Optimization of M₄ positive allosteric modulators (PAMs): The discovery of VU0476406, a non-human primate in vivo tool compound for translational pharmacology

Abstract

This letter describes the further chemical optimization of the 5-amino-thieno[2,3-c]pyridazine series (VU0467154/VU0467485) of M₄ positive allosteric modulators (PAMs), developed via iterative parallel synthesis, culminating in the discovery of the non-human primate (NHP) in vivo tool compound, VU0476406 (8p). VU0476406 is an important in vivo tool compound to enable translation of pharmacodynamics from rodent to NHP, and while data related to a Parkinson’s disease model has been reported with 8p, this is the first disclosure of the optimization and discovery of VU0476406, as well as detailed pharmacology and DMPK properties.

Graphical abstract

To create your abstract, type over the instructions in the template box below. Fonts or abstract dimensions should not be changed or altered.9.0_/0

Muscarinic acetylcholine receptor subtype 4 (M₄) positive allosteric modulators (PAMs) represent a fundamentally new approach to treat multiple symptom domains of schizophrenia, Huntington’s disease and Parkinson’s disease (PD).^1–12 M₄ PAMs (Figure 1) represent the extreme of allosteric modulator caveats in terms of steep SAR, species differences in pharmacology (rat versus human versus cynomolgus (cyno) monkey versus dog M₄ PAM potency, affinity/cooperativity and subtype selectivity), poor solubility, and/or low CNS penetration.^3,10–16 Work in this area has been challenging since the first report by Eli Lilly of 1, a human-preferring M₄ PAM.¹ Subsequent optimization efforts afforded 2, which provided the first in vivo proof-of-concept (POC) for selective M₄ potentiation of endogenous acetylcholine in reversing amphetamine-induced hyperlocomotion (AHL) in rats.³ Further efforts improved physiochemical/DMPK properties and delivered the in vivo rodent POC tool compound, 3.^11,15 Finally, 4 afforded a balance in M₄ PAM activity across species, and was evaluated in depth as our first potential M₄ PAM clinical candidate.¹⁶ Despite these advances, we required an in vivo non-human primate (NHP) POC tool compound for use in translational efficacy studies, primate PD models and biomarker studies that could be freely shared with collaborators. In this Letter, we detail the discovery effort that led to VU0467406, an in vivo M₄ PAM non-human primate (NHP) tool compound.

Figure 1.

Structures of representative M₄ PAMs 1–4, exemplifying an optimized rodent in vivo tool M₄ PAM, VU0467154 (3) and the clinical candidate VU0467485/AZ13713945 (4).

As 4 offered balanced M₄ PAM activity across species, we focused our efforts on this chemotype.¹⁶ A multitude of substituted benzyl amides had previously been explored, and SAR suggested a large lipophilic binding pocket accessible to the 4-position substituent, e.g., the 4-SO₂CF₃ moiety of 3 and related analogs. Interestingly, and despite their prevalence in GPCR ligands, we never synthesized and evaluated any biaryl or heterobiaryl methyl amide congeners; therefore, we focused our search for an NHP in vivo tool within this unexplored chemical space.

The synthesis of analogs 8 proved to be straightforward. Condensation of commercially available 3-chloro-5,6-dimethylpyridazine-4-carbonitrile 5 with methyl thioglycolate under basic conditions smoothly affords the sodium carboxylate 6 in 78% yield. A HATU-mediated amide coupling with 2-, 3- or 4-bromobenzyl amines then delivers analogs 7 in yields ranging from 45–92%. A subsequent Stille or Suzuki cross-coupling reaction with a diverse range of aryl and heteroaryl moieties delivers analogs 8 in yields ranging from 58–90%.¹⁷

