Molecular imaging of σ receptors: synthesis and evaluation of the potent σ1 selective radioligand [18F]fluspidine
- First Online:
- Cite this article as:
- Fischer, S., Wiese, C., Große Maestrup, E. et al. Eur J Nucl Med Mol Imaging (2011) 38: 540. doi:10.1007/s00259-010-1658-z
- 354 Views
Neuroimaging of σ1 receptors in the human brain has been proposed for the investigation of the pathophysiology of neurodegenerative and psychiatric diseases. However, there is a lack of suitable 18F-labelled PET radioligands for that purpose.
The selective σ1 receptor ligand [18F]fluspidine (1′-benzyl-3-(2-[18F]fluoroethyl)-3H-spiro[benzofuran-1,4′-piperidine]) was synthesized by nucleophilic 18F− substitution of the tosyl precursor. In vitro receptor binding affinity and selectivity were assessed by radioligand competition in tissue homogenate and autoradiographic approaches. In female CD-1 mice, in vivo properties of [18F]fluspidine were evaluated by ex vivo brain section imaging and organ distribution of intravenously administered radiotracer. Target specificity was validated by organ distribution of [18F]fluspidine after treatment with 1 mg/kg i.p. of the σ receptor antagonist haloperidol or the emopamil binding protein (EBP) inhibitor tamoxifen. In vitro metabolic stability and in vivo metabolism were investigated by LC-MSn and radio-HPLC analysis.
[18F]Fluspidine was obtained with a radiochemical yield of 35–45%, a radiochemical purity of ≥ 99.6% and a specific activity of 150–350 GBq/μmol (n = 6) within a total synthesis time of 90–120 min. In vitro, fluspidine bound specifically and with high affinity to σ1 receptors (Ki = 0.59 nM). In mice, [18F]fluspidine rapidly accumulated in brain with uptake values of 3.9 and 4.7%ID/g and brain to blood ratios of 7 and 13 at 5 and 30 min after intravenous application of the radiotracer, respectively. By ex vivo autoradiography of brain slices, resemblance between binding site occupancy of [18F]fluspidine and the expression of σ1 receptors was shown. The radiotracer uptake in the brain as well as in peripheral σ1 receptor expressing organs was significantly inhibited by haloperidol but not by tamoxifen. Incubation with rat liver microsomes led to a fast biotransformation of fluspidine. After an incubation period of 30 min only 13% of the parent compound was left. Seven metabolites were identified by HPLC-UV and LC-MSn techniques. However, [18F]fluspidine showed a higher metabolic stability in vivo. In plasma samples ∼ 94% of parent compound remained at 30 min and ∼ 67% at 60 min post-injection. Only one major radiometabolite was detected. None of the radiometabolites crossed the blood-brain barrier.
[18F]Fluspidine demonstrated favourable target affinity and specificity as well as metabolic stability both in vitro and in animal experiments. The in vivo properties of [18F]fluspidine offer a high potential of this radiotracer for neuroimaging and quantitation of σ1 receptors in vivo.
The sigma (σ) receptor was first discovered about 30 years ago as a new subtype of opioid receptors . At present it is known that σ receptors possess specific drug selectivity patterns, differential anatomical distribution and unique properties which are different from opioid and other known neurotransmitter and hormone receptor families. Pharmacological data based on binding studies, anatomical distribution and biochemical features distinguish at least two σ receptor subtypes (σ1 and σ2) which have recently been cloned and characterized [2–6]. The human σ1 receptor is a small transmembrane protein consisting of 223 amino acids, which is located on the outer cell membrane and the endoplasmic reticulum (ER) and was first cloned and functionally expressed by Kekuda and colleagues . It is structurally unrelated to any known mammalian receptor protein. However, a 30% homology exists between the cloned σ1 receptor and the yeast ergosterol-Δ8/Δ7 isomerase . Recently the hallucinogen N,N-dimethyltryptamine was discovered as an endogenous ligand which binds to σ1 receptors and produces a blockade of sodium channels, possibly through receptor translocation from the ER to the plasma membrane region .
