Cannabinoid receptors 1 (CB1) and 2 (CB2) are the best characterized subtypes among the cannabinoid receptors (Console-Bram et al. 2012). After their discovery in mammalian tissues, they were cloned in the 1990s (Matsuda et al. 1990; Munro et al. 1993). CB1 and CB2 belong to the large family of rhodopsin-like G-protein coupled receptors (GPCRs), which negatively regulate adenylate cyclase. CB1 and CB2 are encoded by the genes CNR1 and CNR2, respectively, and share 44% overall homology at the protein level, 68% in transmembrane regions (Hohmann and Herkenham 1999; Price et al. 2003). They differ in tissue localization, and, to some extent, signal transduction mechanisms. Cannabinoid receptors are involved in a broad range of processes including appetite, anxiety, memory, cognition, immune regulation, and inflammation (Pertwee 1997). They are considered as attractive targets for the development of therapeutic agents (Rom and Persidsky 2013) and many CB1 and CB2 agonists/antagonists have been developed and described in the literature (Muccioli and Lambert 2006; Pertwee 2009; Evens and Bormans 2010). CB1 receptors are mainly found in the central nervous system (CNS), whereas CB2 receptors are abundantly expressed in immune cells and tissues, for example in spleen, thymus and tonsils but are negligible in healthy brain tissue (Galiegue et al. 1995; Schatz et al. 1997). However, CB2 increase, linked to neuroinflammation, is observed in the brain of Alzheimer patients where activated microglia cluster at β-amyloid plaques (Ashton and Glass 2007; Benito et al. 2007a,b; Ramirez et al. 2012). CB2 was also increased in the striatum of a rat chronic lesion model of Huntington's disease (HD) and both hypoxia-ischemia and middle cerebral artery occlusion induced the expression of CB2 in proliferating microglia in rat brain (Ashton and Glass 2007; Fernandez-Ruiz et al. 2007). Moreover, the spinal cord of amyotrophic lateral sclerosis (ALS) patients as well as a mouse model of ALS and plaques of demyelination in multiple sclerosis patients showed an increase of CB2 (Benito et al. 2007b; Shoemaker et al. 2007). Therefore, imaging by positron emission tomography (PET) of CB2 may provide a valuable research tool to explore the role and importance of CB2 receptor expression in CNS disorders and to evaluate the therapeutic value of new CB2-related drugs (Kiferle et al. 2011). The high level of CB2 expression in immune cells, CNS disorders and much lower expression in other cell types, particularly in the CNS, makes this receptor a promising target for imaging and monitoring of therapy for neurological diseases. A large number of structurally distinct CB2 agonists and inverse agonists have been synthesized and characterized in the past decade. They include tricyclic pyrazoles, sulfamoyl benzamides, triarylbis-sulfones, arylsulfonamide, indole, and quinolone derivatives (Evens and Bormans 2010). Among all these classes of CB2 ligands, 2-oxoquinoline, and 4-quinolone-3-carboxamide derivatives appear to be the most efficient inverse agonists, with high binding affinity and selectivity for CB2. While several radioligands have been developed for imaging CB1 in the brain, culminating in the successful application of [18F]MK-9470 in clinical PET (Van Laere 2007; Horti and Van Laere 2008; Van Laere et al. 2008; Terry et al. 2010), CB2 receptor imaging in neurodegenerative disease is much less explored (Fujinaga et al. 2010; Horti et al. 2010; Evens et al. 2011; Turkman et al. 2011, 2012; Ruhl et al. 2012). Herein, we report the radiolabeling of one of the most promising 4-quinolone-3-carboxamide derivatives, designated [11C]KD2, and its in vitro/in vivo evaluation as a potential radiotracer for imaging CB2 receptor with PET.
