Design and Synthesis of mGlu₂ NAMs with Improved Potency and CNS Penetration Based on a Truncated Picolinamide Core

Abstract

Herein, we detail the optimization of the mGlu₂ negative allosteric modulator (NAM), VU6001192, by a reductionist approach to afford a novel, simplified mGlu₂ NAM scaffold. This new chemotype not only affords potent and selective mGlu₂ inhibition, as exemplified by VU6001966 (mGlu₂ IC₅₀ = 78 nM, mGlu₃ IC₅₀ > 30 μM), but also excellent central nervous system (CNS) penetration (K_p = 1.9, K_p,uu = 0.78), a feature devoid in all previously disclosed mGlu₂ NAMs (K_ps ≈ 0.3, K_p,uus ≈ 0.1). Moreover, this series, based on overall properties, represents an exciting lead series for potential mGlu₂ PET tracer development.

The group II metabotropic glutamate receptors, mGlu₂ and mGlu₃, signal through G_i/o to diminish cAMP production by the inhibition of adenylyl cyclase.¹⁻⁸ These presynaptic receptors are widely expressed in the mammalian central nervous system (CNS; cerebral cortex, amygdala, hippocampus, and cerebellum).¹⁻⁸ Utilizing dual mGlu_2/3 orthosteric antagonists and negative allosteric modulators (NAMs), therapeutic relevance has been established in multiple neurodegenerative (chronic pain, Alzheimer’s disease (AD), Parkinson’s disease (PD), drug abuse) and psychiatric (schizophrenia, anxiety, and depression) disorders.⁹⁻¹⁴ However, the contribution due to selective mGlu₂ or mGlu₃ inhibition remains elusive; indeed, for the group II mGlu receptors, the field has only had selective mGlu₂ PAMs^15,16 and, only recently, selective mGlu₃ NAM in vivo tool compounds.¹⁷⁻²¹ Furthermore, the only reported mGlu₂ NAMs, 1 and 2 (Figure 1), are P-gp substrates with limited CNS penetration (K_ps < 0.3), precluding their use as tools to dissect the role of selective mGlu₂ inhibition in vivo or as mGlu₂ PET tracers.^22,23 The development of selective and highly CNS penetrant mGlu₂ NAMs is clearly warranted, as mGlu₂, by inhibiting glutamate release, has been proposed to protect neurons from excitotoxicity in key brain regions.²⁴ For example, mGlu₂ is overexpressed in the hippocampus of Alzheimer’s disease patients relative to age-matched controls.²⁵ In this Letter, we will detail an optimization effort based on 1 and 2 in which we undertake a reductionist approach to develop a new, minimum pharmacophore-based series of potent and selective mGlu₂ NAMs with exceptional CNS penetration (K_ps > 1.5) suitable for use as in vivo probes and as leads for PET tracer development.

Figure 1.

Structures, pharmacology, and rat CNS exposure data for mGlu₂ NAMs 1 and 2 reported in the primary literature.^21,22

Based on a reductionist approach to the optimization of an mGlu₃ NAM 3 to provide 4 (Figure 2) with improved potency and physiochemical and DMPK properties,²¹ we elected to employ a similar strategy for development of a highly CNS penetrant tool compound from mGlu₂ NAMs 1 and 2. Here, we again envisioned truncating the bicyclic core of 1 and 2 to a simple picolinamide scaffold with a suitably tethered western tail moiety, represented generically by 5.

Figure 2.

Structures of mGlu₃ NAM 3 and the optimized mGlu₃ NAM 4 via a reductionist approach to the minimum pharmacophore possessing the desired potency, selectivity, and CNS penetration. A similar strategy for 2 will ideally afford a simplified picolinamide-based mGlu₂ NAM 5 with the desired potency, selectivity, and CNS penetration.²¹

