Discovery of VU0431316: A negative allosteric modulator of mGlu₅ with activity in a mouse model of anxiety

Abstract

Development of SAR in an aryl ether series of mGlu₅ NAMs leading to the identification of pyrazine analog VU0431316 is described in this Letter. VU0431316 is a potent and selective non-competitive antagonist of mGlu₅ that binds at a known allosteric binding site. VU0431316 demonstrates an attractive DMPK profile, including moderate clearance and good bioavailability in rats. Intraperitoneal (IP) dosing of VU0431316 in a mouse marble burying model of anxiety, an assay known to be sensitive to mGlu₅ antagonists and other anxiolytics, produced dose proportional effects.

The metabotropic glutamate receptors (mGlus) comprise a family of eight related G-protein-coupled receptors (GPCRs) wherein each receptor acts through binding glutamate, the major excitatory transmitter in the mammalian central nervous system (CNS). In these seven transmembrane spanning (7TM) receptors, the orthosteric binding sites are located in the extracellular domain while known allosteric binding sites are contained in the transmembrane domain.¹ Design of highly selective orthosteric ligands has continually proven difficult due to the extensive homology of the binding sites across the mGlu family. In many instances, the development of allosteric modulators of mGlus has been established as a viable solution to enhancing selectivity among family members.² Among the individual mGlus investigated as potential drug targets, a substantial portion of that attention has been devoted to the design of small molecule negative allosteric modulators (NAMs), or non-competitive antagonists, of mGlu₅.³

The vast majority of the preclinical behavioral work with mGlu₅ NAMs has been conducted using one of two structurally related disubstituted alkyne tool compounds, MPEP⁴ and MTEP⁵ (Fig. 1). Efficacy has been reported with these compounds across a number of different disease models. Examples include pain,⁶ anxiety,⁷ gastroesophageal reflux disease (GERD),⁸ Parkinson’s disease levodopa induced dyskinesia (PD-LID),⁹ fragile X syndrome (FXS),¹⁰ and other autism spectrum disorders.¹¹ Furthermore, both MPEP and MTEP have been used effectively in various animal models of addictive behavior with well-known drugs of abuse, such as cocaine,¹² nicotine,^12g,13 methamphetamine,¹⁴ morphine,¹⁵ and ethanol.¹⁶

Figure 1.

mGlu₅ NAM tool and advanced clinical compounds

Currently, several mGlu₅ NAMs have progressed to human trials, and results from studies in patients with GERD,¹⁷ FXS,¹⁸ and PD-LID¹⁹ have been encouraging. Though structural diversity among mGlu₅ NAMs in the literature has expanded considerably in recent years, the majority of clinical compounds have been from the disubstituted alkyne structure class.^3a Furthermore, the most advanced clinical compounds, mavoglurant (AFQ056) and basimglurant (RG7090, RO4917523), each contain the alkyne moiety (Fig. 1).²⁰ Recently, concerns that such alkyne compounds might be prone to metabolic activation and resultant toxicities have proven warranted, at least in one instance. Pfizer has now disclosed their observation of biliary epithelial hyperplasia in non-human primate regulatory toxicology studies with the disubstituted alkyne compound known as GRN-529. Glutathione conjugation at the alkyne moiety was believed to be related to these adverse findings.²¹

Our mGlu₅ NAM program has long been centered on the identification and optimization of compounds from chemotypes that do not contain a disubstituted alkyne motif. The majority of this effort has been spent on the optimization of hits identified from a functional cell-based high-throughput screen (HTS) of a collection of 160,000 compounds;²² however, rational design approaches²³ and a virtual screening approach also produced new non-alkyne based mGlu₅ NAM tool compounds.²⁴ We recently reported on a lead optimization effort based around hit compound 1 from our functional HTS (Fig. 2).^22a This particular optimization effort, based on 1, culminated in discovery of the in vivo tool compound VU0409106.

Figure 2.

