Challenges in the development of an M4 PAM preclinical candidate: The discovery, SAR, and in vivo characterization of a series of 3-aminoazetidine-derived amides

Abstract

This letter details the continued chemical optimization of a novel series of M₄ positive allosteric modulators (PAMs) based on a 5-amino-thieno[2,3-c]pyridazine core by incorporating a 3-amino azetidine amide moiety. The analogs described within this work represent the most potent M₄ PAMs reported for this series to date. The SAR to address potency, clearance, subtype selectivity, CNS exposure, and P-gp efflux are described. This work culminated in the discovery of VU6000918, which demonstrated robust efficacy in a rat amphetamine-induced hyperlocomotion reversal model at a minimum efficacious dose of 0.3 mg/kg. 2009 Elsevier Ltd. All rights reserved.

Graphical Abstract

Positive allosteric modulators (PAMs) of the muscarinic acetylcholine receptor (M₄) (1-4) have emerged as an exciting potential strategy for the treatment of numerous CNS disorders, including schizophrenia,^1–20 Huntington’s disease,²¹ and Alzheimer’s disease.²² Previous reports from our laboratory have described the discovery and characterization of VU0152100 (ML108, 1), an in vivo tool compound which demonstrated efficacy in rodent models of anti-psychotic efficacy.^3,4 We subsequently reported related congener VU0467154 (2), based on a 5-amino-thieno[2,3-c]pyridazine core, which, despite its robust in vivo activity in multiple preclinical rodent models and a favorable pharmacokinetic (PK) profile, suffered from considerably lower potency at the human M₄ receptor as compared to rat.^11,19 In the course of our medicinal chemistry campaign to identify a compound with improved potency at the human M₄ receptor while maintaining suitable DMPK properties for a clinical candidate, we encountered steep SAR not only in potency at M₄, but in multiple DMPK properties as well.^13,14,19,20 Herein, we describe our efforts to replace the benzylic linker present in compounds 1-4 with substituted 3-amino azetidines.

Observing that small cyclic amides afforded potent analogs in both Eli Lilly’s and our M₄ PAM programs, we wished to examine the introduction of a cyclic linker between the 5-aminothieno[2,3-c]pyridazine amide core and the appended aryl ring. Such a change may serve to decrease the planarity of the molecule, thus reducing its ability to form pi-stacking interactions and thereby improve solubility, restrict the conformations available for the aryl ring to adopt, and remove the benzylic methylene as a potential metabolic soft spot. Diamine linkers would provide a convenient synthetic handle by which to introduce substituents on the cyclic linker. Several potential linkers were examined, including monocyclic and bicyclic diamines; however, 3-amino substituted azetidines yielded the most potent analogs (Figure 2).

Figure 2.

Cyclic diamines examined as alternative amide linkers to thieno[2,3-c]pyridazine core.

Analogs were readily prepared following functionalization of commercially available 3-(Boc-amino)-azetidine via nucleophilic substitution or Buchwald-Hartwig^23,24 cross-coupling reactions, followed by Boc deprotection and amide coupling to the thieno[2,3-c]pyridazine core (Scheme 1). Our initial library examined the effect of tertiary carbamates, sulfonamides, and amides (Table 1). Basic tertiary azetidine amines were poorly tolerated and led to a sharp decrease in human M₄ (hM₄) potency (data not shown). Carbamates proved to be the most potent compounds in this class, with analog 6b displaying an EC₅₀ of 23 nM. However, upon further profiling, 6b was found to have weak activity at human M₂ (hM₂, EC₅₀ = 2.65 μM) and a short elimination half-life in vivo in rat (t_1/2 < 30 min) due to facile hydrolysis of the carbamate, which proved to be the case in general for the carbamate series and thus precluded their advancement. Azetidine sulfonamides (6e), ureas (6d), and amides (6f-h) were also tolerated, albeit with lower potency as compared to the carbamates. Compound 6h was selected for further assessment, which gratifyingly found an improved profile compared to the carbamate series with reduced activity at hM₂ (EC₅₀ > 10 μM) and low in vivo clearance (rat CL_p = 3.1 mL/min/kg). Unfortunately, 6h was found to have low CNS exposure (rat brain:plasma K_p = 0.03, K_p,uu = 0.37 at 0.25 hr post-IV cassette dose) likely due to P-gp efflux (MDCK-MDR1 ER = 96).