SAR was driven using activity at the human M₄ receptor (potency in functional assay), but key compounds were assessed on rat and cyno M₄ as well, as the goal was an NHP tool compound. As shown in Table 1, many potent human M₄ PAMs were discovered and several displayed attractive DMPK profiles and excellent CNS penetration (brain:plasma K_p and K_p,uu) vide infra. Initial evaluation of the 2-, 3- and 4-bromobenzyl derivatives, 7a–c respectively, demonstrated a clear preference for 3- and 4- substitution. Conversion of these into the corresponding biaryl analogs, 8a–c, confirmed the finding, with the 2-phenyl derivative 8a inactive (hM₄ EC₅₀ >10 μM), the 3-phenyl congener 8b weak (hM₄ EC₅₀ = 1.65 μM) while the 4-phenyl derivative 8c displayed good M₄ PAM activity (hM₄ EC₅₀ = 319 nM). Thus, future efforts focused solely on 4-substituted biaryl and heterobiaryl analogs 8d–p. PAM 8c showed low efficacy and poor physiochemical properties and a high cLogP (>4); therefore, we elected to evaluate diverse azaheterocycles in an attempt to improve physiochemical properties, enable salt formation and lower lipophilicity. Exploration of the pyridyl biaryl analogs 8d–f showed robust SAR. Potency increased as the pyridine nitrogen was moved from the 2-position (8d, hM₄ EC₅₀ = 232 nM), to the 3-position (8e, hM₄ EC₅₀ = 103 nM) and finally to the 4-position (8f, hM₄ EC₅₀ = 36 nM). Other heterocycles such as pyridazine (8g), pyrazine (8h), pyrimidine (8i and 8j) were also well tolerated. Attempts to modulate the pyridine basicity in 8d–f by the incorporation of fluorine atoms, as in 8k–p, afforded a broad spectrum of activity, with 8k, 8m and 8p displaying good hM₄ PAM potency and efficacy. From these initial analogs, PAMs 8f, 8k, 8m and 8p were selected for more in-depth DMPK profiling.

Table 1.

Structures and activities for human M₄ PAM analogs 7 and 8.

Cpd	R	hM4 EC50 (nM)a [% ACh Max ± SEM]	hM4 pEC50 (±SEM)
7a		>10,000 [31±2]	>5
7b		185 [84±2]	6.77±0.13
7c		187 [82±2]	6.74±0.07
8a		>10,000 [30±1]	>5
8b		1,650 [41±3]	5.80±0.12
8c		319 [49±6]	6.53±0.12
8d		232 [75±5]	6.67±0.12
8e		103 [63±4]	7.20±0.31
8f		36 [83±8]	7.57±0.19
8g		56 [96±5]	7.26±0.06
8h		159 [63±6]	7.00±0.29
8i		84 [91±9]	7.08±0.04
8j		230 [65±11]	6.67±0.11
8k		58 [92±10]	7.31±0.16
8l		1,716 [86±5]	5.81±0.14
8m		64 [79±2]	7.20±0.06
8n		144 [78±14]	6.84±0.04
8o		994 [68±1]	6.11±0.22
8p		91 [74±3]	7.04±0.11

^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.

As shown in Table 2, all four PAMs showed attractive cLogPs (2.66 to 2.92) and conserved TPSAs with molecular weights below 500. All were highly bound in plasma, except 8f, which also showed low rat clearance in vivo (despite high in vitro predicted rat CL_hep, a likely consequence of excluding fuplasma/fu_mic terms from the well-stirred equation used for prediction) and high rat CNS penetration (brain:plasma K_p = 0.94, K_p,uu = 1.1). However, 8f, as expected due to the naked pyridine ring, was a potent inhibitor of multiple CYP₄₅₀s (IC₅₀s < 10 μM), including 3A4 (IC₅₀ = 340 nM). Similarly, both 8k and 8m were potent inhibitors of 1A2 (IC₅₀s of 2.5 μM and 790 nM, respectively) with low rat in vivo clearance (CL_ps of 3.6 and 1.7 mL/min/kg, respectively) and moderate total brain distribution (K_ps <0.21). Here, 8p stood out, with an improved CYP inhibition profile (all IC₅₀s ≥11 μM) andfavorable rat PK (CL_p = 1.6 mL/min/kg, elim. t_1/2 = 2.7 hr) despite a moderate K_p (0.11), an attractive K_p,uu (1.2). Moreover, 8p was not a human P-gp substrate in vitro (ER = 1.7). However, we noted a lack of an in vitro/in vivo correlation (IVIVC) between the in vitro predicted CL_hep and the in vivo CL_p. Employing a revised prediction method (inclusion of binding terms in the well-stirred model), the predicted rat CL_hep decreased to 0.49 mL/min/kg, and a more robust IVIVC resulted.