The cannabinoid type 2 (CB2) receptor plays an important role in neuroinflammatory and neurodegenerative diseases such as multiple sclerosis, amyotrophic lateral sclerosis, and Alzheimer's disease and is therefore a very promising target for therapeutic approaches as well as for imaging. Based on the literature, we identified one 4-oxoquinoline derivative (designated KD2) as the lead structure. It was synthesized, radiolabeled and evaluated as a potential imaging tracer for CB2. [11C]KD2 was obtained in 99% radiochemical purity. Moderate blood–brain barrier (BBB) passage was predicted for KD2 from an in vitro transport assay with P-glycoprotein-transfected Madin Darby canine kidney cells. No efflux of KD2 by P-glycoprotein was detected. In vitro autoradiography of rat and mouse spleen slices demonstrated that [11C]KD2 exhibits high specific binding towards CB2. High spleen uptake of [11C]KD2 was observed in dynamic positron emission tomography (PET) studies with Wistar rats and its specificity was confirmed by displacement study with a selective CB2 agonist, GW405833. A pilot autoradiography study with post-mortem spinal cord slices from amyotrophic lateral sclerosis (ALS) patients with [11C]KD2 suggested the presence of CB2 receptors under disease conditions. Specificity of [11C]KD2 binding could also be demonstrated on these human tissues. In conclusion, [11C]KD2 shows good in vitro and in vivo properties as a potential PET tracer for CB2.
The cannabinoid type 2 receptor (CB2) plays an important role in neuroinflammatory and neuro–degenerative diseases. [11C]KD2, a new CB2 radioligand, exhibits selectivity towards CB2 receptor in vitro and in vivo in rat. A pilot autoradiography study with spinal cord slices from amyotrophic lateral sclerosis patients with [11C]KD2 suggested the presence of CB2 under disease conditions (Figure). Selective CB2 agonist, GW405833.
amyotrophic lateral sclerosis
bovine serum albumin
cannabinoid receptors 1
cannabinoid receptors 2
positron emission tomography
region of interests
standardized uptake values
Materials and methods
Animal experiments were in accordance with the Swiss Animal Protection Law and the ARRIVE guidelines and were approved by the Veterinary Office of the Canton Zurich. Female NMRI mice and male Wistar rats (350–520 g) were purchased from Charles River (Sulzfeld, Germany) and kept under standard conditions. Frozen post-mortem spinal cord tissues from ALS patients were provided by courtesy of the Cantonal Hospital St. Gallen, Switzerland, with written consent of the donors.
All chemicals, unless otherwise stated, were purchased from Sigma-Aldrich (Zug, Switzerland) or Merck (Buchs, Switzerland) and used without further purification. Solvents for extractions, column chromatography, and thin layer chromatography (TLC) were purchased as commercial grade. Nuclear magnetic resonance (NMR) spectra (1H and 13C NMR) were recorded in Fourier transform mode at the field strength specified on Bruker Avance FT-NMR spectrometers. The measured chemical shifts are reported in δ (ppm) and the residual signal of the solvent was used as the internal standard. Multiplicities in the 1H NMR spectra are described as: s = singlet, d = doublet, t = triplet, m = multiplet, b = broad; coupling constants are reported in Hz. High resolution mass spectrometry (HRMS) was performed with a Bruker's maXis (ESI-Qq-TOF-MS) spectrometer (Bruker Daltonik GmbH, Bremen, Germany).
High-performance liquid chromatography (HPLC) analyses were performed using a reversed phase column (ACE column, C18, 3 μm) with an isocratic solvent system 75% MeCN in water, flow rate: 1 mL/min. Analytical radio-HPLC was performed on a Merck-Hitachi L-2130 system equipped with a L-2450 diode array detector and a radiodetector (Berthold Technologies GmbH & Co. KG, Bad Wildbad, Germany). Semi-preparative HPLC purifications were carried out using a reversed phase column (ACE column, Symmetry C8 5 μm; 7.8 × 50 mm) under the following conditions: 0.1% H3PO4 in H2O (solvent A), MeCN (solvent B); 0.0–1.0 min, 30% B; 1.1–12.0 min, 30–90% B; 12.1–20 min, 90% B; 20.1–40 min, 30% B; flow rate: 4 mL/min. A Merck-Hitachi L2130 system equipped with a radiation detector VRM 202 (Veenstra Instrument, Joure, the Netherlands) was used for semi-preparative HPLC. Specific activity was calculated by comparing ultraviolet peak intensity of final formulated products with calibration curves of corresponding non-radioactive standards of known concentrations.
