**Evaluating the In Vivo Specificity of [18F]UCB-H for the SV2A Protein, Compared with SV2B and SV2C in Rats Using microPET**

**Maria Elisa Serrano 1, Guillaume Becker 1, Mohamed Ali Bahri 1, Alain Seret 1, Nathalie Mestdagh 2, Joël Mercier 2, Frédéric Mievis 3, Fabrice Giacomelli 3, Christian Lemaire 1, Eric Salmon 1, André Luxen <sup>1</sup> and Alain Plenevaux 1,\***


Received: 22 March 2019; Accepted: 29 April 2019; Published: 1 May 2019

**Abstract:** The synaptic vesicle protein 2 (SV2) is involved in synaptic vesicle trafficking. The SV2A isoform is the most studied and its implication in epilepsy therapy led to the development of the first SV2A PET radiotracer [18F]UCB-H. The objective of this study was to evaluate in vivo, using microPET in rats, the specificity of [18F]UCB-H for SV2 isoform A in comparison with the other two isoforms (B and C) through a blocking assay. Twenty Sprague Dawley rats were pre-treated either with the vehicle, or with specific competitors against SV2A (levetiracetam), SV2B (UCB5203) and SV2C (UCB0949). The distribution volume (Vt, Logan plot, t\* 15 min) was obtained with a population-based input function. The Vt analysis for the entire brain showed statistically significant differences between the levetiracetam group and the other groups (*p* < 0.001), but also between the vehicle and the SV2B group (*p* < 0.05). An in-depth Vt analysis conducted for eight relevant brain structures confirmed the statistically significant differences between the levetiracetam group and the other groups (*p* < 0.001) and highlighted the superior and the inferior colliculi along with the cortex as regions also displaying statistically significant differences between the vehicle and SV2B groups (*p* < 0.05). These results emphasize the in vivo specificity of [18F]UCB-H for SV2A against SV2B and SV2C, confirming that [18F]UCB-H is a suitable radiotracer for in vivo imaging of the SV2A proteins with PET.

**Keywords:** SV2A; SV2B; SV2C; microPET; [18F]UCB-H; epilepsy; PBIF; distribution volume; blocking assay; preclinical imaging

#### **1. Introduction**

The synaptic vesicle protein 2 (SV2) is an integral membrane protein with twelve transmembrane domains and three N-glycosylation sites in the intravesicular loop. The SV2 protein is ubiquitously present in the nerve terminals of the central and peripheral nervous systems, and in several types of endocrine cells [1]. This protein is critical for the adequate functioning of the central nervous system, acting as a modulator of synaptic transmission [2,3]. Moreover, it has been associated with the pathophysiology of epilepsy [4–6].

Previous studies have identified three SV2 isoforms: SV2A, SV2B and SV2C, characterized by different expression levels during rodent brain development [7] and adulthood [8]. While the SV2A isoform is present across all brain areas, the SV2C isoform can only be found in specific regions, such as the striatum, pallidum, midbrain, brainstem, substantia nigra, and the olfactory bulb [9]. The SV2B isoform is particularly present in the cerebral cortex, and the cornu ammonis sub-region of the hippocampus [10]. The three isoforms present large similarities in their structure: 65% between isoforms A and B, 62% between A and C, and 57% between B and C [8].

Of these three isoforms, SV2A is the most investigated. The antiepileptic drug levetiracetam (Keppra®) binds to SV2A, suggesting a role for SV2A in the pathology underlying certain forms of epilepsy [11–14]. Several studies have shown a correlation between the brain expression of this isoform and the clinical efficacy of this drug [5,13].

To investigate the role of SV2A in vivo, in 2013 [18F]UCB-H was presented as an imaging agent with a nanomolar affinity for human SV2A [3,15–17]. Since then, other PET radiotracers, such as [ 11C]UCB-J, or [11C]UCB-A, have been synthetized to study this protein [3,18–20] (see Figure 1). These PET radiotracers appear to be more specific than [18F]UCB-H (pIC50 = 7.8) [3,16], based on their respective affinity measured in vitro, with pIC50 = 8.2 for [11C]UCB-J [18] and pIC50 = 7.9 for [ 11C]UCB-A [20]. The three radiotracers have demonstrated potential for use as synaptic density biomarkers not only in animals, but also in humans [3,21–23]. However, despite the valuable properties of [11C]UCB-J and [11C]UCB-A in assessing brain synaptic density in vivo, their clinical application is limited to facilities with a cyclotron due to the short half-life of 11C (20.3 min) compared to the half-life of 18F (110 min). In addition, the use of a PET radiotracer with a longer half-life (such as 18F) allows the evaluation of a greater number of patients per day with just one production. Therefore, different fluorine-18-labelled derivatives of UCB-J are currently being developed and characterized, such as [18F]SDM-8 [24]. The potential of [18F]UCB-H for detecting variations in SV2A has already been demonstrated in vivo [25,26]. Nevertheless, as the actual specificity of [18F]UCB-H for SV2A against SV2B and SV2C has never been addressed in vivo, we consider that this point deserves more careful evaluation.

**Figure 1.** Chemical structures of [18F]UCB-H, [11C]UCB-J and [11C]UCB-A.

This paper, therefore, aims to evaluate for the first time the specificity of [18F]UCB-H for the SV2A isoform against SV2B and SV2C using microPET imaging in rats, by means of a blocking assay between this radiotracer and specific competitors for the three SV2 isoforms. The results will provide highly valuable information about the actual potential of [18F]UCB-H as a radiopharmaceutical candidate to study the SV2A protein with PET in research or clinical practice.

#### **2. Results**

Table 1 summarizes the results obtained from the in vitro binding assays. We observe that SV2BL presents a high affinity for SV2B (pIC50 = 7.8), but also has an affinity for SV2A similar to that of SV2AL (pIC50 = 5.6).

**Table 1.** pIC50 of the competitors used for the different SV2 isoforms. Binding affinities measured for human SV2 proteins at 37 ◦C. Data are presented as mean (n = 3 to 10) from non-linear regression analysis of raw data using a sigmoidal dose-response model. Additional data for SV2BL (UCB5203) solubility: 0.1 mg/mL, route of administration: ip (suspension in 5% DMSO–1% methyl cellulose in water), CEREP @ 10 μM: all targets < 50% inhibition, mouse brain fraction unbound: 37%, mouse brain exposure (3 mg/kg, 30–60 min): ~1.8 μM total → 0.66 μM free → ~100 fold IC50 SV2B. Additional data for SV2CL (UCB0949) solubility: 0.055 mg/mL, route of administration: ip (suspension in 5% DMSO—1% methyl cellulose in water), CEREP @ 10 μM: all targets < 50% inhibition, mouse brain fraction unbound: 54%, mouse brain exposure (3 mg/kg, 30–60 min): ~8 μM total → 4.3 μM free → ~270 fold IC50 SV2C.


In Figure 2, [18F]UCB-H parametric Vt maps are presented for the vehicle group and the three pre-treated groups (SV2AL, SV2BL and SV2CL). These pictures highlight a clear reduction of the [ 18F]UCB-H binding throughout the entire brain induced by levetiracetam (SV2AL) pre-treatment at 10 mg/kg (PET image corresponding to the SV2AL group).

**Figure 2.** Example of an individual parametric Vt map of [18F]UCB-H binding in rat brain (PET), along with the corresponding individual MRI and the overlay of both images (PET + MRI). Rats were pre-treated 30 min before the 60 min PET acquisition with either vehicle, SV2A competitor (levetiracetam [SV2AL] at 10 mg/kg), SV2B competitor (UCB5203 [SV2BL] at 3 mg/kg) and SV2C competitor (UCB0949 [SV2CL] at 3 mg/kg).

In Figure 3A, we can observe the time activity curves (TACs) corresponding to the four different treatments (vehicle, SV2AL, SV2BL and SV2CL), for one of the regions of interest (ROIs): the whole brain. The four TACs reveal a high initial uptake of [18F]UCB-H, which peaks around 5 min post-injection. Subsequently, the radioactivity is quickly washed out of the brain. Some differences can be observed in the kinetic of the TAC after pre-treatment with the respective ligands: The highest peak activity is observed after pre-treatment with the vehicle and with SV2CL. Interestingly, the pre-treatments with SV2AL and SV2BL display the same peak of initial uptake. In the case of SV2BL, the kinetics of the TAC from 15 to 60 min are similar to the kinetics of the radiotracer after pre-treatment with either the vehicle or SV2CL. The TACs for all the ROIs are included in Supplementary Figure S1. In addition, the area under each TAC (the AUC) is represented in Figure 3B, where we can observe the differences between the [18F]UCB-H uptake after pre-treatment with SV2AL, and after pre-treatment with the other compounds.

**Figure 3.** Representative time activity curves (TACs) and AUC (area under each TAC) for the different regions of interest (ROIs). (**A**) TACs extracted from the whole brain as ROI, and normalized by the injected activities and the body weight. Lines represent the [18F]UCB-H uptake over a 60 min acquisition after pre-treatment with the vehicle, SV2AL, SV2BL, or SV2CL. (**B**) The bar plots represent the AUC in the eight ROIs (mean ± SEM, n = 5).

Figure 4 presents the mean Vt values for the eight selected brain structures, calculated from the previous TACs and the population-based input function (PBIF). Comparing Figures 3B and 4, the differences between groups in AUC and in Vt are similar, with the highest value associated to pre-treatment with the vehicle, and the lowest value for pre-treatment with SV2AL.

Table 2 summarizes, for the same regions, the impact on the Vt induced by the blocking experiments, expressed as the relative difference in Vt between the vehicle group and the pre-treated groups. In the whole brain, mean Vt values of 10.4 ± 0.7, 6.0 ± 0.3, 8.3 ± 0.2 and 9.8 ± 0.3 were obtained for the vehicle (control) group, the SV2AL pre-treated group, the SV2BL pre-treated group and the SV2CL pre-treated group, respectively. For the eight ROIs, a statistically significant difference was observed between the SV2AL pre-treated group and all the other groups (*p* < 0.001). Furthermore, for the whole brain, the

cerebral cortex and the inferior and superior colliculus, a statistically significant difference was also detected between the vehicle group and the SV2BL pre-treated group (*p* < 0.05).

**Figure 4.** Vt values for the eight selected brain structures. Bars represent the mean ± SEM (n = 5). One-way ANOVA and Scheffe post-hoc tests were performed, with \*\*\**p* < 0.001 and \**p* < 0.05.



#### **3. Discussion**

The SV2 protein is critical for the adequate functioning of the central nervous system, acting as a modulator of synaptic transmission by priming vesicles in quiescent neurons [4]. The divergent roles of the three isoforms which comprise this family have yet to be clarified, although different pathologies have been associated with them. As previously stated, the SV2A isoform is associated with the physiopathology of epilepsy [5,27]. In contrast, the SV2B isoform is related with prostate small cell carcinoma [28] and the SV2C isoform is generally associated with the correct functioning of basal ganglia nuclei [9,29,30]. Some studies have evaluated the possible relation between SV2C and Parkinson's disease, as SV2C modulates dopamine release [29,31].

This paper's goal was to evaluate, for the first time, in vivo the specificity of the [18F]UCB-H radiotracer in targeting the SV2A isoform compared to SV2B and SV2C. The relevance of such a study stems from the fact that in vivo SV2A quantification can be considered to be an indirect measure of the synaptic density [2,3,18], which is a key parameter for fundamental research and for the clinic.

