Discovery of [¹¹C]MK-8056: A Selective PET Imaging Agent for the Study of mGluR₂ Negative Allosteric Modulators

Abstract

Modification of potent, selective metabotropic glutamate receptor 2 negative allosteric modulator (mGluR₂ NAM) led to a series of analogues with excellent binding affinity, lipophilicity, and suitable physicochemical properties for a PET tracer with convenient chemical handles for incorporation of a ¹¹C or ¹⁸F radiolabel. [¹¹C]MK-8056 was synthesized and evaluated in vivo and demonstrated appropriate affinity, selectivity, and physicochemical properties to be used as a positron emission tomography tracer for mGluR₂.

Glutamate is a major neurotransmitter in the mammalian central nervous system (CNS). Glutamate receptors are classified as either ionotropic (iGlu) or metabotropic (mGlu) and are involved in numerous cellular functions. Metabotropic glutamate receptors are G-protein coupled receptors and are divided into three subgroups: group I receptors (mGluR₁ and mGluR₅), which activate phospholipase C and cause release of calcium, group II (mGluR₂ and mGluR₃), and group III (mGluRs 4, 6, 7 and 8), which inhibit adenylyl cyclase.¹ Metabotropic glutamate receptor 2 (mGluR₂) is predominantly expressed presynaptically and is activated by glutamate resulting in inhibition of further glutamate release.² Negative allosteric modulators (NAMs) of group II receptors have been pursued, as they offer the potential for improved selectivity over other mGlu subgroups as compared to orthosteric antagonists.³ While mGlu receptors are widely expressed in the CNS, mGluR₂ is expressed mainly within the cerebral cortex and the olfactory bulb regions of the brain.⁴ It has been reported that mGluR₂ is involved in numerous brain functions and is a potential target for the treatment of a number of neurological disorders including Alzheimer’s Disease cognition⁵ and depression.⁶⁻⁸ Moreover, antagonists and negative allosteric modulators (NAMs) of mGluR₂ have shown activity in preclinical assays measuring cognition and antidepressant-like effects.⁹⁻¹¹

An mGluR₂ NAM positron emission tomography (PET) tracer would offer a noninvasive in vivo imaging technique that could enable visualization and quantification of mGluR₂ receptors under normal and disease-state conditions as well as provide evidence of target engagement during clinical studies of therapeutic candidates. Preclinically, the PET tracer could also be used to establish rodent and nonhuman primate pharmacokinetics/occupancy/efficacy relationships. An ideal PET tracer should provide high affinity and selectivity over other CNS receptors (>30- to 100-fold) with minimal nonspecific binding. CNS PET tracers have well-defined requirements for their physicochemical properties; this includes high passive permeability (P_app > 20 × 10^–6 cm/s), low susceptibility for P-gp (BA/AB ratio < 3),¹² and moderate lipophilicity (logD range of 2 to 3.5).¹³⁻¹⁵ Structurally, a CNS PET tracer needs to contain a moiety amenable for incorporation of a positron-emitting isotope, preferably ¹¹C or ¹⁸F. Given the short half-lives of PET radionuclides (20 min for ¹¹C and 110 min for ¹⁸F), late-stage incorporation of the isotope via a robust, high-yielding reaction is required to enable rapid synthesis and purification of the PET tracer. In addition, the synthetic route needs to be readily translatable to a clinical setting for GMP synthesis.