Synthesis of N-(1-adamantyl)-8-hydroxy-4-oxo-1-phenyl-1,4-dihydroquinoline-3-carboxamide (precursor compound): boron tribromide (0.5 mL, 1.32 g, 5.3 mmol) was added to a solution of N-(1-adamantyl)-8-methoxy-4-oxo-1-phenyl-1,4-dihydroquinoline-3-carboxamide (420 mg, 1 mmol) in methylene chloride (5 mL) and the mixture was stirred at 23–25°C for 18 h. The reaction mixture was quenched by addition of ice water, extracted with ethyl acetate, the organic layer was washed three times with water, dried over MgSO4 and evaporated. The residual yellow foam was dissolved in chloroform and then purified by silica gel column chromatography using 20–40% ethyl acetate in hexane as eluent. The pure fractions were collected and evaporated to dryness, benzene (5 mL) was added to dissolve the residue, followed by the addition of cyclohexane (25 mL). The crystals were filtered and washed with cyclohexane and dried. Yield: 300 mg (74%) colorless solid. TLC: hexane: ethyl acetate (6:4), Rf = 0.4. 1H-NMR(CDCl3): 10.27 and 10.17 (2 s, 2 H, OH and NH), 8.58 (s, 1 H, N-CH=), 8.05 (br. d, J = 8 Hz, 1 H, ArH), 7.41 (br. d, J = 8 Hz, 1 H, ArH), 7.27 (br. t, J = 8 Hz, 1 H, ArH), 4 .58 (br. t, J = 7, 2 H, NCH2), 2.20 (br. s, 6 H, 3 adamantyl CH2), 2.12 (br. s. 3 H, 3 adamantyl CH), 1.85–1.66 (m, 8 H, incl. 3 adamantyl CH2 and 1 alkyl CH2), 1.24–1.10 (m, 4 H, 2 alkyl chain CH2), 0.75 (br. t, J = 6.6 Hz, alkyl chain CH3). 13C-NMR(CDCl3): 176.41, 164.57, 149.30, 148.16, 130.53, 129.15, 125.85, 117.85, 111.22, 59.71, 52.10, 41.82, 36.46, 31.49, 29.49, 28.46, 26.92, 22.22, 13.82. HRMS: calculated for [M + H]+ C25H33N2O3 409.2486; found 409.2481; Purity by HPLC: 98%.
Radiosynthesis of [11C]KD2
Carbon-11 was produced via the 14N(p,α)11C nuclear reaction at a Cyclone 18/9 cyclotron (18-MeV; IBA, Ottignies-Louvain-la-Neuve, Belgium) in the form of [11C]CO2. [11C]Methyl iodide ([11C]MeI) was generated in a two-step reaction sequence involving the catalytic reduction of [11C]CO2 to [11C]methane and subsequent gas phase iodination. [11C]KD2 was prepared by reaction of the phenolic precursor compound (0.5–1 mg) with [11C]MeI (ca. 40 -50 GBq) in dimethylformamide (DMF) solution in the presence of Cs2CO3 (5 mg) at 90°C for 3 min (Fig. 1). The crude product was diluted with water (1.5 mL) and injected onto a semi-preparative HPLC. The radiolabeled product was collected, diluted with water (10 mL), passed through a pre-conditioned C18 cartridge (Waters, pre-conditioned with 5 mL EtOH and 10 mL water), washed with water (5 mL), and eluted with EtOH (0.5 mL). After adding 9.5 mL of water, the 5% ethanol solution containing [11C]KD2 was then passed through a sterile filter (0.2 μm) and used for all in vitro/in vivo studies.