Using 1 and 2 as leads, our goal was to reduce molecular complexity and enhance CNS penetration in a next generation mGlu₂ NAM suitable for in vivo studies and as a lead for PET tracer development to enable occupancy studies. Based on the success with mGlu₃ NAM 4,²¹ we truncated 2 to a simple 4-phenylpicolinamide with a pendant 5-alkoxy linker to a diverse array of aryl and heteoraryl tails. The initial structure–activity relationship (SAR) exploration evaluated the ethylenedioxy linker of 2 while holding the 4-F phenyl moiety of 1 constant. The chemistry required to access analogues 13 required eight steps (Scheme 1).²⁶ Starting from commercial 2,5-difluoro-4-iodopyridine 6, a chemoselective Suzuki coupling with 4-phenylboronic acid affords 7 in 92% yield. Acidic hydrolysis affords the pyridinone 8, which was smoothly converted to the corresponding triflate 9. A microwave-assisted, palladium-catalyzed cyanation reaction delivers 10 in good yield. An S_NAr reaction with 2-(tetrahydro-2H-pyran-2-yloxy)ethanol and deprotection of the THP-ether provides alcohol 11. Finally, a Mitsunobu reaction with various hydroxypyridines and nitrile hydrolysis to the primary carboxamide gives putative mGlu₂ NAM analogues 13.

Scheme 1. Synthesis of Analogues 13.

Reagents and conditions: (a) 4-fluorophenylboronic acid, 10 mol % PdCl₂(dppf), 1 M aq. Na₂CO₃, DME, 100 °C, 92%; (b) AcOH, H₂O, 130 °C, 99%; (c) N-phenyl triflimide, TEA, DCM, DMF, 0 °C, 93%; (d) Zn(CN)₂, Pd(PPh₃)₄, DMF, microwave 140 °C, 15 min, 73%; (e) 2-(tetrahydro-2H-pyran-2-yloxy)ethanol, NaH, DMF, rt; (f) PTSA, DCM, EtOH, rt, 18 h, 65% for the two steps; (g) R¹OH, PPh₃, D^tBAD, THF, rt,18 h, 9–35%; (h) KOSiMe₃, THF, reflux, 85%.

Gratifyingly, this strategy afforded potent mGlu₂ NAMs 13a–d (Table 1), with high selectivity versus mGlu₃ (IC₅₀s > 30 μM). Whereas 1 and 2 possessed cLogPs in the 3.1 to 3.6 range, analogues 13 proved less lipophilic (cLogPs 2.1 to 2.7). Across the board, analogues 13 displayed good fraction unbound in rat plasma (f_us 0.04 to 0.12) but moderate to high predicted hepatic clearance (rat CL_hep = 36 to 64 mL/min/kg) and high brain homogenate binding (f_u = 0.007).²⁶ However, the most potent analogue in this series, 13a (mGlu₂ IC₅₀ = 200 nM), proved to be highly CNS penetrant with a rat brain/plasma partitioning coefficient (K_p) of 2.8 and a K_p,uu of 0.28, representing a ∼10-fold increase over historical mGlu₂ NAMs 1 and 2 (K_ps ≈ 0.2). Despite this exciting advance, we wished to further simplify the chemotype, reduce the number of synthetic steps, and further improve K_p,uu and disposition.

Table 1. Structures and Activities of Analogues 13^a.

^aCalcium mobilization assays with rmGlu₂/TREx/Ga15-HEKcells or rmGlu₃/TREx/Ga15-HEK cells performed in the presence of an EC₈₀ fixed concentration of glutamate; values represent means from three (n = 3) independent experiments performed in triplicate.

Next, we sought to evaluate if the ethylenedioxy linker of 13 could be contracted to a simple benzyloxy linker. If tolerated, this modification would expedite the synthesis of putative mGlu₂ NAMs, while providing a more diverse array of functionality to modulate disposition. The synthesis of analogues 15 began with advanced intermediate 10 (Scheme 2). An S_NAr reaction sequence with diversely functionalized benzyl and heteroaryl methyl alcohols delivered 14. In some cases, due to the presence of adventitious water, 15 was also produced in small amounts. Hydrolysis of nitrile 14 smoothly provided the desired analogues 15 in good overall yield.

Scheme 2. Synthesis of Analogues 15.

Reagents and conditions: (a) (Het)ArCH₂OH, NaH, DMF, rt, 48–99%; (b) KOSiMe₃, THF, reflux, 80–90%.

