Synthesis and Preliminary Evaluations of a Triazole-cored Antagonist as PET Imaging Probe ([¹⁸F]N2B-0518) for GluN2B Subunit in the Brain

Abstract

GluN2B is the most studied subunit of N-methyl-D-aspartate receptors (NMDARs) and implicated in the pathologies of various central nervous system disorders and neurodegenerative diseases. As pan NMDAR antagonists often produce debilitating side effects, new approaches in drug discovery have shifted to subtype-selective NMDAR modulators, especially GluN2B-selective antagonists. While positron emission tomography (PET) studies of GluN2B-selective NMDARs in the living brain would enable target engagement in drug development and improve our understanding in the NMDAR signaling pathways between normal and disease conditions, a suitable PET ligand is yet to be identified. Herein we developed an ¹⁸F-labeled potent antagonist, 2-((1-(4-[¹⁸F]fluoro-3-methylphenyl)-1H-1,2,3-triazol-4-yl)methoxy)-5-methoxypyrimidine ([¹⁸F]13; also called [¹⁸F]N2B-0518) as a PET tracer for imaging the GluN2B subunit. The radiofluorination of [¹⁸F]13 was efficiently achieved by our spirocyclic iodonium ylide (SCIDY) method. In in vitro autoradiography studies, [¹⁸F]13 displayed highly region-specific binding in brain sections of rat and non-human primate, which was in accordance with the expression of GluN2B subunit. Ex vivo biodistribution in mice revealed that [¹⁸F]13 could penetrate the blood-brain barrier with moderate brain uptake (3.60% ID/g at 2 min) and rapid washout. Altogether, this work provides a GluN2B-selective PET tracer bearing new chemical scaffold and shows high specificity to GluN2B subunit in vitro, which may pave the way for the development of a new generation of GluN2B PET ligands.

Graphical Abstract

INTRODUCTION

N-Methyl-D-aspartate receptors (NMDARs) are ligand-gated ion channel ionotropic glutamate receptors (iGluRs) that are fundamental to excitatory neurotransmission in the mammalian central nervous system (CNS).^{1, 2} NMDARs are involved in numerous neurological disorders and neurodegenerative diseases such as ischemic insults, epilepsy, Parkinson’s disease and Alzheimer’s disease (AD).^{3, 4} Several pan NMDAR antagonists including phencyclidine (PCP), ketamine, MK-801, Memantine, and amantadine were developed and used in the clinical setting;^5–8 however, these drugs often showed debilitating adverse effects and thus had a very narrow therapeutic window.^{7, 9} Of note, NMDARs are typically assembled as tetramers containing two GluN1 subunits and two GluN2 subunits.¹⁰ The GluN2 subunits, which include four subtypes of GluN2A−GluN2D, have distinct developmental expression patterns in CNS and control a variety of functional properties of NMDARs.^11–13 For example, the GluN2B subunit is widespread throughout the brain with strong expression levels around birth; but in adults, becomes progressively restricted to the forebrain while barely detectable in cerebellum.^{11, 12, 14, 15} This subunit endows the NMDARs with various binding sites for both positive allosteric modulators (PAM, e.g., polyamines) and subunit-selective negative allosteric modulators (NAM, e.g., ifenprodil)^{8, 16–19} and thus attracts sufficient attention in drug development as a therapeutic target of high interest for various CNS pathologies, such as cerebral ischemia, acute and chronic pain, schizophrenia, and AD.^{5, 7, 20, 21} The discovery of the first GluN2B-selective NMDAR antagonist, ifenprodil, provided an alternative solution for the treatment of NMDAR-related diseases.²² Compared with the pan NMDAR antagonists, ifenprodil uses a unique activity-dependent mechanism that avoids excessive blockade by leaving transiently activated receptors relatively unaffected, which could mitigate its adverse effect at high drug doses.^{18, 23–25} The finding of ifenprodil has stimulated the continuous development of other potent GluN2B-selective antagonists and several drug candidates have entered human trials, including CP-101606 (traxoprodil),^{26, 27} RGH-896 (radiprodil),⁸ MK-0657 (rislenemdaz),^{28, 29} and EVT-101.³⁰ However, despite multiple clinical trials, no GluN2B-selective antagonists have been approved for use in man by FDA.^{4, 8} Thus, a targeted probe that enables the understanding of in vivo molecular progress of GluN2B subunit is critical in assessing the future prospects for drug development, like facilitating drug candidates by evaluating the receptor occupancy.

Positron emission tomography (PET) imaging of NMDAR GluN2B subunit could help to diagnose and monitor related neurological disorders and further validate treatment intervention and efficacy. In parallel with drug discovery, there is an unmet clinical need for GluN2B-selective NMDAR PET tracers in basic and clinical research. There are continuous research efforts on the development of ¹¹C- or ¹⁸F-labeled probes targeting the GluN2B subunit of NMDAR, which have been applied for PET imaging studies in rats or non-human primates (NHP).^31–33 Among them, ifenprodil-like ligands (±)[¹¹C]methoxy-CP-101606 ((±)[¹¹C]1)³⁴ and [¹¹C]EMD-95885 ([¹¹C]2, Figure 1)³⁵ displayed low brain uptake and homogenous distribution in different brain regions, which was inconsistent with GluN2B expression in the brain.^{11, 12, 14, 15} In 2004, a pyridine derivative [¹¹C]Ro-647312 ([¹¹C]3, Figure 1) also showed homogenous brain distribution in rat brain and exhibited high plasma/brain uptake ratio by PET.³⁶ In 2006, a benzamidine derivative [¹¹C]4 (Figure 1) demonstrated poor brain uptake and rapid metabolism in the rat brain, which impeded its further application in PET imaging.³⁷ Although in vitro studies revealed two ¹⁸F-labeled probes [¹⁸F]trans-5 and [¹⁸F]cis-5 (Figure 1) had promising B_max/K_d ratios (8–37) in different rat brain regions, PET studies failed to observe significant difference of radioactivity signals in different brain regions.^{38, 39} More recently, [¹¹C]Me-NB1 ([¹¹C]6, Figure 1) was generated from a benzo[d]azepin GluN2B subunit antagonist.⁴⁰ [¹¹C]6 had high affinity (K_i = 5.3 nM) to GluN2B subunit and was successfully used in measurement of receptor occupancy by eliprodil (an NMDAR antagonist). It is worthy of note that the specific uptake of [¹¹C]6 in rat brain was also affected by (+)-pentazocine (a σ−1 receptor agonist) and haloperidol (an inverse agonist of the σ−1 receptor) blocking studies, which was explained by the indirect effect of the σ−1 receptor in vivo. Furthermore, Cai et al. reported an interesting sulfur analog of benzo[d]azepin PET tracer [¹¹C]NR2B-SMe ([¹¹C]7, Figure 1).⁴¹ In PET imaging of rats, the brain uptake of [¹¹C]7 was reduced by the pretreatment with eliprodil, ifenprodil, and NR2B-SMe (self-blocking), as well as a σ−1 receptor agonist SA 4503 in a dose-dependent manner. Despite continuous research efforts in the development of PET ligands for this target, to date, no PET tracer of GluN2B subunit has been advanced to human study and PET imaging of GluN2B subunit remains challenging.^31–33

Figure 1.

