Synthesis and evaluation of 6-([¹¹C]methyl(4-(pyridin-2-yl)thiazol-2-yl)amino)benzo[d]thiazol-2(3H)-one for imaging γ-8 dependent transmembrane AMPA receptor regulatory protein by PET

Abstract

Transmembrane AMPA receptor regulatory proteins (TARPs) are a recently discovered family of proteins that modulate AMPA receptors activity. Based on a potent and selective TARP subtype γ−8 antagonist, 6-(methyl(4-(pyridin-2-yl)thiazol-2-yl)amino)benzo[d]thiazol-2(3H)-one (compound 9), we perform the radiosynthesis of its ¹¹C-isotopologue 1 and conduct preliminary PET evaluation to test the feasibility of imaging TARP γ−8 dependent receptors in vivo.

Graphical Abstract

Transmembrane AMPA Receptor Regulatory Proteins (TARPs) are a recently discovered family of proteins that modulate the activity of AMPA receptors in the ionotropic glutamate system. TARPs exhibit regional specific expression in the brain, leading to physiological differentiation and modulation of AMPA activity. Herein we develop a novel PET ligand based on a novel TARP subtype γ−8 dependent antagonist and perform preliminary evaluation by PET.

Ionotropic glutamate receptors (iGluRs) are ligand-gated ion channels, which are responsible for primary excitatory neurotransmission in the mammalian central nervous system (CNS).^1–3 Based on pharmacology and sequence similarity, iGluRs are divided into three subtypes: α-amino-3-hydroxyl-5-methyl-4-isoxazolepropionic acid receptors (AMPARs), N-methyl-d-aspartate receptors (NMDARs), and kainate receptors. It is reported that AMPARs play a critical role in learning and memory,¹ and are involved in the pathoetiology of several excitotoxic diseases, including ischemia and epilepsy.^4–5 These AMPARs have been found recently to function with the required regulative auxiliary proteins rather than alone.^6–8 These auxiliary proteins modulate gating, trafficking and pharmacology of AMPARs, which are classified as transmembrane AMPA receptor regulatory proteins (TARPs, including Type I: γ−2, γ−3, γ −4, γ−8, and Type II: γ−5, γ−7). Genetic analysis demonstrates a potential role of TARPs in depression, epilepsy, schizophrenia, bipolar disorder and neuropathic pain.^7,9 Several TARPs exhibit regiospecific patterns in the brain,^7,8 such as TARP γ−8 enriched in the hippocampus and TARP γ−2 localized primarily in the cerebellum, indicating the potential to regulate AMPAR activity by targeting individual TARP subtypes without generating unwanted side effects, such as sedation, ataxia, and dizziness.^[10] Recently, several TARP γ−8 targeted antagonists have been developed to potentially treat CNS excitotoxic diseases, including JNJ55511118,¹¹ JNJ-61432059,¹² and LY3130481/CERC-611.^13–15 Specifically, LY3130481/CERC-611 has been advanced to phase I clinical trial in 2017 (Fig. 1A).

Figure 1.