Table 2.

In vitro and in vivo DMPK properties of 8f, 8k, 8m and 8p.

Property	8f	8k	8m	8p
MW	398.4	407.4	407.4	407.4
cLogP	2.66	2.92	2.92	2.92
TPSA	93.4	93.8	93.8	93.8
In vitro PK parameters
CLINT (mL/min/kg), rat	415	57.1	81.6	134.0
CLHEP (mL/min/kg), rat	59.9	31.5	37.7	46.0
CLINT (mL/min/kg), human	20.1	19.6	20.2	19.2
CLHEP (mL/min/kg), human	15.2	17.8	17.1	14.8
Rat fuplasma	0.037	0.002	0.004	0.002
Human fuplasma	0.013	0.029	0.023	0.011
Rat fubrain	0.042	ND	ND	0.020
Cytochrome P450 (IC50, μM)
1A2	9.7	2.5	0.79	11
2C9	2.3	25	26	>30
2D6	1.3	>30	>30	19
3A4	0.34	24	25	>30
IV Pharmacokinetics (SD Rat; 0.1–1.0 mg/kg)
CLp (mL/min/kg)	9.0	3.6	1.7	1.6
Elimination t½ (hr)	1.5	1.7	0.48	2.7
Vss (L/kg)	1.1	0.50	0.60	0.34
Brain Distribution (0.25 hr) (SD Rat; 0.1–0.2 mg/kg IV)
Kp, brain:plasma	0.94	0.21	0.13	0.11
Kp,uu, brain:plasm	1.1	ND	ND	1.2
MDCK-MDR1 (Pgp) ER	1.1	ND	ND	1.7

ND = not determined.

While 8p emerged as the most attractive of the analogs of 8 that were surveyed, we wanted to explore modifications to the benzylic phenyl ring before initiating a more exhaustive profiling effort around 8p. While many analogs in which fluorine atoms and/or nitrogen atoms were added retained excellent M₄ PAM potency (EC₅₀s < 200 nM), all suffered from increased cLogP, poor in vitro/in vivo clearance and/or unacceptable CYP inhibition profiles. Thus, 8p was advanced as a potential NHP tool compound.