Chemical structures of representative GluN2B-selective NMDAR PET tracers.

Herein we developed a new series of fluorinated compounds that are amenable for ¹⁸F-labeling based on a novel class of GluN2B antagonists recently disclosed from Janssen Pharmaceutical patents,⁴² with the aim to discover new generation of GluN2B PET ligands with diverse chemotypes and pharmacology. Starting from an efficient parallel synthesis of several methoxypyrimidine compounds (11–14), we performed preliminary pharmacological evaluation and identified two promising molecules (11 and 13) with high potency and selectivity to GluN2B subunit. Molecular docking studies also indicated several key interactions including π-π stacking and hydrogen bonds were found to support our pharmacology findings. As for radiochemistry, our spirocyclic iodonium ylide (SCIDY) method^{43, 44} enables rapid, efficient and modular synthesis of [¹⁸F]11 and [¹⁸F]13 in high radiochemistry yields (RCYs), which is otherwise difficult to access by traditional aromatic nucleophilic substitution. The following in vitro autoradiography (ARG) in brain sections of both rat and NHP demonstrated that [¹⁸F]13 could bind to GluN2B subunit with high specificity. Furthermore, ex vivo biodistribution in mice revealed that [¹⁸F]13 possessed moderate brain uptake, although short retention and fast clearance was also observed. As proof-of-concept, our work provides a preliminary evaluation of new potent and selective triazole-based PET probes targeting GluN2B subunit and may lead to new chemotypes to discover new chemical scaffolds for NMDAR drug discovery and PET tracer development.

RESULTS AND DISCUSSION

Chemistry.

The synthesis of new GluN2B-selective NMDAR antagonists amenable for radiolabeling (11−14) is illustrated in Scheme 1. The synthesis of 11 commenced with the preparation of aromatic azide 9. In acidic conditions, aniline 8 reacted with sodium nitrite and subsequently sodium azide, which resulted in diazonium salts in situ and then converted into azide 9 in a high yield of 89% over two steps. The key triazole intermediate 10 was prepared by copper(I)-induced alkyne−azide cycloaddition ‘click’ reactions between azidobenzene 9 and propargyl alcohol in a high yield of 71% as a white solid. Then the nucleophilic aromatic substitution (S_NAr) reaction between triazole 10 and 2-bromo-5-methoxypyrimidine afforded the target compound 11 in a moderate yield of 29%. In brief, the synthesis of 11 was efficiently achieved in three steps with an overall yield of 18%. In an analogous manner, the other target compounds 12−14 were obtained as white solids in overall yields of 13%−18%.

Scheme 1.

Synthesis of GluN2B-selective NMDAR antagonists (11–14).

Reagents and conditions: (a) NaNO2₂, NaN₃, HCl (6 N), 0–5 °C, 2 h, 89% yield. (b) Propargyl alcohol, DIPEA, CuI, THF, 40 °C, 2 h, 71% yield. (c) 2-Bromo-5-methoxypyrimidine, NaH, THF, 40 °C, 2–3 h, 29% yield.

Pharmacology and Physicochemical Properties.

In most cases, NMDARs are dimer of dimers containing two glycine-binding GluN1 and two glutamate-binding GluN2 subunits, and their functionating relies on joint action of glycine and glutamate.^{2, 10} The potencies of compounds 11−14 as GluN2B-selective antagonists were evaluated via glutamate/glycine (100 μM/100 μM) assays with Xenopus oocytes expressing human GluN1/GluN2B (GenBank NP_015566/GenBank NP_000825) receptors. The current responses of GluN1/GluN2B receptors were inhibited by 11−14 in a dose-dependent manner (Figure 2A). As shown in Table 1, 11 had the highest potency with the IC₅₀ value of 19 nM, followed by 13 with the value of 28 nM. However, the potencies of 12 and 14 (positional isomers of 11 and 13, respectively) significantly decreased to 339 and 89 nM (IC₅₀ values), respectively. We also evaluated the subtype-selectivity of compounds 11 and 13 for GluN2B subunit over other GluN2 subunits. Xenopus oocytes expressing GluN1 with human GluN2A, rat GluN2C, or human GluN2D subunit were used, and current responses to maximal agonists (glutamate/glycine, 100 μM/100 μM) concentrations were recorded in the presence of 11 or 13 (1 μM). The activity of GluN1/GluN2B receptors was substantially inhibited by 11 and 13 with the %current responses of 9.3% and 15.0%, respectively (Table 1 and Figure 2B). In contrast, the current responses of other iGluRs including GluN1/GluN2A, GluN1/GluN2C, GluN1/GluN2D, GluA1, and GluK2 were virtually not affected by 11 or 13 (Table 1 and Figure 2B).

Figure 2.

Pharmacology studies of our GluN2B-selective NMDAR antagonists. (A) Concentration−response curves for antagonists 11−14 (0.03−1.0 μM) on human GluN1/GluN2B were plotted as the percent of the maximal response to glutamate/glycine (100 μM/100 μM) and fit by the Hill equation. (B) %Current responses to glutamate/glycine (100 μM/100 μM for NMDAR) or glutamate (100 μM for AMPAR and KAR) co-applied with compound solution (1 μM) of 11 or 13 were recorded in Xenopus oocytes expressing human GluN1/GluN2A receptors, human GluN1/GluN2B receptors, human GluN1/GluN2D receptors, rat GluN1/GluN2C receptors, rat GluA1(flip) subunit or rat GluK2(Q) subunit. The data were expressed as the percent of the maximal response to agonists. (C and D) Inhibition of triheteromeric receptors by compounds 11 (C) and 13 (D), respectively. Concentration−response curves were generated from the triheteromeric receptors including GluN1/GluN2A/GluN2A (2A/2A), GluN1/GluN2B/GluN2B (2B/2B), and GluN1/GluN2A/GluN2B (2A/2B) upon activated by glutamate/glycine (100 μM/100 μM). Data are mean ± SEM from 10–14 oocytes.

Table 1.

Potency and selectivity of compounds 11−14

entry	potencya	selectivityb
	IC50, mean (nM) pIC50, mean (95% CI)	%control at 1 μM, mean (99% CI)
		GluN2A	GluN2B	GluN2C	GluN2D	GluA1	GluK2
11	19 −1.72 (−1.75, −1.69)	102 (93, 111)	13 (7, 19)	93 (92, 94)	92 (88, 96)	97 (90, 104)	92 (84, 100)
12	339 −0.55 (−0.76, −0.34)	n.d.
13	28 −1.57 (−1.63, −1.51)	99 (93, 105)	17 (11, 23)	95 (92, 98)	96 (94, 98)	98 (91, 105)	91 (80, 102)
14	89 −1.12 (−1.30, −0.94)	n.d.