In recent years, there has been growing need for an mGluR₂ NAM PET tracer to support the increasing efforts to develop new therapeutics.^16,17 Previously, we disclosed the discovery of novel mGluR₂ NAMs, exemplified by compound 1 (Figure 1).¹⁸ Compound 1 is a potent mGluR₂ NAM with a high selectivity over the other mGlu receptor subtypes. The affinity of compound 1 for mGluR₂ was determined with the development of a competitive binding assay (mGluR₂ BIND assay, see Supporting Information), and compound 1 had a K_i of 1.8 nM. While compound 1 is a substrate for rodent P-gp transport (P-gp (h/r/m) = 1.6/7/9.8), it has high passive permeability (P_app (h/r) = 29/31 × 10^–6 cm/s) and moderate lipophilicity (HPLC logD_7.0 = 2.24), making it a reasonable starting point to design a CNS PET tracer for the in vivo imaging of the mGluR₂ binding site. Recently, compound 1 was modified by Liang and co-workers to yield [¹¹C]QCA.¹⁹ This compound showed potential as a mGluR₂ PET tracer based on elevated in vitro binding in rat cerebellum, cortex, and striatum but had limited brain uptake. This was attributed to the compound likely being a P-gp/Bcrp substrate in rodents. Building on previous work,²⁰⁻²² recent efforts²³⁻²⁶ have led to PET tracer MG2N001, a tetrahydro-1,7-naphthyridine analogue, which demonstrated the ability to distribute at high levels into nonhuman primate brains. In this paper, we describe further SAR studies in the quinoline carboxamide series that led to the discovery of mGluR₂ NAM PET tracer [¹¹C]MK-8056.

Figure 1.

In vitro profile of compound 1.

The initial strategy for this mGluR₂ NAM PET tracer effort was to improve CNS permeability by reducing transporter susceptibility across species while improving binding affinity and maintaining subtype selectivity and appropriate physicochemical properties. We hypothesized that the suboptimal brain levels achievable with compound 1 are related to P-gp susceptibility of the compound, which in turn might be attributed to the polar nature of the succinimide group. Hence, our design strategy was to replace the succinyl group with basic heterocycles to maintain physicochemical properties while also reducing P-gp susceptibility. To that end, six-membered nitrogen-containing heterocycles were explored as succinimide replacements, and the corresponding data is summarized in Table 1. The 3-pyridyl and 4-pyridyl analogues (2 and 3) resulted in moderate loss of binding affinity (K_i = 6.7 and 7.0 nM, respectively) while retaining desirable lipophilicity. Moreover, compound 2 showed a significant reduction of transporter susceptibility across species as shown by P-gp transport ratios in human and rat (h/r = 0.6/0.8). The 3,5-pyrimidine analogue 4 resulted in a significant potency loss in binding affinity (40 nM). Concurrent to this effort was an examination of the linker between the quinoline and the appended heterocycle. Interestingly, homologating the linker length by one carbon resulted in improved binding affinity for 3-pyridyl analogue 6 (K_i = 1.1 nM) and 3,5-pyrimidyl analogue 8 (K_i = 1.4 nM), while 2-pyridyl analogue 5 and 4-pyridyl analogue 7 displayed decreased binding affinity (K_i = 15 nM, 19 nM). The series of homologated analogues maintained the improved P-gp transport ratios in humans and rat. With the promising increase in binding affinity observed for the ethyl-linked analogues 6 and 8, we were interested in further exploring the SAR of this subseries.

Table 1. SAR of Succinimide Replacements.

Compound	BIND Ki (nM)a	P-gp ratio h/rb	Papp (10–6 cm/s) h/rc	logD7.0d
1	1.8	1.6/7.0	29/31	2.24
2	6.7	0.6/0.8	24/24	2.62
3	7.0	nd	nd	2.70
4	40	nd	nd	3.00
5	15	nd	nd	3.00
6	1.1	0.5/1.0	21/22	2.85
7	19	0.7/1.1	21/24	2.88
8	1.4	0.5/1.0	30/28	2.31

^aAverage value of two runs.

^bP-gp transport ratio BA/AB determined using LLC-MDR1 cells.

^cPermeability determined using control LLC-PK1 cells.

^dHPLC logD at pH 7.0.

