Development of a Radiofluorinated Adenosine A2B Receptor Antagonist as Potential Ligand for PET Imaging

The adenosine A2B receptor has been proposed as a novel therapeutic target in cancer, as its expression is drastically elevated in several tumors and cancer cells. Noninvasive molecular imaging via positron emission tomography (PET) would allow the in vivo quantification of this receptor in pathological processes and most likely enable the identification and clinical monitoring of respective cancer therapies. On the basis of a bicyclic pyridopyrimidine-2,4-dione core structure, the new adenosine A2B receptor ligand 9 was synthesized, containing a 2-fluoropyridine moiety suitable for labeling with the short-lived PET radionuclide fluorine-18. Compound 9 showed a high binding affinity for the human A2B receptor (Ki(A2B) = 2.51 nM), along with high selectivities versus the A1, A2A, and A3 receptor subtypes. Therefore, it was radiofluorinated via nucleophilic aromatic substitution of the corresponding nitro precursor using [18F]F-/K2.2.2./K2CO3 in DMSO at 120 °C. Metabolic studies of [18F]9 in mice revealed about 60% of radiotracer intact in plasma at 30 minutes p.i. A preliminary PET study in healthy mice showed an overall biodistribution of [18F]9, corresponding to the known ubiquitous but low expression of the A2B receptor. Consequently, [18F]9 represents a novel PET radiotracer with high affinity and selectivity toward the adenosine A2B receptor and a suitable in vivo profile. Subsequent studies are envisaged to investigate the applicability of [18F]9 to detect alterations in the receptor density in certain cancer-related disease models.


Introduction
Adenosine receptors belong to the family of G-protein-coupled receptors and are activated by the purine nucleoside adenosine, a precursor and degradation product of adenine nucleotides such as adenosine 5 -triphosphate (ATP). The effects of adenosine are mediated via the four receptor subtypes A 1 , A 2A , A 2B , and A 3 . Each of these exhibits distinct pharmacological properties, cell and tissue distribution, and intracellular signaling [1][2][3]. The adenosine A 2B receptor can be seen as unique as its activation requires high levels of adenosine with concentrations in the micromolar range [3][4][5],  [14], the selected new bicyclic lead compound 2 [16], and the newly developed A2B receptor PET ligand [ 18 F]9.

Organic Synthesis and In Vitro Binding Studies
In order to investigate the influence of the substitution of the pyrimidine ring in lead compound 2 by the 2-fluoropyridine ring in target compound 9 on the affinity and selectivity toward the A2B receptor, 2 was synthesized as a reference compound. This synthesis was performed according to the procedures described by Eastwood et al. [16,17]. For the synthesis of the new A2B receptor ligand 9, another strategy was chosen, which is shown in Scheme 1. Briefly, the three-fold substituted pyridine core 5 was built up via a two-step synthesis sequence starting from 2-acetyl furane, which was treated with N,N-dimethylformamide dimethyl acetal to obtain compound 3 [18]. Further reaction of 3 with the hydrochloride salt 4 and ammonium acetate gave 5, which was selectively brominated to obtain compound 6. Derivatives 3-6 have been previously described and the reported synthesis procedures [18][19][20] were adopted with slight modifications (see Supplementary material). In the next step, the cyano group was hydrolyzed using hydrogen peroxide under basic conditions to obtain the amide 7, which was then cyclized to 8 by deprotonation with sodium hydride and subsequent reaction with N,N-carbonyldiimidazole. The obtained pyridopyrimidine-2,4-dione 8 was used for a Suzuki coupling reaction [21,22] with 2-fluoropyridine-4-boronic acid to achieve the desired compound 9. The bicyclic intermediate 8 was also used for the synthesis of the precursor compound 10. The required 2-nitropyridine-4-boronic acid pinacol ester was synthesized via a Miyaura borylation [23] of the corresponding bromo-substituted nitropyridine and subsequently coupled in situ with 8 in a Suzuki reaction, affording the nitro precursor 10.  [14], the selected new bicyclic lead compound 2 [16], and the newly developed A 2B receptor PET ligand [ 18 F]9.