In 2007 Hayashi and Su proposed the concept of “receptor chaperones” to explain the physiological role of σ1 receptors [9, 10]. According to this concept receptor activation allows σ1 receptors to act as molecular chaperones to inositol 1,4,5-trisphosphate (IP3) receptors, thereby enhancing Ca2+ signalling from the ER into mitochondria, activating the tricarboxylic acid cycle and increasing the production of ATP, which induces cell hypermetabolization, and ultimately results in neuroprotection and neurite outgrowth [8, 10]. Related to this postulated mechanism are experimental findings that σ1 receptor ligands exert a potent neuromodulation on excitatory neurotransmitter systems, e.g. the cholinergic and glutamatergic system [11–15], which offers potential for the treatment of neuropsychiatric diseases with σ1 receptor ligands [10, 16]. Based on studies reporting an involvement of σ1 receptors in the myelination of neuron-oligodendrocytes co-cultures and in the regulation of oligodendrocyte differentiation, it has been suggested that σ1 receptor ligands may be of therapeutic interest also for the treatment of multiple sclerosis [17, 18].
Of great interest is that certain σ receptor ligands such as ifenprodil not only possess affinity to σ1 and σ2 receptors but also to an additional binding site which shows homology to σ receptors, the emopamil binding protein (EBP), which is a vertebrate sterol isomerase and integral membrane protein of the ER involved in the biosynthesis of cholesterol [19–21]. Low densities of this non-receptor protein have been found in brain tissue . Selective ligands towards this target have been developed and suggested as a new starting point for developing new anti-cancer drugs .
In humans, several brain-related diseases are characterized by a decrease in the density of σ1 receptors as has been demonstrated in Parkinson’s and Alzheimer’s disease by postmortem [3H]DTG binding  and in vivo [11C]SA4503 imaging studies [23, 24]. σ1 receptors might be also implicated in the pathophysiology of psychiatric diseases, as well as in the mechanisms of action of some selective serotonin reuptake inhibitors (SSRIs). Using [11C]SA4503 it has been demonstrated that the SSRI fluvoxamine bound to σ1 receptors in living human brain at therapeutic doses . Thus, the concept of in vivo targeting of cerebral σ1 receptor expression by imaging approaches such as positron emission tomography (PET) might be of potential in both (1) drug development by target validation via correlation of σ1 receptor occupancy and efficacy of potential therapeutics  and (2) pathogenesis research and clinical diagnosis by quantifying alterations in the expression of σ1 receptors in several neuropathological disorders.
The most widely investigated σ1 receptor ligands for PET imaging belong to the piperidine structure such as [18F]FPS , [18F]FBFPA , [18F]FM-SA4503  and [11C]SA4503  as well as [18F]-labelled benzamides . However, while experimental data supported the suitability of e.g. [18F]FPS  or [18F]FM-SA4503 , evaluation of central σ1 receptors is hindered by the failure of [18F]FPS to reach transient equilibrium and brain peak uptake within 3 h post-injection (p.i.) in humans  and by the uptake of peripheral [18F]FM-SA4503 radiometabolites to the brain . Thus, at present the feasibility of clinical studies on σ1 receptor imaging in neurology is limited to the 11C-labelled PET radiotracer [11C]SA4503 demanding on-site availability of a cyclotron. Therefore, the successful application of transferable 18F-labelled radiotracers requires further development.
Moreover, [11C]SA4503 has high affinity not only to σ1 receptors (Ki between 4 and 14 nM; [34–36]) but also to the above-mentioned EBP with Ki of 1.7 nM  and to the vesicular acetylcholine transporter (VAChT, Ki = 50 nM; ), which has not been mentioned in any related PET publication so far but may be of high impact for the interpretation of neuroimaging findings.
In the present paper we describe the radiosynthesis and the biological evaluation of the new spirocyclic 3-(2-fluoroethyl)-2-benzofuran [18F]fluspidine. Its superiority over the recently described spirocyclic 3-(3-fluoropropyl)-2-benzofuran  and the aspect of potential binding to EBP are discussed.
Materials and methods
All reagents and solvents were obtained commercially and used without further purification. In vivo studies were carried out in 3-month-old female CD-1 mice (20–25 g). Animals were obtained from the Medizinisch-Experimentelles Zentrum, Universität Leipzig, maintained on a 12-h light-dark cycle and habituated for at least 2 days before experiments. The animals were deprived of food for 24 h prior to experiments with free access to tap water. All procedures involving animals were approved by the respective State Animal Care and Use Committee and conducted in accordance with the German Law for the Protection of Animals.