As shown in Table 2, all analogues 15 were potent mGlu₂ NAMs (IC₅₀s = 78 to 560 nM) with no activity at mGlu₃ (IC₅₀s > 30 μM), demonstrating significant tolerability for diverse substituents. Calculated physiochemical properties were also highly favorable (cLogPs 1.9 to 3.1, TPSAs 75–90 Ų, and molecular weights averaging 372). These analogues 15 displayed good fraction unbound in rat plasma (f_us = 0.04 to 0.20), but moderate to high predicted hepatic clearance (rat CL_hep = 48 to 65 mL/min/kg) and good brain homogenate binding (f_u = 0.046 to 0.72).²⁶ Of these, 15m (VU6001966) emerged as an mGlu₂ NAM (mGlu₂ IC₅₀ = 78 nM, pIC₅₀ = 7.11 ± 0.10 and 1.90 ± 0.39 Glu min) without activity at mGlu₃ (IC₅₀ > 30 μM) worthy of further evaluation. Thus, when evaluated against the full mGlu receptor family, 15m was completely selective versus mGlu_{1,3,4,5,6,7,8} in our standard 10 μM fold-shift assay.²⁶ We also explored broader ancillary pharmacology beyond the mGlus in a Eurofins radioligand binding panel of 68 GPCRs, ion channels, transporters, and nuclear hormone receptors and found no significant activities (no inhibition >50% @ 10 μM).^26,27 Calculated physiochemical properties were also highly favorable (cLogP 1.92, TPSAs 83 Ų, and molecular weight of 326). These properties translated into a soluble compound (kinetic solubility of 14.7 ± 1.1 μM in PBS buffer at pH 7.4, 24 h time point) and favorable disposition. NAM 15m displayed high fraction unbound in plasma (rat f_u = 0.20; human f_u = 0.11) and brain (rat brain f_u = 0.07), but moderate to high predicted hepatic clearance (rat CL_hep = 60 mL/min/kg and human CL_hep = 16.8 mL/min/kg) with an acceptable CYP₄₅₀ profile (3A4, 2D6, and 2C9 IC₅₀s > 30 μM, 1A2 IC₅₀ = 3.1 μM).²⁶ In our standard rat plasma/brain level (PBL) cassette study, 15m demonstrated a rat K_p of 1.9, with a K_p,uu of 0.78, values much improved over 1 and 2, and an improved K_p,uu over 13a.²⁶ A robust in vitro/in vivo correlation (IVIVC) was noted, with a rat in vivo PK study showing high clearance (CL_p = 118 mL/min/kg), a short half-life (t_1/2 = 20 min), and a good volume (V_ss = 2.6 L/kg). Similarly, 15m showed favorable CNS penetration in mouse as well (K_p = 0.65, K_p,uu = 0.29), and similar PK (CL_p = 136 mL/min/kg, t_1/2 = 34 min, and V_ss = 5.2 L/kg). However, while [¹¹C]QCA, 2, was limited to autoradiography due to the low CNS penetration and rodent P-gp liabilities, 15m possesses an ideal profile for a PET tracer–high CNS penetration coupled with rapid clearance from plasma (following an IV route of administration) and desired physiochemical properties.

Table 2. Structures and Activities of Analogues 15^a.

^aCalcium mobilization assays with rmGlu₂/TREx/Ga15-HEKcells or rmGlu₃/TREx/Ga15-HEK cells performed in the presence of an EC₈₀ fixed concentration of glutamate; values represent means from three (n = 3) independent experiments performed in triplicate.

As this series began to position toward a candidate for PET tracer development rather than for robust in vivo tools suitable for proof-of-concept studies, we next evaluated the preferred aryl moiety of 2 (2-fluoro-4-methoxy phenyl) in the context of the simplified picolinamide scaffold and generated analogues 16 (Table 3) following a subtle variation of the route depicted in Scheme 2. Analogues 16 afforded the most potent mGlu₂ NAMs to date within this series. For example, 16c was a potent mGlu₂ NAM (mGlu₂ IC₅₀ = 26 nM, pIC₅₀ = 7.57 ± 0.07, 1.81 ± 0.35 Glu min) as was 16b (mGlu₂ IC₅₀ = 54 nM, pIC₅₀ = 7.27 ± 0.09, 2.17 ± 0.39 Glu min), and both were without activity at mGlu₃ (IC₅₀ > 30 μM). In addition, both showed high fraction unbound in plasma (16b, rat f_u = 0.08, human f_u = 0.10; 16c, rat f_u = 0.06, human f_u = 0.06), but, as with 15m, moderate to high predicted hepatic clearance (16b, rat CL_hep = 61 mL/min/kg and human CL_hep = 16.0 mL/min/kg; 16c, rat CL_hep = 67 mL/min/kg and human CL_hep = 15.8 mL/min/kg). Finally, both demonstrated high CNS penetration (K_ps of 0.34 and 1.38 and K_p,uus of 0.15 and 0.45 for 16b and 16c, respectively; therefore, like 15m, both represent attractive leads as in vivo PET tracers for mGlu₂ and overcome CNS penetration issues associated with 2.