Competition binding assay
Frozen membrane preparations from CHO-K1 cells transfected with human CB1 (hCB1) and CB2 (hCB2), respectively (PerkinElmer, Waltham, MA, USA), were thawed on ice and diluted to a final protein concentration of 1 μg/mL in assay buffer (50 mM TRIS/HCl, 1 mM EDTA (Applichem, Darmstadt, Germany), 3 mM MgCl2, pH 7.4, containing 0.05% bovine serum albumin, BSA). Membrane dilutions (0.5 μg protein in 550 μL final volume) were incubated at 30°C with KD2 at concentrations between 10−5 and 10−11 M and 1.4 nM [3H]CP-55 940 (PerkinElmer) (Thomas et al. 1998), a hCB1 and hCB2 agonist with Ki values of 0.58 and 0.68 nM, respectively. Nonspecific binding of [3H]CP-55 940 was determined after addition of 5 μM hCB1/hCB2 agonist WIN-55212-2 (Ki 1.89 and 0.28 nM, respectively). All samples were prepared in triplicates. After 90 min, 3 mL ice cold assay buffer was added and samples were immediately filtered through Whatman GF/C filters (pre-soaked in 0.05% polyethylenimine) and washed twice with 3 mL ice cold assay buffer. Bound [3H]CP-55 940 was quantified in a Beckman LS 6500 Liquid Scintillation Counter (Beckman™, Brea, CA, USA) and IC50 values were determined by non-linear regression analysis (Milicevic Sephton et al. 2012). Ki values were determined with the equation from Cheng-Prusoff and are means from three independent experiments.
Determination of logD
The partition coefficient D was determined by the shake flask method as previously reported (Wilson et al. 2001; Baumann et al. 2010). In brief, 2 μL formulated solution of [11C]KD2 (ca 1 MBq, < 0.1 nmol) was added to 500 μL phosphate buffer (pH 7.4) and 500 μL 1-octanol. The sample was equilibrated for 15 min in an overhead shaker before the two phases were separated by centrifugation (3 min, 5000 g) and 50 μL aliquots each of both phases were analyzed in a gamma counter (Wizard; PerkinElmer). LogD is expressed as the logarithm of the ratio between the radioactivity concentrations (Bq/mL) of the octanol and the buffer phase.
Permeation across human P-glycoprotein transfected MDCK (hMDR1-MDCK) cells
Transport experiments hMDR1-MDCK cells (Bakos et al. 1998), kindly provided by the Netherlands Cancer Institute, were performed as described elsewhere (Wanger-Baumann et al. 2011). In brief, cells were cultured on polyethylene terephthalate culture inserts (Falcon #35-3090, (BD Science, Franklin Lakes, NJ, USA) 4.2 cm2, 0.4 μm pores) for 3 days. Formulated [11C]KD2 (400 kBq) was added to the apical or basal (donor) compartments of three inserts each. The plates were incubated at 37°C on a rocking plate at 25 g and 100 μL aliquots were withdrawn from the receiver compartments 15, 30, and 45 min and from both donor and receiver compartments 60 min after addition of the radioligand. The transport of the P-gp substrate [3H]vinblastine (Perkin Elmer, 429.2 GBq/mmol, 20 kBq per insert) was determined in parallel for comparison (Hammerle et al. 2000). Radioactivity of the samples was quantified with a gamma and beta counter, respectively, and apparent permeability coefficients (Permapp) were calculated as described previously (Wanger-Baumann et al. 2011).
Plasma protein binding
Human serum (400 μL) was incubated with 18 MBq (< 1 μM) [11C]KD2 for 10 min at 23–25°C and filtered with a Microcon® YM-3 centrifugal filter device with 3 kDa cut-off (Amicon, Billerica, MA, USA) to produce ca 40 μL filtrate. Aliquots of the filtrate and retentate were analyzed in a gamma-counter (Wizard; PerkinElmer). Filtrations were performed in triplicates. The unbound fraction (fu) was calculated as the ratio between the Bq/mL in the filtrate and the retentate.