^aDetermined with Xenopus oocytes expressing human GluN1/GluN2B receptors in 100 μM glutamate/glycine assay coapplied with increasing concentrations of 11−14 (n = 6−12).

^b%Control response was expressed as the percent of the maximal response to 100 μM glutamate/glycine (for GluN2a-GluN2D subunits), or to 100 μM glutamate (for GluA1 and GluK2 subunits). Xenopus oocytes coexpressing human GluN1/GluN2A receptors, human GluN1/GluN2B receptors, human GluN1/GluN2D receptors, rat GluN1/GluN2C receptors, rat GluA1(flip) subunit or rat GluK2(Q) subunit were used. n.d., not determined.

Two different GluN2 subunits, GluN2A and GluN2B, have been found in most NMDAR-expressing cells and are assembled as triheteromers like GluN1/GluN2A/GluN2B (2A/2B) and diheteromers like GluN1/GluN2A/GluN2A (2A/2A) and GluN1/GluN2B/GluN2B (2B/2B).^{2, 45} Thus, we further compared compounds 11 and 13 inhibition of triheteromeric receptors (2A/2B) to that of diheterometric 2A/2A and 2B/2B receptors. Compounds 11 and 13 inhibited current responses from wild-type 2B/2B with an IC₅₀ of 11 ± 2.3 and 20 ± 3.4 nM, maximal inhibition of 86% ± 1.8% (n = 11) and 88% ± 1.8% (n = 11), respectively, whereas responses from wild-type 2A/2A were insensitive to 1 μM 11 (n = 10) and 13 (n = 11) (Figures 2C and 2D). The inhibition of 2A/2B triheteromeric receptors by 11 and 13 (1 μM) was reduced compared to that of 2B/2B, since IC₅₀ values increased to 141 ± 8.9 nM for 11 and 154 ± 28 nM for 13, and maximal inhibition was 23% ± 2.5% (n = 11) for 11 and 27% ± 2.0% for 13 (n = 13). Similar to the effects of ifenprodil,⁴⁵ our data suggest that reduced sensitivity of triheteromers to these two compounds (11 and 13) is mainly mediated by a diminished maximal inhibition, in addition to reduction in potency. As a result, 11 and 13 are identified as potent GluN2B-selective antagonists with high selectivity over other GluN2 subunits and iGluR units, representing a promising chemical scaffold for GluN2B-selective NMDAR PET ligands.

The lipophilicity with preferred range of 2.0−3.5 in candidate compounds is an empirical factor that could predict the ability of the blood-brain barrier (BBB) penetration.^46–48 The calculated logP (clogP) values of 11−14 are 3.32, 3.32, 2.90, and 2.90, respectively (Table 2). The logD values of [¹⁸F]11 and [¹⁸F]13 (see radiochemistry below, Scheme 2) were determined to be 3.14 and 2.89 by “shake-flask” method,⁴⁹ respectively, which are within the optimal range for BBB permeability. In silico prediction gave a reasonable topological polar surface area (tPSA) value of 71.4 Å (<90 Å) for all the compounds (Table 2), which indicated their great potentials as brain penetrant leads.⁵⁰ We also explored the CNS PET multiparameter optimization (MPO) scores of these compounds. As reported, a higher MPO score of a compound often leads to a higher probability in achieving optimal CNS drug-like properties such as permeability, clearance, and safety.^{51, 52} As shown in Table 2, 13 and 14 had preferred MPO scores >4.9 while 11 and 12 exhibited a lower value of 4.05. It is worth mentioning that compound 13 has the highest MPO score of 5.16, which is situated in the optimal range (>5) that is characteristic of successful PET ligands.⁵³ These results led to further evaluations of [¹⁸F]13 in in vitro ARG and ex vivo biodistribution studies.

Table 2.

Physicochemical properties of compounds 11−14.

compound	clogPa	logD7.4b	tPSA (Å)a	MPOc
11	3.32	3.14 ± 0.037	71.4	4.05
12	3.32	n. d.	71.4	4.05
13	2.90	2.89 ± 0.037	71.4	5.16
14	2.90	n. d.	71.4	4.91

^aClogP and tPSA values were calculated by ChemBioDraw Ultra 14.0.

^bLogD was measured with “shake-flask” method by using corresponding ¹⁸F-labeled tracers (18.5 MBq) and expressed as mean ± SD (n = 3).

^cMPO scores were calculated using the method reported in Zhang, L.et al. ⁵³. n. d., not determined.

Scheme 2.

Synthesis of the precursors (17, 20, and 21) and radiosynthesis of [¹⁸F]11 and [¹⁸F]13.

Reagents and conditions: (a) DIPEA, CuI, THF, 40 °C, 4 h, 25% yield. (b) [¹⁸F]F^-, Kryptofix₂₂₂/K₂CO₃ (7.5/1.5 mg), anhydrous DMF, 140 °C, 10 min, <3% yield (decay corrected). (c) 1) mCPBA, CHCl₃, rt, overnight; 2) SPIAd, Na₂CO₃ (10%), ethanol, pH = 10, rt, 6 h, 36% yield over two steps. (d) [¹⁸F]F^-, TEAB, anhydrous DMF, 120 °C, 10 min. (e) DIPEA, CuI, anhydrous DMF, 100 °C, 10 min, 35% and 39% yield for [¹⁸F]11 and [¹⁸F]13, respectively, over two steps (decay corrected).

Molecular Docking Studies

A 2.6-Å resolution crystal structure of the NMDA receptor subunit GluN1b and GluN2B complex was downloaded (PDB ID: 3QEL)¹⁸ and the original ligand, ifenprodil, was removed before performing the molecular docking studies. The goal is to identify possible and preliminary molecular interaction between our GluN2B inhibitors 11 & 13 and binding domain based on the fact that low nanomolar binding affinities were identified. First, we reproduced the binding pose of the original ligand, ifenprodil, within 3QEL using AutoDock Vina, which yielded a nearly-perfect overlapping between the docked pose and original pose (RMSD = 0.36 Å). Subsequently, candidate compounds 11 and 13 were docked onto the 3QEL structure in the ifenprodil site. A consensus binding pose with highest docking scores was found for these two compounds within the ifenprodil binding channel, which is the interface between subunits GluN1b and GluN2B (Figure 3). Inside the binding pocket, angled π-π stacking interactions were observed between the phenyl group of either compound and nearby Phe429 and Tyr82 side chains (Figure 3). Multiple hydrogen bonds were also found between the pyrimidine moiety of the compounds and Arg88 side chain. In addition, the charged side chain of Glu421 was found ~3Å away from both compounds, indicating possible electrostatic interactions. The high docking scores of 11 and 13 (−9.1 and −9.4, respectively) indicated strong bindings between the compounds and NMDAR subunit, which agree well the experimental IC₅₀ data in Table 1. Overall, this preliminary docking study showed that compounds 11 and 13 interacted with NMDAR subunit at the ifenprodil site with π-π stacking interactions, hydrogen bonds, and possibly electrostatic interactions.