Next, 8p was evaluated as an M₄ PAM across multiple species (Figure 2). Unlike all predecessors (except 4), 8p was a potent M₄ PAM at the rat (rM₄ EC₅₀ = 13.5 nM, pEC₅₀ = 7.93±0.17, 72±4% ACh Max), human (hM₄ EC₅₀ = 91.0 nM, pEC₅₀ = 7.04±0.11, 74±3% ACh Max), dog (dM 4 EC₅₀ = 111 nM, pEC₅₀ = 6.95±0.12, 49±4% ACh Max) and cyno receptors (cM₄ EC₅₀ = 87.3 nM, pEC₅₀ = 7.06±0.25, 64±2% ACh Max). ACh concentration response curve (CRC) fold-shift Ca²⁺ mobilization and ACh competition binding (with [³H-N-methylscopolamine) assays provided operational model parameters for 8p at rM₄ (K_B = 0.37 μM, α = 8.4, β = 9.1),hM₄ (K_B = 2.8 μM, α = 7.0, β = 21), and cyno M₄ (K_B= 2.1 μM, α=28.2, β=2.5); all experiments are n=3–4, performed in duplicate. Moreover, 8p was inactive (EC₅₀ > 30 μM) at both rat and human M1-3,5 (and also inactive at dog and cyno M₂, data not shown). These data were highly noteworthy and indicated that 8p was a potent and highly-selective NHP M₄ PAM. A broad secondary pharmacology panel (Cerep) revealed only one sub-micromolar off-target activity for 8p – a rat GABA_A receptor (benzodiazepine site) binding IC₅₀ = 0.44 μM, which was subsequently de-risked by a functional assay determination (no activity at 10 μM). A functional electrophysiology hERG assay with 8p was likewise clean (IC₅₀ > 33 μM), and a mini-Ames assay (TA98 and TA100 strains ± S9) was negative for mutagenicity. In addition, 8p was devoid of evidence for CYP 1A2/2B6/3A4 induction potential in cryopreserved human hepatocytes (48 hr incubation with enzyme activity readout; EC₅₀s > 50 μM, E_maxs < 2), and in a 4 day sub-chronic dosing rat study (10 mg/kg, QD PO, n = 2), the day 4 versus day 1 AUC_0-∞ ratio was 0.77, indicating little to no auto-induction in vivo in rat.

Figure 2.

M₄ PAM concentration response curves (CRCs) for 8p (VU0476406) at human (hM₄ EC₅₀ = 91.0 nM, pEC₅₀ = 7.04±0.11, 74±3% ACh Max), rat (rM₄ EC₅₀ = 13.5 nM, pEC₅₀ = 7.93±0.17, 72±4% ACh Max), cyno (cM₄ EC₅₀ = 87.3 nM, pEC₅₀ = 7.06±0.25, 64±2% ACh Max) and dog (dM₄ EC₅₀ = 111 nM, pEC₅₀ = 6.95±0.12, 49±4% ACh Max).

Compound 8p was found to possess a largely attractive PK profile across species. In rat, 8p as a 10 mg/kg solution dose achieved good oral bioavailability (61% F), and the HCl salt of 8p, when dosed as a suspension, achieved 54 %F. PAM 8p likewise exhibited favorable IV and PO PK in dog (CL_p = 5.5 mL/min/kg, elim. t_1/2 = 1.0 hour, V_ss = 0.70 L/kg, and 47 %F from 3 mg/kg suspension dose of the HCl salt) and favorable IV PK in cynomolgus monkey (CL_p = 9.3 mL/min/kg, elim. t_1/2 = 1.3 hours, V_ss = 0.87 L/kg), but with low oral bioavailability (4.7 %F from 10 mg/kg solution dose of HCl salt). CNS penetration of 8p was also assessed in dog (brain:plasma K_p = 0.74, K_p,uu = 2.6, C_csf:C_plasma,u = 4.8 at 2.0 hr; via a terminal study) and NHP (brain:plasma K_p = 0.75, K_p,uu = 1.1, C_csf:C_plasma,u = 1.0 at 1 hr; via a PET study measuring brain distribution of [¹⁸F]-VU0476406 after IV administration),¹⁸ providing strong support for the utility of 8p in NHP behavioral models and biomarker studies.

Encouraged, we performed additional studies to evaluate 8p (VU0476406) as a potential candidate for clinical development before releasing it as a public M₄ PAM NHP tool compound. In vitro metabolite identification experiments employing cryopreserved hepatocytes found low turnover and no evidence for human unique metabolites, with rat and cyno anticipated to provide adequate coverage of human metabolites (Figure 3). Human CYP₄₅₀ phenotyping experiments revealed that multiple CYPs (3A4 (predominant), 2D6, 2C19, 2C9 and 1A2) contribute to 8p’s metabolism. Moreover, no significant levels of GSH conjugates were observed in reactive metabolism/bioactivation experiments with human hepatic microsomes, which further bolstered the potential to advance 8p.

Figure 3.

