Discovery of (R)-(2-fluoro-4-((-4-methoxyphenyl)ethynyl)phenyl) (3-hydroxypiperidin-1-yl)methanone (ML337), an mGlu₃ Selective and CNS Penetrant Negative Allosteric Modulator (NAM)

Abstract

A multi-dimensional, iterative parallel synthesis effort identified a series of highly selective mGlu3 NAMs with sub-micromolar potency and good CNS penetration. Of these, ML337 resulted (mGlu₃ IC₅₀ = 593 nM, mGlu₂ IC₅₀ >30 μM) with B:P ratios of 0.92 (mouse) to 0.3 (rat). DMPK profiling and shallow SAR led to the incorporation of deuterium atoms to address a metabolic soft spot, which subsequently lowered both in vitro and in vivo clearance by >50%.

Introduction

G-protein-coupled metabotropic glutamate receptors (mGluRs) have emerged as new drug targets with potential for treatment of a range of CNS disorders.^1-4 Highly subtype-selective allosteric ligands have previously been developed for mGlu₁, mGlu₄, mGlu₅ and mGlu₇.^1-9 While the group II mGluRs (mGlu₂ and mGlu₃) are among the most highly studied of the mGluR subgroups, previous efforts were limited to group II mGluR ligands that act at both mGlu₂ and mGlu₃.^1-4,6 Recently, selective positive allosteric modulators for mGlu₂ have emerged, and demonstrated that mGlu₂ activation is responsible for the antipsychotic efficacy of mGlu_2/3 agonists.⁶ However, despite major advances in understanding the functions of mGlu₂, mGlu₃ remains one of the least understood mGluR subtypes, due in large part to the lack of selective ligands.^1-9 Despite this, numerous studies indicate that mGlu₃ is the key mGluR subtype involved in glialneuronal communication, and inhibition of mGlu₃ is hypothesized to have therapeutic utility in the treatment of cognitive disorders, schizophrenia, depression and Alzheimer's disease.^1-12 Therefore, our laboratory focused attention on the development of selective mGlu₃ negative allosteric modulators (NAMs) as probes to elucidate the role of mGlu₃ in vivo.

To date, only three mGlu₃ NAMs have been reported (Figure 1).^11-13 The first, RO4491533 (1), a dual mGlu₂/mGlu₃ NAM (mGlu₂ IC₅₀ = 296 nM, mGlu₃ IC₅₀ = 270 nM) was efficacious in cognition and depression models.¹¹ About the same time, Lilly disclosed LY2389575 (2), displaying ∼4-fold selectivity for mGlu₃ over mGlu₂ (mGlu₂ IC₅₀ = 17 μM, mGlu₃ IC₅₀ = 4.2 μM).¹² In 2012, we disclosed a potent (IC₅₀ = 649 nM), selective (>15-fold vs. mGlu₂) and CNS-penetrant mGlu₃ NAM (3, ML289), derived from a 0.37 μM mGlu₅ positive allosteric modulator (PAM).¹³ Once again, a subtle ‘molecular switch’,¹⁵ in the form of a p-methoxy moiety, conferred selective mGlu₃ inhibition over mGlu₅ potentiation. While this was a notable advance, we continued to seek an mGlu₃ NAM probe that was devoid of mGlu₂ activity (IC₅₀ >30 μM) in order to enable proof of concept studies.

Figure 1.

Structures and activities of reported mGlu₃ NAMs 1-3.

Results and Discussion

Chemistry

3 became our lead compound from which to develop a more potent and selective mGlu₃ NAM.¹³ As we have previously reported, due to the steep nature of allosteric modulator SAR (especially in series prone to ‘molecular switches’), we pursued an iterative parallel synthesis approach for the chemical optimization of 3, ^2,12,13 which was divided into five quadrants for SAR exploration (Figure 2). First, we wanted to identify replacements for the metabolically labile p-OMe moiety to improve improve disposition.¹³ Second, we hoped to employ the wealth of acetylene replacements from previous mGlu₅ NAM discovery efforts to replace this less than optimal moiety.⁸ Third, we desired to perform a broader amide scan to identify novel amide congeners that eliminate mGlu₂ activity. Finally, we wanted to see if the ‘fluorine walk’ approach² would offer advantages in terms of potency, selectivity or DMPK profiles.

Figure 2.

Library optimization strategy for 3 to improve mGlu₃ NAM activity, eliminate mGlu₂ activity and improve the DMPK profile.