Initially, we investigated substitutions on the 3-pyridyl analogue 6, focusing on groups that contained moieties that could be readily radiolabeled. The data are summarized in Table 2 (compounds 9–13). Methyl or fluoro substitution at the 4-position of the pyridine ring was examined and showed modest improvement in K_i (9, K_i = 1.1 nM; 10, K_i = 0.96 nM). Addition of a 4-methoxy group (compound 11) resulted in a decreased binding affinity (K_i = 13 nM). Additional substitution on the pyridine, including the 4-cyano and 4-fluoromethyl substituents, also displayed an affinity similar to that of the 4-methyl pyridine analogue 9. Unfortunately, substitution of the pyridine ring caused an increase in lipophilicity that was above the desired threshold suitable for a PET tracer. To reduce lipophilicity, we shifted focus to the 3,5-pyrimidyl analogues. Methyl or fluoro substitution at the 4-position of the pyrimidine ring was initially examined. Excitingly, the analogues showed improvement in K_i (14, K_i = 0.80 nM; 15, K_i = 0.78 nM) while maintaining appropriate lipophilicity with logD_7.0 values < 3. Analogous to the pyridine example, addition of a 4-methoxy group diminished binding affinity (16, K_i = 4.5 nM). Addition of a 4-cyano group (17, K_i = 0.32 nM) provided excellent affinity but did not maintain the desirable lipophilicity. The 4-fluoromethyl analogue (18, K_i = 1.5 nM) was equipotent to analogue 8. Overall, substituted pyrimidine analogues showed the most promise, notably compound 14, as it increased the binding affinity as compared to compound 8 while retaining the desirable physicochemical properties needed for a CNS PET tracer.

Table 2. SAR of Ethyl-Linked Heterocycles.

Compound	BIND Ki (nM)a	P-gp ratio h/rb	Papp (10–6 cm/s) h/rc	logD7.0d
9	1.1	nd	nd	3.23
10	0.96	nd	nd	3.39
11	13	nd	nd	3.75
12	1.0	0.6/0.8	21/18	3.28
13	0.70	nd	nd	3.34
14	0.80	0.5/0.7	31/30	2.52
15	0.78	0.5/0.5	27/26	2.99
16	4.5	0.5/0.8	27/23	2.85
17	0.32	0.4/1.0	18/19	3.24
18	1.5	0.6/1.4	28/25	2.70

^aAverage value of two runs.

^bP-gp transport ratio BA/AB determined using LLC-MDR1 cells.

^cPermeability determined using control LLC-PK1 cells.

^dHPLC logD at pH 7.0.

Next, we examined the C4 appended aryl group to identify substitution position and functionality for the introduction of a radiolabel while maintaining the desirable 4-methyl pyrimidine side chain of compound 14. The data are summarized in Table 3. Replacement of the 4-fluoro phenyl moiety with either 4-methoxy phenyl (19) or 4-fluoromethyl phenyl (20) resulted in a minimal loss in binding affinity. Additional substitution on the aryl ring, including the 2-fluoro-4-methyl phenyl analogue 21, maintained binding affinity (K_i = 0.74 nM). Gratifyingly, addition of fluoro substituents to compound 19 resulted in the most potent analogues in this series (compound 22, K_i = 0.19 nM, compound 23, K_i = 0.46 nM), with the added advantage of having a group readily amenable for the incorporation of a ¹¹C radio tracer. Compound 22, containing the 2-fluoro-4-methoxy phenyl group, was a high-affinity mGluR₂ NAM (K_i = 0.19 nM) with desirable lipophilicity (logD_7.0 = 2.46) as well as low P-gp susceptibility in vitro and therefore emerged as the lead candidate for the PET tracer program.

Table 3. SAR of C4 Aryl Substituents.

Compound	BIND Ki (nM)a	P-gp ratio h/rb	Papp (10–6 cm/s) h/rc	logD7.0d
19	2.6	0.6/0.8	27/25	2.30
20	1.7	nd	nd	2.36
21	0.74	0.2/0.5	36/30	2.99
22	0.19	0.5/1.2	27/26	2.46
23	0.46	nd	nd	2.42

^aAverage value of two runs.

^bP-gp transport ratio BA/AB determined using LLC-MDR1 cells.

^cPermeability determined using control LLC-PK1 cells.

^dHPLC logD at pH 7.0.