Before discussing the results obtained during these blocking experiments, we have to address some general considerations. Firstly, the results presented issue from the microPET imaging technique. Like many other microPET cameras, the Focus120 used during this work has a spatial resolution of 1.5 mm, at best hampering the study of brain structures of small size due to the partial volume

effect [32,33]. Secondly, the rat brain distributions of the three SV2 protein isoforms [7–9] indicate that almost all major brain structures express at least two SV2 protein isoforms. SV2A, the most extensively studied, is ubiquitously distributed [1,8]. Like SV2A, SV2B can be found in almost all the rat brain structures with few subtle differential expressions in some hippocampal substructures like CA3 and the dentate gyrus, along with the reticular nucleus of the thalamus and some small areas in the brain stem [8,9,34]. Unfortunately, these regions are far too small to be correctly quantified with microPET. Janz and Sudhof showed that unlike SV2A and SV2B, the SV2C protein is characterized by much more restricted localization in brain regions considered to be evolutionarily well preserved in rats: The olfactory bulb, the striatum, the substantia nigra, and some nuclei in the pons and the medulla oblongata [9]. As we can see, it is impossible to find well defined brain structures for in vivo microPET quantification in which one of the three isoforms is uniquely or even mostly expressed. Another important point is that we do not have a clear picture of the respective proportions of each isoform present in the main rat brain structures. All these considerations will have to be taken into account in the following discussion. Accordingly, we have decided to select eight major ROIs to ensure robust in vivo quantification with microPET: Whole brain, cortex, hippocampus, inferior colliculus, superior colliculus, midbrain, caudate putamen and thalamus.

The [18F]UCB-H Vt values obtained during this study for the vehicle pre-treated group, calculated using the PBIF [35] were in good agreement with those previously published for rats [35,36]. This is important in order to establish the consistency of the proposed methodology. The blocking experiments realized with SV2AL at 10 mg/kg demonstrated a clear significant competition (46.2%) between levetiracetam and [18F]UCB-H in eight selected ROIs. These values are of the same order of magnitude as those previously reported in rats [36]. According to the potency of levetiracetam for SV2A against SV2B and SV2C (Table 2), we can conclude that SV2A is one of the main target of [18F]UCB-H in vivo in rats.

After performing a blocking experiment with SV2CL at 3 mg/kg, we obtained TACs with similar peaks and kinetics to the TAC corresponding to pre-treatment with the vehicle, in all ROIs. The quantification of the radiotracer uptake, using the Vt, highlighted no statistically significant in vivo competition between SV2CL and [18F]UCB-H in any of the eight selected ROIs. From this we can infer that SV2CL pre-treatment has either no impact or an impact of very small size. The population used (n = 5) is not sufficient to demonstrate an effect of small size (f = 0.10), but is optimal to detect medium (f = 0.25) and large effect sizes (f = 0.5). Another important point is that a highly potent SV2C competitor like UCB0949 (pIC50 of 7.8) was unable to modify [18F]UCB-H binding in brain structures with a high expression of SV2C, like the midbrain or the caudate/putamen [9]. The reduction measured in these regions was of the same order of magnitude as that found in the other structures. These considerations support the theory that SV2C does not seem to be the main target of [18F]UCB-H in rats.

The pre-treatment with SV2BL at 3 mg/kg resulted in a TAC with a peak as high as that obtained after pre-treatment with SV2AL. However, it features kinetics which are similar to those obtained after pre-treatment with the vehicle or SV2CL. This lower peak could be attributed to an initial non-specific binding of SV2BL to the SV2A protein, for which it presents an affinity which is similar to that of SV2AL, with a pIC50 = 5.6. After the peak, the SV2BL TAC follows a similar shape and level to those of pre-treatment with the vehicle and SV2CL, indicating a washing out effect of this fraction of non-specific binding of SV2BL to the SV2A protein. In order to confirm this hypothesis, a similar experiment with a SV2BL with a lower affinity for SV2A should be performed. However, the SV2BL used in this paper is currently the only one available. In addition to the previous analysis, we evaluated the changes in Vt after pre-treatment with SV2BL. In these results, we can observe a consistent mean reduction of 20.5% of the Vt values in the eight selected ROIs. The SV2BL is characterized by a pIC50 of 7.8 for the SV2B isoform. Such a highly potent competitor is expected to effectively impede the binding of any radioligand to the SV2B isoform. If SV2B was the main target for [18F]UCB-H, the blocking induced with this highly efficient SV2BL would have been very pronounced and much higher than the 20% measured. In order to explain the 20% reduction of [18F]UCB-H Vt values, we have to take into account

that SV2BL also presents some potency for SV2A. SV2BL has a pIC50 of 5.6, which is of the same order of magnitude as that of levetiracetam. Thus, the SV2BL ligand has some affinity for SV2A, which could lead to partial blocking of SV2A. Hence, the 20% reduction observed is most likely linked to SV2A blocking induced by SV2BL. Accordingly, we can conclude that SV2B does not seem to be the main target of [18F]UCB-H in vivo in rats.

We are aware that the respective affinities of SV2BL and SV2CL are a problem for the interpretation of the data, but we have to consider that today UCB5203 and UCB0949 are the only compounds that can be used for this purpose.

#### **4. Materials and Methods**

#### *4.1. Animals*

Twenty male Sprague Dawley CD rats (five weeks old) were used, bred by Janvier Laboratories (France). The animals were housed in pairs for three weeks under standard 12:12 h light/dark conditions, maintaining room temperature at 22 ◦C, and humidity at approximately 50%. Standard pellet food and water were provided ad libitum.

The experimental procedures and protocols used in this investigation ("Synap-SV2A project" files 14-1753 and 13-1573) were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Liege, according to the Helsinki declaration, and conducted in accordance with the European guidelines for care of laboratory animals (2010/63/EU). Moreover, the Animal Research Reporting In Vivo Experiments (ARRIVE) guidelines [37] were followed as closely as possible to confer a minimal intrinsic quality to the study.

#### *4.2. Radiopharmaceutical Production and Drugs*

[ 18F]UCB-H was produced through one-step radiolabeling of a pyridyliodonium precursor. This method provides 34% <sup>±</sup> 2% of injectable [18F]UCB-H (uncorrected radiochemical yield) from up to 285 GBq (7.7 Ci) of [18F]fluoride (specific activity of 815 <sup>±</sup> 185 GBq/μmol and measured purity of 99.8 ± 0.5 wt %); this has previously been reported in Warnier et al. [17].

The ligand for the SV2A isoform (SV2AL) was purchased as an injectable solution (levetiracetam, Keppra®, UCB Pharma S.A. Brussels, Belgium). At the present time, there are no commercially available specific ligands for the other two SV2 isoforms (SV2B and SV2C). The competitors used were obtained from UCB Pharma s.a.: UCB5203 for SV2B (SV2BL, MW: 236.238 g/mol) and UCB0949 for the SV2C (SV2CL, MW: 281.197 g/mol). The information on these compounds was supplied by UCB Pharma s.a. The respective affinities for the different SV2 isoforms are presented in Table 2.

The competitors were prepared daily in a vehicle composed of distilled water containing 1% methyl cellulose (viscosity: 15 cP, Sigma-Aldrich, Overijse, Belgium) and 5% dimethyl sulphoxide (DMSO, Sigma-Aldrich, Belgium). The concentrations differed depending on the product specifications and their respective pharmacokinetics, provided by UCB Pharma s.a. The dosing used was 10 mg/kg for SV2AL, 3 mg/kg for SV2BL, and 3 mg/kg for SV2CL. All solutions were administered through the intraperitoneal (i.p.) route in a total volume of 1 mL per kg of body weight. The animals used as a control group (vehicle) received an equal volume of vehicle through the same route of administration.

#### *4.3. In Vitro Binding Assays*

Reagents and reference compounds used were of analytical grade and obtained from various commercial sources. All cell culture reagents were obtained from Invitrogen (Merelbeke, Belgium). Radioligands (3H-UCB30889, 1184 GBq/mmol; 3H-UCB1418435, 925 GBq/mmol; and 3H-UCB101275-1, 1110–1480 GBq/mmol) were obtained from G.E Healthcare, Amersham, UK (now Perkin Elmer, Zaventem, Belgium) and reference compounds (levetiracetam, UCB108649-1 and UCB101275-1) were custom synthesized and stored according to manufacturer's recommendations. Test and reference

compounds were dissolved in 100% DMSO or H2O to give 1 or 10 mM stock solution. The final DMSO concentration in assays was 0.1% unless otherwise stated.

Cell lines generated at UCB Biopharma were human embryonic kidney (HEK) 293 cells expressing human SV2A, SV2B or SV2C proteins. Cells were cultured in Dulbecco's Modified Eagle medium. The culture medium was supplemented with foetal bovine serum (FBS, 10%), 2 mM l-glutamine, 50 to 100 U/mL penicillin, 50 to 100 μg/mL streptomycin, and 200 μg/mL hygromycin B. Cells were grown at 37 ◦C with 95% air. Confluent cells were detached by 10 min incubation at 37 ◦C in phosphate buffered saline (PBS) containing 0.02% EDTA. Culture flasks were washed with 15 mL of ice-cold PBS. The cell suspension was centrifuged at 1500× *g* for 10 min at 4 ◦C. The pellet was homogenized in 15 mM Tris-HCl buffer (pH 7.5) containing 2 mM MgCl2, 0.3 mM EDTA, and 1 mM EGTA (buffer A) using a glass/teflon homogenizer. The crude homogenate was subjected to a freeze and thaw cycle in liquid nitrogen and DNAse (1 μL/mL) was then added. The homogenate was further incubated for 10 min at 25 ◦C before being centrifuged at 40,000× *g* for 25 min at 4 ◦C. The pellet was re-suspended in buffer A and washed once under the same conditions. The final crude membrane pellet was re-suspended at a protein concentration of 1–3 mg/mL in 7.5 mM Tris-HCl buffer (pH 7.5 at 25 ◦C) containing 250 mM sucrose and stored in liquid nitrogen until use.

Membranes were incubated in binding buffer (see Table 3) containing test compound or positive control in the presence of the radioligand. The non-specific binding (NSB) was defined as the residual binding observed in the presence of a high concentration (1000 fold its Ki) of a specific unlabeled reference compound. Membrane-bound and free radioligands were separated by rapid filtration through glass fiber filters (GF/C). Samples and filters were rinsed using at least 6 mL of washing buffer. The entire filtration procedure did not exceed 10 s per sample. The radioactivity trapped on the filters was counted by liquid scintillation in a β-counter. To determine the affinity of a compound for a given target, competition curves were performed with at least 10 concentrations of compound spanning at least 5 log units.

**Table 3.** Details of the in vitro binding assay determination. Percentage of inhibition was calculated as follows: % INHIBITION = 100 − [((BI − NSB)/(B0 − NSB)) × 100], where B0 and BI represent the binding observed in the absence and presence of the test compound, respectively (dpm), NSB is the radioligand non-specific binding (dpm). Raw data were analyzed by non-linear regression using XLfitTM (IDBS, London, Great Britain) according to the following generic equation: <sup>B</sup> <sup>=</sup> NSB <sup>+</sup> [(B0 <sup>−</sup> NSB)/(1 <sup>+</sup> (((10X)/(10<sup>−</sup>pIC50))nH))], where B is the radioligand bound in the presence of the unlabeled compound (dpm), NSB is the radioligand non-specific binding (dpm), B0 is the radioligand bound in the absence of unlabeled compound (dpm), X is the concentration of unlabeled compound (log M), pIC50 is the concentration of unlabeled compound that inhibits the radioligand specific binding by 50% (−log M), and nH is the Hill coefficient.


#### *4.4. PET Acquisitions*

The animals (n = 5 per group) were anesthetized using 4% isoflurane in air at a flow rate of 1 L/min during induction and 1.5% to 2% isoflurane in air at 0.6 L/min during maintenance. Respiration rate and rectal temperature were continuously measured using a physiological monitoring system (Minerve, France). The temperature was maintained at 37 ± 0.5 ◦C using an air warming system.

MicroPET scans were performed with a Siemens FOCUS 120 microPET (Siemens, Knoxville, TN, USA). The animals were anesthetized and pre-treated i.p. with vehicle, SV2AL, SV2BL, or SV2CL. Thirty minutes later, they were installed in the microPET scanner and [18F]UCB-H was injected via the lateral tail vein (44.7 ± 3 MBq, 0.55 mL), simultaneously starting a 60 min emission scan, in list mode. Finally, a 10 min transmission scan was performed in a single event acquisition mode, using a 57Co source. The acquired data were then reframed as follows: 6 × 5 s, 6 × 10 s, 3 × 20 s, 5 × 30 s, 5 × 60 s, 8 × 150 s, and 6 × 300 s. For each frame, a total of 95 trans-axial slices were obtained using Fourier rebining (FORE), followed by 2D ramp filtered backprojection (FBP), in 256 × 256 matrix. The slice thickness was 0.796 mm and the in-slice pixel size was 0.433 mm.