Scheme 1.

Synthesis of M₄ PAM analogs 6, 16, 17. Reagents and conditions: (a) R-X, DCM, DIPEA, rt. (b) R-Het-X, Cs₂CO₃, DMF, heat (c) Ar-X, Pd₂(dba)₃, rac-BINAP, Cs₂CO₃, toluene, 100 °C (d) TFA, DCM, rt, 3 hr (e) 5-amino-3,4-dimethylthieno[2,3-c]pyridazine-6-carboxylic acid, HATU, DIPEA, DMF, 2 hr.

Table 1.

Structures and activities for M₄ PAM analogs 6.

Cmpd	R	hM4 EC50 (nM)a [% ACh Max ±SEM]	hM4 pEC50 (±SEM)
6a	CO2Bn	30 [81±8]	7.62±0.19
6b	CO2Ph	23 [96±3]	7.65±0.03
6c	CO2(3-Me)Ph	67 [85±9]	7.23±0.17
6d	C(O)NHPh	217 [89±6]	6.66±0.03
6e	SO2Ph	268 [70±8]	6.58±0.07
6f	C(O)Ph	773 [85±6]	6.22±0.25
6g	C(O)2-pyridyl	564 [91±5]	6.25±0.03
6h	C(O)4-pyridyl	179 [85±7]	6.75±0.03

^aCalcium mobilization assays with hM_4/Gqi5-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.

Encouraged by our initial results within this series, we sought to further improve the azetidine amides by examining N-aryl azetidines. Initial efforts sought to mimic the carbamate functionality by incorporating 1,3-heteroaryl substituents (16a-c). Compound 16c maintained reasonable hM₄ potency (EC₅₀ = 106 nM), exhibited moderate in vitro clearance (predicted rat and human CL_hep = 41 and 12 mL/min/kg, respectively, based on hepatic microsomal CL_int), and achieved moderate CNS exposure (rat brain:plasma K_p = 0.17, K_p,uu = 1.3). Despite lacking mAChR subtype selectivity (hM₂ EC₅₀ = 220 nM), 16c was advanced to a rat amphetamine hyperlocomotion (AHL) reversal study where it exhibited marginal efficacy (16% reversal following 10 mg/kg PO) but provided initial proof-of-concept for the azetidine amide class of compounds. Broadening our scope of N-heteroaryl azetidines led us to identify numerous potent analogs (16d-q). A wide variety of heterocycles was tolerated, with pyridyl, pyrimidyl, and pyrazinyl substituents providing EC₅₀s < 100 nM. Substitution of the heteroaryl ring proved capable of imparting dramatic shifts in potency. While the parent 4-pyridyl azetidine analog possessed micromolar PAM activity (data not shown), introduction of halogen substituents on the 4-pyridyl ring provided exquisitely potent analogs (16p, 16q). Substitution at the meta-position was also tolerated on 3-pyridyl analogs, with methoxy (16l) and fluorine (16m) providing analogs of comparable potency to 16e.

While we were able to achieve excellent in vitro potencies within this series, obtaining both reasonable CNS exposure and metabolic stability proved more challenging. A number of compounds (16e-h, 16l) failed to achieve acceptable CNS exposure (rat brain:plasma K_p < 0.05) and/or were found to be substrates for human P-gp efflux (16e, 16i, 16q,). Additionally, rat in vivo PK studies revealed evidence for extrahepatic non-CYP₄₅₀ metabolism in certain cases (16o, 16p; possibly due to aldehyde oxidase-mediated metabolism alpha to the 4-pyridyl nitrogen). Increasing the lipophilicity of analogs by incorporation of halogen atoms generally led to modest increases in rat CNS exposure and reduced P-gp efflux. Compound 16j was selected for further DMPK profiling, which revealed low predicted clearance (human and rat predicted CL_hep = 5.6 and 18 mL/min/kg, respectively, based on hepatic microsomal CL_int) and low potential for CYP₄₅₀ inhibition (3A4, 2D6, 2C9, 1A2 IC₅₀ > 30 μM). In a rat (male, Sprague-Dawley; n = 2) in vivo PK study, 16j demonstrated low clearance (CL_p = 8.8 mL/min/kg) with a small volume of distribution (V_ss = 0.89 L/kg) and moderate elimination half-life (t_1/2 = 1.3 hr). Total and unbound distribution of 16j to the brain was moderate in rat (brain:plasma K_p = 0.12, K_p,uu = 0.33 at 0.25 hr post-IV cassette dose), and, while it was still a substrate for human P-gp (ER = 8.5 in MDCK-MDR1 cells), its efflux was attenuated compared to related analogs. Given this favorable profile, it was advanced to a dose-response amphetamine hyperlocomotion (AHL) study in rat where it demonstrated robust reversal of AHL (Figure 3). An oral dose of 1 mg/kg provided a 44% reversal of AHL, and a maximal effect of 55% AHL reversal was achieved from 10 and 30 mg/kg dose levels. This level of in vivo efficacy was encouraging for the series and comparable to that previously reported for benchmark compounds 3 and 4. However, due to 16j’s P-gp liability and potentiation of hM₂ (EC₅₀ = 0.96 μM, ACh_Max = 43%), it was deemed not suitable for further development.

