In vitro evaluation of PET radiotracers for imaging synaptic density, the acetylcholine transporter, AMPA-tarp-γ8 and muscarinic M4 receptors in Alzheimer’s disease

Abstract

Several therapeutics and biomarkers that target Alzheimer’s disease (AD) are under development. Our clinical positron emission tomography (PET) research programs are interested in six radiopharmaceuticals to image patients with AD and related dementias, specifically [¹¹C]UCB-J and [¹⁸F]SynVesT-1 for synaptic vesicle glycoprotein 2A as a marker of synaptic density, two vesicular acetylcholine transporter PET radiotracers: [¹⁸F]FEOBV and [¹⁸F]VAT, as well as the transmembrane AMPA receptor regulatory protein (TARP)-γ8 tracer, [¹⁸F]JNJ-64511070, and the muscarinic acetylcholine receptor (mAChR) M4 tracer [¹¹C]MK-6884. The goal of this study was to compare all six radiotracers (labeled with tritium or ¹⁸F) by measuring their density variability in pathologically diagnosed cases of AD, mild cognitive impairment (MCI) and normal healthy volunteer (NHV) human brains, using thin-section in vitro autoradiography (ARG). Region of interest analysis was used to quantify radioligand binding density and determine whether the radioligands provide a signal-to-noise ratio optimal for showing changes in binding. Our preliminary study confirmed that all six radiotracers show specific binding in MCI and AD. An expected decrease in their respective target density in human AD hippocampus tissues compared to NHV was observed with [³H]UCB-J, [³H]SynVesT-1, [³H]JNJ-64511070, and [³H]MK-6884. This preliminary study will be used to guide human PET imaging of SV2A, TARP-γ8 and the mAChR M4 subtype for imaging in AD and related dementias.

Introduction

Losses of neurons, alterations of synapses and neurotransmission deficits are associated with many neurodegenerative disorders including Alzheimer’s disease (AD). The loss of synaptic integrity observed in the brain of AD patients, especially in the frontal cortex and hippocampus regions, has been shown to be an important hallmark of AD, and leads to the downregulation of several presynaptic and postsynaptic proteins [1]. As a result, several efforts have been made to evaluate different targets in the brain and develop high-affinity and brain-penetrant positron emission tomography (PET) radiotracers to quantify changes in both normal and pathological conditions in AD. This study focused on four brain targets: the synaptic vesicle glycoprotein 2 (SV2A), the vesicular acetylcholine transporter (VAChT), the transmembrane AMPA receptor regulatory protein (TARP)-γ8, and the M4 subtype of muscarinic acetylcholine (ACh) receptors which play a crucial role in neurotransmission as their altered expression strongly correlates with severity of AD pathologies.

As one of the membrane proteins on the synaptic vesicles, expressed in high concentration throughout the brain, SV2A is vital to neurotransmission and has been of major interest to detect reduced synaptic density in living patients with AD pathologies using PET [2-4]. A first-generation SV2A radiotracer for PET is [¹¹C]UCB-J (Figure 1), which showed high specific binding signals and high binding kinetics in the human brain and was validated as a viable biomarker for synaptic density imaging [5-7]. With the increased efforts invested in the development of ¹⁸F-labeled analogues of UCB-J for widespread human translation, [¹⁸F]SynVesT-1 (Scheme 1, also known as [¹⁸F]SDM-8 or [¹⁸F]MNI-1126) was developed as a promising SV2A biomarker, presenting similar pharmacokinetics as [¹¹C]UCB-J, with the advantages of ¹⁸F-labeling [4,8,9].

Figure 1.

PET radiotracers and respective targets for clinical research interest in the present study.

Located in presynaptic terminal vesicles of cholinergic neurons, VAChT regulates the synthesis and release of acetylcholine [10]. Degeneration of cholinergic cells in the brain have been shown to correlate with AD pathologies and perturbs cholinergic neurotransmission in the cortex and hippocampus [11,12]. Thus, the quantification of VAChT by PET has been established as a potential tool for diagnosis of AD [13-17]. Vesamicol and derivatives have been evaluated as specific ligands for VAChT [18-20]. Despite efforts focused on optimizing the chemical structure to increase the selectivity, in vivo kinetics, and metabolic stability, only a few ¹¹C- and ¹⁸F-labeled ligands have progressed to PET imaging of non-human primate and human subjects [21-24]. Two promising PET radiopharmaceuticals are (-)-5-[¹⁸F]fluoroethoxybenzovesamicol ([¹⁸F]FEOBV, Figure 1) which showed a decrease in tracer uptake in preclinical models with cholinergic deficit and in AD patients [22,25-28] and [¹⁸F]VAT (Figure 1), which is demonstrated to be a suitable PET radioligand for imaging VAChT in vivo [29-32] and has been recently translated to human studies (https://clinicaltrials.gov/study/NCT05034263).