Figure 3.

Molecular docking structures of compounds 11 (purple in A) and 13 (orange in B) onto NMDAR subunit GluN1b and GluN2B interface.

Radiochemistry.

S_NAr reactions with [¹⁸F]F^- is one of the widely used and highly attractive methods for the synthesis of ¹⁸F-labeled aromatic PET ligands.⁵⁴ In our initial design of the labeling strategy, we took advantage of S_NAr reactions to incorporate fluorine-18 into the desired compounds. As shown in Scheme 2A, compound 17 was prepared by the copper(I)-induced “click” reaction between aromatic azide 15 and intermediate 16 (the synthesis of 15 and 16 is shown in Scheme S1, Supporting Information) and used as the precursor for one-step ¹⁸F-labeling, in which -NO₂ serves as the leaving group. However, ¹⁸F-labeled compound 11 ([¹⁸F]11) was obtained in poor radiochemistry yield (RCY) <3% (decay corrected) and difficult to purify by high-performance liquid chromatography (HPLC) because of overlapping impurities. The low RCY of this method may be caused by the lack of an appropriate leaving group in 17, i.e., an electron-withdrawing group at its ortho or para position.^{55, 56} As our alternative strategy, we explored a step-wise synthesis of [¹⁸F]11 and [¹⁸F]13 using SCIDY method^{43, 44} (Scheme 2B). The corresponding precursors 20 and 21 were prepared from the oxidation of iodinated azidobenzenes (18 and 19, the synthesis is shown in Scheme S1, Supporting Information) with mCPBA, followed by ligand exchange with SPIAd⁴⁴ in the same overall yields of 36%. As illustrated in Scheme 2B, the radiolabeling commenced with the generation of [¹⁸F]9 and [¹⁸F]22, and the subsequent step was accomplished by “click” reaction between ¹⁸F-labeled azidobenzenes ([¹⁸F]9 and [¹⁸F]22) and 16, which afforded [¹⁸F]11 and [¹⁸F]13 in high overall isolated RCYs of 35% and 39% (n = 5, decay corrected), respectively. The products of [¹⁸F]11 and [¹⁸F]13 were generated in high radiochemistry purity >99% and good molar activity of ~19 GBq/μmol (n = 5) for [¹⁸F]13 and ~6 GBq/μmol for [¹⁸F]11 (n = 5). Furthermore, since the 1,2,3-triazole linker showed good biostability in vivo and is used as an isosteric surrogate for peptide bond,⁵⁷ our radiolabeling method may provide a highly-efficient modular synthesis route for the preparation of new ¹⁸F-arene click agents.

In Vitro Autoradiography.

The GluN2B subunit is widespread throughout the CNS with high expression in the forebrain including cortex, hippocampus, and thalamus, while barely detectable in the cerebellum.^{11, 12, 14, 15} To assess the specificity of the new tracer to GluN2B subunit, we used [¹⁸F]13 (with higher MPO score and lower lipophilicity compared to [¹⁸F]11) to perform in vitro ARG on the brain cryosections (20 μm) of both rat and NHP. As shown in Figure 4A, the radioactivity of [¹⁸F]13 in rat brain sections showed brain region-specific localization (heterogeneous distribution), in which the uptake was the highest in hippocampus and cortex (especially orbital, prelimbic, and somatomotor areas), moderate in striatum and thalamus, and low in the cerebellum. This distribution pattern was in accordance with the in vitro autoradiograms of [³H]CP-101606,⁵⁸ (±)[¹¹C]1,³⁴ and [¹⁸F]cis-5,³⁸ as well as with mRNA distribution of GluN2B subunit.⁵⁹ In self-blocking experiments, the uptake of [¹⁸F]13 was substantially reduced (~90%) in the presence of cold compound 13 (10 μM, Figure 4C, also see Figure S3 in SI for self-blocking results at 1 μM) and the specific binding of [¹⁸F]13 was 94%, 94%, 92%, and 88% (n = 3, p <0.001, T-test, Figure 4E) in brain regions of hippocampus, frontal cortex, striatum, and thalamus, respectively. The specificity of [¹⁸F]13 to GluN2B subunit was further confirmed by using a GluN2B-selective NAM, the ifenprodil derivative BMT-108908 (K_i = 1.4 nM),^{60, 61} and the results showed that the binding of [¹⁸F]13 was strongly inhibited by this drug (Figure 4D and 4E) in a similar pattern of self-blocking. This result also indicated that [¹⁸F]13 associated with GluN2B subunit in the same binding site as BMT-108908, which was also confirmed by the molecular docking results (Figure 3).

Figure 4.

In vitro autoradiography results of [¹⁸F]13 (37 kBq/mL, 1.58 nM) in rat brain sections. Representative autoradiograms in rat brain sagittal sections: [¹⁸F]13 alone (A, baseline), and pre-blocked by cold compound 13 (C, self-blocking, 10 μM) and BMT-108908 (D, a NAM drug blocking, 10 μM). (B) GluN2B mRNA expression in mouse brain and the data were retrieved from mouse.brain-map.org (experiment: 69257725). (E) Quantitative analysis of baseline and blocking experiments. The value is expressed as the ratio of the radioactivity of different brain regions to that of the cerebellum (reference region). HIP, hippocampus; FCx, frontal cortex; STR, striatum; THM, thalamus; Cb, cerebellum; OLF, olfactory bulb. Asterisks indicate statistical significance: ***p <0.001, and **p <0.01 vs control.

Given the high specificity of [¹⁸F]13 to GluN2B subunit, we further explored the distribution pattern of GluN2B subunit in the NHP brain. Figure 5A showed that the highest bound activity of [¹⁸F]13 was found in brain regions of the hippocampus, entorhinal cortex, and insular cortex. The activity was moderate in caudate nucleus and putamen, low in the thalamus and claustrum, and no appreciable binding in white matter of forebrain. The nonspecific binding was determined by self-blocking studies using non-radioactive 13 (10 μM, Figure 5B) and the results showed that the radioactivity of [¹⁸F]13 distributed to several brain regions with high specificity including hippocampus (87%, n = 3, p <0.001, T-test, Figure 5D), entorhinal cortex (89%, p <0.01), insular cortex (82%, p <0.001), caudate nucleus (79%, p <0.01), and putamen (73%, p <0.01), but lower in thalamus (53%, p <0.01). The specific binding to NMDAR GluN2B subtype was further confirmed by the blocking experiments with BMT-108908 (10 μM)^{60, 61} and the %blockade results were similar to those of self-blocking studies (Figure 5C and 5D). In summary, [¹⁸F]13 showed high specific binding to GluN2B subunit in the NHP brain sections, which showed characteristic and regional-specific GluN2B subunit expression and target specificity (validated by blocking studies).

Figure 5.