Figure 3.

Reversal of amphetamine-induced hyperlocomotion in rat (male, Sprague Dawley, n = 6 per dose group) by 16j (VU0477806). 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 hr interval (t = 60–120 min) and used to calculate %reversal of AHL for each dose group. Data represent means ± SEM.

Expanding on the scope of azetidine analogs we had already investigated, we synthesized a library of N-aryl analogs (Table 3). Phenyl congener 17a retained the excellent hM₄ potency that could be attained with N-heteroaryl analogs. While 17a did suffer from higher in vivo clearance (rat CL_p = 45 mL/min/kg), a consequence of phenyl hydroxylation based on in vitro metabolic soft-spot experiments (data not shown), as well as moderately potent potentiation of hM₂ (EC₅₀ = 1.78 μM, 58% ACh_Max), it showed a promising divergence from the N-heteroaryl azetidine SAR with improved CNS exposure (rat brain:plasma K_p = 0.57, K_p,uu = 0.73). Fluorine substitution(s) on the phenyl ring (17b, 17h-j) generally maintained good hM₄ potency and reduced in vivo clearance (17i, 17j; rat CL_p = 4.4 and 16 mL/min/kg, respectively). Notably, the azetidine congeners (17f, 17g) of benzylic-linked M₄ PAMs previously reported by our group^19,20 exhibited significantly weaker hM₄ activity compared to that of 17j (~10-fold and 40-fold, respectively). In light of its favorable potency and rat PK, compound 17j, bearing a 2,3-difluorophenyl substituent, was selected for additional characterization. Operational model parameters were determined for 17j from ACh CRC fold-shift experiments (Ca²⁺ mobilization assays), which revealed a rat M₄ K_B of 120 nM and αβ of 63, a cynomolgus monkey M₄ K_B of 890 nM and αβ of 120, and a human M₄ K_B of 2000 nM and αβ of 380. Gratifyingly, 17j displayed broadly attractive properties including an absence of P-gp efflux (MDCK-MDR1 ER = 1.3), acceptable selectivity versus hM₂ (~230-fold based on potency), and high brain distribution (rat brain:plasma K_p = 0.77, K_p,uu = 0.86 at 0.25 hr post-IV cassette dose).

Table 3.

Structures and activities for M₄ PAM analogs 17.

Cmpd	R	hM4 EC50 (nM)a [% ACh Max ±SEM]	hM4 pEC50 (±SEM)
17a		25 [85±8]	7.61±0.08
17b		28 [82±9]	7.56±0.03
17c		282 [77±8]	6.55±0.04
17d		141 [77±11]	6.88±0.11
17e		1289 [81±8]	5.90±0.06
17f		934 [86±1]	6.03±0.04
17g		325 [75±13]	6.63±0.29
17h		37 [94±9]	7.49±0.12
17i		30 [84±5]	7.54±0.06
17j		19 [88±5]	7.73±0.06

^aCalcium mobilization assays with hM_4/Gqi5-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.

Based on these findings, compound 17j was then evaluated in a rat AHL dose response study (Figure 4) where it demonstrated statistically significant AHL reversal (18%) from a low oral dose of 0.03 mg/kg and reached maximal reversal (74%) from a 3 mg/kg dose, with a resulting in vivo plasma EC₅₀ of 74 nM (0.66 nM unbound) based on terminal concentrations measured in the study animals (1.5 hr post-administration of 17j).