The transmembrane α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionic acid (AMPA) receptor regulatory protein γ8 (TARP-γ8) is a subunit of the AMPA receptors, ligand gated ion channels subgroup of the ionotropic glutamate receptors. TARP-γ8, predominantly found in the hippocampus, plays a crucial role in glutamatergic synaptic neurotransmission and plasticity, and has emerged as a promising target for PET [33-39]. Despite considerable efforts, the early generations of TARP-γ8 ligands suffered from poor CNS permeability, and high non-specific binding in vivo [40,41]. Encouraging results were obtained for the TARP-γ8 candidates JNJ-55511118 [42] and LY-3130481, and the latter was advanced to Phase I clinical trials in 2017 [43-46]. More recently, the synthesis of selective TARP-γ8 ligands based on a benzothiazolone or an indazole scaffolds were reported and have high brain uptake and high specific binding to TARP-γ8 [47,48]. [¹⁸F]JNJ-64511070 (Scheme 1) is one of the promising candidate PET radiotracers, which exhibits desirable characteristics for the quantification of TARP-γ8 in the brain [49].

Muscarinic acetylcholine receptors (mAChRs) play a crucial role in the cholinergic pathway in mediating the actions of ACh [50]. Targeting mAChRs has demonstrated potential in the treatment of CNS disorders by alleviating psychosis and behavioral disorders [51]. Albeit, with five distinct mAChRs subtypes (M1-M5), ligand selectivity is crucial to ensure success in further clinical development [50]. The mAChR M4 subtype is expressed in key brain regions such as the striatum, hippocampus, and cortex [50]. Previous studies on this M4 subtype have demonstrated changes in its expression in neurodegenerative diseases such as AD, making it a viable target for disease diagnosis [50,52]. The PET ligand [¹¹C]MK-6884 has shown selectivity for mAChR M4 over the other subtypes and a high brain uptake [53]. Recent clinical research in AD patients revealed the ability of [¹¹C]MK-6884 to show changes in mAChR M4 density and may provide crucial insights into neurodegenerative diseases diagnosis [54].

The objectives of this in vitro autoradiography (ARG) [55] study are: i) to compare the binding properties of two SV2A selective PET tracers, UCB-J and SynVesT-1, in post-mortem human brain tissues and quantify the reduction of synaptic density in AD cases; ii) to compare FEOBV and VAT for assessment of VAChT expression and presynaptic density in AD diagnosis; iii) to characterize JNJ-64511070 as a PET ligand for the quantification of TARP-γ8 in AD brains; iv) to determine the specific distribution of muscarinic M4 receptors via MK-6884 binding in AD tissues. To achieve these objectives, thin-section ARG binding assays were performed to compare the regional distribution of [³H]UCB-J, [³H]SynVesT-1, [¹⁸F]FEOBV, [³H]VAT, and [³H]JNJ-64511070 in post-mortem tissue from five AD cases, three mild cognitive impairment (MCI) cases and three normal healthy volunteer (NHV) subjects in the cerebellum (CRB), the prefrontal cortex (PFC) and the hippocampus (HIP).

Results

Post-mortem healthy rodent (Figure S1) and human brain tissues were used to establish the distribution of the binding of all ligands in MCI and AD cases. Information regarding the gender and age of the subjects, as well as brain regions provided are summarized in Table 1. For this ARG binding study, the ¹⁸F-labeled FEOBV was synthesized, and the four other radiotracers were obtained as tritium-labeled ligands.

Table 1.