In vitro autoradiography results of [¹⁸F]13 (37 kBq/mL, 1.58 nM) in NHP brain sections. Representative autoradiograms in NHP brain sections: [¹⁸F]13 alone (A, baseline), and pre-blocked by cold compound 13 (B, self-blocking, 10 μM) and BMT-108908 (C, a NAM drug blocking, 10 μM). (D) Quantitative analysis of baseline and blocking experiments. The value is expressed as the ratio of the radioactivity of different brain regions to that of the white matter of forebrain. HIP, hippocampus; ECx, entorhinal cortex; ICx, insular cortex; CdN, caudate nucleus; CLA, claustrum; PUT, putamen; THM, thalamus; FWM, white matter of forebrain. Asterisks indicate statistical significance: ***p <0.001, and **p <0.01 vs control.

Whole Body Ex Vivo Biodistribution Studies and Radiometabolite Analysis.

GluN2B antagonist 13 exhibited a favorable MPO score (5.16), proper lipophilicity (logD_7.4 2.89) and reasonable polar surface area (71.4 Å), indicating a high probability to be a successful PET tracer. We next performed whole body ex vivo biodistribution studies to examine the uptake, distribution and clearance of [¹⁸F]13 in organs of interest. The studies were conducted in mice at four different time points (2, 10, 30, and 60 min) after the injection of [¹⁸F]13. As shown in Figure 6A, several organs including heart, lung, kidney, liver, and small intestine showed high initial uptakes (> 5% ID/g) of [¹⁸F]13 at 2 min postinjection. After initial uptake, the radioactivity in organs/tissues of heart, lung, kidney, spleen, pancreas, and muscle displayed rapid washout with the uptake <0.6% ID/g at 60 min postinjection, while the levels in liver (20.84% ID/g) and small intestine (39.15% ID/g) remained high until 10 min post injection and then decreased rapidly (Figure 6A and Table S1 in Supporting Information). These results indicated that [¹⁸F]13 probably underwent hepatobiliary elimination. Of note, [¹⁸F]13 displayed reasonable initial brain uptake (brain_2min) with the value of 3.60 %ID/g, followed by a very rapid washout to 0.72 %ID/g at 10 min post injection, then to 0.018% ID/g at 60 min post injection (Figure 6B). The uptake in blood was higher than that in the brain at all time points (Figure 6). As a result, the observed abnormal rapid clearance of [¹⁸F]13 in the brain (brain_2min/brain_10min = 5, and brain_2min/brain_60min = 200) diminished our enthusiasm in pursuing in vivo PET evaluation in rodents. Notably, no substantial bone uptake of [¹⁸F]13 was observed (<2 %ID/g, Figure 6B), indicating little defluorination occurred in vivo. In addition, probe [¹⁸F]11 with higher lipophilicity (logD_7.4 3.14) displayed similar biodistribution results as [¹⁸F]13 (Table S1) in different organs. In particular, [¹⁸F]11 showed lower initial brain uptake (brain_2min = 3.27% ID/g) and slower brain washout (brain_2min/brain_60min = 55).

Figure 6.

(A) Ex vivo biodistribution in mice at four different time points (2, 10, 30, and 60 min) post injection of [¹⁸F]13. The results are expressed as the percentage of the injected dose per gram of wet tissue (% ID/g) with the exception that stomach uptake was expressed as % ID/organ. (B) The values of the brain, blood, and bone uptakes of [¹⁸F]13. The values for remaining organs of interest are included in Table S1 of Supporting Information.

To further understand the pharmacokinetics of [¹⁸F]13, the metabolism studies in ICR mice (n = 2) were carried out by radioHPLC analysis of plasma, brain, and liver extracts at 15 min post injection. The results showed that [¹⁸F]13 displayed moderate stability in the brain with 56.6% parent fraction left. In contrast, this probe degraded to a greater extent in plasma and liver with 11.8% and 5.1% parent fraction remained, respectively. Only polar radiometabolites of [¹⁸F]13 were found in plasma, brain, and liver. In addition, [¹⁸F]11 exhibited better stability than [¹⁸F]13 in mice with 84.8%, 32.0%, and 17.7% parent tracer remained in brain, plasma, and liver, respectively, at 15 min post injection. The results indicated that both [¹⁸F]11 and [¹⁸F]13 underwent moderate metabolism in the brain but rapid degradation in plasma and liver, which may contribute to the moderate brain uptake seen in the ex vivo biodistribution studies.

Conclusion

In this report, we successfully designed and synthesized a series of GluN2B-selective NMDAR antagonists (11–14) with the chemical scaffold containing triazole and methoxypyrimidine groups. Two compounds, 11 and 13, showed high potency and high selectivity to GluN2B subunit in in vitro glutamate/glycine assays with Xenopus oocytes, which identified them as potent candidates for PET ligand development. Molecular docking studies also identified the binding of both compounds to the interface between GluN1 and GluN2B subunits. In silico prediction showed that 11 and 13 had optimal clogP and tPSA values for CNS drug discovery and 13 exhibited the highest MPO score (5.16), indicating its great potential for PET imaging studies. Our SCIDY method enables an efficient radiosynthesis of [¹⁸F]11 and [¹⁸F]13 (also called [¹⁸F]N2B-0518) in high RCYs (>35%). The following in vitro ARG in rat brain sections unveiled the specific binding of [¹⁸F]13 in brain regions of the hippocampus, frontal cortex, striatum, and thalamus, but not cerebellum, which is in accordance with the expression of GluN2B subunit. Furthermore, autoradiograms of [¹⁸F]13 in NHP brain sections demonstrated that the GluN2B subunit expressed strongly in the hippocampus, entorhinal cortex, insular cortex, caudate nucleus, and putamen, moderately in claustrum and thalamus. Ex vivo biodistribution in mice showed that [¹⁸F]13 penetrated the brain with moderate initial uptake and rapid washout, which may be attributed to poor biostability of [¹⁸F]13 in the plasma. Further PET imaging work of [¹⁸F]13 in higher species will be planned to investigate if inter-species differences exist in the brain uptake and metabolic rate in the plasma. In summary, we provided a potent and selective PET ligand for in vitro detection of GluN2B subunit, which not only widens the landscape of PET probes targeting GluN2B subunit but also offers a new structural framework for medicinal chemistry design.

METHODS

General Information.

All chemicals were purchased from commercial vendors and used without further purification unless otherwise indicated. Thin-layer chromatography (TLC) was conducted with 0.25 mm silica gel plates (⁶⁰F₂₅₄) and visualized by exposure to UV light (254 nm) or by staining with potassium permanganate. Column chromatography purification was performed using silica gel (SiliCycle Inc., 230−400 mesh, 40−63 μm). ¹H, ¹³C, and ¹⁹F NMR were obtained at 300, 75, and 282 MHz, respectively, on a Bruker spectrometer in CDCl₃ or d₆-DMSO solutions at room temperature, and the chemical shifts were quoted in δ values (parts per million, ppm) downfield relative to the internal TMS. The multiplicities are abbreviated as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad signal, and dd = doublet of doublets. For LC-MS, the ionization method is ESI using Agilent 6430 Triple Quad LC/MS. The compounds 11–14 did not show any promiscuous moieties in the Pan Assay Interference Compounds Assay (PAINS) using the in silico filter (http://www.swissadme.ch/index.php).^{62, 63} The animal experiments were approved by the Institutional Animal Care and Use Committee of Massachusetts General Hospital. CD1 (ICR) mice (female; 8 weeks, 23–27 g) were kept on a 12 h light/12 h dark cycle and were allowed food and water adlibitum.