Immediately after the PET acquisition, the anesthetized rats were transferred into a 9.4 Tesla MRI horizontal bore system (Agilent Technologies, Palo Alto, CA, USA), with a 72 mm inner diameter volumetric coil (Rapid Biomedical GmbH, Würzurg, Germany). Anatomical T2-weighted brain images were obtained using a fast spin echo multi-slice sequence with the following parameters: TR = 2000 ms, TE = 40 ms, matrix = 256 × 256, FOV = 45 × 45 mm, 30 contiguous slices of thickness = 0.80 mm and in-plane voxel size = 0.176 × 0.176 mm.

#### *4.5. Imaging Data Processing*

PMOD software (Version 3.6, PMOD Technologies, Zurich, Switzerland) was used to process the imaging data. The structural MRI images were firstly co-registered to the corresponding PET images, and subsequently spatially normalized into the PMOD MRI T2 template. Finally, the inverse normalization parameters were calculated and applied to the PMOD rat brain atlas to bring it in the individual PET space. From this atlas, eight relevant regions of interest (ROIs) were chosen according to their differential expression of SV2A, SV2B and SV2C: whole brain, cortex, caudate/putamen, hippocampus, inferior colliculus, superior colliculus, midbrain and thalamus.

Individual time-activity curves (TACs) were extracted for each ROIs and normalized by the body weight and the injected dose of radiotracer to be expressed as standardized uptake value (SUV). A population-based input function (PBIF) published by our laboratory [35] was used to avoid arterial blood sampling during the acquisitions. The distribution volume (Vt), was determined by Logan plot kinetic modelling using the TACs and the PBIF. The equilibration time (t\*) was fixed at 15 min (starting point of the range used in the multi-linear regression analysis).

#### *4.6. Statistical Analysis*

The results are presented as mean (Vt) ± standard error of the mean (SEM). All the data were tested for normal distribution with Levene's test for homogeneity, and with a Kolmogorov–Smirnov test for normality. Data were analyzed using one-way analysis of variance (ANOVA) followed by Scheffe post-hoc tests.

All statistical analyses were performed with the statistics software Statistica 12 (Statsoft, France) and GraphPad Prism (version 6, GraphPad software, Inc., San Diego, CA, USA). The critical level of statistical significance was always set at *p* < 0.05.

#### **5. Conclusions**

For the first time, the specificity of a radiopharmaceutical compound for the three SV2 protein isoforms was assessed in vivo, in rats. The results obtained clearly indicated that SV2A was the main target of [18F]UCB-H, and confirmed that [18F]UCB-H is a suitable radiotracer for in vivo imaging of the SV2A proteins with PET. Consequently, [18F]UCB-H is an interesting candidate to study SV2A-associated pathologies.

**Supplementary Materials:** The following are available online. Figure S1: TACs extracted from the eight ROIs, and normalized by the injected activities and the body weight. Lines represent the [18F]UCB-H uptake over a 60 min acquisition after pre-treatment with the vehicle, SV2AL, SV2BL, or SV2CL (mean ± SEM; n = 5).

**Author Contributions:** Conceptualization, M.E.S., G.B., A.S. and A.P.; methodology, M.A.B., A.S.; formal analysis, M.E.S., G.B., M.A.B., A.S. and A.P.; investigation, M.E.S. and G.B.; resources, N.M., J.M., F.M., F.G., C.L. and A.L.; data curation, M.A.B.; writing—original draft preparation, M.E.S.; writing—review and editing, G.B., M.A.B., A.S., N.M., J.M., F.M., F.G., C.L., E.S., A.L. and A.P.; visualization, M.E.S., M.A.B. and A.P.; supervision, A.L. and A.P.; project administration, E.S. and A.P.; funding acquisition, A.P.

**Funding:** This research was funded by the University of Liège grant 13/17-07 and UCB Pharma s.a. as partners. The SV2BL (UCB5203) and SV2CL (UCB0949) ligands were provided by UCB Pharma s.a., compound structures and specifications being confidential. M.E.S. is supported by the University of Liège grant 13/17-07. A.P. is research director from F.R.S.-FNRS Belgium. J.M. and N.M. are UCB Pharma s.a. employees. F.M. and F.G. are Nucleis s.a. employees.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


**Sample Availability:** Not available.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **In Vitro and In Vivo Characterization of Dibenzothiophene Derivatives [125I]Iodo-ASEM and [ 18F]ASEM as Radiotracers of Homo- and Heteromeric** α**7 Nicotinic Acetylcholine Receptors**

### **Cornelius K. Donat 1,2,\*, Henrik H. Hansen 1, Hanne D. Hansen 1, Ronnie C. Mease 3, Andrew G. Horti 3, Martin G. Pomper 3, Elina T. L'Estrade 1,4,5, Matthias M. Herth 4,5, Dan Peters 6, Gitte M. Knudsen <sup>1</sup> and Jens D. Mikkelsen 1,\***


#### Academic Editor: Peter Brust

Received: 24 January 2020; Accepted: 27 February 2020; Published: 20 March 2020

**Abstract:** The α7 nicotinic acetylcholine receptor (α7 nAChR) is involved in several cognitive and physiologic processes; its expression levels and patterns change in neurologic and psychiatric diseases, such as schizophrenia and Alzheimer's disease, which makes it a relevant drug target. Development of selective radioligands is important for defining binding properties and occupancy of novel molecules targeting the receptor. We tested the in vitro binding properties of [125I]Iodo-ASEM [(3-(1,4-diazabycyclo[3.2.2]nonan-4-yl)-6-(125I-iododibenzo[b,d]thiopentene 5,5-dioxide)] in the mouse, rat and pig brain using autoradiography. The in vivo binding properties of [18F]ASEM were investigated using positron emission tomography (PET) in the pig brain. [125I]Iodo-ASEM showed specific and displaceable high affinity (~1 nM) binding in mouse, rat, and pig brain. Binding pattern overlapped with [125I]α-bungarotoxin, specific binding was absent in α7 nAChR gene-deficient mice and binding was blocked by a range of α7 nAChR orthosteric modulators in an affinity-dependent order in the pig brain. Interestingly, relative to the wild-type, binding in β2 nAChR gene-deficient mice was lower for [125I]Iodo-ASEM (58% <sup>±</sup> 2.7%) than [125I]α-bungarotoxin (23% <sup>±</sup> 0.2%), potentially indicating different binding properties to heteromeric α7β2 nAChR. [18F]ASEM PET in the pig showed high brain uptake and reversible tracer kinetics with a similar spatial distribution as previously reported for α7 nAChR. Blocking with SSR-180,711 resulted in a significant decrease in [18F]ASEM binding. Our findings indicate that [125I]Iodo-ASEM allows sensitive and selective imaging of α7 nAChR in vitro, with better signal-to-noise ratio than previous tracers. Preliminary data of [18F]ASEM in the pig brain demonstrated principal suitable kinetic properties for in vivo quantification of α7 nAChR, comparable to previously published data.

**Keywords:** alpha 7; nicotinic acetylcholine receptors; PET; nAChR; autoradiography

#### **1. Introduction**

The α7 nicotinic acetylcholine receptor (α7 nAChR) belongs to the superfamily of ligand-gated ion channels and is expressed across all mammalian species [1–4]. The receptor plays an important role in cognition [5], mood [6] and consistent with this, α7 nAChR are particularly abundant in hippocampus and prefrontal cortex [7,8]. Furthermore, α7 nAChR are implied in neuro-immune [9] and immune functions [10] under homeostatic conditions.

Changes in protein and mRNA levels of α7 nAChR have been reported in a number of neuropsychiatric and neurodegenerative diseases [1,11–15]. Notably, certain polymorphisms in the promoter region of the α7 nAChR gene (*CHRNA7*) [16] are probable risk factors for neuropsychiatric diseases, such as major depression [17] and schizophrenia [18] and are associated with developmental disorders and cognitive impairments [19]. Additionally, α7 nAChRs are expressed by several central and peripheral immune cells and activation via agonists and positive allosteric modulators showed neuroprotective and immunomodulatory efficacy in different preclinical disease models [20–24].

Changes of α7 nAChR in the healthy and diseased brain can only be detected in vivo by molecular imaging, such as positron emission tomography (PET) using specific radiotracers. A clinically usable radiotracer requires sufficient selectivity, specificity and suitable affinity, depending on the target [25]. Most of the previously described α7 nAChR PET tracers, among those [18F]NS14490, [11C]NS14492, [ 11C]CHIBA-1001 and [11C]A-582941 (Table 1), studied in mice, pigs and non-human primates exhibited some shortcomings, such as poor specific and/or high nonspecific binding or radiometabolites crossing the blood–brain barrier [26–32]. Furthermore, the specificity of novel α7 nAChR tracers has not always been tested in respective gene-deficient mice, e.g., using in vitro autoradiography.

While development of novel tracers from different lead structures is still ongoing [33–37], tilorone [38] provided a lead-structure for a number of derivatives subsequently developed into α7 nAChR PET tracers. From those, [18F]ASEM (JHU82132) [39] and the structurally related [18F]DBT-10 (JHU82108) [40] have been most widely investigated. Initial studies have shown that ASEM is a potent antagonist [39] with subnanomolar affinity and high selectivity [39,41], further substantiated by the radiolabelled compounds [18F]ASEM [26] and [125I]Iodo-ASEM [42] as tested in human and rat recombinant α7 nAChR. [18F]ASEM and [125I]Iodo-ASEM readily enter the mouse brain, are displaceable, and accumulate in regions with highest α7 nAChR density [26,39,42,43].

More recent studies using [18F]ASEM and [18F]DBT-10 further supported the suitability of the tracers, showing high and reversible brain uptake with a regional binding pattern consistent with the distribution of α7 nAChR receptors in the non-human primate brain [39,44,45]. Favourable brain pharmacokinetics, excellent test-retest reproducibility and regional uptake [18F]ASEM pattern consistent with post-mortem α7 nAChR distribution have been reported in human PET studies [43,46]. Several recent studies extended the available data in human subjects, showing good agreement with previous distribution volumes (VT) and test-retest values in nonhuman primates and healthy volunteers [44]. A study in ageing subjects showed a significant positive correlation between age and [18F]ASEM VT in striatum and several cortical regions [47], however without any correlation between VT and cognitive measures. A small sample of individuals with schizophrenia on stable antipsychotic medication showed lower [18F]ASEM VT in cingulate cortex and hippocampus [46] and individuals with recent onset of psychosis were also reported to show lower [18F]ASEM VT in hippocampus, after controlling for age [48]. Interestingly, patients with mild cognitive impairment showed higher [18F]ASEM VT when adjusted for age as compared to the control group [49], consistent with post-mortem findings from patients and animal models. Additionally, [18F]ASEM has been employed in a rat 6-OHDA lesion model of Parkinson's disease, showing an initial increase of [ 18F]ASEM SUVr in the ipsilateral striatum and substantia nigra between 3 and 7 days, which coincided with several histology markers of glia activation [50].

While this data shows the general applicability of [18F]ASEM, binding properties and interpretation of novel α7 nAChR tracers might be complicated by the fact that α7 subunits can form heteromeric receptors together with other subunits, specifically β2 [51]. These receptors can be heterologously expressed in oocytes and are found in the rodent and human basal forebrain and cortex [52,53]. While these heteromeric receptors display different pharmacological properties [52,54], it is not clear how this translates to radiotracer binding. In vitro binding studies of [18F]ASEM or [125I]ASEM in gene-deficient mice could answer the question, and would also reveal the suitability of [125I]ASEM for in vitro autoradiographic studies. The latter would offer a better comparability to in vivo PET data over the current gold-standard tracer [125I]α-bungarotoxin. We therefore investigated the potential of [125I]ASEM for in vitro studies of the α7 nAChR, by comparing binding of [125I]Iodo-ASEM in rat, mouse and pig brain sections. Furthermore, [18F]ASEM was characterized for in vivo brain uptake and target selectivity in a PET study conducted in the pig.


