N-Acyl-N′-arylpiperazines as negative allosteric modulators of mGlu₁: Identification of VU0469650, a potent and selective tool compound with CNS exposure in rats

Abstract

Development of SAR in an N-acyl-N′-arylpiperazine series of negative allosteric modulators of mGlu₁ using a functional cell-based assay is described in this Letter. Characterization of selected compounds in protein binding assays was used to aid in selecting VU0469650 for further profiling in ancillary pharmacology assays and pharmacokinetic studies. VU0469650 demonstrated an excellent selectivity profile and good exposure in both plasma and brain samples following intraperitoneal dosing in rats.

l-Glutamic acid (glutamate) is the major excitatory neurotransmitter in the mammalian central nervous system (CNS). Glutamate produces its effects through binding to both ionotropic and metabotropic glutamate receptors. The metabotropic glutamate receptors (mGlus) belong to family C of the G protein-coupled receptors (GPCRs). Further classification of the mGlus discovered to date has been according to their structure, preferred signal transduction mechanisms, and pharmacology (Group I: mGlu₁ and mGlu₅; Group II: mGlu₂ and mGlu₃; Group III: mGlu₄, mGlu₆, mGlu₇, and mGlu₈).¹ Many of these receptors have attracted significant attention as promising targets for the treatment of a variety of CNS related disorders. The design of drug-like compounds that selectively target a specific mGlu has been complicated by the fact that the orthosteric binding site of the mGlus is highly conserved. An alternative strategy that has proven successful for overcoming such selectivity hurdles has been the optimization of compounds that modulate the function of the receptor through interaction with an allosteric binding site.²

An area within the mGlu allosteric modulator field that has garnered substantial interest has been the design and application of selective small molecule negative allosteric modulators (NAMs) of mGlu₁.³ Several tool compounds have been discovered during recent years further establishing the link between mGlu₁ antagonism and the potential treatment of CNS related disorders such as addiction,⁴ anxiety,⁵ epilepsy,⁶ pain,^5a,7 and psychotic disorders. ^5a,8 Compounds 1–6 (Fig. 1) are representative of the mGlu₁ NAM chemotypes that have been discovered to date and examples of molecules with demonstrated efficacy in vivo. For example, JNJ16259685 (1) was efficacious in a rat model of anxiety⁹ as well as rat¹⁰ and squirrel monkey¹¹ models of addiction. A-841720 (2) was established as an analgesic in rat models of neuropathic pain¹² and post-operative pain.¹³ Inhibition of nociceptive pain was observed with 3 in rats¹⁴ and with YM298198 (4) in mice.¹⁵ Antipsychotic activity as measured by acoustic prepulse inhibition models was noted with 5¹⁶ and 6¹⁷ in rats and with 1⁸ and 4⁸ in mice. Also of interest, recent research has identified a potential role for mGlu₁ inhibition in the treatment melanoma¹⁸ and certain types of breast cancer.¹⁹

Figure 1.

mGlu₁ NAM compounds with in vivo efficacy.

We have recently become interested in the discovery and development of structurally novel mGlu₁ NAMs with properties suitable for evaluation in rodent models of a variety of CNS disorders. A potential starting point for a hit to lead optimization program focused on mGlu₁ NAMs was identified through cross screening during the course of our recent work in the mGlu₅ NAM field (Fig. 2).²⁰ The adamantyl carboxamides 7 and 8 were mGlu₅ NAMs that were also found to behave as mGlu₁ NAMs in a functional cell based assay.²¹ Our mGlu₁ assay measures the ability of the compound to block the mobilization of calcium by an EC₈₀ concentration of glutamate in cells expressing human mGlu₁.

Figure 2.

mGlu₁ NAM starting points for hit to lead optimization.

One attractive feature of this scaffold was the fact that robust chemistry would allow to us to rapidly vary three distinct portions of the chemotype and evaluate considerable synthetic diversity (Scheme 1). For instance, commercially available and enantiopure piperazines 9 (R¹ = CH₃) were useful for evaluating substitution of the piperazine core. For the purpose of evaluating SAR about the aryl portion of the chemotype (R²), substituted piperazines 9 were reacted with adamantoyl chloride to afford amide intermediates 10. Removal of the carbamate protecting group under acidic conditions gave amine intermediates 11. Target compounds were accessible from amines 11 through nucleophilic aromatic substitution reactions with aryl fluorides or Buchwald–Hartwig²² amination reactions with suitable aryl halides.²³ For evaluation of SAR around the amide portion of the scaffold, a simple reorganization of the synthetic transformations was required. Reaction of amines 9 with either 2-fluoropyridine or 2-bromopyridine under appropriate conditions provided intermediate 12. Following cleavage of the protecting group, amines 13 were readily converted to the target amide compounds using well established methods.²⁴

Scheme 1.

