The discovery of VU0486846: steep SAR from a series of M₁ PAMs based on a novel benzomorpholine core

Abstract

This letter describes the chemical optimization of a new series of M₁ positive allosteric modulators (PAMs) based on a novel benzomorpholine core, developed via iterative parallel synthesis, and culminating in the highly utilized rodent in vivo tool compound, VU0486846 (7), devoid of adverse effect liability. This is the first report of the optimization campaign (SAR and DMPK profiling) that led to the discovery of VU0486846 and details all of the challenges faced in allosteric modulator programs (both steep and flat SAR, as well as subtle structural changes affecting CNS penetration and overall physiochemical and DMPK properties).

Introduction

M₁ (muscarinic acetylcholine receptor subtype 1) positive allosteric modulators (PAMs) represent an exciting therapeutic strategy to treat multiple domains of cognitive dysfunction in CNS disorders such as schizophrenia and Alzheimer’s disease.^1–5 However, the great potential of this target has been hindered, originally by non-selective orthosteric ligands (1),⁶ and then by potent ago-PAMs 2–5 (in cell lines (M₁ PAM EC₅₀s < 100 nM) and M₁ agomist potency <1 μM) and native tissues) that proved to be cognitive disrupting and proconvulsive due to over stimulation of the M₁ receptor (Fig. 1).^7–12 Recently, we disclosed the first M₁ PAM free from adverse events, VU6004256 (6),¹³ and with robust efficacy in NMDA receptor 1 (NR1) knock-down mice.¹⁴ With the goal of providing the community with an improved, and much needed, M₁ PAM in vivo tool compound, here we report on VU0486846 (7). PAM 7 (M₁ PAM EC₅₀ ~250 nM for human and rat), based on a novel benzomorpholine core, is a significant structural departure within the M₁ PAM field that has proven to be a valuable rodent in vivo M₁ PAM tool compound, devoid of agonist activity in native systems, free from adverse effects and highly efficacious in multiple rodent cognition models.¹⁵ Here, we describe for the first time the optimization campaign (SAR and DMPK profiles).

Fig. 1.

Structures representative of a classical ‘M₁ agonist’ (1), that engenders cholinergic side effects, M₁ ago-PAMs 2–5, that are M₁ agonists in native systems and engender adverse effects, and M₁ PAMs 6 and 7, which are devoid of adverse effects.

As briefly detailed in the initial disclosure of 7,¹⁵ we scaffold-hopped from the prototypical 6,6-fused ring system of 2 to a novel benzomorpholine core (as in 7) wherein we simultaneously increased sp³ character while bringing the Lewis basic oxygen of the quinolone into the 6,6-ring system. These modifications of the core also created a new stereogenic center, which, in the case of 7, the (R)-enantiomer displayed enantiopreference.¹⁵ Here, we will detail the synthesis, SAR, and DMPK profiles of a multi-dimensional optimization campaign within the benzomorpholine series that ultimately led to the discovery of 7, an M₁ PAM with a balance of overall properties as a new in vivo tool compound, free from adverse effect liability.

The synthesis of diverse analogs 13 was straightforward and starting materials were readily available from commercial sources (see Scheme 1).¹⁵ Ethyl 2,3-dibromopropanoate 8 was condensed with various substituted 2-aminophenols to provide the racemic heterocyclic cores 9 in 52–80% yields. Alkylation with substituted 4-bromo benzyl bromides or the analogous heterocyclic congeners proceeded smoothly delivering 10 inyields ranging from 66 to 74%. A quantitative hydrolysis of the ester gave 11, which then underwent a HATU-mediated amide coupling with diverse primary and secondary amines to provide derivatives 12 as mixtures of diastereomers. Finally, either a copper-mediated coupling with heterocycles, or a Suzuki coupling protocol, produced the putative M₁ PAMs 13 in 35–86% yield. To facilitate more rapid SAR, we initially screened analogs 13 as racemates, and used the racemic version of 7, 14 (VU0484043, M₁ EC₅₀ = 0.92 μM, pEC₅₀ = 6.10 ± 0.13, ACh Max 77 ± 7; M₂-M₅ EC₅₀s > 30 μM) as a reference compound.¹⁵

Scheme 1.

Synthesis of M₁ PAM analogs 13.^a. ^aReagents and conditions: (a) substituted 2-aminophenols, CH₃CN, K₂CO₃, 60 °C, 52–80%; (b) 4-bromoAr(Het)bromide, K₂CO₃, CH₃CN, 80 °C, 66–74%; (c) KOH, THF/H₂O (2:1), rt, 95–98%; (d) HN₂R₃, HATU, DIEA, DMF, rt, 66–80%; (e) Het-NH, (1S,2S)-N¹,N²-dimethylcyclohexane-1,2-diamine, CuI, K₃PO₄, dioxane, rt, 35–50%, or Ar(Het)-B(OH)₂, 5 mol% Pd(PPh₃)₄, THF:H₂O, 45 °C, 54–86%.