Representative TARP γ8-dependent AMPAR antagonists and PET tracer

Positron emission tomography (PET), as a non-invasive imaging technology, in combination with selective TARP radioligands, would offer a unique opportunity to study these receptors in vivo, measure biochemical and pharmacological process under normal and disease conditions, and facilitate drug development and target engagement study. Compared with continuous effort in the advancement of AMPAR PET ligands,^16–22 subtype-selective TARP PET tracer development is still at a nascent stage. Our group has been engaged in the development of TARP γ8-dependent PET tracers for years,²³ synthesizing a series of ¹¹C-labeled 1,3-dihydro-2H-benzo[d]imidazole-2-ones were developed, yet with moderate-to-high nonspecific binding and low brain uptake (Fig. 1B), representing an unmet clinical need. Herein we evaluate a novel molecule 1 (Fig. 1C) as a subtype-selective TARP γ−8 radioligand. Radioactive compound 1 is derived from ‘cold’ (non-radioactive) compound 9 (IC₅₀ = 16 nM), which is a close analog of TARP γ−8 antagonist LY3130481/CERC-611 in the clinical trial. Although it is disputable that a more potent compound may be needed to demonstrate high binding potential and high specific binding, based on favorable pharmacological and physiochemical characteristics of compound 9,²⁴ synthesis and preliminary studies of radioligand 1 were carried out as an entry point for feasibility study. Two carbon-11 labeling strategies were carried out, namely [¹¹C]CO₂ fixation for 10 and [¹¹C]CH₃I for 1. We then evaluated the brain permeability and binding specificity by PET imaging in vivo and autoradiography studies in vitro. Furthermore, the comparison between two labeling strategies for the desired product provides a radiochemical roadmap for the future design of new TARP γ−8 imaging agents.

Recently we developed a novel fully-automated method for the direct incorporation of [¹¹C]CO₂ into organic molecules.²⁷ Our ‘in-loop’ [¹¹C]CO₂ fixation method is simple, efficient and operates at ambient temperature and pressure. As an application of this method, we first designed the radiosynthetic strategy of 10 using [¹¹C]CO₂ fixation. The precursor 8 and standard compound 9 were prepared according to previous literatures with some modifications.^{13,14,25–27} As shown in Scheme 1, thiourea 3 was prepared via HCl-mediated condensation reactions between aniline 2 and potassium thiocyanate, followed by the treatment with 2-bromo-1-(pyridi-2-yl)ethan-1-one to afford intermediate 4 in 64% yield over 2 steps. The methylation of compound 4 mediated by sodium hydride gave compound 5 in 95% yield. The following K₂CO₃ promoted substitution reaction between 5 and para-methoxyl toluenethiol (PMBSH) generated key intermediate 6 in high yield (99%). In the presence of hydrogen (1 atm) and 10% Pd/C, compound 6 was reduced to amine 7 in 91% yield. PMBS group was then removed by trifluoroacetic acid to afford disulfide 8 (87% yield) without the detection of sulfide product 8a. In order to reduce the disulfide 8, we chose tributyl phosphine (PBu₃) among the reported reductants^28–38 avoiding the use of metal reductants. The sulfide 8a was successfully generated under the reduction of PBu₃ in situ, validated by the subsequent carboxylation to give product 9 in the presence of 1,1’-carbonyldiimidazole (CDI).³⁹ By means of the ‘shake flask method’, namely liquid-liquid partition between water and n-octanol,⁴⁰ the LogP value of 9 was determined to be 3.58 ± 0.04 (n = 3), slightly more lipophilic than desired range (1–3.5).^41–43 To our surprise, our PBu₃ method was not successfully translated from non-radioactive to radioactive conditions. Specifically, radiolabeled product 10 was merely formed in low radiochemical yields (RCYs <1%), in spite of several attempts using reported [¹¹C]CO₂ fixation conditions.^{24, 44–45}

Scheme 1.

Initial synthetic route to standard and labeling precursor for [¹¹C]CO₂ fixation. Reagents and conditions: (i) HCl, KSCN, 80 °C, 16 h; (ii) 2-bromo-1-(pyridin-2-yl)ethan-1-one hydrobromide, EtOH, reflux, 2 h, 64% over 2 steps; (iii) CH₃I, NaH, DMF, 0 °C - rt, 3 h, 95%; (iv) PMBSH, K₂CO₃, DMF, 100 °C, 16 h, 99%; (v) 10% Pd/C, H₂, EtOH, rt, 91%; (vi) TFA/CH₃SO₃H/anisole, 70 °C, 16 h, 87%; (vii) PBu₃, tol. Ar, rt, 10 min, then CDI, 100 °C, 24 h, 78%; (viii) PBu₃, tol. Ar, rt, 5 min, then [¹¹C]CO₂, BEMP, 100 °C, 5 min, <1% RCY. DMF = N,N-dimethylformamide; PMBSH = 4-methoxy-toluenethiol; Ar = argon; CDI = 1,1’-Carbonyldiimidazole; BEMP = 2-tert-Butylimino-2-diethylamino-1,3-dimethylperhydro-1,3,2-diazaphosphorine.

