VU6007477, a Novel M₁ PAM Based on a Pyrrolo[2,3-b]pyridine Carboxamide Core Devoid of Cholinergic Adverse Events

Abstract

Herein, we report the chemical optimization of a new series of M₁ positive allosteric modulators (PAMs) based on a novel pyrrolo[2,3-b]pyridine core, developed via scaffold hopping and iterative parallel synthesis. The vast majority of analogs in this series proved to display robust cholinergic seizure activity. However, by removal of the secondary hydroxyl group, VU6007477 resulted with good rat M₁ PAM potency (EC₅₀ = 230 nM, 93% ACh max), minimal M₁ agonist activity (agonist EC₅₀ > 10 μM), good CNS penetration (rat brain/plasma K_p = 0.28, K_p,uu = 0.32; mouse K_p = 0.16, K_p,uu = 0.18), and no cholinergic adverse events (AEs, e.g., seizures). This work demonstrates that within a chemical series prone to robust M₁ ago-PAM activity, SAR can result, which affords pure M₁ PAMs, devoid of cholinergic toxicity/seizure liability.

Muscarinic acetylcholine receptor subtype 1 (M₁) positive allosteric modulators (PAMs) hold great promise for the treatment of cognitive impairment, schizophrenia, and Alzheimer’s disease.¹⁻⁷ Originally, the clinical promise of M₁ as a target was derailed by M₁ agonists lacking true M₁ selectivity,^1,2,8 and later, by robust M₁ ago-PAMs that overstimulated the M₁ receptor, resulting in cholinergic toxicity and adverse events (AEs).⁹⁻¹⁸ While these ago-PAMs, represented by 1–4 (Figure 1), dampened enthusiasm for the translational utility of selective M₁ activation, next generation M₁ PAMs 5 and 6, devoid of agonism in cells and native systems (and without cholinergic toxicity/seizures), provided a new path to the clinic.^14,19 However, due to the voluminous literature with ago-PAMs 1–4, multiple new M₁ PAMs are required for the community to evaluate the safety and efficacy of the “pure” M₁ PAM mechanism and independently embrace the therapeutic potential of M₁ PAMs. Herein, a novel series of M₁ PAMs featuring a pyrrolo[2,3-b]pyridine carboxamide core is described, including the design, SAR, DMPK properties, pharmacology, and cholinergic adverse effect profiles to provide the community with another new, pure M₁ PAM tool for in vivo efficacy and tolerability studies.

Figure 1.

Structures of representative M₁ PAMs 1–6. Of these, 1–4 are robust ago-PAMs in high expressing cell lines and native systems, resulting in severe seizures/cholinergic AEs due to overstimulation of M₁. M₁ PAMs 5 and 6 are devoid of seizures/cholinergic AEs and show little to no M₁ agonism in high expressing cell lines and native systems.

Scaffold-hopping has been a major strategy in the M₁ PAM arena, and this approach led to the discovery of the benzomorpholine-based pure M₁ PAM 6.¹⁹ We once again held the key carboxamide and heterobiaryl tail moieties of 3, 5, and 6 constant while surveying new heterobiaryl cores, as in 7, that might engender pure M₁ PAM pharmacology (Figure 2). From this exercise, a novel pyrrolo[2,3-b]pyridine carboxamide core, represented generically by 8, resulted, with a spectrum of M₁ pharmacology from potent ago-PAMs to pure PAMs.

Figure 2.

Scaffold hopping approach based on M₁ PAMs 3, 5, and 6, which led to the discovery of a novel pyrrolo[2,3-b]pyridine carboxamide-based series of M₁ ago-PAMs and pure PAMs, 8.

The synthesis of diverse analogs 13 (Scheme 1) proved straightforward with readily available starting materials from commercial sources.²⁰ 4-Chloro-1H-pyrrolo[2,3-b]pyridine-6-carbonitrile 9 could be alkylated with various R₁ moieties, but we first explored methyl. Conversion to the boronate ester under standard conditions provided 10 in >90% yield for the two steps. Suzuki coupling with either benzyl halides or heteroaryl methyl halides 11 generated derivatives 12 in 50–84% yield. Finally, hydrolysis of the nitrile and standard HATU amide coupling reactions delivered the putative M₁ PAMs 13 in 35–66% yield.²⁰

Scheme 1. Synthesis of M₁ PAM Analogs 13.