Reagents and conditions: (a) 1-adamantoyl chloride, DIEA, CH₂Cl₂; (b) HCl, MeOH, dioxanes; (c) R²F, NMP, μ wave, 250 °C, 10 min; (d) R²X (X = Cl, Br, or I), Pd₂(dba)₃ or Pd(OAc)₂, Xantphos, NaO^tBu or Cs₂CO₃, dioxanes, μ wave, 120 °C, 10 min or 100 °C, 18 h; (e) R³COCl, DIEA, CH₂Cl₂; (f) R³CO₂H, HATU, DIEA, CH₂Cl₂.

Evaluation of methyl substituted piperazines was one of our first areas of investigation (Table 1). New compounds were prepared with the 2-pyridyl group found in hit 7 (14–17). Evaluation of these enantiopure analogs yielded some interesting results. A preference with regard to the mGlu₁ NAM activity was observed for one enantiomer versus the other in the case of both the 2-methyl and 3-methyl analogs. With the 2-methyl analogs a 15- fold preference was observed for the R-enantiomer (15 vs 14), while with the 3-methyl analogs a sixfold preference was observed for the R-enantiomer (17 vs 16). Furthermore, both 15 and 17 were approximately fourfold more potent than the unsubstituted comparator 7. Moreover, an equally substantial discovery was noted upon testing of 15 and 17 in our functional mGlu₅ assay.²⁵ Both compounds were only weak antagonists in this assay (IC₅₀ >10 ¼M***). In short order, we had moved from a hit compound with little to no group I mGlu selectivity to two compounds with more than 60-fold selectivity for mGlu₁ versus mGlu₅.

Table 1.

Chiral piperazine analogs

Compd	R1	R2	mGlu1 pIC50a (±SEM)	mGlu1 IC50 (nM)	% Glu maxa,b (±SEM)
7	H	H	6.20 ± 0.21	630	1.1 ± 0.8
14	(S)-CH3	H	5.62 ± 0.18	2420	−1.4 ± 2.5
15	(R)-CH3	H	6.79 ± 0.15	161	2.5 ± 0.5
16	H	(S)-CH3	6.01 ± 0.13	988	1.8 ± 1.0
17	H	(R)-CH3	6.78 ± 0.14	166	2.0 ± 0.4

^aCalcium mobilization mGlu₁ 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.

We next chose to evaluate SAR around the amide portion of the chemotype in the context of both the (R)-2-methyl piperazine (series I) and the (R)-3-methyl piperazine (series II) cores (Table 2). A number of analogs (18–35) were prepared in an attempt to identify replacements for the 1-adamantyl group. This was not generally successful; however, there were some notable exceptions. In the context of the (R)-2-methyl piperazine but not the (R)-3-methyl piperazine, the cyclooctyl (26) and the trans-(tert-butyl)cyclohexyl (30) groups demonstrated weak to moderate NAM activity against mGlu₁. Additionally, the bicyclo[3.3.1]nonan-3-yl (32) group was essentially equipotent to 15, though this was again not effective in the case of the (R)-3-methyl piperazine (33 vs 17). It has been reported that the cubyl group can function as an effective and less lipophilic isosteric replacement for the adamantyl group.²⁶ Unfortunately, such a modification was not effective in our chemotype (34 and 35).

Table 2.