As the [¹¹C]carboxylation labeling strategy failed to afford the desired radioactive product 10 in a reasonable amount, we shifted our attention to the formation of product 1 via ¹¹C-methylation (Scheme 2). Specifically, chemical scaffold of 3-(methoxymethyl)-6-((4-(pyridin-2-yl)thiazol-2-yl)amino)benzo [d]thiazol-2(3H)-one (16) offers an opportunity to utilize nucleophilic [¹¹C]CH₃-labeling methodology.⁴⁶ Thus we adopted this method to carry out radiosynthesis of 1 from compound 16 as the precursor. To label nitrogenous CH₃ of 9 selectively, a benzothiazolone protecting group was used to avoid the problem of radiolabeling the two unwanted nitrogens. As shown in Scheme 2, an acid-labile protecting group, methoxymethyl (MOM), was introduced in the first step, which yielded the intermediate 12 in 95%, followed by reduction of nitro group to afford amine 13. Benzoyl isothiocyanate was used to functionalize 13 to afford 14 as a thiourea. After the removal of benzoyl group, condensation reactions between 15 and 2-bromo-1-(pyridin-2-yl)ethan-1-one hydrobromide provided intermediate 16 in 32% yield over 4 steps according to the literature with some modification in this work.¹⁶ Methylation of secondary amine 16 with methyl iodide gave nonradioactive labeling intermediate 17 in 79% yield, which was converted to product 9 by the removal of MOM protecting group at prolonged reaction time. ¹¹C-Methylation occurred smoothly in the presence of sodium hydroxide and [¹¹C]CH₃I in DMF for 5 min to afford radioactive intermediate 18. It is noteworthy that TFA¹³ is not an ideal acid to remove MOM group for radiosynthesis attributed to low conversion and long reaction time (32% yield over 3 days in reflux). Therefore, our focus turned into the optimization for MOM deprotection in the subsequent radiosynthetic work.

Scheme 2.

Second synthetic approach of standard and labeling precursor for ¹¹CH₃I method. Reagents and conditions: (i) chloromethyl methyl ether, NaH, DMF, 0 °C, 2 h, 95%; (ii) 10% Pd/C, H₂, EtOH, rt, 16 h; (iii) benzoyl isothiocyanate, acetone, rt, 3 h; (iv) NaOH, CH₃OH/H₂O, rt, 3 h; (v) 2-bromo-1-(pyridin-2-yl)ethan-1-one hydrobromide, DIPEA, DMF, 100 °C, 1 h, 32% over 4 steps; (vi) CH₃I, NaH, DMF, 0 °C, 3 h, 79%; (vii) TFA, reflux, 3 d, 32%. DMF = N,N-dimethylformamide; DIPEA = N,N-Diisopropylethyl-amine; TFA = Trifluoroacetic acid.

We first tested a series of acid-mediated deprotection methods under non-radioactive conditions to provide guidance for the MOM group removal (see ESI, Table S1). Of all non-radioactive conditions tested, we selected three representative acids to test their deprotection efficiency under radioactive conditions. Unsurprisingly, TFA failed even though we increased the temperature (80 °C) and ratio of acid (1:1 ratio of TFA/DMF, Table 1, entry 1). Hydrochloride (6N) gave the similar results as TFA (Table 1, entry 2). Fortunately, TfOH was able to remove the MOM protecting group efficiently, as indicated in non-radioactive tests (Table S1, entry 7) and under radioactive conditions in one-pot manner (Table 1, entry 3). The reaction mixture was then quenched and neutralized by addition of 1M NaOH aqueous solution, and purified by a reverse phase semi-preparative HPLC, followed by reformulation. Ultimately, compound 1 was synthesized in 39% decay-corrected radiochemical yield (RCY) based on the starting [¹¹C]CO₂ with >99% radiochemical purity (n = 5). The molar activity was greater than 74 GBq/μmol (2.0 Ci/μmol) at the end-of-synthesis (EOS). No sign of radiolysis was observed up to 90 min after re-formulation. The successful radiosynthesis of 1 with high radiochemical purity and excellent molar activity enabled our subsequent in vitro and in vivo evaluation.