Initial SAR around 14 with analogs 15–20 showed overall ‘flat’ SAR in terms of M₁ PAM potency, but significant impact on physiochemical and DMPK properties. Of the structural changes in Fig. 2, introduction of a quaternary carbon at the chiral center, as in 16, led to a diminution in potency (M₁ EC₅₀ = 7.3 μM, 61% ACh Max); however, all other modifications were within 2- to 3-fold of 14. A major finding was that the des-oxy tetrahydroisoquinoline analog 15 was essentially equipotent to 14, suggesting that the key intramolecular hydrogen bond (IMHB) of PAMs 4–7^7–14 may not be a key tenet in this new series (however, this will be disclosed in a subsequent publication).¹⁶ Other changes, represented by 17–20, while active as M₁ PAMs, negatively impacted plasma protein binding (f_u < 0.001) and/or decreased CNS penetration (brain/plasma K_ps < 0.05) relative to 14 (f_u = 0.11, rat K_p = 0.17, mouse K_p = 0.7).

Fig. 2.

Initial SAR around 7, surveying multiple dimensions with analogs of 13, M₁ PAMs 15–20.

Based on these data, we next explored the ‘fluorine walk’, a strategy that has been highly successful in allosteric modulator optimization, particularly for other M₁ PAM scaffolds.^17,18 This exercise led to intriguing SAR (Fig. 3) relative to 14. While incorporation of a fluorine atom at the 5-positon (21) led to an approximate 3-fold decrease in activity (M₁ PAM EC₅₀ = 3.2 μM, pEC₅₀ = 5.51 ± 0.12, ACh Max 76 ± 2), installation at the 6-position (22) led to a ~3-fold increase in M₁ PAM potency (M₁ PAM EC₅₀ = 0.32 μM, pEC₅₀ = 6.51 ± 0.08, ACh Max 88 ± 1). Further movement to the 7-position (23) led to a diminution in potency (M₁ PAM EC₅₀ = 2.2 μM, pEC₅₀ = 5.66 ± 0.03, ACh Max 80 ± 2). An almost 7-fold increase in M₁ PAM potency (M₁ PAM EC₅₀ = 0.14 μM, pEC₅₀ = 6.85 ± 0.06, ACh Max 89 ± 1) was realized by incorporation of a single fluorine atom at the 8-position (24). Moreover, the rat CNS penetration was improved, relative to 14 (K_p = 0.17), with K_ps ranging from 0.3 to 0.4. Despite the improvement in M₁ PAM potency and modest gains in CNS penetration, incorporation of a single (see Fig 4).Fluorine atom negatively impacted the DMPK profiles of 21–24. Unlike 14 (with good fraction unbound, f_u > 0.11 in rat and human plasma, and moderate predicted hepatic clearance, CL_hep = 11.1 and 43.2 mL/min/kg for human and rat, respectively), analogs 21–23 all showed unacceptable fraction unbound (f_u < 0.01) and high predicted hepatic clearance (CL_heps of ~20 and >65 mL/min/kg, for human and rat, respectively). Thus, the ‘fluorine walk’ improved potency, but at the cost of disposition; therefore, a ‘balanced’ M₁ PAM in vivo tool compound would not be derived from this sub-series, and alternate regions of 14 needed to be evaluated.

Fig. 3.

Impact of the ‘fluorine walk’ around the benzomorpholine core of 14, to deliver M₁ PAM analogs 21–24.

Fig. 4.

Divergent metabolic pathways leading to high predicted human and rat hepatic clearance for M1 PAMs 14 and 25j.

We returned to the unsubstituted benzomorpholine core of 14, and surveyed alternative amide moieties for the (1S,2S)-2-aminocyclohexan-1-ol amide with analogs 25 (Table 1).

Table 1.

Structures and human activities for M₁ PAM analogs 25.

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

^bTotal brain:plasma partition coefficients determined at 0.25 h post-administration of an IV cassette dose (0.20–0.25 mg/kg) to male, SD rats (n = 1), ND = not determined.