The first libraries were aimed at identifying a replacement for the p-methoxy moiety or electronically perturbing the aryl ring, rendering P450-mediated O-dealkylation less facile.¹³ Following the synthetic route depicted in Scheme 1, a library of 24 analogs was readily prepared via standard amide and Sonogashira couplings, and screened against both mGlu₃ and mGlu₂ in kinetic assays (See supplemental information). All compounds possessed purity exceeding 95% as judged by ¹H NMR and analytical LCMS (214 nM, 254 nM and ELSD). SAR in this region was found to be shallow, as all attempts to increase steric bulk on the ether or electronically deactivate the aromatic ring (Figure 3) led to a significant loss of mGlu₃ activity (IC₅₀s >10 μM); thus the p-methoxy moiety was discovered to be an essential component of the biarylacetylene pharmacophore.

Scheme 1. Synthesis of Aryl Analogues 6^a.

^aReagents and conditions: (a) (R)-3-hydroxymethyl piperidine, EDC, DMAP, DCM, DIPEA, 95%; (b) 20 mol% CuI, 5 mol% Pd(PPh₃)₄, arylacetylene (1.1 equiv.), DMF, DIEA, 60 °C, 1 h, 15-90%.

Figure 3.

Representative Ar moieties surveyed to replace the p-OMe phenyl group. All lost significant activity against mGlu₃ (IC₅₀s >10 μM).

From the literature regarding acetylene replacements in related mGlu₅ NAM biaryl acetylene ligands, we synthesized and screened a diverse array of reported bioisosteres (Figure 4);⁸ unfortunately, only a few weak NAMs were identified, with most inactive (mGlu₃ IC₅₀s >10 μM). Therefore, the p-methoxy phenyl acetylene component was crucial for mGlu₃ activity. Based on these data, we elected to survey alternative amide moieties in an effort to improve mGlu₃ NAM activity and selectivity while holding the p-OMe phenyl acetylene pharmacophore constant. Key acid 7 was readily prepared by Sonogashira coupling as shown in Scheme 1, and amide analogues were prepared in high yield under standard conditions (Scheme 2).^13,15 This library proved far more productive, yielding a number of active analogues, and for the first time, robust SAR and a general lack of activity at mGlu₂ (Table 1).

Figure 4.

Representative acetylene biosiosteres surveyed to replace the p-OMe phenyl acetylene group.⁸ All were weak to inactive on mGlu₃ (IC₅₀s >10 μM).

Scheme 2. Synthesis of Amide Analogues 8^a.

^aReagents and conditions: (a) HNR₁R₂, EDC, DMAP, DIPEA, CH₂Cl₂, rt, 16 h, 70-95%.

Table 1.

Structures and Activities of Analogues 8.

Entry	NR1R2	mGlu3 pIC50*	Glu Min* (%)	mGlu2 IC50 (μM)
8a		5.87 ±0.04	0.4±3.0	>30
8b		5.26±0.05	0.0±3.2	>30
8c		5.77 ±0.04	1.7±3.2	>30
8d		6.18±0.02	2.0±1.5	>30
8f		5.12±0.11	1.6±39.8	>30
8g		4.56±1.61	---	>30
8h		4.99±0.09	0.0±16.1	>30
8i		5.12±0.11	0.0±10.3	>30
8j		5.96±0.06	-0.1 ±4.4	>30
8k		5.56±0.07	1.6±5.9	>30
8l		5.26±0.11	1.7±11.2	>30

^*mGlu₃ pIC₅₀ and Glu Min data reported as averages ±SEM from our calcium mobilization assay; n = 3

A racemic 3-hydroxy piperidine congener (8a) showed significant activity (mGlu₃ IC₅₀ = 760 nM), and upon synthesis of the pure enantiomers, enantioselective inhibition was noted. Here, the (R)-enantiomer (8d) was more potent (mGlu₃ IC₅₀ = 650 nM) than the (S)-enantiomer (8c, mGlu₃ IC₅₀ = 1.1 μM). When the hydroxy group was capped as a methyl ether in 8b, mGlu₃ NAM activity was lost. Interestingly, the [3.3.0] piperidine mimetic was active (8k and 8l), and was a reasonably effective surrogate for the piperidine ring. Contraction to a pyrrolidine ring, as in 8g-i, led to a significant diminution in potency, as did an acyclic congener 8f. Based on the potency of the tertiary hydroxyl analogue 8j (IC₅₀ = 711 nM), we prepared the ethyl and allyl congeners as well, and resolved the enantiomers via chiral SFC.¹⁵ Only modest ∼2-fold increases in mGlu₃ NAM potency were noted for the (+)-enantiomers (Supplemental Figure 1). Finally, following the synthetic routes depicted in scheme 1 and 2, we incorporated fluorine atoms into the benzoic acid moiety of 8d, and discovered two additional sub-micromolar mGlu₃ NAMs 9 and 10 worthy of further profiling (Figure 5).¹⁵

Figure 5.