HTS hit 1 and mGlu₅ NAM in vivo tool VU0409106

Concomitant to the recently described work that led to the discovery of VU0409106, we were also pursuing additional analogs of 1. Reasoning that a potential route of metabolism for analogs of 1 might include amide bond cleavage, we immediately sought to identify compounds that would not produce electron rich anilines should the amide bond indeed be cleaved in vivo. In the case of VU0409106 and associated analogs, we achieved this goal by reversing the orientation of the amide bond; however, the work described herein centers on the replacement of the phenyl core with heteroaryl rings. Preparation of the initial heteroaryl ether analogs of 1 was executed according to one of the two general methods outlined here (Scheme 1).²⁵

Scheme 1. Reagents and conditions: (a) For 2 (Z = N; V = Q = CH), 3 (V = N; Q = Z = CH), 4 (Q = N; V = Z = CH), and 5 (V = Z = N; Q = CH); RCO₂H (R = 3-chlorophenyl), EDC, DMAP, CH₂Cl₂, (83–94%); (b) 3-hydroxypyridine, CuI, Cs₂CO₃, Me₂NCH₂CO₂H·HCl (35–55%); (c) For 14 (W = N; Q = Z = CH; X = F; Y = Br), 15 (Q = W = N; Z = CH; X = Y = Cl), and 16 (Q = Z = N; W = CH; X = Y = Cl); 3-hydroxypyridine, K₂CO₃, DMF, microwave, 150 °C (60–93%); (d) RCONH₂ (R = 3-chlorophenyl), NaO^tBu, Pd(OAc)₂, Xantphos, PhMe, 100 °C (50-78%).

Certain pyridine (10-12) and pyrimidine (13) analogs were prepared by first coupling the commercial amines 2-5 with 3-chlorobenzoic acid to afford the corresponding amides 6-9 (Route I). Reaction with 3-hydroxypyridine in the presence of copper (I) iodide and dimethylglycine afforded the desired compounds 10-13. Alternatively, pyridine 20, pyrazine 21, and pyrimidine 22 were prepared via a route relying on initial installation of the aryl ether (Route II). Reaction of the commercial monomers 14-16 with 3-hydroxypyridine in a microwave assisted nucleophilic aromatic substitution (S_NAr) reaction provided heteroaryl halide intermediates 17-19. The final analogs were prepared directly through a Buchwald-Hartwig coupling with 3-chlorobenzamide in moderate to high yields.²⁶

Evaluation of these initial analogs against mGlu₅ yielded clear SAR (Table 1). Our functional assay measures the ability of the compound to block the mobilization of calcium induced by an EC₈₀ concentration of glutamate in HEK293A cells expressing rat mGlu₅.²⁷ Among the pyridine analogs, compounds 12 and 20 were superior to compounds 10 and 11. In fact both 12 and 20 exhibited potency at a level near hit 1. Pyrimidine analogs 13 and 22 were weak antagonists; however, pyrazine 21 exhibited the best potency in this set of analogs. Having established the pyrazine core as a favorable group for further SAR development, lead optimization continued in that area.

Table 1. Core SAR.

Cpd	Core	mGlu5 pIC50a (± SEM)	mGlu5 IC50 (nM)	% Glu Maxa,b(± SEM)
1		6.55 ± 0.19	284	1.3 ± 0.2
10		6.00 ± 0.21	994	1.8 ± 0.7
11		5.43 ± 0.26	3700	−4.9 ± 4.7
12		6.44 ± 0.19	359	2.0 ± 0.7
13		5.53 ± 0.23	2930	1.1 ± 0.5
20		6.39 ± 0.07	403	1.1 ± 0.4
21		6.89 ± 0.04	129	1.0 ± 0.3
22		<5.0c	>10,000	44 ± 8

^aCalcium mobilization mGlu5 assay; values are average of n ≥ 3

^bAmplitude of response in the presence of 30 μM test compound as a percentage of maximal response (100 μM glutamate); average of n ≥ 3

^cConcentration response curve (CRC) does not plateau

Ongoing research has identified the 5-fluoropyridin-3-yl and pyrimidin-5-yl ethers as optimal groups in the northern portion of the chemotype.^22a Thus, much of the SAR was developed in the context of one or both of these moieties. Compounds containing the pyrimidine moiety are less lipophilic than their 5-fluoropyridine counterparts,²⁸ a feature that can often provide advantages with respect to drug-like properties. Though the synthesis outlined in Scheme I (Route II) was utilized to prepare some new pyrazine analogs, a new synthetic route allowing for the preparation of a broader diversity of amides was utilized in most cases (Scheme 2).²⁹ This route also begins with a similar S_NAr reaction, providing ethers 23-24. A Buchwald-Hartwig coupling with t-butyl carbamate was employed to afford intermediates 25-26.³⁰ Cleavage of the protecting group was carried out under acidic conditions to yield amines 27-28. Conversion to the desired amide products was accomplished using standard coupling conditions.