Reagents and conditions: (a) R₁I, NaH, DMF, 0 °C–rt, 85–90%; (b) bis(pinacolato)diboron, KOAc, Pd(dppf)Cl₂, 1,4-dioxane, 100 °C, 16 h, >98%; (c) benzyl chloride, Cs₂CO₃, Pd(dppf)Cl₂, THF/H₂O, 90 °C, 16 h; 50–84%; (d) conc. HCl, reflux, 2 h, >98%; (e) amine, HATU, DIEA, DMF, rt, 20 min, 35–66%.

We then surveyed this new core with a variety of heterobiaryl tails and amide congeners, grouped based on internal knowledge regarding SAR to engender robust M₁ ago-PAM versus PAM activity. The first amide moiety surveyed was the (3S,4R)-3-hydroxy-4 amino tetrahydropyranyl (THP) amide, well documented to engender potent M₁ PAMs, but with undesired potent M₁ agonist activity.^{11,13−17,19} As we previously reported, M₁ PAM potencies below 100 nM, coupled with M₁ agonist activity, generally lead to robust cholinergic AEs.¹⁹ As anticipated, analogs 14 (Table 1) with diverse southern heterobiaryl tails provided potent M₁ ago-PAMs. Analog 14b stands out as a 14 nM M₁ PAM (86% ACh max), with an agonist EC₅₀ of 575 nM (67% ACh max) and a direction not worthy of further pursuit, as overstimulation of M₁ would definitely result.

Table 1. Structures and M₁ Pharmacology for Analogs 14^a.

^aCalcium mobilization assays with rat M₁-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine for PAM, and no exogenous ACh for the agonist EC₅₀; values represent means from three (n = 3) independent experiments performed in triplicate.

As the analogous cyclohexyl congener 15 of THP 14 generally afforded a lower degree of M₁ agonism, we next evaluated such analogs. As shown in Table 2, analogs 15 were also potent M₁ ago-PAMs (15a–c), but with reduced M₁ agonism relative to 14,^{11,13−17,19} as well as a moderately potent M₁ PAM (15d). Interestingly, both 15b and 15c were not CNS penetrant in rat, with K_p values below level of quantitation (BLQ), despite favorable CNS MPO scores (4–5).²¹ Previously, we have shown that the secondary hydroxyl moiety engenders poor CNS penetration in rodents, and future analogs would avoid this moiety.¹⁹

Table 2. Structures and M₁ Pharmacology for Analogs 15^a.

^aCalcium mobilization assays with rM₁-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine for PAM, and no exogenous ACh for the agonist EC₅₀; values represent means from three (n = 3) independent experiments performed in triplicate.

Thus, we evaluated the profile of simple, unsubstituted pyranyl amides, as we hoped this would engender a balance between M₁ PAM potency (and diminished M₁ agonism) and CNS penetration. Our initial evaluation produced pyranyl amide 16 and the spiro-cyclic congener 17 (Figure 3); importantly, both proved to be pure M₁ PAMs (agonist EC₅₀s > 10 μM). However, M₁ PAM potency was modest (16: rM₁ PAM EC₅₀ = 560 nM, pEC₅₀ = 6.25 ± 0.05, ACh max = 89 ± 1; 17: rM₁ PAM EC₅₀ = 604 nM, pEC₅₀ = 6.22 ± 0.04, ACh max = 91 ± 1). Still, similar pyranyl amides (lacking the secondary hydroxyl moiety) in other series (such as 5 and 6) were inactive or weak, suggesting this is a unique core. To further optimize M₁ PAM potency with the pyranyl amide, we next explored alternative heterobiaryl tails.

Figure 3.

SAR evaluation of nonhydroxylated amides 16 and 17.

Surprisingly, while holding the pyranyl amide moiety constant in analogs 18 and varying the heterobiaryl tail (Table 3), the lack of M₁ agonism in these des-hydroxy congeners proved to be a generally conserved pharmacological property. Of these, 18c emerged as an attractive M₁ PAM (EC₅₀ = 230 nM, 93% ACh max; log K_b = −4.82; log αβ = 2.6) with minimal to no agonism in our high expressing cell line (M₁ agonist EC₅₀ > 10 μM).²⁰ Replacement of the N-Me pyrazole in 16 for an oxazole analog (e.g., 18f) provided an M₁ PAM of comparable potency (EC₅₀ = 760 nM, 88% ACh max) to 16 and still devoid of M₁ agonism. Regioisomeric pyridyl pyrazole heterobiaryls 18b and 18c showed similar M₁ PAM activity, but 18b displayed weak M₁ agonism (∼34% @ 10 μM). An N-linked pyrazole congener, 18a, displayed similar PAM potency and weak M₁ agonism (∼26% at 10 μM) as well, highlighting the sensitivity of this series for M₁ agonist activity.