Summary of the demographic data for the post-mortem brain samples used in the experiments

Conditions	Age (years)	Brain regions
AD	96	CRB, PFC
	67	CRB, PFC, HIP
	86	CRB, PFC
	88	HIP
	90	HIP
Average	85.4	-
MCI	86	CRB, PFC, HIP
	102	CRB, PFC, HIP
	79	CRB, PFC, HIP
Average	89	-
NHV	85	CRB, PFC, HIP
	72	CRB, PFC, HIP
	100	CRB, PFC
Average	85.67	-

Synaptic density

We evaluated the two tritium-labeled PET tracers, [³H]UCB-J and [³H]SynVesT-1, targeting SV2A in the three brain regions: HIP, PFC and CRB. Thin-section ARG was performed to evaluate the specific signal of the radiotracers to determine whether the radioligands provide a signal-to-noise ratio optimal for showing changes in radioligand binding between NHV, MCI and AD tissues (Figure 2). Both radiotracers displayed high specific binding when displaced using levetiracetam (200 μM), a known ligand for the SV2A receptor [56] in NHV (93.64 ± 1.94%, n = 2 for [³H]UCB-J and 89.35 ± 2.50%, n = 2 for [³H]SynVesT-1), MCI (93.41 ± 2.51%, n = 3 for [³H]UCB-J and 90.18 ± 8.26%, n = 3 for [³H]SynVesT-1) and AD (93.85 ± 2.55%, n = 3 for [³H]UCB-J and 86.71 ± 10.49%, n = 3 for [³H]SynVesT-1) tissues (Figures 2, S2, S6 and S10).

Figure 2.

Representative autoradiograms of SV2A total binding sites in the HIP, PFC and CRB of NHV, MCI and AD patients for [³H]UCB-J (left) and [³H]SynVesT-1 (right); Quantification of the ARG signal in HIP, PFC, and CRB, for [³H]UCB-J and [³H]SynVesT-1 in NHV (dark blue), MCI (light blue) and AD (orange) post-mortem human brain sections. These slides were exposed to a phosphor sensor sheet for 4 days for [³H]UCB-J and [³H]SynVesT-1, and then scanned for visualizing and quantifying the ARG signal.

At equal ligand concentration, [³H]UCB-J displayed a higher binding compared to [³H]SynVesT-1 in AD, MCI and NHV tissues (Figure 2, in NHV (n = 2, dark blue), specific binding for [³H]UCB-J: 145.56 ± 18.71 μCi/g versus 45.99 ± 5.72 μCi/g for [³H]SynVesT-1). The ROI analysis showed a decrease in signal generated in MCI and AD tissues compared with NHV tissue sections for both [³H]UCB-J and [³H]SynVesT-1 in HIP. A depletion of 11% and 17% was observed in MCI and AD sections respectively for [³H]UCB-J (Figure 2, specific binding: 145.56 ± 18.71 μCi/g in NHV tissues (n = 2, dark blue), 129.51 ± 35.17 μCi/g in MCI samples (n = 3, light blue) and 120.49 ± 16.40 μCi/g in AD cases (n = 3, orange)). The decrease was 5% and 37% in the same tissues for [³H]SynVesT-1 (Figure 2, specific binding mean: 45.99 ± 5.72 μCi/g in NHV tissues (n = 3, dark blue), 43.59 ± 10.71 μCi/g in MCI samples (n = 3, light blue) and 29.12 ± 12.62 μCi/g in AD cases (n = 3, orange)).

VAChT

We evaluated [¹⁸F]FEOBV and [³H]VAT which both target VAChT, in the HIP, PFC and CRB (Figures 3, S3, S7 and S11). With previously reported conditions for the in vitro autoradiography binding techniques [26,32], [¹⁸F]FEOBV at 0.1 nM and [³H]VAT at 10 nM displayed higher binding in all investigated gray matter regions than in the white matter. In NHV tissues, the binding for [¹⁸F]FEOBV reached 0.65 ± 0.02 μCi/g in the PFC (n = 3), 0.73 ± 0.03 μCi/g in the CRB (n = 3) and was highest in the HIP at 1.26 ± 0.49 μCi/g (n = 2) (Figure 3). Similarly for [³H]VAT in NHV tissues, the highest binding was found in the HIP at 16.27 ± 2.16 μCi/g (n = 2) and reached 6.59 ± 0.36 μCi/g in the PFC (n = 3) and 4.53 ± 0.56 μCi/g in the CRB (n = 3). Additionally, high specific binding was observed for [¹⁸F]FEOBV and [³H]VAT towards VAChT in the HIP (84.28 ± 3.25% for [¹⁸F]FEOBV and 74.08 ± 7.29% for [³H]VAT), the PFC (57.93 ± 10.25% for [¹⁸F]FEOBV and 45.92 ± 6.36% for [³H]VAT) and the CRB (63.56 ± 2.79% for [¹⁸F]FEOBV and 42.29 ± 5.69% for [³H]VAT). In the sample size used for this study (n = 3), no significant reductions in binding were observed for either [¹⁸F]FEOBV nor for [³H]VAT, in the MCI or AD cases in comparison to the NHV cases.