Amide SAR

Compd	Series	R	mGlu1 pIC50a (±SEM)	mGlu1 IC50 (nM)	% Glu maxa,b (±SEM)
15	I		6.79 ± 0.15	161	2.5 ± 0.5
17	II		6.78 ± 0.14	166	2.0 ± 0.4
18	I		<4.5	>30,000	—
19	II		<4.5	>30,000	—
20	I		<4.5	>30,000	—
21	II		<4.5	>30,000	—
22	I		<5.0c	>10,000	51.8 ± 3.3
23	II		<4.5	>30,000	—
24	I		<5.0c	>10,000	29.9 ± 2.1
25	II		<5.0c	>10,000	62.6 ± 3.3
26	I		5.18 ± 0.06	6630	0.0 ± 8.2
27	II		<5.0c	>10,000	52.9 ± 3.3
28	I		<5.0c	>10,000	32.1 ± 3.1
29	II		<4.5	>30,000	—
30	I		5.44 ± 0.08	3610	0.3 ± 0.4
31	II		<5.0c	>10,000	44.1 ± 3.7
32	I		6.78 ± 0.11	167	1.3 ± 0.2
33	II		<5.0c	>10,000	12.2 ± 10.6
34	I		<5.0c	>10,000	62.7 ± 9.6
35	II		<4.5	>30,000	—

^aCalcium mobilization mGlu₁ 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.

Our attention next moved from the amide to the aryl portion of the template (Table 3). Using an approach similar to that previously described, we again observed only weak antagonists (37, 41, 43, 45) and a compound (39) with no activity at the top concentration tested (30 µM) in the context of the (R)-3-methyl piperazine. Fortunately, the same was not observed with the (R)-2- methyl piperazine analogs. Though 4-pyridyl analog 38 demonstrated only weak mGlu₁ NAM activity, 3-pyridyl analog 36 exhibited good potency. Good to moderate potency was also noted with pyrimidines 40 and 44 and pyrazine 42. Finally, the 2-cyanophenyl analog 46 had potency within twofold of that observed with 15 and represented an improvement over the unsubstituted comparator piperazine 8 (Fig. 2).

Table 3.

First generation aryl SAR

Compd	Series	R	mGlu1 pIC50a (±SEM)	mGlu1 IC50 (nM)	% Glu maxa,b (±SEM)
15	I		6.79 ± 0.15	161	2.5 ± 0.5
17	II		6.78 ± 0.14	166	2.0 ± 0.4
36	I		6.22 ± 0.06	596	1.8 ± 0.2
37	II		<5.0c	>10,000	32.3 ± 5.2
38	I		<5.0c	>10,000	49.2 ± 4.5
39	II		<4.5	>30,000	—
40	I		6.15 ± 0.07	714	0.9 ± 0.4
41	II		<5.0	>10,000	17.0 ± 5.7
42	I		5.93 ± 0.08	1190	−0.3 ± 0.8
43	II		<5.0	>10,000	34.0 ± 5.1
44	I		5.88 ± 0.07	1330	−1.0 ± 2.2
45	II		<5.0	>10,000	34.7 ± 7.2
46	I		6.58 ± 0.06	264	2.1 ± 0.3
47	II		5.61 ± 0.09	2460	5.3 ± 1.7

^aCalcium mobilization mGlu₁ 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.

Having seen little progress in identifying improved analogs with the (R)-3-methyl piperazine core, we chose to focus our remaining efforts on attempting to improve upon the results observed with (R)-2-methyl piperazine analogs 15 and 46 (Table 4). Knowing that very subtle structural modifications can dramatically affect activity with allosteric modulators, we prepared fluorinated analogs of 15 (48–50). Though all three compounds exhibited good potency, 6-fluoropyridine 48 was the most potent and only slightly less potent than 15.We also prepared heteroaryl analogs (51–54) of 46 as a means to reduce lipophilicity. Most of these derivatives demonstrated good (51 and 53) to moderate (52) potency; however, analog 54 represented a nearly threefold improvement relative to 46 and the most potent compound from within the chemotype identified to date.

Table 4.

Second generation aryl SAR

Compd	R	mGlu1 pIC50a (±SEM)	mGlu1 IC50 (nM)	% Glu maxa,b (±SEM)
15		6.79 ± 0.15	161	2.5 ± 0.5
48		6.59 ± 0.11	255	2.0 ± 0.4
49		6.38 ± 0.07	414	1.3 ± 0.9
50		6.15 ± 0.05	712	2.4 ± 0.7
46		6.58 ± 0.06	264	2.1 ± 0.3
51		6.46 ± 0.14	344	3.3 ± 1.0
52		5.86 ± 0.04	1400	−0.8 ± 3.3
53		6.40 ± 0.06	398	0.3 ± 2.2
54		7.01 ± 0.03	99	2.8 ± 0.3

^aCalcium mobilization mGlu₁ 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.