In vitro autoradiography
In vitro autoradiography with [11C]KD2 was performed on 20 μm tissue slices (rat/mouse spleen or human spinal cord tissues) adsorbed to SuperFrost Plus slides (Thermo Scientific, Runcorn, UK). Slices were thawed (10 min) on ice before pre-incubation with incubation buffer (TRIS/HCl 50 mM, pH 7.4, containing 5% BSA) at 4°C for 10 min. Excess solution was carefully removed and slides were incubated with 0.6 or 0.2 nM [11C]KD2 alone or together with full unselective CB agonist (WIN 55212-2, 5 μM) or specific CB2 agonist (GW405833, 1 μM) in incubation buffer. After incubation for 15 min at 23–25°C in a humid chamber, slides were washed twice with ice cold washing buffer (TRIS/HCl 50 mM, pH 7.4, 1% BSA, 5% EtOH) for 2 min each and dipped twice in water. Dried slides were exposed to a phosphor imager plate for 30 min and the plate was scanned in a BAS5000 reader (Fujifilm, Dielsdorf, Switzerland).
In vivo characterization
PET scanning was performed with the GE VISTA eXplore PET/CT tomograph (Sedecal, Madrid, Spain). The scanner is characterized by high sensitivity (absolute central point source sensitivity of 4% for the 250–700 eV energy windows) and an axial field of view of 4.8 cm (Wang et al. 2006). Male Wistar rats were immobilized by isoflurane inhalation and the tracer was injected into a tail vein on the tomographic bed (Honer et al. 2004). Tracer accumulation in the brain and spleen were recorded by dynamic one-bed position PET scans over 60 min starting with the injection of 20–61 MBq (0.06–0.2 nmol) of [11C]KD2. In the displacement study, 1.5 mg/kg GW405833 was injected i.v. over 1 min, 15 min after scan start. PET data were reconstructed by 3-dimensional FORE/2-dimensional OSEM in user-defined time frames with a voxel size of 0.3875 × 0.3875 × 0.775 mm. Singlets and random corrections but no attenuation correction was applied. Image files were evaluated with the software PMOD v3.4 (PMOD Technologies Inc., Zurich, Switzerland). Time-activity curves (TACs) of brain regions were generated with the implemented rat brain region of interest (ROI) template and TACs of the abdominal region with the respective ROIs generated by the PMOD segmentation tool. Spleen, liver, and background TACs were confirmed by analysis of manually drawn regions of interest. Standardized uptake values (SUV) were calculated as tissue activities (Bq/cm3), normalized to the injected dose per body weight (Bq/g).
Permapp values of the in vitro transport assay were compared by Student's two-tailed homoscedastic t-test. Differences with p-values < 0.05 were considered significant.
Chemistry and radiochemistry
The non-radioactive standard reference compound was synthesized based on the procedure described by Pasquini (Pasquini et al. 2011). Briefly, diethyl 2-[(2-methoxy-phenylamino)-methylene]malonate was obtained by condensation of commercially available 2-methoxyaniline with diethyl ethoxymethylenemalonate according to the Gould–Jacobs reaction. The intermediate enaminoesters were directly cyclized in refluxing diphenyl ether to afford the quinolone esters. The ester was used for further N-alkylation to introduce pentane, followed by hydrolysing to their corresponding acids. Finally, coupling reaction of the acid with adamantylamine free base provided N-(1-adamantyl)-8-methoxy-4-oxo-1-phenyl-1,4-dihydroquinoline-3-carboxamide, the standard reference compound. Demethylation of the reference compound was performed with BBr3 to afford the phenolic precursor for carbon-11 labeling. The newly synthesized compounds were characterized by mass spectrometry and NMR, and their chemical purities were assessed by HPLC. KD2 was successfully labeled with carbon-11 in a one-step reaction by reacting the phenolic precursor with [11C]-methyl iodide (Fig. 1). [11C]KD2 (ca. 3–5 GBq) was obtained in 99% radiochemical purity after semi-HPLC purification. The total radiolabeling time was around 40 min after delivery [11C]CO2 from the cyclotron to the hot-cell. Specific activity was high and ranged from 80 to 320 GBq/μmoL at the end of synthesis. The radiochemical yield was 30–40% (decay corrected).