Removal of the chiral hydroxyl moiety (25a) led to a complete loss of M₁ PAM activity (EC₅₀ > 10 μM). However, the rat K_p for 25a was >1, suggesting that this key hydroxyl pharmacophore engendered rat P-gp susceptibility and was responsible for the generally low brain levels (K_p < 0.4) observed with this moiety across multiple M₁ PAM chemotypes (note, the ER for human is generally <1.2, suggesting this is rat specific). Interestingly, all these analogs possess favorable CNS MPO scores (>4.5), despite rodent CNS exposure issues.¹⁹ The ring in 14 could be contracted to a cyclopentyl congener (25b) with comparable activity, but the well-known pyran homolog (25c) afforded a substantial increase in M₁ PAM potency (M₁ PAM EC₅₀ = 0.19 μM, pEC₅₀ = 6.73 ± 0.13, ACh Max 91 ± 4). Deletion of the hydroxyl group on the pyran ring as in 25d similarly led to a ~34-fold loss in activity. Replacement of the hydroxyl with either a primary amine (25e) or a ketone (25f) retained reasonable M₁ PAM activity. Acyclic congeners (25 g–i) of the (1S,2S)-2-aminocyclohexan-1-ol amide of 14 could begin to buy back M₁ PAM activity as steric bulk increased, but proved inferior to their cyclic counterparts. We then evaluated functionalized pyridinyl amides (25j-l), and were pleased to see that these too retained M₁ PAM activity despite the significant structural departure. Of these, the 3-hydroxy pyridinyl amide (25j) was equipotent to 14. For all of these new analogs 25, their DMPK profiles were inferior compared to the parent 14, displaying an undesired combination of high predicted hepatic clearance (CL_hep ~20 and >65 mL/min/kg in human and rat, respectively), unacceptable plasma protein binding (human and rat f_u < 0.01)and/or poor CNS penetration (K_ps < 0.1). Interestingly, metabolite identification studies with representative analogs, 14 and 25j, showed that the route of metabolism differed between the saturated amides (driven by P₄₅₀-mediated debenzylation to afford 26) and pyridinyl amides (amide hydrolysis by esterases liberating 27). Of all the analogs evaluated thus far in this benzomorpholine series, 7 was an outlier with a balance of M₁ PAM potency, pharmacological, and DMPK properties that made it an ideal in vivo tool compound.

As the pyridine biaryl derivative 18 (Fig. 2) was equipotent to 14, we elected to evaluate analogs of this motif in an effort to identify an in vivo tool (see Table 2). Simple substituents, such as Cl, Br and N(CH₃)₂ (28a-c) demonstrated that such modifications were indeed tolerated as M₁ PAMs, but a methyl group (28d) proved equipotent (M₁ PAM EC₅₀v= 0.85 μM, pEC₅₀ = 6.09 ± 0.07, Ach Max 85 ± 3) to 14, but with reduced molecular weight and hydrogen bond acceptors. Interestingly, bromine (28b) was more potent than chlorine (27a), yet CF₃ (28e) and cyclopropyl (28f) were weaker than methyl (28d), suggesting size alone was not a driving factor for M₁ PAM potency. A simple vinyl group (28 g) was also equipotent (M₁ PAM EC₅₀ = 0.70 μM, pEC₅₀ = 6.17 ± 0.10, Ach Max 89 ± 3) to 14. However, similar to analogs 25, the DMPK profiles of these compounds precluded further advancement as in vivo tools. Thus, we decided to evaluate additional heterobiaryl derivatives of 18. Moving from the N-linked pyrazole (18) to an N-linked imidazole (28 h), and introduction of basic amine, led to an increase in M₁ PAM potency (M₁ PAM EC₅₀ = 0.29 μM, pEC₅₀ = 6.55 ± 0.12, ACh Max 87 ± 2), as did replacement with a pyrrole (28j). Non-N-linked, N-methyl pyrazoles (28 k and 28 l) also proved to be active, but regioisomeric preference was noted. Here, 28 l was the most potent analog in the series (M₁ PAM EC₅₀ = 0.22 μM, pEC₅₀ = 6.75 ± 0.07, ACh Max 84 ± 5). Once again, however, none of these analogs possessed a balanced profile that positioned them to serve as M₁ PAM in vivo tools, and 7 (VU0486846) remained the most attractive compound from the overall series, offering robust efficacy without adverse effect liability or cholinergic side effects in multiple preclinical species.¹⁵ As recently detailed, 7 was devoid of M₁ agonism in cells and native systems, was devoid of cholinergic toxicity in mice, rats and non-human primates, afforded robust efficacy in a rat novel object recognition assay (MED = 3 mg/kg p.o.) and for the first time, demonstrated that M1 PAMs can reverse the cognitive deficits induced by atypical antipsychotics, such as risperidone.¹⁵

Table 2.

Structures and human activities for M₁ PAM analogs 28.

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

In summary, we have detailed for the first time the progression and SAR from the prototypical quinolone-based M₁ PAM core to a novel benzomorpholine core. A multi-dimensional optimization campaign extensively surveyed diverse regions of the racemic counterpart 14 of 7 (VU0486846). While many new pharmacophores were identified that enhanced M₁ PAM potency, and challenged some notions regarding the critical need for an IMHB, physiochemical properties and DMPK profiles preclude their advancement as in vivo tool compounds. Importantly, the scaffold hopping exercise from the quinolone core to the benzomorpholine core gave rise to the highly valuable rodent M₁ PAM in vivo tool compound 7 (VU0486846), with a balanced profile; this compound demonstrated utility in target validation studies without adverse effect liability. Further optimization efforts in route to M₁ PAM clinical candidates, and further evaluation of the tetrahydroisoquinoline core, will be reported in due course.