Figure 3.

Representative autoradiograms of VAChT total binding sites in the HIP, PFC and CRB of NHV, MCI and AD patients for [¹⁸F]FEOBV (left) and [³H]VAT (right); Quantitative analysis of specific binding (μCi/g) in HIP, PFC, and CRB for [¹⁸F]FEOBV (left) and [³H]VAT (right), in NHV (dark blue), MCI (light blue) and AD (orange) human post-mortem brain sections. These slides were exposed to a phosphor sensor sheet for 20 min for [¹⁸F]FEOBV or 14 days for [³H]VAT, and then scanned for visualizing and quantifying the ARG signal.

TARP-γ8

To characterize [³H]JNJ-64511070 distribution in AD, MCI and NHV tissues, ARG was performed at a 2.5 nM radioligand concentration. The binding of [³H]JNJ-64511070 was evaluated in the same sample cases as previously described. To the best of our knowledge, this is the first report of the binding of this tracer using in vitro thin section ARG. High specific binding was observed for [³H]JNJ-64511070 towards TARP-γ8 in the presence of 10 μM of JNJ-64511070 in the HIP (92.32 ± 2.94%), PFC (67.88 ± 4.60%) and CRB (35.75 ± 11.04%) tissues (Figures 4, S4, S8 and S12).

Figure 4.

Representative autoradiograms of TARP-γ8 total binding sites in the HIP, PFC and CRB of NHV, MCI and AD patients; Quantification of the ARG signal for [³H]JNJ-64511070 in the HIP, PFC, and CRB, in NHV (dark blue), MCI (light blue) and AD (orange) post-mortem human brain sections. These slides were exposed to a phosphor sensor sheet for 5 days for [³H]JNJ-64511070, and then scanned for visualizing and quantifying the ARG signal.

When quantifying the binding, the TARP-γ8 compound revealed the lowest brain binding in AD and MCI brain tissues compared to NHV and showed 41% and 56% losses in the hippocampus in MCI and AD tissues, respectively (Figure 4, HIP, specific binding: 98.83 ± 46.25 μCi/g in NHV tissues (n = 2, dark blue), 58.56 ± 19.75 μCi/g in MCI samples (n = 3, light blue) and 43.27 ± 4.69 μCi/g in AD cases (n = 3, orange)). Binding in MCI and AD PFC sections was also found to be lower than in NHV tissues with a decrease of 8% and 15%, respectively (Figure 4, PFC, specific binding: 18.90 ± 0.44 μCi/g in NHV tissues (n = 3, dark blue), 17.30 ± 2.22 μCi/g in MCI samples (n = 3, light blue) and 16.10 ± 2.31 μCi/g in AD cases (n = 3, orange)).

mAChR M4 subtype

The [³H]MK-6884 distribution in MCI and AD tissues was demonstrated by ARG at 2.5 nM radioligand concentration in the presence of the cholinergic agonist carbachol (10 μM). Carbachol is a cholinergic agonist like acetylcholine activating muscarinic receptors increasing the affinity of [³H]MK-6884 for the mAChR M4 subtype [53]. Due to a low overall binding, a moderate specific binding was observed for [³H]MK-6884 towards M4 in the presence of 10 μM of MK-6884 in the hippocampus (59.48 ± 9.25%), PFC (60.60 ± 4.16%) and CRB (21.31 ± 4.31%) tissues (Figures 5, S5, S9 and S13). Nevertheless, a decrease in binding of 27% and 41%, respectively, was observed in MCI and AD tissues compared to NHV tissues in the hippocampus region (Figure 5, HIP, specific binding: 4.31 ± 0.83 μCi/g in NHV tissues (n = 2, dark blue), 3.15 ± 0.39 μCi/g in MCI samples (n = 3, light blue) and 2.55 ± 0.53 μCi/g in AD cases (n = 3, orange)).