Table 1.

One-pot reaction for radiosynthesis of 1

Conditions:
entry	Acid/solvent(v/v)	T	t	results of 1
1	TFA/DMF (1/1)	80 °C	5 min	NDa
2	6N-HCI/DMF (1/1)	100 °C	5 min	NDa
3	TfOH/DMF (1/1)	120 °C	5 min	39% (decay-corrected)

notes:

^aND = not observed.

The in vivo brain kinetics of 1 was evaluated by PET dynamic scans in Sprague-Dawley rats for 60 min. Representative PET images of 1 (coronal and sagittal, summed images 0–60 min) and time-activity curves of four brain regions are shown in Figure 2. The results showed an initial brain uptake with a SUV close to 1, but with marginal regional difference between the TARP γ−8 enriched hippocampus, and TARP γ−8 deficient cerebellum. High non-specific binding was observed and confirmed by self-blocking experiments in Fig. S1 (see ESI), which showed no substantial changes between baseline and blocking conditions.

Figure 2.

Representative PET/MR summed images (0–60 min) of tracer 1 in rat brain. (A) baseline and (B) time activity curves of regional brain at baseline.

We next carried out autoradiography studies to evaluate target binding in vitro. As shown in Fig. 3, autoradiograms on sagittal sections of rat brains (baseline) indicated minimal regional distribution. Self-blocking studies with cold compound 9 (10 μM) resulted in 30–50% reduction in the hippocampus, cerebral cortex, striatum and cerebellum. Another structurally-distinct TARP γ−8 antagonist, JNJ-55511118,¹⁴ failed to demonstrate any obvious change in quantitative measurement of radioactivity (Fig. 3B). Autoradiography results showing low binding specificity of 1 in vitro, were consistent with PET imaging results. Our results suggest that 1 might bind to a different binding pocket on the TARP γ−8 protein from JNJ-5551118, or that it binds to additional target(s).

Figure 3.

In vitro autoradiography of tracer 1 in rat brain sections. (A) Representative autoradiograms in rat brain sagittal sections: 1 (baseline), pretreatment by compound 9 and JNJ-55511118. (B) Quantitative analysis of control and blocking experiments. CCx, Cerebral cortex; HIP, hippocampus; STR, striatum; Cb, cerebellum. Statistical Analysis: Statistical analysis was performed by a student’s two-tailed t-test, and asterisks were used to indicate statistical significance: *p < 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001.

In summary, we have evaluated two radiochemical methods to prepare a ¹¹C-labeled labeled TARP γ−8 antagonist (compound 1; also known as TARP-1903, IC₅₀ 16 nM) based on a lead drug scaffold LY3130481/CERC-611. ¹¹C-Methylation methods, albeit in two steps, outperformed the [¹¹C]CO₂ fixation method due to challenge associated with the sulfide precursor 8a. Ultimately, the desired compound 1 was labeled by [¹¹C]CH₃I in high radiochemical yield (ca. 40%), high molar activity (>74 GBq/μmol) and high radiochemical purity (>99%). While the PET ligand showed sufficient brain penetration, a relatively homogeneous brain distribution indicated low specific binding, which was confirmed by the subsequent in vitro autoradiography. Because the ligand demonstrated low specific binding and moderate brain permeability, further search to obtain new lead to visualize the TARP γ−8 proteins in the brain is needed.