Scheme 2. Reagents and conditions: (a) 3-fluoro-5-hydroxypyridine or 5-hydroxypyrimidine, K₂CO₃, DMF, microwave, 120 °C (51% for 23; 62% for 24); (b) H₂NCO ^tBu, NaO₂ ^tBu, Pd₂(dba)₃·CHCl₃, ^tBuXPhos, PhMe (46% for 25; 64% for 26); (c) For 25 27, TFA, CH₂Cl₂ (96%); (d) For 26 28, 4N HCl in dioxane (100%); (e) RCO₂H, EDC, DMAP, CH₂Cl₂ or RCO₂H, HATU, DIEA, CH₂Cl₂, DMF or RCO₂H, POCl₃, pyridine or RCOCl, DMAP, CH₂Cl₂ (30–80%).

Evaluation of various substituted benzamides revealed some additional potent compounds (Table 2). As anticipated, both the 5-fluoropyridin-3-yl (35) and pyrimidin-5-yl (29) ethers proved competent replacements for the simple pyridine-3-yl (21) ether. Furthermore, the importance of the 3-chloro substituent on the benzamide was established through preparation of unsubstituted analogs 30 and 36. Many additional 3-substituted analogs were prepared and tested (31-34 and 36-41); however, only the 3-methyl analogs 32 and 38 demonstrated potency comparable to 29 and 35. 3-Cyanobenzamides 33 and 39 were approximately six fold less potent than 3-chlorobenzamide comparators 29 and 35. Additional monosubstituted benzamides demonstrated moderate to weak antagonist activity. Several disubstituted benzamides were also evaluated (42-50), with only 2-fluoro-5-chlorobenzamide 42 demonstrating an IC₅₀ less than one micromolar.

Table 2. Benzamide SAR.

Cpd	A	R	mGlu5pIC50a (± SEM)	mGlu5 IC50 (nM)	% Glu Maxa,b (± SEM)
29	N	3-Cl	6.94 ± 0.10	116	1.2 ± 0.0
30	N	H	<5.0c	>10,000	12 ± 11
31	N	3-F	5.29 ± 0.17	5140	1.5 ± 2.7
32	N	3-Me	6.67 ± 0.16	212	1.5 ± 0.5
33	N	3-CN	6.15 ± 0.17	709	1.5 ± 0.2
34	N	3-OMe	5.39 ± 0.15	4110	−4.3 ± 2.7
35	CF	3-Cl	6.93 ± 0.10	118	2.0 ± 0.3
36	CF	H	5.06 ± 0.09	8650	−9.4 ± 2.3
37	CF	3-F	5.42 ± 0.09	3820	−0.4 ± 1.7
38	CF	3-Me	6.70 ± 0.18	199	2.1 ± 0.4
39	CF	3-CN	6.18 ± 0.16	662	1.2 ± 0.3
40	CF	3-OMe	5.36 ± 0.17	4400	−5.6 ± 5.1
41	CF	3-NMe2	<5.0c	>10,000	52 ± 11
42	CF	2-F, 3-Cl	<4.5	>30,000	—
43	CF	3-Cl, 4-F	5.34 ± 0.15	4590	−2.8 ± 6.2
44	CF	3-Cl, 5-F	<5.0c	>10,000	48 ± 10
45	CF	2-F, 5-Cl	6.05 ± 0.13	894	1.0 ± 0.8
46	CF	3,5-di-F	<4.5	>30,000	—
47	CF	3,5-di-Cl	<4.5	>30,000	—
48	CF	3,5-di-Me	5.18 ± 0.18	6600	6.6 ± 2.0
49	CF	3,5-di-OMe	<4.5	>30,000	—
50	CF	3-CN, 5-F	<4.5	>30,000	—

^aCalcium mobilization mGlu5 assay; values are average of n ≥ 3

^bAmplitude of response in the presence of 30 μM test compound as a percentage of maximal response (100 μM glutamate); average of n ≥ 3

^cCRC does not plateau

Turning our attention from benzamide analogs to analogs with a heteroaryl amide moiety identified a mixture of weak antagonists and compounds that were inactive up to the top concentration tested of 30 μM (Table 3). Recognizing the importance of substitution that was observed in the case of the benzamide analogs, particularly at the 3-position, we decided to prepare some additional analogs of the weak antagonists identified from this initial set (Table 4). Methyl substitution resulted in modest potency enhancement in the case of thiophene 70 relative to 55. More dramatic potency improvement was noted with picolinamides 67 and 75 when compared to 59. Preparation of additional 4-substituted picolinamides (68, 69, and 76) identified 4-chloropicolinamide 68 (VU0431316) as the most potent mGlu₅ NAM in this series.