Potent and selective mGlu₃ NAMs for further profiling.

Molecular Pharmacology

The four leading mGlu₃ NAMs 8d, 8j, 9 and 10 proved to be potent and highly selective versus mGlu₂ (Figure 6). Based on DMPK and ancillary pharmacology profiles (vide infra), 9 was favored for further characterization. As shown in Figure 6C, 9 displayed classical non-competitive antagonism with respect to the orthosteric agonist glutamate in a progressive fold shift assay.^2,3,13,15 For certain electrophysiology studies, an exogenous agonist may be required in order to engender selective group II mGluR activation; we therefore examined the probe dependence of 9, and noted no differences between glutamate and LY379268¹⁶ (Figure 6D). Considering 9 was inactive against the remaining mGluRs (no activity at mGlu_{1,2,4,5,6,7,8} up to 30 μM) we declared ML337 an MLPCN probe.¹⁷

Figure 6.

Molecular pharmacology profile of 9 and related mGlu₃ NAMs. A) mGlu₃ EC₈₀ antagonist CRC. All four compounds are potent and fully efficacious mGlu₃ NAMs (n = 3). B) mGlu₂ EC₈₀ CRC. All four compounds are inactive up to 30 μM. C) Progressive fold shift analysis with 9 and glutamate displayed a non-competitive decrease in the EC₈₀, indicating 9 is acting allosterically. D) Evaluating probe dependence. 9 is equipotent and efficacious in inhibiting mGlu₃ activation by both glutamate and LY379268 (Supplemental Figure 2).

DMPK Disposition Attributes

9 was subsequntly profiled in a battery of in vitro and in vivo DMPK assays to assess its utility as in vivo probe (Table 2). Although 9 was found to be unstable in rat and human microsomes, it possessed free fractions in both mouse and human plasma approaching 0.03 (97% PPB), as well as a favorable P₄₅₀ inhibition profile and solubility (7.8 μM in PBS). In a Ricerca radioligand binding panel of 68 GPCRs, ion channels and transporter,¹⁸ displayed significant activity (>50% inhibition @10 μM) at only 2 targets (DAT, 71% and 5-HT_2B, 74%), but no functional activity at these targets. To rapidly assess the extent of CNS penetration, we performed a mouse tissue distribution study in which 8b, 8j, 9, and 10 were administered as a cassette via an IP route, followed by LC/MS/MS analysis of plasma and brain tissue. All four compounds afforded acceptable CNS exposure, producing brain-to-plasma ratios (B:P) ranging from 0.59 to 0.92 in mice (Supplemental Table 1). 9 demonstrated a B:P ratio approaching unity (B:P, 0.92), with a Brain_AUC of 3.37 μM and a corresponding plasma_AUC of 3.71 μM. A subsequent rat study demonstrated a good overall CNS exposure for 9, producing a B:P ratio of 0.3 with high plasma exposures (Supplemental Table 2).

Table 2. DMPK Characterization of 9.

Parameter	9
MW	353.38
TPSA	59.7
cLogP	3.51
In Vitro Pharmacology	IC50(μM)
CYP (1A2, 2C9, 3A4, 2D6)	>30, >30, >30, >30
In Vitro PK
Rat CLHEP (mL/min/kg)	54.1
Human CLHEP (mL/min/kg)	18.9
Rat PPB (fu)	0.005
mPPB (fu)	0.027
In Vivo Rat PK (IP, 10 mg/kg, 0-6 h)
Plasma AUC0-6 (μM*h)	33.1
Brain AUC0-6 (μM*h)	9.6
Brain:Plasma	0.3

The major metabolite of 9, as with 3, was P450-mediated O-demethylation.¹⁴ As mentioned above, all efforts to replace this group synthetically proved futile, resulting in inactive compounds. In an attempt to improve the PK in rodents, we elected to introduce deuterium atoms into the methoxy substituent (D₃) of both 8d and 9 in order to increase the metabolic stability of these mGlu3 NAMs (providing 11 and 12, respectively).¹⁹ As shown in Table 3, introduction of the D₃CO moitety led to an analog with a substantially lower intrinsic clearance (CL_int) and predicted hepatic clearance value (CL_hep) in vitro. Indeed, the deuteration strategy resulted in an approximate 50% lowering of the plasma clearance (CL_p) in rats while providing mGlu₃ NAMs of comparable potency and selectivity (Supplemental Figure 3). Importantly, identification of the principal metabolites of the deuterated analogs revealed there to be no metabolic shunt from P₄₅₀-mediated O-demethylation (data not shown). Thus, employing the apparent kinetic isotope effect as a means to combat the shallow SAR of these allosteric modulators led to improved disposition in vivo. ¹⁹

Table 3.