Synthesis of reference compound
Radiochemical and chemical purity and specific activity of [18F]fluspidine were analysed by analytical radio-HPLC (Multosorb RP18 AQ, 250 × 5 mm, 5 μm i.d. + Multosorb 120 RP18 AQ 40 × 5 mm as pre-column; eluent: 42.5% MeCN with 20 mM NH4OAc; flow rate: 1.0 ml min−1, tR [18F]fluspidine: 27 min) and radio-TLC [SIL G/UV254 precoated plates; ethyl acetate:petroleum ether: NH3 = 7:3:0.1 (v/v/v) and ethyl acetate:MeOH = 95:5 (v/v); spots visualized under UV light 254 nm].
In vitro stability and distribution coefficients of [18F]fluspidine
[18F]Fluspidine was incubated in 0.9% NaCl and several buffer systems (0.01 M phosphate buffer, Dulbecco’s buffered saline, 0.01 M TRIS-HCl; pH 7.2 each) at 40°C for 2 h. Moreover, it was stirred in MeCN (reaction medium) at 60°C for 2 h. At various times, aliquots were analysed by radio-HPLC and radio-TLC for determination of [18F]fluspidine and radioactive degradation products.
Distribution coefficients (log D, pH 7.2) of [18F]fluspidine were determined by shake-flask methods (multiple distribution in duplicates) in n-octanol and Sørensen’s citrate/phosphate buffer (50 mM, pH 7.2), and n-octanol and Dulbecco’s phosphate-buffered saline (pH 7.2). The determined log D values from RP-HPLC retention on three selected columns (Multosorb 100 RP18 AQ, 250 × 5 mm, particle size 5 μm; Kromasil 100 RP18, 250 × 5 mm, particle size 5 μm; Multosorb RP18-7 μm, 250 × 5 mm; 42.5% MeCN + 20 mM NH4OAc) were compared with log D values, calculated according to ChemDraw (Crippen’s and Viswanadhan’s fragmentation).
Receptor affinity and selectivity of fluspidine
The σ receptor affinities of the spirocyclic compounds were determined in competition experiments with radioligands. Homogenates of guinea pig brains (Harlan-Winkelmann, Borchen, Germany) were used as receptor material in the σ1 assay, and the σ1 selective ligand [3H]-(+)-pentazocine (42.5 Ci/mmol; PerkinElmer) was employed as radioligand. The non-specific binding was determined in the presence of a large excess of non-tritiated (+)-pentazocine. In the σ2 assay homogenates of rat liver (Harlan-Winkelmann, Borchen, Germany) served as source for σ2 receptors. The non-selective radioligand [3H]-1,3-di(o-tolyl)guanidine (ARC; specific activity 50 Ci/mmol) was employed in the presence of an excess of non-tritiated (+)-pentazocine for selective occupation of σ1 receptors. The non-specific binding of the radioligand was determined by performing the σ2 assay in the presence of an excess of non-tritiated 1,3-di(o-tolyl)guanidine. The affinity of fluspidine towards the phencyclidine binding site of the N-methyl-D-aspartate (NMDA) receptor and towards μ, κ and δ opioid receptors has also been investigated using protocols which have previously been published [40–42].
Affinity to the VAChT was determined by radioligand displacement studies on homogenates of PC12 cells stably transfected with rVAChT (Ali Roghani, Texas Tech University, Lubbock, TX, USA) by using (-)-[3H]vesamicol (PerkinElmer; specific activity 1,296 GBq/mmol). Cell homogenates were incubated in 50 mM TRIS-HCl, pH 7.4, at room temperature. Incubation was terminated after 60 min by filtration [GF-B filter, pre-incubated in 0.3% polyethyleneimine (PEI) at room temperature for 90 min; Brandel Cell Harvester]. Non-specific binding was determined in the presence of 10 μM (±)-vesamicol. We had previously determined the KD of (-)-[3H]vesamicol (2.3 nM) and used this value for calculation of Ki values according to the Cheng-Prusoff equation.
Affinity to the EBP was determined by radioligand displacement studies on microsomal preparations of Y00788 yeast transfected with human EBP plasmid (Hartmut Glossmann, Institute of Biochemical Pharmacology, Innsbruck, Austria; Swetlana König, Saxon Institute for Applied Biotechnology, Leipzig, Germany). Microsomal preparations were incubated in 50 mM TRIS-HCl, pH 8.3, at room temperature. Incubation was terminated after 60 min by filtration (GF-B filter, pre-incubated in 0.3% PEI at room temperature for 90 min; Brandel Cell Harvester). Non-specific binding was determined in the presence of 10 μM tamoxifen. The KD of (-)-[3H]emopamil has been determined by homologous competition experiments with a value of 218 nM used for calculation of Ki values according to the Cheng-Prusoff equation.