Table 3. Heteroaryl Amide SAR.

Cpd	A	R	mGlu5 pIC50a (± SEM)	mGlu5 IC50 (nM)	% Glu Maxa,b (± SEM)
51	N		<5.0c	>10,000	20d
52	N		<5.0c	>10,000	15 ± 8
53	N		<5.0c	>10,000	35 ± 6
54	CF		5.2 ± 0.11	6370	7.8 ± 2.8
55	CF		5.5 ± 0.15	3450	0.1 ± 2.4
56	CF		<4.5	>30,000	—
57	CF		<5.0c	>10,000	40 ± 9
58	CF		5.07 ± 0.05	8430	22 ± 11
59	CF		<5.0c	>10,000	40 ± 15
60	CF		<4.5	>30,000	—
61	CF		<4.5	>30,000	—
62	CF		<4.5	>30,000	—
63	CF		<4.5	>30,000	—
64	CF		<4.5	>30,000	—

^aCalcium mobilization mGlu5 assay; values are average of n ≥ 3

^bAmplitude of response in the presence of 30 μM test compound as a percentage of maximal response (100 μM glutamate); average of n ≥ 3

^cCRC does not plateau

^daverage of n = 2

Table 4. Substituted Heteroaryl Amide SAR.

Cpd	A	R	mGlu5 pIC50a (± SEM)	mGlu5 IC50 (nM)	% Glu Maxa,b (± SEM)
65	N		<4.5	>30,000	—
66	N		<4.5	>30,000	—
67	N		6.72 ± 0.22	193	2.1 ± 0.8
68	N		7.20 ± 0.06	62.4	1.6 ± 0.1
69	N		7.00 ± 0.29	100	2.3 ± 0.6
70	CF		6.00 ± 0.04	990	1.1 ± 0.75
71	CF		<5.0c	>10,000	20 ± 10
72	CF		5.33 ± 0.17	4640	19 ± 6
73	CF		<4.5	>30,000	—
74	CF		<4.5	>30,000	—
75	CF		6.51 ± 0.15	311	1.7 ± 0.4
76	CF		6.95 ± 0.29	111	2.0 ± 0.3

^aCalcium mobilization mGlu5 assay; values are average of n ≥ 3

^bAmplitude of response in the presence of 30 μM test compound as a percentage of maximal response (100 μM glutamate); average of n ≥ 3

^cCRC does not plateau

Further in vitro characterization of 68 (VU0431316) was subsequently initiated. Competitive displacement of the established radioligand [³H]3-methoxy-5-(pyridin-2-ylethynyl)pyridine³¹ confirmed the interaction of the compound with a known mGlu₅ allosteric binding site (mGlu₅ K_i = 37 nM (n=1)).³² Evaluation of 68 (VU0431316) in cell based functional assays against the other seven mGlus showed no detectable activity at 10μM.³³ Additionally, the functional activity of 68 (VU0431316) at human mGlu₅ was determined and was essentially identical to that at the rat receptor (human mGlu₅ IC₅₀ = 85 nM (n=1)).³⁴ Finally, 68 (VU0431316) was tested in a commercially available radioligand binding assay panel of 68 clinically relevant GPCRs, ion channels, kinases, and transporters,³⁵ and no significant responses were found at 10 μM compound.³⁶

The pharmacological profile of 68 (VU0431316) warranted further evaluation of its drug-likeness and potential utility as an in vivo tool (Table 5). Evaluation of the compound’s propensity to bind to plasma proteins revealed similar results across multiple species. Nonspecific binding to mouse brain homogenates was also evaluated to enable the estimation of the fraction unbound in the CNS.³⁷ Bidirectional permeability was assessed in both Madin Darby canine kidney (MDCK) and human intestinal epithelial (Caco-2) cells, and permeability was high with no evidence of efflux.³⁸ Determination of the cytochrome P450 (CYP) inhibition profile of 68 (VU0431316) indicated potent inhibition of CYP1A2 with no inhibition of other isoforms up to the top concentration tested of 30 μM.³⁹ As was the case with the related tool compound VU0409106, the common pyrimidine ether moiety resulted in a major non-P450 mediated metabolic pathway for 68 (VU0431316).⁴⁰

Table 5. In vitro DMPK Profile for 68 (VU0431316).