Table 3. Structures and Activities of Analogues 16^a.

^aCalcium mobilization assays with rmGlu₂/TREx/Ga15-HEKcells or rmGlu₃/TREx/Ga15-HEK cells performed in the presence of an EC₈₀ fixed concentration of glutamate; values represent means from three (n = 3) independent experiments performed in triplicate.

Before leaving 15m, we elected to assess a dose-range mouse PBL study via intraperitoneal (IP) dosing to determine if more meaningful exposure could be achieved by bypassing first-pass metabolism. Thus, 15m was dosed IP in male CD-1 mice at 3, 10, and 30 mg/kg (in 10% Tween80/H₂O vehicle), and plasma and brain levels were determined at a 25 min time point. As shown in Figure 3, excellent 15m exposure was achieved in both plasma and brain across this dose-range with consistent K_ps (K_p = 1.05 @ 3 mg/kg; K_p = 1.2 @ 10 mg/kg; K_p = 1.3 @ 30 mg/kg). Total brain concentrations at the highest dose in this study were 14 μM (@30 mg/kg IP), which corresponds to ∼180-fold above the mGlu₂ IC₅₀ of 15m, and 1.1 μM free brain (∼14-fold above the mGlu₂ IC₅₀ of 15m). Even at 10 mg/kg IP, free brain levels were 3-fold above the mGlu₂ IC₅₀ of 15m. Thus, 15m could serve as an excellent tracer candidate with IV dosing and as an in vivo proof of concept for mGlu₂ NAM via IP dosing.

Figure 3.

Exposures of 15m (VU6001966) in plasma and brain from an IP PBL dose range study in male CD-1 mice. (A) Free and total plasma and brain concentrations achieved (3, 10, and 30 mg/kg IP). (B) Plot of plasma and brain levels achieved in the three dose groups (8–10 mice/per group).

In summary, we have developed the next generation of highly selective mGlu₂ NAMs by a reductionist strategy that also provided the first highly CNS penetrant mGlu₂ NAMs in both mice and rats. Excitingly, these new mGlu₂ NAMs possess profiles that render them attractive leads for mGlu₂ PET tracer development via IV dosing paradigms and have utility as in vivo proof of concept compounds via IP dosing. Further refinements and progress toward mGlu₂ NAM in vivo tool compounds and PET tracers are underway, and results will be reported in due course.

Glossary

ABBREVIATIONS

mGlu

metabotropic glutamate receptor

PAM

positive allosteric modulator

NAM

negative allosteric modulator

mGlu₂

metabotropic glutamate receptor subtype 2

PBL

plasma/brain level

K_p

plamsa/brain partitioning coefficient

K_p,uu

unbound plasma/unbound brain partitioning coefficient

IP

intraperitoneal

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.7b00279.

• General methods for the synthesis and characterization of all compounds, methods for the in vitro and in vivo DMPK protocols, and supplemental figures (PDF)

Author Present Address

^⊥ Department of Pharmaceutical Sciences, UNT System College of Pharmacy, University of North Texas Health Science Center, 3500 Camp Bowie Boulevard, Fort Worth, Texas 76107, United States.

Author Contributions

C.W.L. wrote the manuscript and oversaw the medicinal chemistry. K.A.E. designed compounds, and J.L.E., K.A.B., A.S.F., and C.J.B. performed chemical synthesis. P.J.C., A.L.R., H.P.C., and C.M.N. performed and analyzed molecular pharmacology data. R.L.W. performed molecular pharmacology assays. S.C. and A.L.B. oversaw and analyzed in vitro and in vivo DMPK data. C.K.J. and M.B. performed and analyzed the mouse tail suspension assay/data. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.