In vitro metabolism of fluspidine
Frozen rat livers from male Wistar rats were thawed in phosphate buffer pH 7.4 with 0.25 M sucrose and 5 mM ethylenediaminetetraacetate (EDTA), cut into small pieces and homogenized with a Potter (Elvehjem Potter, B. Braun Biotech International). The homogenization was carried out at 4°C. The resulting suspension was centrifuged at 10,000 g for 15 min at 4°C. Supernatant fat floating at the surface was removed by absorption with conventional pulp paper (cellulose material). The pellet was resuspended in buffer, the mixture was centrifuged again and afterwards both supernatants were combined. This suspension was transferred into an ultracentrifuge for 60 min at 100,000 g and 4°C. The resulting supernatant (cytosol) was discarded, the pellet washed carefully with buffer, resuspended and the centrifugation repeated. Finally the supernatant was removed, the pellet resuspended in a small amount of phosphate buffer pH 7.4 and the microsome suspension was stored at −80°C.
Six incubations of fluspidine with microsomes were carried out as duplicates in phosphate buffer pH 7.4 at room temperature in a circular shaker containing rat liver microsomes (1.5 mg/ml protein), 0.86 mM MgCl2 and 2.6 mM reduced nicotinamide adenine dinucleotide phosphate (NADPH)/H+. The concentrations of fluspidine were 320 μM in a total volume of 0.9 ml. After an incubation period of 5, 10, 30, 40, 60 and 90 min, two incubations were stopped by addition of cold acetonitrile (−20°C), respectively; 200 μl of a praziquantel solution (0.6 mg/ml) was added as internal standard resulting in a final concentration of 220 μM in a total volume of 1.1 ml. The samples were stored for 30 min at −20°C to complete protein precipitation. After thawing, the samples were centrifuged (10,000 g). The supernatant was decanted, filtered with a 0.45-μm (pore size) syringe filter made from regenerated cellulose and finally analysed. The calibration was carried out with the same matrix except NADPH/H+. All calibration standards were treated in the same way (90 min on the shaker, protein precipitation with acetonitrile, centrifugation etc.).
All samples were subjected to an HPLC system from Merck Hitachi consisting of UV detector: L-7400; autosampler: L-7200; pump: L-7150; interface: D-7000. A 50-μl volume of the prepared incubation solution was injected onto a LiChrospher® RP-Select B 5-μm (250 x 4 mm) column (Merck, Darmstadt, Germany) at a flow rate of 1.0 ml/min. The mobile phase was composed of (A) 15% acetonitrile in water and (B) 60% acetonitrile in water; 0.05% trifluoroacetic acid was added to both components. The following gradient was applied (A%): 0 min: 100%, 20 min: 0%, 23 min: 0%, 24 min: 100%, 30 min: 100%.
Structure elucidation of in vitro metabolites
Structure elucidation of metabolites was achieved using an LC-MSn system consisting of an LTQ Orbitrap® XL with Accela® HPLC pump and autosampler (Thermo Fisher), and 20 μl of the prepared incubation solutions was injected onto a LiChrospher® RP-Select B column (5 μm, 250 × 4.0 mm) with guard column LiChroCART® RP-Select B (4 × 4.0 mm) at a flow rate of 1.0 ml/min. The mobile phase was composed of (A) 15% acetonitrile in water and (B) 60% acetonitrile in water. Formic acid (0.1%) was added to both components (gradient see above).
The flow rate that was applied to the MS was reduced from 1 ml/min to 250 μl/min by a post-column splitter (Acurate, LC Packings, Dionex). The MS parameters were as follows: ion spray voltage: 4 kV in positive mode, sheath gas flow: 40 arbitrary units, auxiliary gas flow: 15 arbitrary units, sweep gas flow: 10 arbitrary units, capillary temperature: 275°C, capillary: voltage 40.5 V.
First a total ion current (TIC) spectrum was recorded. In order to elucidate the structures of the metabolites collision-induced dissociation (CID) fragmentation experiments were carried out. Therefore in the Orbitrap MS system fragmentations up to three generations (MS3) were performed. The fragmentation patterns of the metabolites are available as Electronic supplementary material.