Figure 5.

Representative autoradiograms of M4 total binding sites in the HIP, PFC and CRB of NHV, MCI and AD patients; Quantification of the ARG signal for [³H]MK-6884 in HIP, PFC and CRB, in NHV (dark blue), MCI (light blue) and AD (orange) post-mortem human brain sections. These slides were exposed to a phosphor sensor sheet for 5 days for [³H]MK-6884, and then scanned for visualizing and quantifying the ARG signal.

Discussion

Changes in synaptic integrity in the hippocampus and prefrontal cortex is considered to be an important hallmark of AD [1]. This translates to the downregulation of several synaptic proteins [1]. In previous studies, [¹¹C]UCB-J detected reduced synaptic density in living patients with AD and MCI with an uptake reduction of 41% compared to NHV [57], while [¹⁸F]SynVesT-1 was shown to have high brain uptake and high specific binding comparable to UCB-J [58]. [³H]UCB-J and [³H]SynVesT-1 were evaluated in NHV, MCI and tissues in our in vitro ARG study. All tissues showed reduction in binding for both tracers, with the greatest reduction in AD tissues for [³H]SynVesT-1 (-37%) in comparison to [³H]UCB-J (-17%). These results support previous ARG studies with [³H]UCB-J including by Patel et al. who reported a reduction in cortex and hippocampus [61] as well as Mikkelsen et al. [62] who showed an 11% decrease in PFC. Our work is also consistent with the high specific binding shown by Kumar et al. [63], but contrasts with their data and work by Metaxas et al. [64] which did not find decreased binding of [³H]UCB-J, albeit different experimental conditions need to be considered (large vs. small brain sections, different blocking agents and/or concentrations, etc.). Our study reinforces the important role of [¹⁸F]SynVesT-1 for synaptic density evaluation in the living brain with applications in AD. [¹⁸F]SynVesT-1 also offers the advantage of a ¹⁸F-labeled tracer, and facilitates longer scanning protocols and multi-center clinical trials, and is being evaluated for human PET imaging studies at several laboratories, including ours [59,60].

[¹⁸F]FEOBV and [¹⁸F]VAT are two promising VAChT-specific radiotracers with potential for imaging AD. As VAChT expression is considered to correlate with neurodegenerative disease progression, PET imaging with a suitable ligand would be a useful tool to assess the loss of cholinergic neurons in living subjects. Another goal of this study was to directly compare the binding ability of both [¹⁸F]FEOBV and [³H]VAT in the same tissue samples using in vitro ARG, to guide our clinical translation studies. ARG with [¹⁸F]FEOBV has previously demonstrated a decrease in binding of 33% in the PFC (11 AD, 13 NHV) and 20% to 25% in hippocampus regions (8 AD, 13 NHV) in AD subjects compared to healthy tissues [65]. However, it is important to note the 20% to 40% overlap between AD and NHV tissues values. As some AD hallmarks have also been shown to correlate with ageing in rodents, including the synaptic dysfunction and changes in cholinergic neurons [66,67], binding variation within the same group is variable as expected. This data is consistent with the present study, as we also observed an overlap between AD and NHV values for both [¹⁸F]FEOBV and [³H]VAT tracer binding levels between AD, MCI, and NHV brain sections. However, in our limited sample, these tracers were not sufficiently sensitive to differentiate AD from NHV.

Using the same human tissues, we evaluated the binding ability of [³H]JNJ-64511070 in showing changes in TARP-γ8 expression between AD, MCI and NHV. To our knowledge, this is the first study investigating the binding of this tracer using ARG. [³H]JNJ-64511070 was demonstrated to target TARP-γ8 with high specificity, with the highest binding in the hippocampal region. In our sample selection, the decrease in binding of [³H]JNJ-64511070 in NHV, MCI and AD tissues was consistent with the degeneration in the glutamatergic pathway expected in AD pathologies. This approach may provide a direct assessment of TARP-γ8 expression in neurodegenerative diseases, our in vitro data supports the use of [¹⁸F]JNJ-64511070 a promising radiotracer for AD diagnosis in vivo.