Calculation of the ligand-lipophilicity efficiency (LLE) values for our most potent analogs identified analog 54 as the compound with the best balance of potency and lipophilicity, two factors linked to drug-likeness (Table 5).²⁷ We also decided to examine these same compounds for their propensity to non-specifically bind to rat plasma proteins.²⁸ Binding to proteins can limit the amount of drug available to interact with the target, and thus impact efficacy in an in vivo setting. Since we are primarily interested in CNS applications of an mGlu₁ NAM, we also chose to further profile interesting compounds by measuring their binding to rat brain homogenates. Not surprisingly, the measured fraction unbound tracked well with the calculated logP values. In the instances where brain homogenate binding was measured, there was reduced fraction unbound in brain homogenates relative to plasma. Our experience is that such a relationship between the unbound fraction in brain homogenates versus plasma is often but not always the case. Gratifyingly, the most potent analog 54 also demonstrated an increased fraction unbound compared to most of the other compounds tested.

Table 5.

Protein binding results

Compd	cLogPa	mGlu1 IC50 (nM)	LLEb	Rat PPBc (Fu)	Rat BHBc (Fu)
15	4.49	161	2.30	0.016	0.011
17	4.49	166	2.29	0.017	0.008
46	4.76	264	1.82	0.003	—
48	4.74	255	1.85	0.006	—
49	4.65	414	1.73	0.012	—
51	4.22	344	2.24	0.016	—
53	3.51	398	2.89	0.081	—
54	3.60	99	3.41	0.071	0.016

^aCalculated using ADRIANA. Code (www.molecular-networks.com).

^bLLE (ligand-lipophilicity efficiency) = pIC₅₀—cLogP.

^cF_u = fraction unbound.

Compound 54 (VU0469650) was also examined in cell based functional assays for its selectivity against the other members of the mGlu family and was determined to be inactive against mGlu_2–8 at 10 µM.²⁹ Additionally, we tested the compound in a functional cell based assay against rat mGlu₁ and found little difference across species.³⁰ With over 100-fold selectivity against the other mGlus, we also wanted to assess the selectivity of the compound against a wider variety of targets. As such, we submitted the compound to a commercially available radioligand binding assay panel of 68 clinically relevant GPCRs, ion channels, kinases, and transporters,³¹ and no significant responses were found at a concentration of 10 µM.³²

Given the results summarized herein, we chose to evaluate VU0469650 in a rat pharmacokinetic study (Table 6).³³ In order to understand the rate of metabolism in vivo, a study was conducted using intravenous (IV) dosing. VU0469650 exhibited moderate to high clearance, a moderate volume of distribution, and a half-life of approximately 1 hour. We reasoned that such a clearance profile was supportive of further studies and proceeded to evaluate the compound in a plasma-brain level experiment using intraperitoneal (IP) dosing. Given the moderate to high clearance observed in the IV experiment, the IP dosing route was chosen in an attempt to maximize exposure. The 30 min post-dose time point was selected as it would be a relevant time point for many behavioral experiments of interest. We were pleased to see that the CNS penetration for VU0469650 was excellent and that the absolute levels in the brain were quite good. Taking into account the aforementioned brain homogenate binding data, we estimate the unbound fraction in the brain to be 32 nM. If that is indeed the case, a threefold increase in dose (30 mg/kg) should provide unbound drug levels near the functional IC₅₀ for mGlu₁.

Table 6.

Rat PK results for 54 (VU0469650)

IV PKa,c		IP PKb,c
Dose	0.2 mg/kg	Dose	10 mg/kg
Half-life	55 min	Time point	30 min
MRT	43 min	Plasma concd	509 ng/mL
CL	52 mL/min/kg	Brain concd	727 ng/mL
VSS	2.2 L/kg	B/P ratio	1.4

^aFormulation = 10% EtOH, 40% PEG400, 50% DMSO.

^bFormulation = 10% Tween80 in water.

^cMale Sprague–Dawley rats (n = 2).

In conclusion, we have identified a new mGlu₁ NAM tool compound from within a series of N-acyl-N′-arylpiperazines using a hit to lead optimization plan initiated from a cross screening hit. VU0469650 can be prepared using a simple three step procedure and readily available starting materials. The compound is also potent and selective with excellent CNS penetration in rats. Future plans include evaluation of the compound in rodent behavioral assays associated with CNS disorders linked to mGlu₁ signaling. Results and observations from such studies will be the subject of future communications.