In vivo metabolism of [18F]fluspidine
The in vivo metabolism of [18F]fluspidine was investigated using brain and liver samples homogenized in ice-cold 50 mM TRIS-HCl, pH 7.4. Plasma samples were obtained by centrifugation of heparinized blood. The sample preparation was performed using twofold extraction by protein precipitation with ice-cold MeCN (1:5), centrifugation of precipitates and concentration of the organic extracts (∼ 60°C, argon flow). The percentages of the parent radiotracer and radiometabolites were analysed by radio-HPLC (Multosorb RP18 AQ column, see above; low MeCN gradient), radio-TLC (SIL G/UV254 precoated plates; ethyl acetate:petroleum ether:NH3 = 7:3:0.1) and by measuring the radioactivity of the respective HPLC fractions using a calibrated gamma counter (Wallac WIZARD, PerkinElmer, Waltham, MA, USA).
Spots on radio-TLC plates were visualized under UV light and by radioluminescence recording (BAS-1800 II Bioimaging Analyzer, Fuji Film, Tokyo, Japan) followed by evaluation with AIDA 2.31 software (raytest, Straubenhardt, Germany).
Organ distribution and ex vivo assessment of brain binding site occupancy of [18F]fluspidine
Awake female CD-1 mice (10–12 weeks old, 20–25 g) received tail vein injection of ∼ 200 kBq [18F]fluspidine (specific activity >150 GBq/μmol at the end of the synthesis) dissolved in 200 μl 0.9% saline. At 5, 20 and 60 min p.i. of the radiotracer, the animals (n = 4 per time) were anaesthetized, blood samples were taken, the animals were euthanized by luxation of the cervical spine, organs of interest were removed and weighed, and radioactivity was measured in a calibrated gamma counter. Percent injected dose per gram of wet tissue (%ID/g) was calculated by comparison of samples to standard dilutions of the initial dose.
Specificity of brain uptake of [18F]fluspidine was investigated in a blocking experiment by injecting i.p. 1 mg/kg haloperidol (n = 5), a σ receptor antagonist or tamoxifen (n = 4), a ligand with high affinity to the EBP, dissolved in saline at 10 min before the radiotracer. Ex vivo autoradiography was performed on frozen sagittal brain sections at 60 min p.i. of 250 MBq [18F]fluspidine. The brain was removed, dissected hemispheres were frozen in isopentane (−35°C) and 12-μm sagittal sections were cut on a cryostat microtome (Microm, Walldorf, Germany), mounted onto microscope slides, dried in a stream of cold air for 10 min and exposed to 18F-sensitive storage phosphor screens (Fuji Film, Tokyo, Japan) for 60 min. The image plates were processed as described above. The brain structures were confirmed by Nissl staining of the exposed sections.
Data given in the figures and the table are mean values ± standard deviation (SD). Comparisons between groups were performed using the two-tailed t test. Differences were considered significant when p < 0.05.
Radiosynthesis and radiotracer characterization
Dried K[18F]F-K2.2.2 complex, prepared from a solution of aqueous K2CO3 (3.6 mg, 150 μl) and Kryptofix 2.2.2 (11 mg, 200 μl MeCN) was added to 2.5–3 mg precursor (Fig. 2) dissolved in 1.0 ml anhydrous MeCN and reacted for 25 min at 85°C. Under these optimized reaction conditions reproducible labelling yields of 60–70% were achieved (n = 10).
After purification via semi-preparative RP-HPLC and SPE using a Sep-Pak C18 cartridge, [18F]fluspidine was obtained with a radiochemical yield of 35–45%, a radiochemical purity of ≥ 99.6% and a specific activity of 150–350 GBq/μmol (n = 6) within a total synthesis time of 90–120 min. The radiosynthetic procedure is transferable to a commercially available automated synthesis module with minor modification (TraceLab FxN® synthesis module; GE Healthcare-Nuclear Interface, Münster, Germany).
[18F]Fluspidine is chemically stable for at least 2 h in 0.9% NaCl solution at 40°C and in MeCN at 60°C. However, in Sørensen’s buffer and phosphate buffer solutions the radiotracer is apparently less stable according to radio-TLC and analytical radio-HPLC. The amount of residual radiotracer [18F]fluspidine was 95, 92 and 89–90% after 30-, 60- and 120-min incubation periods, respectively.