[¹¹C]MK-6884, a promising radiotracer targeting the mAChR M4 subtype, was evaluated in the same AD, MCI and NHV samples. Selectivity for M4 was shown as the tracer was displaced in the presence of MK-6884. A decrease in muscarinic M4 receptor density was observed with [³H]MK-6884 in human AD and MCI hippocampus tissues compared to NHV. These results confirmed the previous study obtained in vivo in patients with AD by Li et al. [54], in which an average decrease of 20% was obtained in individuals with AD compared to healthy adults. These data reveal the potential of [¹¹C]MK-6884 in providing crucial insights into neurodegenerative diseases diagnosis.

Limitations of the present study include tissue availability; a larger sample size would allow further interpretation of the radiotracers binding to AD pathology, as this work reveals variability in binding of several radiotracers between NHV, MCI and AD tissues. It would also be of value to evaluate sex differences and additional brain regions for analysis in future studies to explore PET tracer binding beyond the hippocampus, prefrontal cortex, and cerebellum in AD, for a greater representation of whole-brain imaging achieved with in vivo PET imaging studies.

Conclusions

Our preliminary work reveals that [³H]UCB-J, [³H]SynVesT-1, [³H]JNJ-64511070, and [³H]MK-6884, showed a decrease in their respective target density in human AD hippocampus tissues compared to NHV. This work will be used to guide our clinical research PET imaging programs and will continue to explore synaptic density via SV2A, TARP-γ8 and the mAChR M4 subtype for imaging in AD and related dementias.

Materials and methods

(2R,3R)-5-(2-tosyloxyethoxy)benzovesamicol and (2R,3R)-5-fluoroethoxy-benzovesamicol were provided by Eisai Inc (Nutley, NJ, USA). The reference compounds and associated tritiated radioligand [³H]VAT (49 Ci/mmol), [³H]JNJ-64511070 (34 Ci/mmol), [³H]UCB-J (79 Ci/mmol) were provided by Enigma Biomedical Group (Boston, USA), MedChem Imaging (Boston, USA) and/or Novandi Chemistry AB (Södertälje, Sweden), and [³H]SynVesT-1 (93 Ci/mmol) was purchased from Pharmaron (Cardiff, UK). TARP-γ8 antibody was purchased from Synaptic Systems (Goettingen, Germany).

[¹⁸F]FEOBV was synthesized based on a method described previously [68,69], with minor modifications. Briefly, reaction of the tosyloxy precursor, (2R,3R)-5-(2-tosyloxyethoxy)benzovesamicol, with [¹⁸F]fluoride ion in dimethylsulfoxide followed by purification using preparative high-pressure liquid chromatography (45% MeCN:55% aq. 50 mM ammonium acetate, 6 mL/min, t_R = 23 min) provided the desired [¹⁸F]FEOBV with a decay-corrected radiochemical yield of 8 ± 2% (n = 5), a high radiochemical purity (>99%) and a molar activity of 209 ± 66 GBq/μmol at the end of synthesis. The synthesis was fully automated using a GE TRACERlab FX_FN synthesis module, and the synthesis time was 60 ± 5 min from the end of bombardment.

In vitro autoradiography

Frozen brain samples from NHV, MCI and AD-positive brains were obtained from Eisai Inc. or in-house, in accordance with the guidelines put forth by the Centre for Addiction and Mental Health Research Ethics Board (protocol 036-2019). In total, 26 samples from 11 individuals between 67 and 102 years old were provided: 8 samples from the hippocampus (2 NHV, 3 MCI, 3 AD), 9 from the prefrontal cortex (PFC) (3 NHV, 3 MCI, 3 AD) and 9 from the cerebellum (3 NHV, 3 MCI, 3 AD). Using a freezing sliding microtome (CryoStar NX50) at -16°C, each block of tissue was cut into serial 10 μm-thick sections, which were then thaw-mounted on coated microscope slides (Fisher Superfrost Plus Gold) and stored at -80°C for later use.

The optimal conditions for the in vitro ARG binding techniques such as incubation time, washing procedure, and time of exposition to the phosphor imaging plate were obtained after preliminary experiments on rodent brain. The sections were then exposed on a radioluminographic imaging plate (Fujifilm BAS-TR2040, GE Healthcare) for the indicated time. The autoradiography signal was visualized on an Amersham Typhoon phosphorimager (GE Healthcare, USA). The radioactivity was quantified using MCID 7.0 imaging suite (Interfocus Imaging, Cambridge, UK) by drawing regions of interest (ROIs). The binding was expressed in μCi/g for [¹⁸F]FEOBV and for the ³H-labeled tracers, using commercial tritium standards (American Radiolabeled Chemicals Inc.; St. Louis, USA). Percent specific binding (% Specific binding = ((Total signal - Non-specific signal)/(Total signal)) *100) is reported.