In vitro characterization
In vitro competitive binding assays were performed with membranes containing human CB1 and CB2, respectively, using [3H]CP-55940. The relative binding affinity of the non-radioactive KD2 obtained from three independent experiments was 1.7 ± 2.0 nM towards human CB2 and > 10 000 nM towards CB1, similar to the reported data (Pasquini et al. 2011). LogD of [11C]KD2 in octanol/buffer at pH 7.4 was 3.29 ± 0.04 (n = 6). Autoradiography with slices from rat and mouse spleen, an organ with high CB2 levels (Gong et al. 2006), demonstrated high binding which was blocked by excess WIN 55212-2 (full unselective CB agonist) and GW4058233 (CB2 specific agonist), respectively, indicating specific binding of [11C]KD2 to CB2 (Fig. 2). The free fraction fu of [11C]KD2 in human plasma was 0.001 and 0.017 in 4% BSA. The high plasma protein binding was confirmed in an equilibrium dialysis experiment (data not shown).
Permeation of [11C]KD2 across hMDR1-MDCK cells
The permeation of [11C]KD2 across P-gp-transfected Madin Darby canine kidney cells (MDCK) cells was investigated to evaluate the ability of the tracer to cross the blood–brain barrier (BBB) and to determine whether the compound is a substrate of the efflux transporter P-gp. Permeation of the P-gp substrate [3H]vinblastine, that was used as a control, was sixfold higher from basal to apical (corresponding to efflux transport in vivo) than from apical to basal (corresponding to blood to brain direction in vivo; Fig. 3). The respective ratio for [11C]KD2 was < 1, excluding net efflux transport by P-gp. Permeation coefficient of [11C]KD2 from apical to basal was (1.24 ± 0.18) × 10−6 cm/s, 3.6-fold higher than apical-to-basal permeation of [3H]vinblastine (no BBB permeation in vivo) and 1.7-fold lower than basal-to-apical permeation of [3H]vinblastine (Fig. 3). For comparison, apical to basal permeation of FDEGPECO, an mGluR5 PET probe with good in vivo BBB permeability, was 3.7-fold higher than that of [3H]vinblastine under the same experimental conditions (Wanger-Baumann et al. 2011).
In vivo characterization of [11C]KD2
Figure 4 shows PET coronal sections of consecutive time frames of the spleen and liver region of a rat, fused to the respective CT. [11C]KD2 accumulated in spleen, liver, and intestines. The high spleen uptake is in accordance with the expression pattern of CB2 while the high accumulation in liver suggests a hepatobiliary elimination pathway. Background radioactivity in muscle tissue was relatively low. The TACs of liver, spleen, and peripheral tissue outside the rips are shown in Fig. 5a. To confirm that accumulation in spleen resulted from specific binding to CB2, a second scan was performed where the selective CB2 agonist GW405833 was injected in excess (1.5 mg/kg) into a tail vein from 15 to 16 min after tracer injection. The resulting TACs are shown in Fig. 5b. Radioactivity clearance from spleen was accelerated after injection of GW405833, indicating competitive displacement of KD2 from the binding to CB2, confirming the specificity of the tracer in vivo.
Distribution of the tracer in rat brain was evaluated by PET. Region of interest TAC analysis with two rats showed reproducible uptake of radioactivity in all investigated brain regions including cortex, cerebellum, caudate putamen, pons, and hippocampus, regions with known CB2 expression in healthy rat brain (Sugiura and Waku 2002; Gong et al. 2006; Onaivi 2009; Rossi et al. 2010). Maximal SUVs were reached between 8 and 20 min after tracer injection (Fig. 6). Except for the first time frame, when radioactivity is mainly in blood, peripheral radioactivity levels were higher than levels in brain (data not shown). Note that spillover from periphery to ROIs of cortex, cerebellum and pons may have resulted in an overestimation of the corresponding SUVs. Results from two blocking experiments with GW4058233 (1.5 mg/kg, i.v.) to show whether brain uptake was CB2-related were not conclusive (data not shown).
Pilot experiment to evaluate CB2 expression in spinal cords of ALS
In recent years, CB2 gained interest for diagnosis and new therapy approaches in ALS. Both would benefit from a highly specific CB2 PET tracer. We investigated in a first pilot experiment with human post-mortem ALS spinal cord slices whether [11C]KD2 shows specific binding to this tissue. Figure 7 confirms not only the presence of CB2 but also specific binding of [11C]KD2. These are encouraging results towards non-invasive imaging of CB2 in ALS patients.