To determine the CNS selectivity of compound 22 for mGluR₂, further profiling of compound 22 was performed, and the data is detailed in Table 4. Compound 22 was found to be >1000-fold selective against all tested mGlu receptors including the group II mGluR₃ receptor (IC₅₀ > 19 μM). Off-target screening against a panel of 119 receptors and enzymes by Ricerca Biosciences, LLC (Concord, Ohio USA) including a wide range of neuroscience targets revealed that compound 22 was >2800-fold selective for mGluR₂. Additionally, compound 22 was weakly active against the adenosine transporter (K_i = 2.26 μM) and the GABA_A receptors (K_i > 2.9 μM). Affinities in this range are expected to neither interfere with imaging mGluR₂ using labeled compound 22 at tracer levels nor pose a safety concern. Additional screening in ion channel assays was also performed and revealed IC₅₀ > 5 μM against Ca_v 1.2 (IC₅₀ > 5 μM), Na_v 1.5 (IC₅₀ > 30 μM), and Ik_r (IC₅₀ > 28 μM) ion channels, which demonstrated low cardiovascular safety risk for compound 22 at tracer concentrations.

Table 4. In Vitro Profile of Compound 22.

While HPLC logD_7.0 was initially used as a high-throughput screening measurement of lipophilicity, logD was determined by the standard shake flask method (see Supporting Information) for key analogues. The logD of compound 22 as measured by the shake flask method was determined to be 3.61 ± 0.12 (n = 4), higher than the desired value (<3.5) to minimize nonspecific binding. Although compound 22 had a higher measured logD value than targeted, we envisioned the compound as a viable CNS PET tracer due to high affinity, selectivity, and passive permeability in combination with low in vitro P-gp susceptibility. To test this hypothesis, compound 22 was selected to be radiolabeled and evaluated in vivo in a rhesus monkey.

The synthesis of [¹¹C]22 is shown in Scheme 1. Oxidation of 4,7-dichloroquinoline with 3-chloroperoxybenzoic acid followed by treatment of oxide 25 with dimethylcarbamyl chloride and trimethylsilyl cyanide at reflux yielded carbonitrile 26. Selective palladium-mediated Suzuki coupling^27,28 of 4,7-dichloroquinoline-2-carbonitrile 26 with (2-fluoro-4-methoxyphenyl)boronic acid provided nitrile 27, which was converted to methyl ester 28 by heating in a saturated solution of HCl in methanol. Suzuki coupling of 28 with potassium vinyl trifluoroborate followed by treatment of olefin 29 with 9-BBN and subsequent Suzuki coupling with 5-bromo-2-methylpyrimidine yielded intermediate 30. Ammonolysis of ester 30 to the primary amide was accomplished with ammonia in methanol at elevated temperature to provide amide 22, which was subsequently treated with BBr₃ for deprotection of the methoxy moiety to afford penultimate phenol 31, which was used as a precursor for radiolabeling. Lastly, treatment of phenol 31 with [¹¹C]methyl iodide provided the PET tracer [¹¹C]22 in 27% radiochemical yield (uncorrected from [¹¹C]methyl iodide) in a high molar activity of 3292 Ci/mmol (EOS) with radiochemical and chemical purity of >99%. Details of the synthesis, including the radiosynthesis, are provided in the Supporting Information.

Scheme 1. Synthesis of [¹¹C]22.

Reagents and conditions: ^a3-Chloroperoxybenzoic acid, dichloromethane, 85%.

Dimethylcarbamyl chloride, trimethylsilyl cyanide, dichloromethane, reflux, 70%.

(2-Fluoro-4-methoxyphenyl)boronic acid, 2 M sodium carbonate, Pd(PPh₃)₄, toluene/ethanol, 75° C, 52%.

Sat HCl/MeOH, 60° C, 90%.

Potassium vinyl trifluoroborate, Pd(OAc)₂, S-Phos, Cs₂CO₃, water/dioxane, 85° C, 71%.