[³H]UCB-J, procedure adapted from Patel et al. [61]

Brain sections were warmed up to room temperature, and preincubated in the binding buffer (50 mM Tris-HCl buffer containing 140 mM NaCl, 5 mM KCl, 1.5 mM MgCl₂.6H₂O and 1.5 mM CaCl₂.H₂O at pH 7.4) for 15 min. Tissues were incubated with 2 nM of [³H]UCBJ (for total binding), or with the addition of 200 μM of levetiracetam as a blocking agent (for non-specific binding) for 1 h at room temperature. After incubation, the slides were washed twice for 1 min in ice-cold binding buffer, rinsed in ice-cold distilled water, air-dried, and exposed for 4 days.

[³H]SynVesT-1, procedure adapted from Patel et al. [61]

Brain sections were warmed up to room temperature, and preincubated in the binding buffer (50 mM Tris-HCl buffer containing 140 mM NaCl, 5 mM KCl, 1.5 mM MgCl₂.6H₂O and 1.5 mM CaCl₂.H₂O at pH 7.4) for 15 min. Tissues were incubated with 2 nM of [³H]SynVesT-1 (for total binding), or with the addition of 200 μM of levetiracetam as a blocking agent (for non-specific binding) for 1 h at room temperature. After incubation, the slides were washed twice for 1 min in ice-cold binding buffer, rinsed in ice-cold distilled water, air-dried, and exposed for 4 days.

[¹⁸F]FEOBV, procedure adapted from Parent et al. [65]

Brain sections were warmed up to room temperature, and preincubated in a phosphate-buffered saline solution for 20 min. Tissues were then incubated with 0.1 nM of [¹⁸F]FEOBV (for total binding), or with addition of 10 μM of cold self as a blocking agent (for non-specific binding), in the same buffer solution for 20 min at room temperature. After incubation, the slides were rinsed in ice-cold distilled water, air-dried, and exposed for 5 minutes.

[³H]VAT, procedure adapted from Liang et al. [32]

Brain sections were warmed up to room temperature, and preincubated in the binding buffer (50 mM Tris-HCl buffer containing 1.2 mM NaCl and 1 mM EDTA at pH 7.4) for 20 min. Tissues were incubated with 10 nM of [³H]VAT (for total binding), or with the addition of 10 μM of cold “self” as a blocking agent (for non-specific binding) for 1 h at room temperature. After incubation, the slides were washed twice for 4 min in ice-cold binding buffer, rinsed in ice-cold distilled water, air-dried, and exposed for 8 days.

[³H]JNJ-64511070, procedure adapted from Chen et al. [40]

Brain sections were warmed up to room temperature, and preincubated in a phosphate-buffered saline solution for 20 min. Tissues were incubated with 2.5 nM of [³H]JNJ-64511070 (for total binding) in the binding buffer (50 mM Tris-HCl buffer containing 1.2 mM NaCl and 2 mM MgCl₂.6H₂O at pH 7.4), or with the addition of 10 μM of cold “self” as a blocking agent (for non-specific binding) for 1 h at room temperature. After incubation, the slides were washed 3 times for 2 min in ice-cold binding buffer, rinsed in ice-cold distilled water, air-dried, and exposed for 5 days.

[³H]MK-6884

Brain sections were warmed up to room temperature, and preincubated in the binding buffer (50 mM Tris-HCl buffer containing 1.2 mM NaCl, 2 mM KCl, 2 mM MgCl₂.6H₂O and 1 mM CaCl₂ at pH 7.4) for 20 min. Tissues were incubated with 2.5 nM of [³H]MK-6884 (for total binding), or with the addition of 10 μM of cold self as a blocking agent (for non-specific binding) for 1 h at room temperature. After incubation, the slides were washed twice for 4 min in ice-cold binding buffer, rinsed in ice-cold distilled water, air-dried, and exposed for 5 days.

Disclosure of conflict of interest

N.V. is a co-founder of MedChem Imaging, Inc., S.K. was an employee of Eisai Inc. when the work was undertaken, and A.C. and P.S. are currently employees of Eisai Inc.