**Table 1.** Common α7 nAChR ligands and their structure, previously evaluated as radiotracers.

#### **2. Results**

#### *2.1. In Vitro Autoradiography*

Cerebral binding of [125I]Iodo-ASEM was investigated across several mammalian species, i.e., rat (Figure 1A, upper row), mouse (Figure 2A) and pig brain (Figure 1A, lower row) and compared to [ 125I]α-bungarotoxin (Figure 1B/2B). Total cortical [125I]Iodo-ASEM binding was highest in the pig, and lower in the rat and mouse (Figures 1C and 2C), as compared to [125I]α-bungarotoxin. [125I]Iodo-ASEM binding was displaceable with (-)-nicotine (1 mMol/L, data not shown) and SSR-180,711 (10 μMol/L, Figures 1A and 2A, nonspecific binding) in all species.

**Figure 1.** (**A**,**B**) Representative autoradiographs showing [125I]Iodo-ASEM and [125I]α-bungarotoxin total and non-specific binding (determined with 1 mMol/L (-)-nicotine for [125I]α-bungarotoxin and 10 μMol/L SSR-180,711 for [125I]Iodo-ASEM) in 12 μm sections of the rat (upper row) and pig brain (lower row). Arrowheads indicate residual white matter binding. (**C**) Comparative quantitative analysis of specific binding (<sup>±</sup> S.E.M.) of [125I]Iodo-ASEM and [125I]α-bungarotoxin from autoradiography in the rat (n = 1) and pig cortex (*n* = 2). All autoradiographic experiments and quantifications are carried out in 3-4 sections per animal.

A non-specific binding component remained detectable under the described experimental conditions at low levels in white matter structures (arrowheads, Figures 1A and 2A). The distribution pattern of [125I]Iodo-ASEM binding in the rat, mouse and pig brain was comparable to that of [ 125I]α-bungarotoxin (Figures 1B and 2B). In the pig, [125I]Iodo-ASEM showed a laminar binding pattern in the frontal cortex, with highest density in cortical layers (1–3) (Figure 1A, lower row), while in the rat, binding in motor, cingulate and somatosensory cortex was more prominent in layers 5–6. However, this species difference was also observed for [125I]α-bungarotoxin (Figure 1B).

**Figure 2.** (**A**,**B**) Representative autoradiographs showing total [125I]Iodo-ASEM (**A**) and [125I]a-bungarotoxin (**B**) and non-specific (determined with 1 mMol/L (-)-nicotine for [125I]α-bungarotoxin and 10 μMol/L SSR-180,711 for [125I]Iodo-ASEM) binding in 12 μm brain sections of α7 and β2 nAChR wild-type vs. corresponding gene-deficient (def) mice (*n* = 1 each). Arrowheads indicate residual white matter binding. (**C**) Comparative quantitative analysis of specific binding (<sup>±</sup> S.E.M.) of [125I]Iodo-ASEM and [ 125I]a-bungarotoxin in α7 and β2 nAChR wild-type vs. corresponding gene-deficient mice (*n* = 1). All autoradiographic experiments and quantifications are carried out in 3-6 sections per animal.

The specificity of [125I]Iodo-ASEM to α7 nAChR is further substantiated by tracer binding experiments in α7 nAChR gene-deficient mice. Specific [125I]Iodo-ASEM binding was lacking in α7 nAChR gene-deficient mice (Figure 2A), as indicated by the overall reduction in total binding by 93% <sup>±</sup> 1.7%, compared to wild-type animals (Figure 2C). Similarly, [125I]α-bungarotoxin total binding (Figure 2B) was 96% ± 0.4% lower in α7 nAChR gene-deficient mice (Figure 2C). In wild-type mice, no difference in [125I]Iodo-ASEM and [125I]α-bungarotoxin binding was observed (Figure 2A,B). However, traces of nonspecific binding were again noted in white matter structures (arrowheads in Figure 2A).

[ 125I]Iodo-ASEM binding in β2 nAChR gene-deficient mice was different compared to [ 125I]α-bungarotoxin. An overall 58% <sup>±</sup> 2.7% lower specific [125I]Iodo-ASEM binding was observed, as compared to corresponding wild-type controls (Figure 2C). In contrast, [125I]α-bungarotoxin binding was reduced by 23% ± 0.2%, being less affected by β2 nAChR gene-deficiency as compared to the reduction in [125I]Iodo-ASEM binding.

Saturation binding in rat and pig brain sections indicated that [125I]Iodo-ASEM binding was saturable. In the rat, non-linear regression analysis revealed an equilibrium dissociation constant (*K*d) of 1.14 nM (cortex, Figure 3B) and 1.17 nM (hippocampus, Figure 3A) with corresponding receptor density (Bmax) of 0.70 fmol/mg protein (cortex) and 1.44 fmol/mg protein (hippocampus), respectively (Figure 3A,B). In comparison, the pig cortex showed a *K*<sup>d</sup> of 1.21 nM with a Bmax of 5.47 fmol/mg protein (Figure 3C). The non-specific binding of [125I]Iodo-ASEM at concentrations near the *K*<sup>d</sup> was low (rat hippocampus, 20%; rat cortex, 30%; pig cortex, 10%).

**Figure 3.** Saturation binding of [125I]Iodo-ASEM (0.02-10 nMol/L) to 12 μm sections from the rat hippocampus and cortex (**A**, **B**, *n* = 1) and pig frontal cortex (**C**, *n* = 1) brain. Non-specific binding was determined in the presence of 10 μMol/L SSR-180,711. Optical density of the autoradiograms was converted into ligand binding (fmol/mg protein ± S.E.M.) from a representative experiment. Data from saturation binding experiments were analysed by non-linear regression. Individual *K*<sup>d</sup> and Bmax values are indicated in Section 2.1. All autoradiographic experiments and quantifications are carried out in 2–4 sections per animal using 10 radioligand concentrations.

A range of selective α7 nAChR ligands (10 μMol/L each), including the α7 nAChR preferring antagonist methyllycaconitine (MLA), were used to test whether in vitro [ 125I]Iodo-ASEM binding (0.5 nMol/L) could be blocked in the pig cortex (Table 2). The partial agonists, NS14492, TC-5619, EVP-6124, A-582941, and SSR-180,711, showed almost complete (>90%) blocking of [125I]Iodo-ASEM binding in receptor dense areas of the cortex, e.g. layers 1–3. In contrast, GTS-21 (weak α7 nAChR agonist, ~70% reduction) and MLA (α7 nAChR preferring antagonist, ~80% reduction) exhibited less efficacious blockade of [125I]Iodo-ASEM binding in the pig cortex.



Results are given in % remaining binding of total binding (mean ± S.E.M.).

#### *2.2. In Vivo PET Imaging in the Pig Using [18F]ASEM*

[ 18F]ASEM readily entered the pig brain and highest tracer accumulation was found in the thalamus followed by cortex, striatum and cerebellum (Figure 4A,C). [18F]ASEM uptake in the white matter was initially lower than in the grey matter regions, however the tracer kinetics were also slower, resulting in lower grey to white matter ratio at the end of the scans. The metabolism of [ 18F]ASEM in pigs was relatively slow, with 60% of the radioactivity at 120 min still being parent radioligand (data not shown). Kinetic modelling was performed to quantify the tracer uptake. Baseline VT values varied between animals but after correcting for free fraction in plasma(fP), there was only a 5% difference in VT/fP values between the two baseline animals (Table 3). This also suggest that VT/fP values are unaffected by relatively large differences in injected mass (0.35 μg and 1.78 μg).


**Table 3.** Kinetic modelling of [18F]ASEM with the Logan Graphical Analysis model in different pig brain regions.

In a third animal, we evaluated the specificity of [18F]ASEM binding in vivo, by administering SSR-180,711 (1 mg/kg) prior to injection of [18F]ASEM. Compared to the baseline studies, we found an increase in [18F]ASEM uptake in all brain areas investigated (Figure 4A). Quantification of uptake and subsequent correction for fP revealed that SSR-180,711 administration decreased the VT/fP compared to baseline (Table 3). Occupancy was computed with the Lassen plot using VT/fP values comparing baseline data from animal 1 and blocking data from animal 3 (0-150 min scan data). We found that the 1 mg/kg SSR-180,711 dose resulted in a 49% occupancy (Figure 4B).

From the Lassen plot, the volume of non-displaceable binding (VND/fP) was found to be 9.2 mL/cm3. When comparing the VND/fP to the VT/fP in the thalamus, we found that 78% of the signal observed in the thalamus is specific binding, leaving 22% as non-displaceable binding.

In one animal, [18F]ASEM acquisition time was 240 min, which allowed subsequent analysis of the time-stability of the parameters estimated with kinetic modelling. Again, the LGA model was used to determine VT with different scan length and VT values were found to decrease with more time included in the kinetic modelling. Using all data (0–240 min), VT values were 5.4 mL/cm3 (thalamus), 5.0 mL/cm3 (frontal cortex) and 4.0 mL/cm<sup>3</sup> (cerebellum).

The upper half of the table shows the baseline distribution volumes (VT) values with and without correction for free fraction in plasma (fP) in two different animals. Bottom part of the table describes VT values with and without correction for (fP) at baseline (animal 1) and after pre-treatment with SSR-180,711 (animal 3)**.** See Table 4 (Material and Methods) for fP values in the individual animals. Because animal 2 was only scanned for 90 min, the acquisition time of animal 1 was truncated to 90 min to allow for comparison.

**Figure 4.** [ 18F]ASEM binding in the pig brain. (**A**) Time-activity curves of [18F]ASEM in three different animals: Animal 1, bsl, blue; Animal 2, bsl, black; Animal 3, SSR-180,711 pre-treated, red. The regions shown are: Thalamus (tha, triangles) and cerebellum (cb, circles). (**B**) Lassen plot with total distribution volumes (VT) corrected for free fraction in plasma (fP) using values from animal 1 and animal 3. Each point represents one region of interest (ROI), please refer to the method section for the complete list of ROIs. (**C**) Summed PET image (0–240 min) from animal 1 showing the distribution of [18F]ASEM in the pig brain. SUV: standard uptake value. Bsl: baseline. SSR: SSR-180,711 (1 mg/kg).

#### **3. Discussion**

In this study, we investigated binding properties of radiolabelled ASEM in vitro ([125I]Iodo-ASEM] and in vivo ([18F]ASEM). Autoradiography was used to determine the applicability of [125I]Iodo-ASEM for in vitro assessment of α7 nAChR receptor distribution and occupancy in the mammalian brain. [ 125I]Iodo-ASEM showed high-affinity and specific binding to α7 nAChR in the rat, mouse and pig brain. Specific binding was absent in α7 gene-deficient mice, indicating high specificity and selectivity. Saturation binding experiments in rat and pig brain sections revealed low nanomolar *K*<sup>d</sup> values (approximately 1 nM) in both species. Bmax in the pig cortex was considerably higher as compared to the binding found in the mouse and rat brain cortex. Such species differences are well documented in the literature, e.g., for metabotropic glutamate 5 receptors and the 18 kDa translocator protein between monkey and humans [55,56]. As affinity and selectivity are major criteria for radiotracers, our data further substantiates the suitability of ASEM derivatives as favourable α7 nAChR tracers [57,58]

We found that the affinity of [125I]Iodo-ASEM in the pig brain as determined with autoradiography was in a similar range as [3H]NS14492 [59]. In contrast, higher affinities and receptor densities are reported for radioligand binding assay in brain homogenates for a number of different α7 nAChR ligands, such as [3H]NS14492 and [3H]A-585539 [60], including other dibenzothiophenes [26,38].

Across the brain and specifically regions with high α7 nAChR expression, such as hippocampus and superficial cortical layers, non-specific binding of [125I]Iodo-ASEM at concentrations near the *K*<sup>d</sup> was low (10%–30% of total binding) and produced a robust specific signal. However, a consistent nonspecific binding component in white matter was observed in all investigated species, in particular the corpus callosum and subcortical tracts. This is supported by previously reported in vivo findings in human and non-human primate subjects [39,43], where tracer uptake was lowest in white matter structures, such as the corpus callosum. Furthermore, we also observed slower in vivo kinetics in white matter structures in the pig brain. This could be caused by lower perfusion or kinetics may be different when the tracer interacts with lipid membranes, compared to interaction with the receptor. As in vitro binding conditions are distinctly different due to absent metabolism and blood flow, it is possible that these effects may limit pronounced non-specific white matter binding in vivo or that the nonspecific binding exhibits much slower kinetics. Under the employed incubation conditions, [ 125I]α-bungarotoxin shows no white matter residual binding. However, under the same conditions, the overall non-specific binding in grey matter for [125I]α-bungarotoxin is approximately 45% in human (data not shown) and 55% in pig brain tissue, where it is much lower for [125I]Iodo-ASEM (~10%–30%).