Table 3. Structures and M₁ Pharmacology for Analogs 18^a.

^aCalcium mobilization assays with rM₁-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine for PAM, and no exogenous ACh for the agonist EC₅₀; values represent means from three (n = 3) independent experiments performed in triplicate.

A larger amide scan of the pyrazole regioisomers 19 based on the 18b heterobiaryl tail motif (Table 4) similarly identified a number of pure M₁ PAMs and ago-PAMs (with novel amides), but none showed advantages over analogs highlighted in Tables 1–3. However, unlike predecessor series 1–4, this scaffold afforded pure M₁ PAMs, devoid of agonist activity in high expressing M₁ cell lines.

Table 4. Structures and M₁ Pharmacology for Analogs 19^a.

^aCalcium mobilization assays with rM₁-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine for PAM, and no exogenous ACh for the agonist EC₅₀; values represent means from three (n = 3) independent experiments performed in triplicate.

Prior to assessing physiochemical and DMPK properties, we wanted to ensure good species alignment for key M₁ PAMs at both human and rat M₁. Gratifyingly, there was high species conservation (Table 5), indicating translational potential for this novel series of M₁ PAMs.

Table 5. Human and Rat M₁ PAM Activities for Select Analogs 14–18.

Cpd	rat M1 PAM EC50 (μM)a [% ach Max ± SEM]	rat M1 pEC50 (±SEM)	hum M1 PAM EC50 (μM)a [% ACh max ± SEM]	hum M1 pEC50 (±SEM)
14a	0.04, [87 ± 0]	7.41 ± 0.06	0.06, [83 ± 3]	7.22 ± 0.10
14c	0.10, [90 ± 1]	7.00 ± 0.03	0.09, [80 ± 8]	7.05 ± 0.20
15a	0.18, [87 ± 1]	6.74 ± 0.03	0.18, [84 ± 2]	6.74 ± 0.10
15b	0.10, [87 ± 1]	7.02 ± 0.04	0.12, [79 ± 4]	6.91 ± 0.10
16	0.56, [89 ± 1]	6.25 ± 0.05	0.66, [87 ± 3]	6.18 ± 0.10
18c	0.23, [93 ± 0]	6.64 ± 0.07	0.32, [84 ± 7]	6.49 ± 0.05

^aCalcium mobilization assays with rM₁-CHO or hM₁-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine for PAM; values represent means from three (n = 3) independent experiments performed in triplicate.

The M₁ PAMs highlighted in Table 5 displayed favorable physiochemical properties (MWs < 450, cLogPs between 1.2 and 4.0, TPSAs 89–102 Ų, CNS MPO scores between 3.2 and 4.3) and acceptable in vitro DMPK profiles (Table 6).²⁰ The majority show modest predicted hepatic clearance in human and rat, good free fraction (human, rat, and rat brain homogenate binding), but low to modest CNS penetration in rat (rat brain/plasma K_ps and K_p,uus ≤ 0.3). Moreover, as we have seen significant variations between brain penetration in mice and rats, we also assessed mouse K_p at 100 mg/kg ip. Despite low but measurable K_p in mice, brain levels were high (both total and unbound). The low K_ps were driven by very high plasma concentrations (Supplemental Figure 1).²⁰

Table 6. In Vitro DMPK Data and PBL Data for Select M₁ PAMs 14–18^a.

property	14a	14c	15b	15c	16	18c
MW	445	446	443	444	427	430
cLogP	2.40	1.22	4.02	2.84	5.21	1.77
TPSA	89.7	102.1	80.5	92.9	60.3	81.9
In Vitro PK Parameters
rat CLHEP (mL/min/kg)	39.1	42.0	49.6	27.3	47.9	44.2
human CLHEP (mL/min/kg),	6.97	10.6	17.8	9.3	10.7	6.78
rat fu (plasma)	0.03	0.02	0.03	0.06	0.04	0.06
human fu (plasma)	0.02	0.02	0.12	0.04	0.02	0.04
rat fu (brain)	0.01	0.01	0.04	0.07	0.01	0.08
Rat PBL (IV, 0.2 mg/kg)
Kp	0.11	0.21	BLQ	BLQ	BLQ	0.28
Kp,uu	0.05	0.13	BLQ	BLQ	BLQ	0.32
Mouse PBL (IP, 100 mg/kg)
Kp	0.09	ND	0.28	0.03	ND	0.16
Kp,uu	0.04	ND	0.17	0.04	ND	0.18

^aBLQ: brain concentration is below limit of quantitation (BLQ) in low dose rat cassette (0.2 mg/kg). ND = not determined.