9-BBN, THF, 60° C, then 5-bromo-2-methylpyrimidine, (X-Phos) palladium(II) phenethylamine chloride, K₂CO₃, water/THF, 90° C, 50%.

7 M NH₃/MeOH, 60 °C, 100%.

BBr₃, dichloromethane, 0 to 25 °C, 37%.

[¹¹C]Methyl iodide, Cs₂CO₃, DMF, 0 to 70 °C.

[¹¹C]22 was evaluated in rhesus monkey PET studies to evaluate the potential of the tracer for quantifying mGluR₂ NAM occupancy in brain. Baseline PET scans with [¹¹C]22 in rhesus monkey demonstrated suitable brain penetration and heterogeneous distribution consistent with the known distribution of mGluR₂ receptors (Figure 2, left panel). The highest uptake was in the cerebellum and striatum with slightly lower levels in the cortex, consistent with the distribution of mGluR₂ NAM PET ligands in the rat brain. The PET data from studies when coadministering with selective mGluR₂ NAM therapeutic candidate MK-8768²⁹ resulted in a substantial reduction of [¹¹C]22 uptake (Figure 2, right panel) demonstrating the in vivo specificity of [¹¹C]22 for mGluR₂. An average BP_ND (nondisplaceable binding potential) of 0.70 ± 0.09 (n = 6 rhesus monkeys) was measured in the cerebellum, using the thalamus as a pseudoreference region. While the thalamus was not completely devoid of displaceable binding signal in vivo, it was among the regions with the lowest levels of tracer retention and was used as a pseudoreference region for tracer quantification. While this may introduce some bias in receptor occupancy determination, this has been used successfully for other PET tracers.³⁰ Importantly, it eliminates the need for invasive arterial blood sampling, which is better for animal welfare and operationally demanding. Comparison of BP_ND determinations resulted in an average test–retest variability of 16.5% (see Supporting Information). These results demonstrate that [¹¹C]22 has potential for quantification of mGluR₂ receptor occupancy in human brain.

Figure 2.

Coregistered sagittal PET/MRI summed image (0–90 min) of [¹¹C]22 in rhesus monkey brain: (left panel) baseline data; (right panel) data obtained post administration of MK-8768. The scale is shown in standardized uptake value (SUV) units, which are normalized for the injected radioactive dose and the mass of the monkey. Ctx = cortex; Cb = cerebellum; WM = white matter; Th = thalamus; and St = striatum.

In summary, the identification, synthesis, and in vivo evaluation of an mGluR₂ NAM PET tracer have been described. Compound 22 was found to have high mGluR₂ binding affinity and excellent selectivity over other CNS receptors including mGlu subtypes. [¹¹C]22 was produced in high radiochemical yield, purity, and molar activity with a robust and rapid synthesis suitable for clinical translation. In vivo PET imaging studies in rhesus monkey demonstrated that [¹¹C]22 possessed appropriate properties with high brain uptake and displayed specific and displaceable binding for the mGluR₂ binding site in the rhesus brain. Based on the favorable profile, [¹¹C]22 was designated [¹¹C]MK-8056. The utility of this molecule for measuring receptor occupancy for mGluR₂ NAMs in rhesus monkey and human brain will be reported in due course.

Glossary

Abbreviations

CNS

central nervous system

mGluR₂

metabotropic glutamate receptor 2

NAM

negative allosteric modulator

PET

positron emission tomography

P-gp

P-glycoprotein

HPLC

high-performance liquid chromatography

BCRP

breast cancer resisitance protein

GABA_A

gamma-aminobutyric acid A receptor

SUV

standard uptake value

BP_ND

nondisplaceable binding potential

EOS

end of synthesis

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.3c00175.

• Full experimental details, HRMS data, and copies of NMR spectra for synthesis of MK-8056; procedure for radiochemical synthesis; protocols for in vitro binding assays, and animal procedures (PDF)

Author Contributions

All authors were employees of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA, at the time of their contribution to this work.

These studies were funded by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA.

The authors declare no competing financial interest.