The distribution coefficients calculated by ChemDraw provided consistent clog D values of 3.36 (according to Crippen) and 3.29 (according to Viswanadhan). These values agree well with the experimentally determined log D values obtained by HPLC: Multosorb 100 RP18: 3.61 ± 0.09 (MeCN gradient; n = 3), Kromasil 100 RP18: 3.64 ± 0.03 (MeCN gradient; n = 5) and Multosorb RP18-7: 3.49 (isocratic mode; 42.5% MeCN + 20 mM NH4OAc, n = 2). The data suggest that [18F]fluspidine has sufficient lipophilic properties to pass the blood-brain barrier. Notably, the log D values determined by the shake-flask method resulted in both significantly lower and more scattered values: log Dn-octanol/Sørensen’s citrate/phosphate buffer = 1.9 ± 0.71 and log Dn-octanol/Dulbecco’s phosphate buffer = 1.65 ± 0.19. This is attributed to solvent association effects rather than to decomposition.
Receptor affinity and selectivity
In the receptor binding studies with the radioligand [3H]-(+)-pentazocine it was clearly shown that the fluoroethyl derivative fluspidine is a very potent σ1 receptor ligand with a Ki value of 0.59 ± 0.20 nM (n = 3).
In addition to the high σ1 affinity, a high selectivity of fluspidine against the σ2 subtype (Ki = 785 nM) and the VAChT (Ki = 1.4 ± 0.2 μM, n = 3) has been observed. The affinity of fluspidine towards the phencyclidine binding site of the NMDA receptor and towards μ, κ and δ opioid receptors has also been investigated. At a concentration of 1 μM fluspidine did not inhibit radioligand binding at the NMDA and the δ opioid receptor. At κ and μ opioid receptors Ki values of 372 nM and 456 nM were found, respectively. Altogether, the potent σ1 receptor ligand fluspidine represents a very selective ligand with at least 600-fold selectivity against the investigated receptor systems. Also the affinity towards the human EBP (tamoxifen: Ki = 56 ± 33 nM, n = 3) of fluspidine was low (Ki = 211 ± 181 nM, n = 3).
In vitro metabolism of fluspidine
The rate of the metabolic degradation of fluspidine was investigated using rat liver microsomes and quantification of the remaining amount of fluspidine by HPLC-UV. The structure of the formed metabolites was elucidated with LC-MS using the LTQ Orbitrap XL® system, which allows diverse fragmentation experiments (MSn).
In Fig. 3 the degradation of fluspidine by rat liver microsomes is shown. After an incubation period of 30 min only 13% of the parent compound remained unchanged.
Further oxidation of the hydroxybenzyl metabolite 1b generated the dihydroxylated metabolite 1d. The secondary amine 1e with an hydroxy moiety in the side chain was formed either by oxidation of the secondary amine 1c or by N-debenzylation of 1a. Additionally the formation of the N-oxide 1f was observed. The structure of the metabolite 1g with two additional O atoms and without fluorine was not exactly determined. However, the molecular formula C2H23NO3+H+ was unequivocally proved by determination of the exact mass.
In vivo metabolism of [18F]fluspidine
In the brain, ∼ 98% of total radioactivity was represented by [18F]fluspidine both at 30 and 60 min p.i. According to HPLC and TLC, between 65 and 50% of non-metabolized radiotracer along with one major radiometabolite (tR ∼ 26 min) were found in plasma samples.
In liver homogenates, 72 and 57% of [18F]fluspidine were detected after 30 and 60 min, respectively, as well as two hydrophilic main metabolites (tR ∼ 31 and 37 min) (Fig. 6), which are expected to be identical with metabolites identified after incubation in rat liver microsomes. Interestingly, the two radiometabolites were not detected in plasma samples.
Evidence of strong metabolic degradation was observed in urine samples. Already after 30 min only ∼ 1% of the non-metabolized radiotracer was recorded. HPLC analyses resulted in several hydrophilic radiometabolites with highly variable percentages at both times. Three major peaks were detected at tR ∼ 4 min (double peak: radiofluoride and hydrophilic radiometabolite), tR ∼ 23 and tR ∼ 26 min.