As we have documented previously,^{11,13−17,19} robust ago-PAM activity can lead to over stimulation of the M₁ receptor and/or induce M₁ agonist-dependent long-term depression in the prefrontal cortex leading to cognitive dysfunction and, in mice (the most susceptible species to excessive M₁ activation), Racine scale 4 to 5 seizures.^11,14,17,19 Thus, a quick phenotypic screen in mice (at 100 mg/kg IP) to assess seizure liability has become a proven method of choice to eliminate compounds from the lead progression flow-chart and deem them nondevelopable. Interestingly, this pyrrolo[2,3-b]pyridine carboxamide-based series showed a pronounced tendency for cholinergic adverse effect liability in the mouse seizure model (Figure 4), with all analogs showing robust Racine scale seizures within 30 min of IP administration (100 mg/kg), save 18c (VU6007477). These data further highlight the range of clean versus adverse event (AE) in vivo pharmacology that can occur within a highly conserved chemical series of M₁ PAMs (and the value of an informative phenotypic triage screen).^14,19

Figure 4.

Phenotypic mouse seizure assay with compounds dosed at 100 mg/kg i.p. VU6007477 (18c) was devoid of seizure liability (mouse brain levels: 7 μM total, 546 nM unbound), whereas 14a, 14c, 15b, and 15c elicited robust Racine scale 4/5 seizures.²⁰

Due to the significant AE liability within this series, we were hesitant to further advance this series down the lead optimization flow-chart; however, additional characterization quickly led to a no-go decision for both 18c and this series. While 18c was highly selective for M₁ (M₂–M₅ EC₅₀s > 30 μM, Supplemental Figure 2),²⁰ it proved to be a human P-gp substrate (ER = 4.5), with only moderate permeability (Papp = 1.2 × 10^–5 cm/s). Thus, 18c is a new in vitro/in vivo rodent M₁ “pure” PAM tool compound but is not suitable for translation to the clinic.

In summary, a scaffold-hopping exercise identified a novel pyrrolo[2,3-b]pyridine carboxamide-based series of M₁ PAMs that displayed a range of potent ago-PAM and pure PAM pharmacology. While congeners possessing the prototypical (3S,4R)-3-hydroxy-4 amino tetrahydropyranyl (THP) amide moiety (or the cylcohexyl analog) engendered a range of M₁ ago-PAM activity, they also resulted in severe Racine scale 4/5 seizures that correlated with the presence of M₁ agonist activity (which is predicted to result in overstimulation of M₁). In contrast, a simple pyranyl amide derivative, 18c (VU6007477), was a pure M₁ PAM in high expressing cell lines (M₁ agonist EC₅₀ > 10 μM), displayed improved CNS penetration over the hydroxylated congeners, and was devoid of seizure liability in mice. While 18c was not advanceable as a clinical candidate, it remains an important new rodent tool compound to study selective M₁ activation without concern for cholinergic toxicity and related AEs. The optimization of other pure M₁ PAMs, devoid of cholinergic toxicity and adverse effect liability, with translational potential, will be reported in due course.

Glossary

ABBREVIATIONS

PAM

positive allosteric modulator

PBL

plasma/brain level

AE

adverse event

dppf

1′-ferrocenediyl-bis(diphenylphosphine)

HATU

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

DIEA

diisopropylethyl amine

DMPK

drug metabolism and pharmacokinetics

MPO

multiparameter optimization

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.8b00261.

• General methods for the synthesis and characterization of all compounds, and methods for the in vitro and in vivo DMPK protocols and supplemental figures (PDF)

Author Contributions

C.W.L., C.M.N., P.J.C., H.P.C., and J.M.R. drafted/corrected the manuscript. J.L.E., E.S.C., M.F.L., R.A.C., and D.W.E. performed the chemical synthesis. C.W.L., P.J.C., C.M.N., J.K.R., and H.P.C. oversaw the medicinal chemistry and target selection and interpreted the biological data. V.B.L. and H.P.C. performed the in vitro molecular pharmacology studies. A.L.B. performed the in vitro and in vivo DMPK studies. J.M.R. oversaw the in vivo experiments. J.W.D. performed the in vivo studies. All authors have given approval to the final version of the manuscript.

The authors declare the following competing financial interest(s): Hold IP on M1 PAMs.