[ 125I]Iodo-ASEM binding enables an important distinction between grey and white matter structures, e.g., the distinct cortical laminar binding pattern observable in the pig. While [ 125I]Iodo-ASEM binding was prominent in the deeper cortical layers in the mouse and rat, superficial cortical layers were intensely labelled in the pig. Using in vitro autoradiography, similar laminar cortical binding pattern in the pig brain has also recently been reported for a structurally different α7 nAChR radioligand, [3H]NS14492 [61]. Binding of both tracers was matching the pattern of [ 125I]α-bungarotoxin, the in vitro gold-standard radioligand for α7 nAChR. However, the spatial binding pattern in the rodent brain was only similar between [125I]Iodo-ASEM and [125I]α-bungarotoxin, but not for [3H]NS14492, suggesting different binding profiles of antagonists and agonists, or species differences in affinity.

[ 125I]Iodo-ASEM proved specific to the α7 nAChR, as evidenced by the lack of specific binding in α7 nAChR gene-deficient mice and a virtually complete block of cortical [125I]Iodo-ASEM binding by a wide range of structurally different α7 nAChR selective ligands and MLA, with the rank order being NS14492=TC-5619=EVP-6124=A-582941=SSR-180,711>MLA>GTS-21. While this corresponds well with the individual high affinities in the nanomolar range (NS14492, TC5619, EVP-6124, A-582941, SSR-180,711) [31,62–65], as compared to the lower affinity of a partial agonist (GTS-21) [66], it could also reflect the general differences between antagonists and agonist in terms of binding sites and kinetics.

When comparing to [125I]α-bungarotoxin, the specific binding of [125I]Iodo-ASEM was lower in the rat and mouse brain, but higher in the pig cortex. While species differences in receptor structure may account for the discrepancies, it should also be considered that different incubation protocols were used for the determination of optimal [125I]Iodo-ASEM and [125I]α-bungarotoxin binding. Hence, a relatively high detergent concentration was required in the experiments to obtain optimal total tissue binding of [125I]Iodo-ASEM, which may potentially affect binding of the radioligand in the mammalian species tested, e.g., through differences in lipid content and myelination. For example, [ 125I]Iodo-ASEM showed some degree of non-displaceable binding to white matter structures, which could be caused by the ligands' lipophilicity and/or different kinetics in white matter structures.

Interestingly, radioligand binding in β2 nAChR gene-deficient mice was more strongly reduced for [125I]Iodo-ASEM than [125I]α-bungarotoxin. This observation suggests different binding properties and subtype selectivity to heteromeric α7β2 nAChR, compared to the homomeric receptors. In the CNS, heteromeric α7β2 nAChR are identified in the mouse forebrain and hippocampal neurons, rat basal forebrain cholinergic neurons, as well as in the human basal forebrain and cerebral cortex. Importantly, α7β2 nAChR display distinct functional properties as compared to homomeric α7 nAChR [51,67], owing to their slower whole cell decay kinetics and current amplitudes in both transfected cell systems and native rodent neurons [53,54,68–70]. Accordingly, co-expression of α7 and β2 nAChR subunits in *Xenopus* oocytes also results in lower maximal responses (evoked current amplitudes) of selective α7 nAChR agonists but does not shift pharmacology to a more β2-like profile [52–54,71]. These in vitro studies in transfected cell systems therefore suggest that α7 nAChR agonists bind to the α7-α7 subunit interface, and β2 subunits likely do not contribute to the ligand binding site on heteromeric α7β2 nAChR [67,71]. When using selective α7 nAChR antagonists, including MLA and α-bungarotoxin, to alter the response to some nicotinic agonists in either homomeric α7 and heteromeric α7β2 nAChR, results have been less consistent, as they show unaltered [54] or reduced potency [52] and efficacy [53] in comparison to homomeric α7 nAChR expressed in *Xenopus* oocytes. The functional significance of heteromeric α7 nAChR expression is not well understood, with recent work suggesting that this subtype combination might be more sensitive to inhibition by oligomeric amyloid β1–42 [68,69] and isoflurane [72], as compared to homomeric α7 nAChR. Our finding that binding of [125I]Iodo-ASEM, an antagonist, was markedly reduced in the forebrain of β2 gene-deficient mice therefore may suggest that [125I]Iodo-ASEM binds to heteromeric α7β2 nAChR in the brain, as opposed to [125I]α-bungarotoxin. Whether this is due to different affinity for homomeric α7 and heteromeric α7β2 nAChR requires further in vitro studies. Although speculative, this may offer a chance to probe the binding of amyloid β1–42 to heteromeric α7β2 nAChR in vivo using PET.

In vivo uptake of [18F]ASEM into the pig brain occurred rapidly within the first 10–20 min and a reversible but slower washout was found, as observed in human and non-human primate subjects [39,43]. The in vivo distribution of [18F]ASEM found here is very similar to that of [11C]NS14492 and importantly, also in accordance with the distribution of α7 nAChR in the pig brain [31,73]. Furthermore, our data with [18F]ASEM matches previous reports with the structurally similar analogue [ 18F]DBT-10 in piglets [40].

We found variations in brain uptake and fP in the two baseline animals, and this result is consistent with the interpretation that lower fP will lead to lower brain uptake [74]. Due to the limited number of animals in this study, this observation merits further investigations. Our finding is however supported by PET studies in non-human primates with [18F]ASEM and [18F]DBT-10, where VT/fP was shown to be a more stable outcome measure than VT [44,45]. This has also been shown for radiotracers binding to other neurotransmitter receptors [75].

VT was found to increase slightly when the scan time was prolonged. This phenomenon was most pronounced in the thalamus and least pronounced in the white matter and is evident from the TACs (Figure 4A), where the ratio between e.g., thalamus and cerebellum was lower at 240 min than at 90 min. This finding is in contrast with the non-human primate and human data, where VT was underestimated when reducing the PET data from 180 to 60 min [44]. Given that ASEM is an antagonist, it is unlikely that internalization of the receptor-ligand complex is an explanation for the decrease in VT. We cannot exclude that other receptor adaptations, such as (de)sensitization, could be responsible for this observation. Desensitization could occur if experiments were not conducted at tracer dose, i.e., but we did not attempt to identify the mass dose limit of unlabelled ASEM. Although the injected doses of ASEM varied in the two baseline animals, we only found a 5% difference in the calculated VT/fP, which suggest that the studies were conducted at tracer doses. The injected doses in this study (0.007–0.085 ug/kg) are higher than the doses used in the non-human primate evaluation of [18F]ASEM, where injected doses ranged from 0.009 to 0.056 μg/kg [44]. Further studies are needed to identify the mass dose limit of unlabelled ASEM.

While pre-treatment with 1 mg/kg SSR-180,711 resulted in an increased uptake of [18F]ASEM, kinetic modelling for quantification of tracer uptake showed that SSR-180,711 at this dose resulted in 49% occupancy. A similar phenomenon has also been reported in piglets when [18F]DBT-10 was blocked by the weak agonist NS6740, which was ascribed to a potential blood flow-driven effect of NS6740 leading to greater central uptake of [18F]DBT-10 [40]. The increased tracer uptake could also be a result of peripheral α7 nAChR binding sites having been blocked by SSR-180,771. The occupancy found by us is in line with previous work of Horti et al., reporting 39% and 81% occupancy for doses of 0.5 and 5 mg/kg SSR-180,711, respectively [39]. The occupancy computed in this study should be interpreted with care, as the baseline and blocking study is conducted in two different pigs. Due to the half-life of [18F]ASEM, it was not possible to conduct the study in the same animal on the same day. A further limitation to this in vivo study is the low number of PET scans and animals and thus we can only provide a descriptive presentation of the data, without statistical evaluations.

From our results, [125I]Iodo-ASEM therefore offers several advantages over [125I]α-bungarotoxin: 1) low nonspecific binding, 2) similar high affinity and selectivity and 3) in vivo applicability and direct comparison of PET data with autoradiographic data. The lower nonspecific binding of [125I]Iodo-ASEM is advantageous primarily in vitro, as it allows for a better signal-to-noise ratio over [125I]α-bungarotoxin at very similar affinities. While a low nonspecific binding would also be favourable under in vivo conditions (e.g., PET), having two nearly identical molecules as tracers offers interesting avenues, especially for preclinical studies. In vivo PET/SPECT imaging data can be acquired through [18F]ASEM, [ 18F]DBT10 or [123I]Iodo-ASEM and results can be validated or extended by using the advantages of in vitro autoradiography (e.g. resolution) with [125I]Iodo-ASEM.

In conclusion, [125I]-Iodo-ASEM is applicable for visualizingα7 nAChR binding in vitro, its binding is different between species, and may potentially bind to heteromeric α7β2 nAChR. In addition, [ 18F]ASEM is demonstrated to have suitable kinetic properties for in vivo quantification of α7 nAChR in the pig.

#### **4. Materials and Methods**

#### *4.1. Compounds and Radioligands*

[ 125I]Iodo-ASEM [(3-(1,4-diazabycyclo[3.2.2]nonan-4-yl)-6-(125I-iododibenzo[b,d]thiopentene 5,5-dioxide)] was labelled according to previously published procedures [42]. Mean molar activity was 59.94 <sup>±</sup> 6.25 TBq/mmol. [125I]Tyr-54-mono-Iodo-α-bungarotoxin (81.4 TBq/mmol) was purchased from Perkin-Elmer (Skovlunde, Denmark). (-)-nicotine tartrate was purchased from Sigma-Aldrich (St. Louis, MO). Unlabelled ASEM and precursor for radiosynthesis was provided by DanPET (Malmoe, Sweden). The α7-selective ligands were purchased from Sigma-Aldrich (MLA) or provided by DanPET (NS11492) or NeuroSearch A/S (Copenhagen, Denmark) (SSR-180,711, TC-5619, EVP-6124, A-58294, and GTS-21.

#### *4.2. Tissue Origin and Sectioning for In Vitro Autoradiography*

All animal procedures were approved by the Danish Animal Experimentation Inspectorate (J. No. 2012-15-2034-00156) and treated in concordance with the European Communities Council Directive of 24th November 1986 (86/609ECC).

One female Sprague-Dawley rat (250 g, obtained from Charles River, Sulzfeld, Germany) was euthanized with an intraperitoneal overdose of pentobarbital, the brain was quickly removed and snap-frozen in −50 ◦C 2-methylbutane, then stored at −80 ◦C until further processing.

Mice deficient for the α7 subunits (The Jackson Laboratory) and β2 (Institut Pasteur, Paris, France) and their corresponding wild-type littermates were bred (C57BL/6J background) in an animal care facility at Virginia Commonwealth University. Brains from α7 and β2 gene-deficient mice and corresponding wild-type littermates were kindly provided by Dr. M. Imad Damaj (Dept. of Pharmacology and Toxicology, Virginia Commonwealth University, Richmond, VA, USA).

One two-month old female Danish domestic pig (Landrace x Yorkshire x Duroc, 22 kg) was euthanized with an intravenous injection of pentobarbital, the brain was quickly excised, separated in two hemispheres and frozen on dry ice, before being stored at −80 ◦C.

All brain specimens were cut in 12 μm serial sections on a cryostat (Microm HM 500 OM, Walldorf, Germany), thaw-mounted onto Super Frost slides (Thermo Scientific, Hvidovre, Denmark), briefly air dried and stored at −80 ◦C until further processing. Protein concentration was determined from single or three sections with the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA) based on the method of Bradford [76].