Effect of deuterium incorporation on in vitro and in vivo rat PK with 8d and 9.

	R = CH3,8d	R = CD3, 11	R = CH3, 9	R = CD3, 12
Rat CLINT (mL/min/kg)	214	97.3	239	73.7
Rat CLHEP (mL/min/kg)	52.7	40.7	54.1	35.9
Rat lV PK CLp (mL/min/kg)	6.2	3.3	5.2	2.9
Rat IV PK Vss (L/kg)	0.22	0.21	0.21	0.18
mGIU3 IC50 (μM)	0.65	0.31	0.59	0.45

Conclusion

In summary, we have developed the most potent (mGlu₃ IC₅₀ = 593 nM, 1.9% Glu min) and selective (>30 μM versus mGlu_{1,2,4,5,6,7,8}) mGlu₃ NAM, 9, described to date. ML₃₃₇ possesses a favorable DMPK and ancillary pharmacology profile, and is centrally penetrant. The major metabolic soft spot was identified to be P450-mediated O-demethylation, a fate that could not be overcome through standard steric or electronic perturbations, due to extremely shallow allosteric ligand SAR. However, by exploiting apparent kinetic isotope effects, we were able to combat the shallow SAR within this allosteric modulator series and discover an mGlu3 NAM with improved disposition. Electrophysiology and in vivo studies with 9, and its deuterated analogue 12, are in progress and will be reported in due course.

Experimental Section

Chemistry

The general chemistry, experimental information, and syntheses of all other compounds are supplied in the Supporting Information. (R)-(2-Fluoro-4-((4-methoxyphenyl)ethynyl)(3-hydroxypiperdin-1-yl)methanone, 9: To a solution of 2-fluoro-4-((4-methoxyphenyl) ethynyl) benzoic acid (675 mg, 2.5 mmol) in 20 mL DMF, was added DIPEA (1.07 g, 8.25 mmol) while stirring. EDC (560 mg, 3 mmol), HOBt (337 mg, 2.5 mmol), and (R)-3-hydroxypiperidine hydrochloride (342 mg, 2.5 mmol) were then added. The reaction was allowed to stir for 4 hours at room temperature, then quenched with a solution of saturated NaHCO₃ (20 mL), washed with 5% LiCl (aqueous, 2 × 20 mL), and brine (20 mL). The reaction was extracted into dichloromethane (50 mL), and solvent was removed under vacuum. HPLC purification afforded 9 as an ivory solid (420 mg, 47%). ¹H NMR (500 MHz, d6-DMSO, 75° C) δ (ppm): 7.50 (m, 2H); 7.39 (m, 3H); 6.99 (m, 2H); 4.06 (s, 1H); 3.82 (s, 3H); 3.53 (s, 1H); 3.29 (m, 2H); 2.93 (m, 1H); 1.87 (m, 1H); 1.74 (s, 1H); 1.44 (m, 2H). ¹³C NMR (125 MHz, d6-DMSO, 75° C) δ (ppm): 163.3, 159.7, 158.0, 156.0, 132.7, 127.2, 125.2 (d, J = 9.3 Hz), 124.3 (d, J = 16.7 Hz), 117.7 (d, J = 22.7 Hz), 114.2, 113.4, 91.1, 85.9, 64.7, 55.0, 53.2, 48.2, 32.2, 28.9. [α]_D²³ = -27.6° (c = 1, MeOH). LC (254 nm) 0.704 min (>99%); MS (ESI) m/z = 354.1. HRMS (TOF, ES+) C₂₁H₂₀FNO_3.[M+H]⁺ calc. mass 354.1505, found 354.1507.

Abbreviations Used

mGlu₃

metabotropic glutamate receptor subtype 3

CRC

concentration-response-curve

IP

intra-peritoneal

MLPCN

Molecular Libraries Probe Production Centers Network

RCF

relative centrifugal force