We have recently identified from the literature a 4-oxoquinoline derivative (designated KD2), as our lead structure for proof of concept studies. KD2 belongs to a new type of structure cluster compared to the recently published CB2 ligands such as 2-oxo-quinoline derivatives and N-arylamide oxadiazole derivatives (Ruhl et al. 2012; Turkman et al. 2012; Ahmad et al. 2013). Among them, [11C]NE40 was considered as the most promising CB2 radioligand for further evaluation, our new ligand KD2 exhibits more than five times higher binding affinity and 10 times higher selectivity towards CB2 than NE40 (KD2: Ki = 1.7 nM for CB2 and > 10 000 for CB1; NE40: Ki = 9.6 nM for CB2 and > 1000 for CB1). In vitro/in vivo characterization of [11C]KD2 revealed acceptable albeit not ideal properties as a PET tracer. The affinity and selectivity for CB2 are ideal for CB2 imaging. In vitro autoradiography with slices from rat and mouse spleen, which contain high levels of CB2, as well as in vivo PET of the rat spleen, demonstrated high specific and selective reversible binding.
However, the relatively high plasma protein binding with a fu of 0.001 may not be favorable for brain imaging as plasma protein binding competes with BBB passage and may, therefore, reduce brain uptake of the tracer (Dishino et al. 1983). The high plasma protein binding is probably a consequence of the high lipophilicity of KD2. The quinolin-4(1H)-one structure is presumably uncharged at physiological pH (software ACD/pKa v11), resulting in the relatively high logD value at pH 7.4 of 3.29 (equal to logP). In addition, in vitro transport experiments predicted several fold lower in vivo BBB permeation than observed for our recently published successful mGluR5 brain PET probe FDEGPECO (Wanger-Baumann et al. 2011). Based on these in vitro results, we expected moderate-to-low brain uptake. Efflux of [11C]KD2 by (human) P-gp, which would further reduce BBB passage, was excluded in our in vitro transport assay.
As expected, distribution to brain was relatively low in the in vivo PET experiments with rats. Radioactivity levels in brain were lower than in peripheral tissues in general. Nevertheless, minute accumulation was observed in all investigated brain regions as concluded from the increase in radioactivity after the initial peak during the first time frame. The latter is assigned to brain perfusion with high initial tracer levels in blood. The general but weak uptake is in agreement with the ubiquitous but low CB2 expression in brain (Van Sickle et al. 2005; Gong et al. 2006). The higher uptake in cortex, cerebellum, and pons as compared to caudate putamen and hippocampus may result from spillover from the periphery. Future experiments will show whether [11C]KD2 accumulation is significantly increased under conditions where brain CB2 levels are elevated. Uptake of a 11C-labeled thiazol derivative was approximately doubled in several brain regions in a lipopolysaccharide (LPS)-induced mouse model of neuroinflammation (Horti et al. 2010).
The specific binding of [11C]KD2 to spinal cord tissue from ALS patients confirmed the presence of CB2 in this tissue as reported (Yiangou et al. 2006). The significant difference between the bound radioactivity under baseline and blocking conditions looks promising for non-invasive imaging with a suitable PET probe. Further studies, in particular with tissues from control patients without ALS are now required to evaluate whether the high tracer binding is unique to tissue from ALS patients or whether CB2 is also present in spinal cord tissue under normal conditions.
In conclusion, as a lead structure for CB2 PET imaging, [11C]KD2 showed promising in vivo characteristics with a good potential for improvement, in particular with respect to lipophilicity and plasma protein binding. In vitro autoradiography with spinal cord tissue from ALS patients showed specific tracer accumulation, an encouraging finding towards the non-invasive imaging of spinal cord CB2 levels.
This study was partially funded by the Swiss ALS Foundation. We thank Mr. Bruno Mancosu for his support with carbon-11 radiolabeling and Ms. Claudia Keller for performing the PET/CT scans. The authors have no conflict of interest to declare.