#### *4.3. In Vitro Autoradiography with [125I]Iodo-ASEM*

Initial optimization of assay conditions was performed to maximize total binding while keeping non-specific binding low. Adjustments included buffer composition and pH, detergent concentration, wash and incubation time and temperature. An assay buffer with 50 mMol/L Tris-HCl pH 7.4, 21 ◦C (termed Tris-HCl buffer) provided best preserved tissue integrity and lowest non-specific binding, as compared to physiologic Tris, Tris-EDTA-EGTA or HEPES-KRH buffer (data not shown). For all further experiments, tissue from 1–2 animals was used, with experiments and quantifications carried out using 3–4 sections for pig and rat tissue and 3–6 sections for mouse tissue. Adjacent sections were used for autoradiography for all similar experiments (e.g. saturation binding). Sections were brought to room temperature and pre-incubated for 20 min in Tris-HCl buffer (pH 7.4, 21 ◦C), then incubated for 60 min in the same buffer (21 ◦C) containing 1.5% Triton X-100 (v/v) and 0.5-1.0 nMol/L [ 125I]Iodo-ASEM in a humidified chamber. Non-specific binding of [125I]Iodo-ASEM was assessed in the presence of 10 μMol/L SSR-180,711, added to the buffer. Blocking of [125I]Iodo-ASEM binding was investigated with a series of individual α7 nAChR selective compounds (10 μMol/L) added to the incubation buffer, i.e., methyllycaconitine (MLA) [77], SSR-180,711 [63], NS14492 [31], TC-5619 [64], EVP-6124 [65], A-582941 [62], and GTS-21 [66]. Following incubation, slides were rinsed in Tris-HCl buffer (pH 7.4, 21 ◦C), washed 2 × 5 min in Tris-HCl buffer (pH 7.4, 4 ◦C) and rinsed (5 sec) in ice-cold distilled water. Slides were gently dried under an air stream and exposed to 4% paraformaldehyde vapour overnight at 4 ◦C, followed by another drying step in a desiccator for 1 h. Adjacent sections were used for autoradiography.

#### *4.4. In Vitro Autoradiography with [125I]*α*-bungarotoxin*

Slides were thawed at room temperature (21 ◦C) for 30 min, followed by 30 min of rehydration in 50 mMol/L Tris buffer with 0.1% BSA (w/v), pH 7.3 (binding buffer). For assessment of α-bungarotoxin binding, the binding buffer contained 0.5 mMol/L [125I]α-bungarotoxin and 4.5 nMol/L unlabelled α-bungarotoxin (Tocris, Denmark) yielding a total of 5 nMol/L α-bungarotoxin (incubation buffer). Total binding was determined using one set of slides incubated with the radioligand for 2 h at room temperature in a humidified chamber. Non-specific binding was determined in the presence

of 1 mMol/L (-)-nicotine added to the incubation buffer. Afterwards, slides were briefly rinsed in binding buffer, followed by 2 × 30 min of washing in ice-cold binding buffer (4 ◦C). Finally, slides were briefly rinsed (5 sec) in ice-cold distilled water, dried under a gentle air stream and exposed to 4% paraformaldehyde vapour overnight at 4 ◦C. On the next day, the slides were dried for 1 h in a desiccator.

#### *4.5. Saturation Binding and Kinetic Analysis Using In Vitro Autoradiography*

Saturation binding was carried out in rat and pig brain sections as described above (Section 4.3). Sections were incubated with ten serial dilutions of [125I]Iodo-ASEM ranging from 0.02 to 10 nMol/L, with concentrations measured by gamma-counting. Non-specific binding was determined in adjacent sections in the presence of 10 μMol/L SSR-180,711 for each radioligand concentration. Binding was terminated by washing the sections in ice-cold binding buffer. The equilibrium dissociation constant (*K*d) and maximum number of binding sites (Bmax) were determined by non-linear regression analysis of a one-site saturation binding model using GraphPad Prism 6.0 (GraphPad Software, Inc., San Diego, CA, USA).

#### *4.6. Autoradiographic Image Acquisition and Analysis*

BAS SR2040 phosphor imaging plates (Fujifilm, Toyko, Japan) were exposed to the samples along with [125I] standards (ARI 0133A; American Radiolabeled Chemicals, St. Louis, USA) for 24-72 hours. Imaging plates were scanned using a Phosphor Imager BAS-2500 (Fujifilm Europe GmbH, Düsseldorf, Germany). Images were converted to TIF-files using the manufacturer's software and analysed in QuantityOne (BioRad, Waltham, MA, USA). Regions of interest (ROIs) were drawn over grey and white matter structures, depending on the investigated species. High intensity circular spots were occasionally observed in [125I]Iodo-ASEM autoradiographs and were excluded from the analysis. For α7 and β2 gene-deficient mice, only one ROI was drawn over the whole brain, again excluding spots and irregular white matter binding. In the rat brain, the ROIs were drawn over cortex and hippocampus. From the pig brain, only frontal cortex sections were cut, the ROIs therefore contained the frontal cortex and white matter tracts. The mean values of optical density per mm<sup>2</sup> (averaged from the replicates) were converted to radioactive concentration using a linear regression derived from the [ 125I] radioactive standards. A global background of the imaging plate and individual non-specific binding were subtracted. Final values were expressed as fmol/mg protein, based on the protein measurements from individual sections.

#### *4.7. Radiosynthesis of [18F]ASEM*

The radiosynthesis of [18F]ASEM was performed as previously published [26]. No-carrier-added aqueous 18F-fluoride from the target was collected at a non-conditioned activated (10 mL ethanol, 20 mL water and dried with air) anion-exchange cartridge (QMA). A solution of 20 mg of 1,10-diaza-4,7,13,16,21,24-hexaoxabicyclo[8.8.8]hexacosane (Kryptofix-222) and 3.3 mg of K2CO3 dissolved in a 0.65 mL methanol-water mixture (97/3 v/v) was used to elute the 18F-fluoride off the cartridge. The elute was thereafter dried by evaporation at 90 ◦C under nitrogen and then further dried twice with 1 mL dry acetonitrile. To the dried Kryptofix®222/[ 18F]fluoride complex, 2.4 mg (0.006 mmol)/L of 3-(1,4-Diazabicyclo[3.2.2]nonan-4-yl)-6-nitrodibenzo[b,d]- thiophene 5,5-Dioxide dissolved in 0.8 mL DMSO was added. The reaction was performed at 160 ◦C for 15 min and afterwards the crude was quenched with 3.5 mL H2O. Reactants and by-products were separated from [18F]ASEM by semi-preparative HPLC [Luna column, Phenomenex Ltd. Aschaffenburg, Germany; 10 μm C18(2) 10×250 mm column, flow rate 6 mL/min, eluent: Ethanol/0.1% H3PO4 in water (25:75) with 6 mM ascorbic acid to prevent radiolysis]. The retention time for [18F]ASEM was 400-450 s and the product was collected into a vial containing 9 mL of PBS (phosphate-buffered saline). The product was visually inspected for clarity, absence of colour and visible particles. Chemical and radiochemical purities were assessed by analytical HPLC [Kinetex column, Phenomenex Ltd. Aschaffenburg, Germany; 2.6μ C18

4.60 <sup>×</sup> 50 mm, eluent: ACN/0.1% H3PO4 in water (25:75) RT: [18F]ASEM <sup>=</sup> 1.3 min; nitro precursor <sup>=</sup> <sup>1</sup> min; flow rate 1.5 mL/min]. Molar activity (Am) of the radiotracer was determined as follows: the area of the UV absorbance peak corresponding to the radiolabelled product was measured (integrated) on the HPLC chromatogram. This value was then converted into a molar mass by comparison with an average of integrated areas (triplet) of a known standard of the reference compound.

#### *4.8. In Vivo Imaging in the Pig*

Three female pigs (21, 22 and 23 kg) were used for in vivo PET imaging on a HRRT PET scanner (Siemens Healthcare, Erlangen, Germany). All animal procedures were approved by the Danish Council for Animal Ethics (journal no. 2012-15-2934-00156).

#### 4.8.1. Animal Procedures

Before scanning, anaesthesia was induced with i.m. injection of 0.13 mL/kg Zoletil veterinary mixture (Virbac, Kolding, Denmark; 10.87 mg/kg xylazine + 10.87 mg/kg ketamine + 1.74 mg/kg methadone + 1.74 mg/kg butorphanol + 10.87 mg/kg tiletamine + 10.87 mg/kg zolezepam). Hereafter, anaesthesia was maintained with constant propofol infusion (1.5 mg/kg/h intravenous (i.v.); B. Braun, Melsungen, Germany). An arterial i.v. catheter was employed for blood sampling from the right femoral artery and two venous i.v. catheters for injections were placed in the left and right mammary veins. During anaesthesia, animals were endotracheally intubated and ventilated. Vital parameters (heart rate, body temperature, blood pressure, oxygen saturation and end tidal CO2) were continuously monitored during the scan.

#### 4.8.2. PET Scanning

[ 18F]ASEM was given as intravenous i.v. bolus, with experimental details described in Table 4.


**Table 4.** Experimental details of [18F]ASEM PET scans in pigs.

#### 4.8.3. Blood Sampling

During the first 30 min of the scans, radioactivity in the whole blood was continuously measured using an ABSS autosampler (Allogg Technology, Mariefred, Sweden) counting coincidences in a lead-shielded detector. Concurrently, arterial whole blood was sampled manually at times 2.5, 5, 10, 20, 30, 40, 50, 70, 89, 91, 120 and 150 min after injection of [18F]ASEM. Total radioactivity in plasma (500 μL) and whole blood (500 μL) was measured in a well counter (Cobra 5003; Packard Instruments, Meriden, CT, USA), which was cross-calibrated to the HRRT scanner and autosampler. All measurements of radioactivity were decay corrected to the time of radioligand injection.

#### 4.8.4. Metabolite Analysis

Radiolabelled parent compound and metabolites were determined by direct injection of plasma into a radio-HPLC system (Dionex Ultimate 3000; Thermo Fisher Scientific, Hvidovre, Denmark) configured for column switching. Manually drawn arterial whole blood samples were centrifuged (1500 g, 7 min, 4 ◦C), and plasma was filtered through a syringe filter (Whatman GD/X 13 mm or 25 mm, PVDF membrane, 0.45 μm pore size; Frisenette ApS, Knebel, Denmark) prior to the analysis by HPLC

as previously described [78]. To increase sensitivity on gamma counts from samples with low levels of radioactivity, eluent from the HPLC was collected into fractions (10 mL) using a fraction collector (Foxy Jr FC144; Teledyne, Lincoln, NE, USA) and counted offline in a well counter (2480 Wizard<sup>2</sup> Automatic Gamma Counter, Wallac Oy, Turku, Finland).

#### 4.8.5. Determination of Free Fraction

The free, non-protein bound fraction of [18F]ASEM in pig plasma, fp, was estimated using an equilibrium dialysis chamber method as previously described [79].

#### 4.8.6. Reconstruction and Pre-Processing of PET Data

150 -minute list-mode PET data were reconstructed in 58 dynamic frames (6 × 10, 6 × 20, 6 × 30, 6 × 60, 4 × 120, 14 × 300, 8 × 150, 8 × 300 s). One animal was scanned for 240 min using the mentioned framing protocol but adding 9 frames of 600 s). Images consisted of 207 planes of 256 × 256 voxels of 1.22 × 1.22 × 1.22 mm. A summed picture of all counts in the 150-min scan was reconstructed for each pig and used for co-registration to a standardized MRI-based atlas of the domestic pig brain, similar to that previously published [80]. The time activity curveds (TACs) were calculated for the following volumes of interest (VOIs): thalamus, striatum, hippocampus, cerebellum, white matter, frontal cortex, somatosensory cortex, occipital cortex, rest of the cortex. Radioactivity in all VOIs was calculated as the average of radioactive concentration (Bq/mL) in the left and right sides. Outcome measure in the TACs was calculated as radioactive concentration in VOI (in kBq/mL) normalized to the injected dose corrected for animal weight (in kBq/kg), yielding standardized uptake values (g/mL).