Chemistry.

4-Azido-2-bromo-1-fluorobenzene (9).

Compound 9 was prepared by a previously reported procedure.⁶⁴ Dark brown oil, yield 89%. ¹H NMR (300 MHz, CDCl₃) δ 7.22 (d, J = 2.6 Hz, 1H), 7.10 (t, J = 8.2 Hz, 1H), 6.95 – 6.93 (m, 1H). ¹³C NMR (75 MHz, CDCl₃) δ 156.38 (d, J = 245.6 Hz), 136.85, 123.70, 119.29 (d, J = 7.1 Hz), 117.28 (d, J = 24.0 Hz), 109.97 (d, J = 22.6 Hz). ¹⁹F NMR (282 MHz, CDCl₃) δ −108.03 – −108.05 (m).

(1-(3-Bromo-4-fluorophenyl)-1H-1,2,3-triazol-4-yl)methanol (10).

To a solution of compound 9 (2.0 g, 9.3 mmol) in THF (20 mL) was added propargyl alcohol (1.0 g, 18.0 mmol) and DIPEA (244 mg, 1.9 mmol), and then CuI (88 mg, 0.5 mmol) was added under stirring. The reaction mixture was stirred at 40 °C for 2 h. Upon completion the reaction mixture was filtered, and the solvent was removed under vacuum. The residue was purified via silica gel column chromatography using a gradient mixture of hexane–ethyl acetate (50%–100%) to obtain the pure compound of 10 as white solid (1.8 g, 6.6 mmol, 71%). ¹H NMR (300 MHz, d₆-DMSO) δ 8.73 (s, 1H), 8.30 (d, J = 2.7 Hz, 1H), 7.99 (d, J = 4.2 Hz, 1H), 7.61 (t, J = 8.7 Hz, 1H), 5.34 (s, 1H), 4.60 (s, 2H). ¹³C NMR (75 MHz, d₆-DMSO) δ 158.26 (d, J = 246.1 Hz), 149.76, 134.39 (d, J = 3.1 Hz), 125.24, 121.79, 121.69, 118.26 (d, J = 24.0 Hz), 109.54 (d, J = 22.6 Hz), 55.34. ¹⁹F NMR (282 MHz, d₆-DMSO) δ −104.77 – −104.80 (m).

2-((1-(3-Bromo-4-fluorophenyl)-1H-1,2,3-triazol-4-yl)methoxy)-5-methoxypyrimidine (11).

Compound 10 (108.7 mg, 0.4 mmol) was dissolved in 10 mL THF, along with 2-bromo-5-methoxypyrimidine (50 mg, 0.3 mmol), and then NaH (70 mg, 1.7 mmol, 60% dispersion in mineral oil) was added in small portion under stirring. The reaction mixture was stirred at 40 °C for 3 h, and subsequently quenched with saturated solution of ammonium chloride. After that, the mixture was extracted with dichloromethane, and the combined organic extracts were dried over anhydrous magnesium sulfate, filtered, and concentrated in vacuo. The residue was purified by silica gel column chromatography using a gradient mixture of hexane–ethyl acetate (30%–80%), and the product was obtained as white solid (29 mg, 0.08 mmol, 29%). ¹H NMR (300 MHz, CDCl₃) δ 8.23 (s, 2H), 8.08 (s, 1H), 7.98 (d, J = 2.5 Hz, 1H), 7.66 (d, J = 4.8 Hz, 1H), 7.30 – 7.27 (m, 1H), 5.59 (s, 2H), 3.87 (s, 3H). ¹³C NMR (75 MHz, CDCl₃) δ 159.23, 158.95 (d, J = 250.2 Hz), 149.80, 145.46, 145.17, 133.76 (d, J = 3.6 Hz), 125.84, 121.28, 121.07 (d, J = 7.7 Hz), 117.40 (d, J = 24.1 Hz), 110.22 (d, J = 22.8 Hz), 61.07, 56.52. ¹⁹F NMR (282 MHz, CDCl₃) δ −102.37. HRMS (m/z): [M + Na]⁺ calculated for C₁₄H₁₁BrFN₅O₂Na⁺, 401.9972, found 401.9978.

Compounds 12–14 were synthesized by following the same procedure for 11 described above.

2-((1-(4-Bromo-3-fluorophenyl)-1H-1,2,3-triazol-4-yl)methoxy)-5-methoxypyrimidine (12).

White solid, yield 30%. ¹H NMR (300 MHz, d₆-DMSO) δ 8.96 (s, 1H), 8.40 (s, 2H), 8.06 (d, J = 9.7 Hz, 1H), 7.95 (t, J = 7.5 Hz, 1H), 7.79 (d, J = 8.7 Hz, 1H), 5.46 (s, 2H), 3.84 (s, 3H). ¹³C NMR (75 MHz, d₆-DMSO) δ 159.15, 159.00 (d, J = 245.4 Hz), 150.02, 146.14, 144.54, 137.46 (d, J = 10.0 Hz), 135.14, 123.50, 117.86 (d, J = 3.5 Hz), 109.37 (d, J = 27.5 Hz), 108.25 (d, J = 20.7 Hz), 60.48, 56.97. ¹⁹F NMR (282 MHz, d₆-DMSO) δ −100.92 – −101.00 (m). HRMS (m/z): [M + Na]⁺ calculated for C₁₄H₁₁BrFN₅O₂Na⁺, 401.9972, found 401.9975.

2-((1-(4-Fluoro-3-methylphenyl)-1H-1,2,3-triazol-4-yl)methoxy)-5-methoxypyrimidine (13).

White solid, yield 30%. ¹H NMR (300 MHz, CDCl₃) δ 8.23 (s, 2H), 8.05 (s, 1H), 7.58 (s, 1H), 7.46 (s, 1H), 7.13 (t, J = 8.3 Hz, 1H), 5.59 (s, 2H), 3.87 (s, 3H), 2.35 (s, 3H). ¹³C NMR (75 MHz, CDCl₃) δ 160.98 (d, J = 249.8 Hz), 159.32, 149.77, 145.45, 144.74, 132.93, 126.78 (d, J = 19.0 Hz), 123.89 (d, J = 5.5 Hz), 121.42, 119.64 (d, J = 8.6 Hz), 116.10 (d, J = 24.3 Hz), 61.22, 56.53, 14.66 (d, J = 3.3 Hz). ¹⁹F NMR (282 MHz, CDCl₃) δ −112.34. HRMS (m/z): [M + Na]⁺ calculated for C₁₅H₁₄FN₅O₂Na⁺, 338.1024, found 338.1029.