Protein Binding (Fu)a,b				MDCK-WT Permeability
rat PPB		0.179		Papp (A-B)	50.4 × 10−6 cm/sec
cynomolgus PPB		0.167		Papp (B-A)	45.0 × 10−6 cm/sec
human PPB		0.120		efflux ratio	0.89
mouse BHB		0.054
				Caco-2 Permeability
CYP Inhibition IC50 (μM)c				Papp (A-B)	66.8 × 10−6 cm/sec
3A4	2D6	2C9	1A2	Papp (B-A)	71.4 × 10−6 cm/sec
>30	>30	>30	0.6	efflux ratio	1.07

^aFu = fraction unbound

^bPPB = plasma protein binding; BHB = brain homogenate binding

^cAssayed in pooled HLM in the presence of NADPH

The in vitro DMPK profile of 68 (VU0431316) was deemed supportive of in vivo evaluation (Table 6). Pharmacokinetic parameters were calculated from an IV dosing of 68 (VU0431316) to male Sprague-Dawley rats; hepatic clearance was moderate with an approximate three hour half-life. Bioavailability from a single oral (PO) dose was also encouraging, approaching fifty percent. A satellite tissue distribution study was conducted one hour after a 10 mg/kg PO dose. Seventy percent of the compound was detected in plasma relative to the hepatic portal vein, indicative of a low first-pass effect and consistent with the previously observed clearance. CNS penetration was also excellent with a brain to plasma ratio (B/P) of 1.6.⁴¹

Table 6. Rat PK Results for 68 (VU0431316).

IVa,b		POa,c
half-life	193 min	plasma Cmax	1.12 μM
clearance	30.2 mL/min/kg	plasma Tmax	240 min
VSS	2.7 L/kg	AUC	8.37 μM·h
AUC	1.73 μM·h	F	48%

PO tissue distribution studya,c
time	plasma	HPV	brain	B/P ratio
60 min	0.92 μM	1.31 μM	1.45 μM	1.6

^aMale Sprague-Dawley rats (n=2 per time point)

^b1 mg/kg; vehicle = 10% ethanol, 90% PEG400

^c10 mg/kg; vehicle = 10% Tween 80 in 0.5% MC

It is recognized that naïve mice will bury foreign objects such as glass marbles in deep bedding. Pretreating such mice with low doses of anxiolytic benzodiazepines will consistently inhibit this behavior.⁴² Additionally, the well-known and thoroughly characterized mGlu₅ NAMs MPEP and fenobam each inhibit marble burying in this model.^7a,d Furthermore, we have used this model successfully to evaluate several of our own novel mGlu₅ NAM tool compounds.^{22a, 23b, 24} More recent reports have raised the possibility that marble burying reflects a repetitive and perseverative behavior as opposed to novelty-induced anxiety.⁴³ Still, given the convenience and reliability of the marble burying assay, it has served as the frontline behavioral assay for our mGlu₅ NAM discovery program.⁴⁴ Thus, a dose response study with 68 (VU0431316) using a 15 minute pretreatment time following intraperitoneal (IP) administration was carried out in this model (Fig. 3).⁴⁵ Statistically significant inhibition of marble burying was noted at all doses greater than or equal to 10 mg/kg. A satellite tissue distribution experiment in mice at the top dose of 30 mg/kg (IP) showed considerable concentrations of compound in the brain.⁴⁶ Using the results from this study to extrapolate exposures at 10 mg/kg along with the aforementioned brain homogenate binding data indicates that unbound compound concentrations in the brain were likely near the functional IC₅₀ at the minimum effective dose.

Figure 3.

Dose dependent inhibition of marble burying with 68 (VU0431316); n = 7-8 per dose; *, P < 0.05 vs. vehicle control group, Dunnett’s test. Bars denote marbles buried. Vehicle = 10% Tween 80

In conclusion, a second in vivo tool compound has been developed from HTS hit 1. VU0431316 offers an advantage to VU0409106 in that it is orally bioavailable in rats. VU0431316 is a potent and selective mGlu₅ NAM binding to a known allosteric site. CNS exposure in both mice and rats is quite good and efficacy has been established in a proven anxiolytic model. A host of additional behavioral models associated with mGlu₅ are either planned or already underway with VU0431316 and will be the subject of future communications.