#### 4.8.7. Kinetic Modelling of PET Data

The PET imaging data were analysed with the Logan graphical analysis (LGA) model, using the metabolite corrected arterial plasma concentration to calculate the primary outcome measure: total distribution volume (VT). The secondary outcome measure was VT values corrected for free fraction in plasma (VT/fP). The parent fraction curve for [18F]ASEM was fitted with a Watabe fit. Both curves were constrained to 1.0 at time = 0. Kinetic modeling was performed in PMOD version 3.0 (PMOD Technologies).

**Author Contributions:** Conceptualization: C.K.D., J.D.M., A.G.H., M.G.P., D.P; Methodology: C.K.D., H.D.H., G.M.K., M.M.H.; Formal analysis: C.K.D.; H.D.H., H.H.H.; Investigation: C.K.D.; H.D.H., E.T.L.; Resources: D.P., R.C.M., M.M.H., A.G.H.; Writing—original draft preparation: C.K.D.; H.H.H., H.D.H.; Writing—review and editing: C.K.D., H.D.H., J.D.M., A.G.H., M.G.P., G.M.K., M.M.H.,; Visualization: C.K.D., H.D.H.; Supervision: J.D.M., G.M.K.; Funding acquisition: J.D.M., G.M.K., M.G.P.; All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by the Danish Strategic Research Council (project COGNITO), EB024495, the Augustinus, Elsass and The Lundbeck Foundations.

**Acknowledgments:** The authors would like to thank the animal facilities at the Department of Experimental Medicine, University of Copenhagen.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Sample Availability:** Samples of the compounds are not available.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Brief Report* **A New Positron Emission Tomography Probe for Orexin Receptors Neuroimaging**

**Ping Bai 1,2,3, Sha Bai 2, Michael S. Placzek 2, Xiaoxia Lu 1, Stephanie A. Fiedler 2, Brenda Ntaganda 2, Hsiao-Ying Wey <sup>2</sup> and Changning Wang 2,\***


Received: 17 December 2019; Accepted: 20 February 2020; Published: 25 February 2020

**Abstract:** The orexin receptor (OX) is critically involved in motivation and sleep−wake regulation and holds promising therapeutic potential in various mood disorders. To further investigate the role of orexin receptors (OXRs) in the living human brain and to evaluate the treatment potential of orexin-targeting therapeutics, we herein report a novel PET probe ([11C]CW24) for OXRs in the brain. CW24 has moderate binding affinity for OXRs (IC50 = 0.253 μM and 1.406 μM for OX1R and OX2R, respectively) and shows good selectivity to OXRs over 40 other central nervous system (CNS) targets. [ 11C]CW24 has high brain uptake in rodents and nonhuman primates, suitable metabolic stability, and appropriate distribution and pharmacokinetics for brain positron emission tomography (PET) imaging. [11C]CW24 warrants further evaluation as a PET imaging probe of OXRs in the brain.

**Keywords:** orexin receptors; PET; radiotracer; imaging

#### **1. Introduction**

Orexin (OX) is a hypothalamic hypocretin peptide that mediates multiple functions such as arousal, attention, neuroendocrine, water balance, and pain modulation [1–6]. The orexin system has two peptide members, orexin-A and orexin-B with 33 amino acids and 28 amino acids, respectively [7]. Orexin receptors (OXRs) also have two subtypes (OX1R and OX2R). Orexin-A has similar binding affinities to OX1R and OX2R, while orexin-B has a 10-fold higher affinity for OX2R compared with OX1R. OXRs are differently expressed throughout the brain. OX1R is mainly expressed in the prefrontal and infralimbic cortex, hippocampus, paraventricular thalamic nucleus, and locus coeruleus [8]. OX2R is mainly distributed in the cerebral cortex, septal nuclei, lateral hypothalamus, hippocampus, and hypothalamic nuclei [9]. The differential distribution of OXRs in the brain may be responsible for the various physiological and psychiatric functions mediated by the OXR system [10–12].

Therapeutics targeting the orexin system have been investigated for various brain disorders such as insomnia [1,3], cluster headache [13], substance abuse [4,14], and maladaptation to stress [15,16]. This has led to the design and evaluation of orexin antagonists, including dual-OX1R and OX2R antagonists (DORA) and selective OX1R and OX2R antagonists (SORA). [7] Almorexant [5] (ACT-426606) [17] from Actelion, SB-649868 [18] from GSK, and Suvorexant [6] from Merck are a few examples demonstrating the efforts from pharmaceutical companies. Among these compounds, many have entered into clinical trials, however, Suvorexant is the only U.S. Food and Drug Administration (FDA) approved orexin drug for the insomnia market. The discovery of ORX

antagonists provided powerful tool compounds to investigate the roles of the orexin system in a variety of biological processes. However, the roles of the orexin system in the brain are not well understood and further investigations with translational methods, such as in vivo imaging, are warranted.

A non-invasive molecular imaging with positron emission tomography (PET) is a unique tool to study the living brain, which provides a distinct advantage to quantify receptor expression and/or drug occupancy of novel ligands in vivo. Over the last decade, though there have been major advances in the development of orexin antagonists [19–25], no PET radiotracer is available for OXR imaging in the human brain. This limits our ability to investigate the orexin system in the human brain and hinders the potential of orexin-targeting neurotherapeutics for brain disorders.

To bridge the gap, several orexin antagonists have been radiolabeled by our group, as well as others. Most of the early discovery attempts resulted in compounds with low brain uptake or strong P-glycoprotein (P-gp) binding, including [11C]MK-1064 [26], [11C]EMPA [27], [11C]BBAC [28], and [11C]CW4 [29] (Figure 1). Towards the development of a successful OXR brain-PET imaging probe, we report the discovery and evaluation of [11C]CW24 with good brain uptake, suitable metabolic stability, and pharmacokinetics based on animal imaging studies in rodents and non-human primates.

**Figure 1.** Structures of orexin receptor radioligands.

#### **2. Results and Discussion**

#### *2.1. Selection of Sca*ff*old for Orexin Imaging Probe Development*

Several small molecules have been reported as DORA or SORA [20,30,31]. Previously, our group has radiolabeled several DORAs based on a 1,4-diazepane scaffold [27,29]. However, the low brain uptake limited their further evaluation as a brain-PET radiotracer. Recently, IPSU was reported as an OX2R selective antagonist with good brain uptake [32,33]. It has been studied in rodents, including blood and brain pharmacokinetics, as well as EEG (electrocorticogram/electroencephalogram) and EMG (electromyogram) recordings following treatment with these OXR antagonists in mice. Based on the structure-activity relationship (SAR) study of ISPU analogs, methylation of the indole imine does not disrupt OXR antagonism [32]. Hence, we chose the indole imine of IPSU as the radiolabeling site and synthesized CW24. The affinity of CW24 for orexin receptors (OX1R and OX2R) in cells was measured in a competition binding assay against [125I] orexin. CW24 displayed potent activity at OXRs (IC50 0.253 μM OX1R, 1.406 μM OX2R). Compared to IPSU binding, this was ~5 fold (OX1R) and ~2 fold (OX2R) increase in potency, warranting further evaluation.

#### *2.2. Physicochemical Properties of CW24*

Lipophilicity (logP) and molecular weight need to be optimized for sufficient blood-brain barrier (BBB) penetration while also avoiding high non-specific binding. Studies have shown that it is preferred for clogP ≤ 4, clogD ≤ 3, total polar surface area (tPSA) between 30–75 and molecular weight (M.W.) < 500 for successful brain-PET imaging probes [34,35]. IPSU is reported as a non-selective OXR antagonist with good brain uptake [33]. CW24, the methylated form of IPSU, had favorable properties for CNS penetration (M.W. = 419.5, Log D(7.4) = 2.3, tPSA = 60.74, IC50 = 0.253 μM and 1.406 μM for OX1R and OX2R, respectively) (Figure 2). Furthermore, we tested the off-target binding of CW24 in vitro, [36] and were pleased to see that CW24 was selective for OXRs compared to 40 other CNS targets(National Institute of Mental Health Psychoactive Drug Screening Program, NIMH PDSP) (Table S1).


**Figure 2.** Structures and physicochemical properties of IPSU and CW24 (n.t. = not tested). IC50 of IPSU, CW24 and suvorexant were measured by a radioligand competition binding assay.

#### *2.3. Chemical Synthesis for CW24 and [11C]CW24*

CW24 was prepared from IPSU via direct methylation in the presence of a base (Figures S1 and S2). [ 11C]CW24 was radiosynthesized from IPSU (0.5 mg) in DMF (1 mL) with KOH (10 mg) and allowed to react with [11C]CH3I for 3 min at 100 ◦C (Scheme 1). The average synthesis time was 30–35 minutes from end of cyclotron bombardment (EOB) to end of synthesis (EOS), the radiochemical yield (RCY) was 10–21% (non-decay corrected, from trapped [11C]CH3I), the specific activity (As) was 1.28 <sup>±</sup> 0.2 mCi/nmol (EOS), the chemical and radiochemical purities of [11C]CW24 were <sup>≥</sup> 97%.

**Scheme 1.** Radiolabeling condition: IPSU (precursor, 0.5 mg), [11C]CH3I, KOH (10 mg), in 1.0 mL DMF, 3 min, 100 ◦C. Radiochemical yield (RCY): 10–21% (non-decay corrected from trapped [11C]CH3I).

#### *2.4. Mouse Imaging with [11C]CW24*

We performed PET-computed tomography (CT) imaging on mice to test [11C]CW24 as a neuroimaging probe for OXR imaging in vivo. [11C]CW24 exhibited high brain uptake (%ID/cc = 3.5% at Cmax, 1.5 min post-injection). In general, %ID/cc above 0.1% in rat or 0.01% in non-human primate within 5 min post-injection is above the threshold for suitable BBB penetration for CNS PET imaging studies [37]. Next, to evaluate the specificity of [11C]CW24 for OXRs, PET-CT imaging studies were performed in mice after 5 min pretreatment with IPSU (0.5 and 2.0 mg/kg; i.v.) or Suvorexant (5.0 mg/kg; i.v.). Time-activity curves (TACs) were normalized for peak uptake (~20 seconds post-injection in these studies) (Figure 3). Approximately 10% reduction in [11C]CW24 uptake with 0.5 mg/kg IPSU administration in 30–60 mins post-injection of [11C]CW24. In addition, approximately 50% reduction was found when the blocking dose of IPSU was increased to 2.0 mg/kg. Suvorxant (5.0 mg/kg) administration at 30–60 mins post-injection of [11C]CW24 also showed ~50% reduction in the mouse brain. Following IPSU pretreatment, [11C]CW24 binding was reduced in a dose-dependent manner in the brain for the relative radioactivity changes after administration. These results strongly support that [11C]CW24 binds selectively to the OXRs in vivo.

**Figure 3.** (**A**) The PET/CT imaging of [11C]CW24 focused on mice brain (20-60 min after intravenous administration (i.v.)); (**B**) baseline and blocking time-activity curve (TAC) of [11C]CW24 and (**C**) TAC after normalized the brain uptake curves with the highest uptake time point (20 seconds post-injection).

#### *2.5. Non-human Primate (NHP) Imaging with [11C]CW24*

Encouraged by the promising imaging results of [11C]CW24 in mice, we performed PET- Magnetic resonance imaging (MR) imaging in a NHP (Figures 4 and 5). In macaque, [11C]CW24 demonstrated high brain uptake. We observed a peak standard uptake value (SUV) (SUV=C(T)/(injected dose/body weight) between 1.5–3.0 for all brain regions examined (Figure 4A). Furthermore, relatively higher uptake was observed in the midbrain and the thalamus and lower uptake in the hippocampus which suggests heterogenous expressions of OXR in the macaque brain. To further investigate the specific binding of [11C]CW24, a blocking study was performed in which 2.0 mg/kg IPSU was administered 5 min prior the PET-MR acquisition. Similar to the blocking results shown in mice, a reduction of radioactivity uptake was found in several brain regions including hippocampus, nucleus accumbens, thalamus, hypothalamus, putamen and cerebellum (Figure 4B). Using the arterial blood data as input function, Logan graphical analysis was applied to quantify volume of distribution (VT, mL/cm3) of [ 11C]CW24 (Figure 5). The metabolite-corrected arterial plasma after [11C]CW24 bolus injection is showed in Figure 5C.