2-((1-(3-Fluoro-4-methylphenyl)-1H-1,2,3-triazol-4-yl)methoxy)-5-methoxypyrimidine (14).

White solid, yield 24%. ¹H NMR (300 MHz, d₆-DMSO) δ 8.89 (s, 1H), 8.39 (s, 2H), 7.73 (dd, J = 26.9, 9.4 Hz, 2H), 7.49 (t, J = 8.2 Hz, 1H), 5.44 (s, 2H), 3.84 (s, 3H), 2.28 (s, 3H). ¹³C NMR (75 MHz, d₆-DMSO) δ 161.04 (d, J = 244.4 Hz), 159.18, 150.00, 146.13, 144.23, 136.00 (d, J = 10.3 Hz), 133.13 (d, J = 6.0 Hz), 125.29 (d, J = 17.2 Hz), 123.33, 116.14 (d, J = 3.3 Hz), 107.74 (d, J = 27.5 Hz), 60.53, 56.95, 14.29. ¹⁹F NMR (282 MHz, d₆-DMSO) δ −110.48 – −110.55 (m). HRMS (m/z): [M + Na]⁺ calculated for C₁₅H₁₄FN₅O₂Na⁺, 338.1024, found 338.1025.

Pharmacology study.

Recombinant receptors were expressed from cDNAs encoding human NMDA receptor subunits GluN1–1a (hereafter GluN1; GenBank NP_015566), GluN2A (GenBank NP_000824), GluN2B (GenBank NP_000825), and GluN2D (GenBankNP_000827.1)⁶⁵ Rat GluN1–1a (U11418, U08261) and GluN2C (M91563), GluA1-flip (P19490), and GluK2(Q) (P42260) were used for the study of subunit selectivity. cRNA was transcribed in vitro from plasmids containing NMDAR cDNAs.⁶⁶ Xenopus laevis stage VI oocytes were prepared from commercially available ovaries (Xenopus one Inc, Dexter, MI, USA). The ovary was placed in a petri plate at room temperature, gently pulled apart with two fine forceps and transferred into Collagenase Type 4 (Worthington-Biochem, Lakewood, NJ, USA) solution (850 μg/ml, 15 ml for a half ovary) in Ca²⁺-free Barth’s solution (see below). The ovary was digested on a mixer at room temperature (23 °C) for 2 h. If the oocytes were under digested, they were kept on a mixer for an additional 15 min. The oocytes were then rinsed 5 times with Ca²⁺-free Barth’s on a mixer for 10 min (with 35–40 mL of fresh solution each time), and further rinsed for 4 times with normal Barth’s on the mixer for 10 min (35–40 mL of fresh solution each time). The oocytes were placed in petri plates for sorting and kept in a 16°C incubator afterward. The oocytes were injected with 5–10 ng cRNA and stored at 15 °C in media containing (Barth’s, in mM) 88 NaCl, 2.4 NaHCO₃, 1 KCl, 0.33 Ca(NO₃)₂, 0.41 CaCl₂, 0.82 MgSO₄, 5 HEPES, 1 U/mL penicillin, 0.1 mg/mL gentamicin sulfate, and 1 μg/mL streptomycin (pH 7.4, adjusted with NaOH). Two to seven days after injection, two-electrode voltage-clamp recordings⁶⁶ were performed at room temperature in an extracellular solution containing (in mM) 90 NaCl, 1 KCl, 10 HEPES, 0.5 BaCl₂, and 0.01 EDTA (pH 7.4, adjusted with NaOH). NMDAR current responses from oocytes were recorded at a holding potential of −40 mV. Concentration–response curves for GluN2B antagonists were generated by co-applying 100 μM glutamate and 100 μM glycine with variable concentrations of test compounds up to 1 μM. Test compounds were prepared as 20 mM stock solutions in DMSO and diluted to the final concentration in recording solution. DMSO content was 0.05–0.5% (v/v). The response to test compounds was given as a percentage of the initial response to glutamate and glycine alone. Data for individual cells were fitted in OriginPro (v 9.0) with the Hill equation:

$$Response\left(\%\right)=\left(100-minimum\right)/\left(1+{\left(\left[concentration\right]/I{C}_{50}\right)}^N\right)+minimum$$

where the minimum is the residual response in saturating inhibitor, IC₅₀ is the concentration that reduces the response by half, and N is the Hill slope. Mean fitted IC₅₀ values from individual cells are given with the 95% confidence interval (CI) determined from the log(IC₅₀) was used to determine the CI, which were then transformed back to concentration. For the graphical representation, the data were normalized to the current response to agonist alone, averaged across all cells, and fitted to the Hill equation.

The triheteromeric receptors (2A/2B) and diheterometric receptors (2A/2A and 2B/2B) used here were developed by the method previously reported,⁴⁵ and their current amplitude evoked by agonists glutamate/glycine (100 μM/100 μM) was showed in Figure S2 in SI.

Molecular Docking Studies.

Candidate compounds 11 and 13 were docked onto the selected protein structure (PDB ID: 3QEL) using the AutoDock Vina module in the UCSF Chimera software. Briefly, the mol2 file of protein was prepared by deleting the solvent, adding hydrogen atoms and charges using the default settings. Standard residues were determined according to the AMBER 14 force field, while non-standard residues were ignored. The predicted free energy of binding was calculated in AutoDock Vina using a hybrid scoring function (combined knowledge-based and empirical approaches).⁶⁷

Radiochemistry.

One-step radiolabeling of [¹⁸F]11:

[¹⁸F]F^- on the QMA cartridge was eluted into a 5 mL reaction vial with 1 mL of Kryptofix₂₂₂/K₂CO₃ solution (7.5 mg of Kryptofix₂₂₂ and 1.5 mg of K₂CO₃ in CH₃CN/H₂O, 4/1, v/v). The solvent was removed, and then the residue was dried azeotropically with anhydrous acetonitrile (1 mL × 3) under a stream of nitrogen at 120 ºC. The vessel was cooled to room temperature, and a solution of nitro precursor 17 (2.5 – 3.5 mg) in anhydrous DMF (1 mL) was added. The reaction mixture was heated to 140 ºC and kept for 10 min. After that, the reaction mixture was purified with HPLC (Waters) using a YMC C18 reverse phase column (5 μm, 10 × 250 mm, elution: CH₃CN/H₂O = 45%/55%, 4 mL/min). To the separated solution of [¹⁸F]11, 20 mL of water was added, and then passed through a preconditioned Sep-Pak Plus-C18 cartridge (Waters). The cartridge was washed with 10 mL of water, and the labeled compound [¹⁸F]11 was eluted to a glass vial with 1 mL of ethanol. The ethanol eluent was concentrated under nitrogen flow at 60 ºC and then formulated with saline (containing 5%–8% ethanol) for further studies. The radiochemical yield is ~3% (decay corrected), and the radiochemical purity is > 99%.