In vitro biotransformation of 8p (VU0476406) in cryopreserved hepatocytes from multiple species (rat, dog, cynomolgus monkey, and human).

Human PK prediction, utilizing multiple approaches, suggested that 8p would exhibit low clearance in man (CL_p between 1.3 to 3.6 mL/min/kg) with a 6–17 hour half-life.¹⁹ However, for projected 12-hour daily coverage (targeting an efficacious C_min scaled from rat in vivo pharmacodynamic studies; data not shown), 8p was projected to require moderate to high BID oral doses (370–850 mg) due in large part to moderate predicted human oral F. In parallel, pharmaceutical science work on both the free base and HCl salt was performed, which found 8p to be highly crystalline with low aqueous solubility (< 0.5 μM for free base at pH 7.4) and without a clean melt (free base decomposes at 243–246 °C), suggesting potential challenges to achieving requisite margins in nonclinical safety and toxicology studies. An X-ray crystal structure proved telling (Figure 4), highlighting a network of intra- and inter-molecular hydrogen bonds forming a tight, helix-like packing with highly ordered π-stacking. Efforts to disrupt this network to enable acceptable solubility/dissolution with vehicles suitable for IND-enabling safety and toxicology studies were not successful. Thus, based on the suboptimal physiochemical properties of 8p and the high projected human doses, the program team decided to release 8p as a publicly available NHP M₄ PAM tool compound.

Figure 4.

X-ray crystal structure (CCDC 1538487) of 8p. A) Clear intra- and intermolecular hydrogen bonds are present (H-bond between pyridine and amide is linear, H-bond network forms a helix). B) All aryl/heteroaryl rings are oriented to allow π-stacking on both faces. Unit cell dimensions: 22.3 × 16.8 × 5.04 Å. Combined, this crystalline lattice explains the compound’s poor aqueous solubility.

Upon release, Surmeier and co-workers evaluated 8p (VU0476406) in both mouse and NHP models of L-DOPA-induced dyskinesia (LID), and found that administration of the M₄ PAM ameliorates deficits in synaptic plasticity and behavior in PD-LID mice and NHPs.²⁰ Specifically, 8p dosed at 10 mg/kg IV (route chosen due to low cyno oral F, vide supra) significantly reduced dyskinesia scores and involuntary movements in NHPs, thus providing early proof-of-concept for M₄ PAMs in the management of Parkinson’s disease and highlighting the utility of the compound 8p (VU0476406) in NHP studies.

In summary, we have detailed the discovery and characterization of the first reported M₄ PAM non-human primate in vivo tool compound 8p (VU0476406), with similar M₄ PAM activity across species. The studies presented here produced a required in vivo NHP POC tool compound for use in translational efficacy studies, primate PD models and biomarker studies that could be freely shared with collaborators. Initial results demonstrated that 8p significantly reduces dyskinesia scores and involuntary movements in NHPs, and provided early POC for M₄ PAMs in the management of PD-LID. Further optimization efforts en route to M₄ PAM clinical candidates for the treatment of schizophrenia and other disorders will be reported in due course.

X-ray crystallographic data

The X-ray crystal structure data for 8p was submitted to the Cambridge Crystallographic Data Centre (http://www.ccdc.cam.ac.uk) and assigned CCDC 1538487.

Scheme 1. Synthesis of M₄ PAM analogs 8.^a.

^aReagents and conditions: (a) Methyl thioglycolate, MeOH, 1M aq. NaOH, 150 °C, microwave, 30 min, 78%; (b) bromobenzyl amine, HATU, DMF, DIPEA, 2 h, 45–92%; (c) 5 mol% Pd(dppf)Cl₂·CH₂Cl₂, Bu₃SnAr(Het), THF, 65–80% or 5 mol% Pd(dppf)Cl₂·CH₂Cl₂, (HO)₂BAr(Het), K₃PO₄, THF/H₂O, 100 °C, 58–90%.