Figure 4.

Reversal of amphetamine-induced hyperlocomotion in rat (male, Sprague Dawley, n = 8–24 per dose group) by 17j (VU6000918). M₄ PAM, control M₄ PAM (VU0467154) 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 or vehicle (100% water) was administered subcutaneously 30 min later (t = 60 min). Total ambulations were measured over the subsequent 1 hr interval (t = 60–120 min) and used to calculate % reversal of AHL for each dose group. Data represent means ± SEM.

Compound 17j was further studied to evaluate its suitability for progression into IND-enabling studies. An Ames test found no evidence of mutagenesis, a large secondary pharmacology panel (Cerep) revealed a fairly clean profile (all IC₅₀s/EC₅₀s >25–30 μM except for human DAT binding IC₅₀ = 0.45 μM), and CYP₄₅₀ inhibition was acceptable (1A2 IC₅₀ = 25 μM, 3A4 IC₅₀ = 11 μM, 2D6 IC₅₀ = 14 μM, 2C9 IC₅₀ = 3.8 μM, 2C19 IC₅₀ = 6.3 μM; with no evidence for time-dependent inhibition). IV and PO PK studies with 17j in rat (male, Sprague Dawley, n = 1–3) and dog (female, mongrel, n = 1–2) found the compound to possess low to moderate in vivo clearance (rat and dog CL_p = 16 and 17 mL/min/kg, respectively) with a small to moderate volume of distribution (rat and dog V_ss = 0.67 and 1.6 L/kg, respectively), and moderate oral bioavailability (rat and dog F = 38% and 20%, respectively, from a 2–3 mg/kg solution dose). However, lower oral bioavailability in dog was observed (11%) from a suspension dose (2 mg/kg) in a low-excipient vehicle. These findings, coupled with a low kinetic solubility (solubility in FaSSIF at pH 6.5 after 1 hr = 0.015 mg/mL; aqueous solubility at pH 7.4 = 1.8 μM), as well as suboptimal predicted human PK (~1–2 hr t_1/2 based on moderate and small predicted CL and V_ss, respectively) led us to deprioritize 17j for further evaluation as a preclinical candidate and focus efforts on overcoming the evident solubility-limited absorption.

In summary, substitution of the benzyl linker with a 3-aminoazetidine moiety afforded facile entry into an extremely potent series of M₄ PAMs. By varying the substitution pattern of N-aryl and N-heteroaryl groups, we were able to optimize subtype selectivity, clearance, CNS exposure, and P-gp efflux. Compound 17j demonstrated robust AHL reversal in a rodent model; however, an unacceptably short projected human half-life and low oral bioavailability (due in part to solubility-limited absorption) precluded its advancement as a clinical candidate. Further study and optimization within this series is ongoing, and will be reported in due course.

Figure 1.

Structures of representative M₄ PAMs 1-4, highlighting the optimized rodent in vivo tool M₄ PAM, VU0467154 (2), the clinical candidate VU0467485/AZ13713945 (3) and the non-human primate in vivo tool VU0476406 (4).

Table 2.

Structures and activities for M₄ PAM analogs 16.

Cmpd	R	hM4 EC50 (nM)a [% ACh Max ±SEM]	hM4 pEC50 (±SEM)
16a		439 [87±4]	6.36±0.04
16b		419 [89±6]	6.45±0.19
16c		106 [81±4]	7.04±0.16
16d		124 [81±9]	6.99±0.18
16e		34 [96±5]	7.48±0.09
16f		78 [80±11]	7.13±0.11
16g		41 [89±11]	7.41±0.11
16h		38 [83±10]	7.48±0.17
16i		29 [82±5]	7.74±0.22
16j		76 [75±6]	7.12±0.08
16k		Inactive	Inactive
16l		17 [88±11]	7.77±0.01
16m		16 [90±7]	7.81±0.03
16n		185 [87±4]	6.80±0.19
16o		51 [86±6]	7.30±0.00
16p		8 [86±3]	8.09±0.05
16q		11 [84±5]	7.97±0.09

^aCalcium mobilization assays with hM_4/Gqi5-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.