Ex vivo evaluation of [18F]fluspidine in mice
Radioactivity uptake (%ID/g wet weight) in major organs of female CD-1 mice at 5, 30, 60 and 120 min after intravenous administration of 200–300 kBq [18F]fluspidine. Values are means ± SD
5 min (n = 5)
30 min (n = 5)
60 min (n = 5)
120 min (n = 3)
0.73 ± 0.08
0.51 ± 0.12
0.27 ± 0.06
0.15 ± 0.01
3.88 ± 0.92
4.71 ± 1.39
3.83 ± 0.74
2.82 ± 0.19
5.81 ± 1.07
5.50 ± 1.50
4.56 ± 1.18
4.44 ± 0.72
22.21 ± 3.06
15.60 ± 3.68
10.30 ± 3.89
11.32 ± 3.23
2.86 ± 1.22
3.83 ± 1.97
2.38 ± 1.25
2.36 ± 0.89
9.80 ± 5.18
18.96 ± 3.30
14.59 ± 5.67
9.10 ± 1.78
0.73 ± 0.26
1.22 ± 0.21
1.56 ± 0.77
1.89 ± 0.36
2.94 ± 0.53
3.19 ± 0.36
2.41 ± 0.69
1.85 ± 0.26
10.41 ± 3.22
9.50 ± 2.29
6.88 ± 0.95
6.12 ± 1.16
1.61 ± 0.75
2.40 ± 0.71
1.80 ± 0.66
1.85 ± 0.29
3.31 ± 0.29
6.37 ± 1.65
5.27 ± 0.54
4.35 ± 0.99
2.58 ± 0.53
3.06 ± 0.48
2.94 ± 0.61
2.41 ± 0.53
6.67 ± 2.54
11.34 ± 2.24
9.18 ± 0.63
8.78 ± 0.42
8.65 ± 3.25
7.08 ± 0.98
6.35 ± 0.58
4.40 ± 0.30
2.41 ± 0.48
3.04 ± 0.33
3.68 ± 1.19
2.86 ± 1.32
0.94 ± 0.25
1.75 ± 0.64
1.32 ± 0.19
2.02 ± 0.56
1.25 ± 0.42
1.66 ± 0.22
1.47 ± 0.47
1.18 ± 0.12
0.97 ± 0.30
1.91 ± 0.22
1.81 ± 0.81
1.79 ± 0.66
0.85 ± 0.51
1.08 ± 1.03
2.21 ± 1.18
2.56 ± 1.20
The time-activity data of the combined bladder and urine samples (∼ 5 and ∼ 90%ID/g at 5 and 120 min p.i., respectively) point to renal elimination as the major excretory pathway of the radiotracer and radiometabolites. The gastrointestinal accumulation of radioactivity peaked at 30 min p.i. and decreased thereafter. The high initial uptake in the lung was followed by a rapid washout (22.2 vs 11.3%ID/g at 5 and 120 min p.i., respectively). The amount of radioactivity detected in the femur remained almost constant, indicating that defluorination of the radiotracer did not occur.
Neuroimaging of σ1 receptors by PET in humans is currently being performed by the application of [11C]SA4503 which has been used in patients with Alzheimer’s and Parkinson’s disease [23, 24, 45]. This radioligand has major drawbacks because the clinical application is restricted to the on-site availability of a cyclotron, it has rather low selectivity towards the VAChT  and it displays affinity to the EBP . Therefore, the aims of our study were to design and biologically evaluate a highly selective 18F-labelled σ1 receptor ligand and to address the question of potential EBP binding.
We have successfully labelled the new σ1 receptor ligand fluspidine with 18F and investigated affinity and selectivity in vitro as well as target specificity in vivo. The ligand belongs to our recently described class of spirocyclic piperidines, which combine high σ1 receptor affinity with high selectivity [38, 46]. In fact, the affinity of fluspidine (Ki = 0.59 nM) and the selectivity (Kiσ2/Kiσ1 = 1,200) to σ1 receptors is about twofold higher than that of our previously described PET radioligand, the corresponding fluoropropyl derivative , and 7- to 20-fold higher than that of [11C]SA4503 [34–36].
Also, only moderate selectivity of [11C]SA4503 versus the VAChT (factor 11) has recently been reported , and fluspidine provides a much better selectivity (factor 2,370). Interestingly, haloperidol has an even higher affinity to VAChT than [11C]SA4503 . This fact needs to be considered for interpretation of previous preclinical and human studies with [11C]SA4503 and using haloperidol for the investigation of radioligand selectivity [30, 47, 48]. The low VAChT affinity of fluspidine as shown in our study in vitro excludes the possibility of significant in vivo VAChT binding of our radioligand. Therefore, in our case, the inhibition of brain uptake observed after pre-injection of haloperidol is expected to be caused by σ receptor blockade.
