EC₅₀ = 650 nM). 281 has similar properties to others in its series (hM₄ EC₅₀ = 140 nM, rat CL_p = 13 mL/min/kg, rat brain:plasma K_p,uu = 0.22, hM₂ EC₅₀ > 10 μM), but ultimately suffered from relatively low brain penetration and poor physicochemical properties. Compound 28n while a potent M₄ PAM, was also active at human M₂ (hM₂ EC₅₀ = 600 nM), typifying the unpredictable SAR with respect to hM₂ for this series. Finally, 28p is a potent M₄ PAM (hM₄ EC₅₀ = 110 nM) and had low predicted clearance (rat CL_hep = 14 mL/min/kg based on microsomal CL_int), low in vivo clearance (rat CL_p = 15 mL/min/kg), and moderate brain distribution and low unbound brain concentration (rat brain:plasma K_p = 0.47, K_{p, uu} = 0.04).²⁸ 28p was evaluated in a bidirectional MDCK-MDR1 assay for its potential to be a substrate for human P-gp efflux and showed a low efflux ratio (ER = 1.2). While 28p did show moderate potency at human M₂ (hM₂ EC₅₀ =1.2 μM), we chose to move it forward to reversal studies of amphetamine-induced hyperlocomotion (AHL) in SD rats.

Table 3.

Structures and activities for M₄ PAM 1,3,5-triazolo tricyclic analogs 28.^a

Cmpd	R1	R2	hM4 EC50 (μM)a [% ACh Max]
28a	H		0.037 [85]
28b	H		0.086 [76]
28c	H		0.095 [80]
28d	CH3		0.046 [80]
28e	CH3		0.060 [75]
28f	CH3		0.12 [63]
28g	CH3		0.084 [84]
28h	CH3		0.16 [90]
28i	CH3		1.9 [56]
28j	CH3		0.14 [89]
28k	CH3		0.25 [90]
28l	CH3		0.14 [91]
28m	CH3		0.16 [94]
28n	CH3		0.12 [86]
28o	CH3		0.27 [87]
28p	CH3		0.11 [91]
28q	CH3		0.41 [97]
28r	CH3		0.62 [86]

^aFor SAR determination, calcium mobilization human M₄/G_qi5 assays were performed n = 1 independent times in triplicate with an EC₂₀ fixed concentration of acetylcholine

Compound 28p (VU6001852) was fonnulated in 10% Tween 80 in water and dosed orally (10 mg/kg) in rats with 0.75 mg/kg (SC) challenge of amphetamine. When compared to control compound VU0467154, 28p showed superior reversal of AHL at 75%, compared to 55% for the positive control (Figure 4). This is the largest reversal we have reported for this class of compounds. Overall, many of these tricyclic series had desirable potencies and DMPK properties but suffered from moderate activity at the M₂ receptor and some demonstrated P-gp efflux. Despite these liabilities, 28p showed excellent reversal of AHL, and the promise of continued optimization of this novel series.

Figure 4.

Reversal of amphetamine-induced hyperlocomotion in SD rats (male, n ≥ 4 per dose group) by 28p (VU6001852). M₄ PAM or vehicle (10% tween-80 90% water [v/v]) was administered orally 30 min after habituation in the chamber, and then 0.75 mg/kg amphetamine was administered subcutaneously 30 min later (t = 60 min). Total ambulations were measured over the subsequent 1 h interval (t = 60–120 min) and used to calculate % reversal of AHL for each dose group. Data represent means ± SEM and were analyzed by a one-way ANOVA; post hoc comparisons were made by Dunnett’s test compared to amphetamine-vehicle conditions with statistical significance determined as p < 0.05.

In summary, we identified several novel tricyclic scaffolds with excellent potency at M₄, derived from a putative AO metabolite.²⁴ While some of the SAR data are reminiscent of our previous work highlighting the importance of certain structural motifs in the ß-amino carboxamide class of M₄ PAMs, there are disconnects based on the nature of the tricyclic core. Efforts towards development candidates in this, and other, novel scaffolds are underway, and will be reported in due course.