Two-step radiolabeling of [¹⁸F]11 and [¹⁸F]13:

[¹⁸F]F^- on the QMA cartridge was eluted into a 5 mL reaction vial with 1 mL of TEAB solution (3.5 mg, CH₃CN/H₂O, 7/3, v/v). After [¹⁸F]F^- was dried under a nitrogen stream and cooled to room temperature (same procedure as above), a solution of SCIDY precursor 20 or 21 (~5 mg) in anhydrous DMF (400 μL) was added. After the reaction at 120 ºC for 10 min, 16 (2–3 mg), CuI (2–3 mg), and DIPEA (5 μL) were added and the reaction mixture was kept at 100 ºC for another 10 min. Then the reaction mixture was quenched with 10 mL water, passed through a preconditioned Sep-Pak Plus-C18 cartridge (Waters), washed with 10 mL water and eluted with 1–1.5 mL acetonitrile. The eluent was purified with HPLC (Waters) using a YMC C18 reverse phase column (5 μm, 10 × 250 mm, 5 mL/min, elution: CH₃CN/H₂O = 45%/55% and 40%/60% for [¹⁸F]11 and [¹⁸F]13, respectively). The fraction containing [¹⁸F]11 (or [¹⁸F]13) was mixed with 20 mL of water and then passed through a preconditioned Sep-Pak Plus-C18 cartridge (Waters). The cartridge was washed with 10 mL of water, and the labeled compound [¹⁸F]11 (or [¹⁸F]13) was eluted to a reaction vial with 1 mL of ethanol. The ethanol eluent was concentrated under nitrogen flow at 60 ºC and then formulated with saline (containing 5%–8% ethanol) for further studies. The identities of [¹⁸F]11 and [¹⁸F]13 were confirmed with HPLC by the co-injection with corresponding cold compounds of 11 or 13, respectively (Figure S1 in Supporting Information).

In Vitro Autoradiography of [¹⁸F]13 in Brain Sections of Rat and NHP.

Rat brain cryosections (20 μm) were preincubated for 20 min in 50 mM Tris-HCl buffer (pH 7.4) at room temperature. These sections were incubated for 60 min at room temperature with Tris-HCl buffer containing [¹⁸F]13 (37 kBq/mL, 1.58 nM). After incubation, the brain sections were washed with cold buffer (0.5 mL) and dried with cold air. Then the sections were opposed to imaging plates (GE) overnight and autoradiograms were obtained using an Amersham Typhoon 5 (GE) analyzer system with the resolution of 10 μm and sensitivity of 4000. Self-blocking experiment with cold compound of 13 (10 μM, 0.1% DMSO, 1% ethanol in Tris-HCl buffer) was used to determine the nonspecific binding of [¹⁸F]13: before incubation with [¹⁸F]13, the sections were incubated with the solution of 13 for 30 min and then washed with cold Tris-HCl buffer (1 mL). In blocking experiments with BMT-108908 (10 μM, 0.1% DMSO, 1% ethanol in Tris-HCl buffer), the same procedure as self-blocking was used. The in vitro ARG in NHP brain sections including baseline, self-blocking, and BMT-108908 blocking experiments were carried out by following the same procedure as described above.

Distribution of radioactivity on the plates was analyzed by ImageJ, in which the regions of interest (ROIs) were placed on the hippocampus, frontal cortex, striatum, thalamus, and cerebellum in rat brain sections, and hippocampus, entorhinal cortex, insular cortex, caudate nucleus, putamen, thalamus and white matter of forebrain in NHP brain sections. The radioactivity in these regions was expressed as gray value/unit area, which is in linear with the radioactivity. The relative radioactivity was represented as the ratio of gray value/unit area (radioactivity) of different brain regions to that of the cerebellum (for rat) or white matter of forebrain (for NHP), where only slight radioactivity was observed.

Whole Body Biodistribution Study and Radiometabolite Analysis in Mice.

The CD1 (ICR) mice (female, 8 weeks, 23–27 g) were intravenously injected with [¹⁸F]13 (740 kBq/100 μL in saline containing 5% ethanol) through the tail vein. The mice were sacrificed at various time points of 2, 10, 30, and 60 min post-injection. The organs of interest were removed and weighed, and the radioactivity was measured with an automatic γ-counter (WALLAC/Wizard 1470, USA). The percent dose per gram (% ID/g) or percent dose per organ (% ID/organ) of wet tissue values were calculated by relating the tissue radioactivity counts to that of suitably diluted aliquots of the injected dose. All radioactivity measurements were decay-corrected.

After intravenous injection of [¹⁸F]13 through tail vein, the mice (n = 2) were sacrificed at 15 min. Blood sample was collected and centrifuged at 14000 rpm for 3 min at 4 °C to separate the plasma. The supernatant was collected and added to an ice-cooled test tube containing 100 μL of CH₃CN. After vortex for 10 s, the mixture was centrifuged at 14000 rpm for 3 min at 4 °C for deproteinization. The supernatant was collected, and the process was repeated until no precipitations were observed when CH₃CN was added. The mice brain was immediately dissected, homogenized with 400 μL of ice-cooled CH₃CN, and then centrifuged at 14000 rpm for 5 min at 4 °C. The supernatant was collected into a test tube containing 100 μL of ice-cooled CH₃CN, and the process (vortex and centrifuge) was repeated until no precipitations were observed when CH₃CN was added. The resulting supernatants of plasma and brain were mixed with 50 μL of a solution of cold compound [¹⁸F]13 and then analyzed on a radioHPLC system using a Phenomenex C18 column (5 μm, 10 × 250 mm, 5 mL/min, elution: CH₃CN/H₂O = 55%/45%). The radioactivity was collected and counted using 1480 Wizard gamma counter (PerkinElmer, USA). The radioactivity overlapped with [¹⁸F]13 in the UV signal corresponded to that of [¹⁸F]13(t_R = 6.90 min). The percentage of [¹⁸F]13 to total radioactivity was calculated as (counts for [¹⁸F]13/total counts) × 100. The whole body biodistribution study and radiometabolite analysis in mice of [¹⁸F]11 were conducted by following the same procedures for [¹⁸F]13.

ABBREVIATIONS

AD

Alzheimer’s disease

ARG

autoradiography

BBB

blood-brain barrier

CI

confidence interval

CNS

central nervous system

HPLC

high-performance liquid chromatography

iGluRs

ionotropic glutamate receptors

MPO

multiparameter optimization

NAM

negative allosteric modulators

NMDAR

N-methyl-D-aspartate receptors

NHP

non-human primates

PAINS

Pan Assay Interference Compounds Assay

PAM

positive allosteric modulators

PCP

phencyclidine

PET

positron emission tomography

RCY

radiochemistry yield

ROI

region of interest

SCIDY

spirocyclic iodonium ylide

S_NAr

nucleophilic aromatic substitution

TLC

thin-layer chromatography

tPSA

topological polar surface area