**Figure 4.** (**A**) Positron emission tomography (PET)-Magnetic resonance (MR) Imaging (macaque brain). Summed PET images (30–90 min after [11C]CW24 injection) superimposed structural MRI from the same macaque. (**B**) Time−activity curves for brain regions of interest from baseline and blocking scans are shown (2.0 mg/kg IPSU) (a) hippocampus, (b) nucleus accumbens, (c) thalamus, (d) hypothalamus, (e) putamen and (f) cerebellum.

**Figure 5.** Kinetic modeling results with [11C]CW24 in the macaque brain. (**A**) The total volume of distribution (VT) images derived from Logan plot with arterial blood data as input function (color bar indicates VT values from 1–2.5 mL/cm3 (blue to red); (**B**) Regional VT values showed that the expression difference in brain regions; (**C**) Arterial plasma analysis showed that [11C]CW24 radioactivity was rapidly cleared from blood and [11C]CW24 stability evaluated in plasma over time showed lasting presence of ~47% of parent compound at 30 min.

Based on our evaluation both in rodents and NHPs in vivo, [11C]CW24 is a promising OXR PET radioligand candidate. It has been reported that OXR was typically present in the cerebral cortex, septal nuclei, hippocampus, medial thalamic groups, raphe nuclei, and abundantly expressed in hypothalamic nuclei that regulate the homeostasis in rodents [12]. In addition, OX2R was found to have expression in the hypothalamus that enables its unique role of sleep-wake regulation [11]. Compare to the previous OXR probes we developed [27,29] [11C]CW24 had higher brain uptake, increased isoform selectivity over IPSU, appropriate kinetics and distribution, which warrants further development of OXR PET radioligands based on [11C]CW24.

#### **3. Materials and Methods**

#### *3.1. Synthesis of Compound CW24*

A mixture of IPSU (5 mg, 0.012 mmol, purchased form MedChemExpress), KOH (4 mg, 0.06 mmol), and CH3I (5 mg, 0.036 mmol) in DMF (0.1 mL) were reacted at room temperature overnight. The reaction mixture was purified by reverse phase chromatography (H2O: CH3CN = 1:1) to give the desired product (2 mg, 0.005 mmol, 39% yield) as a white solid. 1H-NMR (500 MHz, Chloroform-d) δ 8.04 (d, *J* = 5.6 Hz, 1H), 7.64 (d, *J* = 7.9 Hz, 1H), 7.29 (d, *J* = 8.2 Hz, 1H), 7.22 (t, *J* = 7.7 Hz, 1H), 7.10 (t, *J* = 7.4 Hz, 1H), 7.02 (s, 1H), 5.93 (s, 1H), 4.72 (s, 2H), 4.40 (dt, *J* = 13.7, 4.7 Hz, 2H), 3.87 (s, 3H), 3.75 (s, 3H), 3.38 (m, 2H), 3.24 (t, *J* = 6.1 Hz, 2H), 2.24 (m, 2H), 1.83–1.77 (m, 2H), 1.73 (m, 2H), 1.50 (m, 2H).13C-NMR (125 MHz, Chloroform-d) δ 174.50, 169.94, 161.78, 157.98, 137.12, 128.79, 127.61, 121.89, 119.68, 119.41, 110.84, 109.29, 95.96, 53.00, 46.79, 41.77, 40.27, 39.80(2), 33.84(2), 32.82, 30.61, 18.91. LC-MS [M]<sup>+</sup> 419.85.

#### *3.2. Radiosynthesis of [11C]CW24*

[ 11C]CH3I was trapped in a reactor (TRACERlab FX-M synthesizer, General Electric) preloaded with the precursor (0.5 mg), KOH (5.0 mg) in 1.0 mL dry DMF. The mixture was stirred for 3 min at 100 ◦C and followed by adding water (1.2 mL). The product was separated by reverse phase semi-preparative HPLC (Phenomenex Luna 5u C8(2), 250 mm × 10 mm, 5 μm; 5.0 mL/min; 60% H2O + ammonium formate (0.1 M)/ 40% CH3CN; isocratic). The collected final product was loaded onto a C-18 sep-pak cartridge, and rinsed with water (5 mL), eluted with EtOH (0.3 mL), and saline (0.9%, 2.7 mL). The average time required for the synthesis from EOB to EOS was 30–35 min. The average radiochemical yield (RCY) was 10%–21% (non-decay corrected to trapped [11C]CH3I). Chemical and radiochemical purities were ≥ 95 % (measured with HPLC equipped with a UV detector and a gamma detector) with a specific activity (As) of 1.28 ± 0.2 mCi/nmol (EOS).

#### *3.3. Assessment of Lipophilicity (Log D; pH 7.4)*

Log D was determined according to methods identical to those we previously reported [38].

#### *3.4. Human Orexin GPCR Binding (Agonist Radioligand) Assay*

Radioligand competition binding assays were performed by Panlabs, Eurofins Pharma Discovery Services. Briefly, human recombinant orexin OX1 receptors expressed in CHO-S cells were used in modified HEPES buffer pH 7.4. An aliquot was incubated with 0.1 nM [125I] Orexin A for 60 minutes at 25 ◦C. Receptors were filtered and then counted to determine [125I] Orexin A specifically binding. Human recombinant OX2 receptors were used in modified HEPES buffer pH 7.4. An aliquot was incubated with 0.04 nM [125I] Orexin A for 180 minutes at 25 ◦C. Membranes were filtered and then counted to determine [125I] Orexin A specifically bound.

#### *3.5. Rodent PET*/*CT Acquisition*

The Subcommittee on Research Animal Care (SRAC) serves as the Institutional Animal Care and Use Committee (IACUC) for the Massachusetts General Hospital (MGH). SRAC reviewed and approved all procedures detailed in this paper. B6C3F1/J mice (male, 18-month old; *n* = 4 total) were utilized in this study. IPSU and Suvorexant were purchased form MedChemExpress and dissolved in 1.0 % DMSO + 1.0 % Tween 80 + 98.0 % saline to make a solution of 1.0 mg/mL.

Animals were anesthetized with 1–1.5% isoflurane and maintained with isoflurane during the imaging scan. In a single imaging session, mice were arranged in a Triumph PET/CT scanner (Gamma Medica, Northridge, CA). The mice were administered [11C]CW24 (3700–7400 KBq per animal) after 5-min pre-treatment of IPSU (0.5 and 2.0 mg/kg; *n* = 1 for each dose; i.v.), Suvorexant (5.0 mg/kg; *n* = 1; i.v.), or vehicle (*n* = 1) via a lateral tail vein catheter. Animals underwent a 60 min dynamic PET scan followed by computed tomography (CT).

#### *3.6. Rodent PET*/*CT Image Analysis*

PET data were reconstructed using a 3D-MLEM method resulting in a full width at half-maximum resolution of 1 mm. PET and CT images in DICOM format were imported to PMOD (PMOD Technologies, Ltd. Zürich, Switzerland) and co-registered to the brain atlas. Volumes of interest (VOIs) were drawn as spheres in brain regions guided by CT structural images and summed PET data. Time-activity curves (TACs) were exported as activity per unit volume (%ID/cc) for analysis.

#### *3.7. Macaque PET-MR Acquisition*

A male rhesus macaque (14.4 kg) was included in this study. After endotracheal intubation, the macaque was catheterized antecubitally for radiotracer injection and a radial arterial line was placed for arterial blood sampling and radiometabolite analysis. The animal was anesthetized with 1.2–1.3% Isoflurane in medical oxygen throughout the imaging session. PET-MR data was acquired on a 3T Siemens MRI scanner (Munich, Germany) with a BrainPET insert. A baseline and a blocking scan were performed on the same animal. Dynamic PET image acquisition was initiated followed by a bolus administration of [11C]CW24 (179 MBq for the baseline scan and 192 MBq for the blocking scan). To characterize the specific binding of [11C]CW24, a blocking study was carried out in which 2.0 mg/kg IPSU (1.0 % DMSO + 1.0 % Tween 80 + 98.0 % saline to make a solution of 1.0 mg/mL) was administered 5 min prior the acquisition. A T1-weighting multi-echo magnetization-prepared rapid gradient-echo (MEMPRAGE) sequence began as the PET scan for anatomic co-registration. Dynamic PET data were corrected for decay, scatter, and attenuation and reconstructed using methods similar to our previous studies [39].

#### *3.8. PET-MR Imaging Data Analysis for the Macaque Study*

Motion-corrected PET data were registered to the INIA19 macaque MRI Template. VOIs were selected according to the INIA19 Template and NeuroMaps Atlas [40]. TACs from the whole *Macaque* brain as well as a few brain regions were exported. With the availability of a metabolite-corrected arterial plasma data, the regional volume of distribution (VT) was calculated using Logan Plot analysis (with a fixed t\* of 40 min).

#### *3.9. Plasma and Metabolite Analysis*

Briefly, arterial blood samples collected (every 10 s at first 3 min post-administration of [11C]CW24, then 5 min, 10 min, 20 min, 30 min, 45 min, 60 min, and 90 min) during macaque imaging were centrifuged to separate plasma, which was then removed and placed in an automated gamma counter [41]. Metabolite analysis of blood samples collected from 5 to 90 min was conducted on a custom automated robot fitted with Phenomenex SPE Strata-X 500 mg solid phase extraction cartridges that were primed with ethanol (2 mL) and deionized water (20 mL). Protein precipitation was achieved by the addition of plasma (300 μL) to acetonitrile (300 μL), which was centrifuged for 1 min to obtain protein-free plasma (PFP). Three hundred microliters of PFP/acetonitrile solution was diluted into deionized water (3 mL), loaded onto the C18 cartridge, and removed of polar metabolites with 100% water. Next, a series of extractions was performed using water and acetonitrile. Each sample was counted in a WIZARD2 Automatic Gamma Counter to determine the presence of radiolabeled metabolites.

#### **4. Conclusions**

In summary, our developed PET imaging radio-ligands, [11C]CW24, provides a non-invasive quantitative imaging tool for evaluating OXR expression in the brain. Based on the in vitro and in vivo evaluations, [11C]CW24 had high brain uptake and good target-selectivity, but suffers from high non-specific binding. Therefore, [11C]CW24 could be used as a lead compound to develop brain-PET probes for OXRs to investigate not only the roles of OXR in a variety of disease applications, but also the development of OXR neurotherapeutics. In future studies, it is important that we characterize and optimize [11C]CW24 to improve specific binding for developing potential PET imaging probes for human imaging.

**Supplementary Materials:** The following are available online, Figure S1: Representative 1H, 13C-NMR and HPLC data for CW24; Figure S2: The HPLC profile of CW24; Table S1: Off-target binding data of CW24.

**Author Contributions:** Investigation, P.B., S.B.; writing—original draft preparation, P.B.; writing—review and editing, M.S.P., B.N., S.A.F., H.-Y.W.; supervision, X.L.; project administration, C.W.; funding acquisition, C.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by National Institutes of Health (NIH), grant number DA048123 and the APC was funded by NIH-DA048123.

**Acknowledgments:** This work was supported by NIH funding (DA048123) and pilot funding from the Athinoula A. Martinos Center for Biomedical Imaging at the Massachusetts General Hospital (C.W. and H.-Y.W.). The imaging studies were carried out at the Martinos Center, using resources provided by the *Center for Functional Neuroimaging Technologies, P41EB015896*, a P41 Regional Resource supported by the National Institute of Biomedical Imaging and Bioengineering (NIBIB), National Institutes of Health. This work also involved the use of instrumentation supported by the NIH Shared Instrumentation Grant Program and/or High-End Instrumentation Grant Program; specifically, grant numbers: S10RR017208, S10RR026666, S10RR022976, S10RR019933, S10RR023401. The author Ping Bai gratefully acknowledges financial support by China Scholarship Council (CSC) for this training at the Martinos Center. The authors are grateful to (Judit Sore, the Martinos Center radioligand lab, PET/MR imaging staff (Grae Arabasz, Shirley Hsu and Regan Butterfield), and Helen Deng for assistance with non-human primate imaging.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Sample Availability:** Samples of the compounds (IPSU and CW24) are available from the authors.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