It is known that [11C]SA4503 displays an even higher affinity to EBP (1.7 nM; ) than to σ1 receptors. Although the expression and distribution of EBP in brain is unknown so far, there is strong evidence of colocalization of EBP and σ1 receptors at the ER and nuclear envelope level . This prompted us to assess if [18F]fluspidine binds to EBP in vitro and in vivo. We have established an in vitro binding assay using a microsome fraction of yeast cells expressing EBP and determined an affinity of fluspidine to EBP of about 100 nM, which is lower than that of (-)-emopamil or tamoxifen. In biodistribution studies in mice we have used tamoxifen, a ligand with high affinity (Kd = 3 nM) to EBP and lower affinity (Ki = 34 nM) to σ1 receptors , to block potential EBP binding sites. No significant effect of tamoxifen on biodistribution of [18F]fluspidine was observed (Fig. 7), indicating that fluspidine has low if any affinity to EBP in vivo.
The initial brain uptake at 5 min p.i. of [18F]fluspidine is somewhat higher than that of a previously reported fluoropropyl derivative . In contrast to this ligand and to [11C]SA4503 , the brain uptake of [18F]fluspidine increased further. At 30 min p.i. it is 40–50% higher than that of the fluoropropyl derivative or [11C]SA4503 which is probably related to the higher receptor affinity. This conclusion is supported by ex vivo autoradiography of mice brain at 45 min after radiotracer injection. The facial nucleus to olfactory bulb ratio is significantly higher for [18F]fluspidine (4.69 ± 0.7) than for the fluoropropyl derivative (1.97 ± 0.99) at 45 min after injection of comparable tracer amounts. The respective pattern of distribution of [18F]fluspidine in mouse brain resembles the pattern of σ1 receptor expression detected by immunolabelling and receptor autoradiography in mouse , rat [51, 52], guinea pig  and primate brain : the reticular nucleus showed the highest density of binding sites, moderate levels were detected in cerebellar as well as cortical structures, and subcortical regions such as the striatum and thalamus were associated with low concentrations of radioactivity.
The radiotracer uptake in spleen and pancreas nearly doubled between 5 and 30 min p.i., and a high radiotracer retention was also observed in heart and lung, which is consistent with the high expression of σ1 receptors in these organs [43, 55–57] and previous studies on biodistribution of σ1 receptor radiotracers [30, 58, 59]. In consistence, besides brain, the highest blockade by haloperidol was observed in heart, lung and spleen (Fig. 7).
As it has been previously reported for the fluoropropyl derivative, an increasing amount of radioactivity was determined in the femur (data not shown), suggesting potential defluorination of [18F]fluspidine. However, in the current study radioactivity in bone and bone marrow has been measured separately. The increased radioactivity uptake refers only to bone marrow (Table 1), probably caused by a specific albeit slow receptor binding in this tissue as already assumed for the fluoropropyl derivative . Thus, no evidence of major defluorination of [18F]fluspidine was obtained. This is supported by the results of [18F]fluspidine metabolism studies where only a small amount of a metabolite (< 4%) was detectable in plasma samples at tR ∼ 3.7 min, which could be attributed to radiofluoride and/or another compound. In the 60-min sample this peak was absent or could not be detected. Furthermore, we have shown that the percentage of non-metabolized [18F]fluspidine accounted for >98% and ∼ 55% of total radioactivity in brain and plasma at 60 min p.i. Hence, radiometabolites of [18F]fluspidine formed in the periphery do not cross the blood-brain barrier. The metabolism of [18F]fluspidine is somewhat faster than that of the fluoropropyl derivative where non-metabolized radiotracer accounted for >85% of total plasma radioactivity at 60 min p.i. . This in vivo result observed in plasma corresponds to the in vitro metabolism studies with fluspidine being faster degraded by rat liver microsomes than the fluoropropyl derivative. Although those microsome studies provide early hints (before radiolabelling) about the potential in vivo stability of drugs, they do not allow conclusions about radiotracer metabolism as a whole. Liver microsomes mainly reflect the metabolic activity of cytochrome P450 enzymes and not of other drug-metabolizing enzymes . Furthermore, the particular regulation of P450 enzymes differs between rats and humans and may account for species differences in drug metabolism [61, 62].
Altogether we have shown that the spirocyclic benzofuran-based radioligand [18F]fluspidine binds with high affinity and specificity to σ1 receptors, shows a fast and sufficiently high brain uptake and accumulates in a target-specific manner in brain and various organs in mice. It has the potential for neuroimaging and quantitation of σ1 receptors in vivo.
This work was supported by a grant of the Deutsche Forschungsgemeinschaft, which is gratefully acknowledged. The authors thank Tina Ludwig for her excellent technical assistance, the staff of the cyclotron facility at the department of Nuclear Medicine, University of Leipzig, for [18F]F− production and the Saxon Institute for Applied Biotechnology for supply of the EBP.
Conflicts of interest