Organic Synthesis and In Vitro Binding Studies
In order to investigate the influence of the substitution of the pyrimidine ring in lead compound 2 by the 2-fluoropyridine ring in target compound 9 on the affinity and selectivity toward the A 2B receptor, 2 was synthesized as a reference compound. This synthesis was performed according to the procedures described by Eastwood et al. [16,17]. For the synthesis of the new A 2B receptor ligand 9, another strategy was chosen, which is shown in Scheme 1. Briefly, the three-fold substituted pyridine core 5 was built up via a two-step synthesis sequence starting from 2-acetyl furane, which was treated with N,N-dimethylformamide dimethyl acetal to obtain compound 3 [18]. Further reaction of 3 with the hydrochloride salt 4 and ammonium acetate gave 5, which was selectively brominated to obtain compound 6. Derivatives 3-6 have been previously described and the reported synthesis procedures [18][19][20] were adopted with slight modifications (see Supplementary material). In the next step, the cyano group was hydrolyzed using hydrogen peroxide under basic conditions to obtain the amide 7, which was then cyclized to 8 by deprotonation with sodium hydride and subsequent reaction with N,N-carbonyldiimidazole. The obtained pyridopyrimidine-2,4-dione 8 was used for a Suzuki coupling reaction [21,22] with 2-fluoro pyridine-4-boronic acid to achieve the desired compound 9. The bicyclic intermediate 8 was also used for the synthesis of the precursor compound 10. The required 2-nitropyridine-4-boronic acid pinacol ester was synthesized via a Miyaura borylation [23] of the corresponding bromo-substituted nitropyridine and subsequently coupled in situ with 8 in a Suzuki reaction, affording the nitro precursor 10.
The binding affinities of 2 and 9 at the four human adenosine receptor subtypes A 2B , A 2A , A 1 , and A 3 , presented in Table 1, were determined in competitive binding assays using appropriate radioligands and membrane preparations of Chinese hamster ovary (CHO) cells or human embryonic kidney (HEK) cells, respectively, recombinantly expressing the corresponding human receptor subtype. Scheme 1. Synthesis of the new A2B receptor ligand 9 and its precursor for radiofluorination 10.  [18][19][20] and was performed accordingly with slight modifications, as described in the Supplementary material.
The binding affinities of 2 and 9 at the four human adenosine receptor subtypes A2B, A2A, A1, and A3, presented in Table 1, were determined in competitive binding assays using appropriate radioligands and membrane preparations of Chinese hamster ovary (CHO) cells or human embryonic kidney (HEK) cells, respectively, recombinantly expressing the corresponding human receptor subtype.
The in-house-determined binding data of compound 2 corresponded to the values reported by Eastwood et al. [16] and thus confirmed its high affinity and selectivity for the A2B receptor. The substitution of the pyrimidine ring in 2 by a 2-fluoropyridine ring in the new derivative 9 did not affect the binding affinity for the A2B receptor, as indicated by the Ki value of 2.51 nM obtained for both compounds. Accordingly, compound 9 possesses a two-fold higher binding affinity for the A2B receptor in comparison to our recently developed ligand 1. Moreover, the selectivity of 9 regarding the two adenosine receptor subtypes A1 and A2A was considerably higher than that of 1 (Ki ratios of 43 and 59 versus 13 and 4.5, resp.; see Table 1). Due to a slight increase in the binding affinity of 9 for the A3 receptor, the selectivity regarding this off-target was slightly decreased but still in a suitable range for the purpose of specific imaging (Ki ratio of 114).  [18][19][20] and was performed accordingly with slight modifications, as described in the Supplementary material.
The in-house-determined binding data of compound 2 corresponded to the values reported by Eastwood et al. [16] and thus confirmed its high affinity and selectivity for the A 2B receptor. The substitution of the pyrimidine ring in 2 by a 2-fluoropyridine ring in the new derivative 9 did not affect the binding affinity for the A 2B receptor, as indicated by the K i value of 2.51 nM obtained for both compounds. Accordingly, compound 9 possesses a two-fold higher binding affinity for the A 2B receptor in comparison to our recently developed ligand 1. Moreover, the selectivity of 9 regarding the two adenosine receptor subtypes A 1 and A 2A was considerably higher than that of 1 (K i ratios of 43 and 59 versus 13 and 4.5, resp.; see Table 1). Due to a slight increase in the binding affinity of 9 for the A 3 receptor, the selectivity regarding this off-target was slightly decreased but still in a suitable range for the purpose of specific imaging (K i ratio of 114). Table 1. K i values of 1, lead compound 2, and the new fluorinated derivative 9 at the four human adenosine receptor subtypes along with the corresponding selectivity ratios.

Compound
K i in nM a Selectivity Ratio

Automated Radiosynthesis and Characterization of [ 18 F]9
The radiolabeling of [ 18 F]9 was performed via a nucleophilic aromatic substitution reaction of the nitro group of precursor 10 and [ 18 F]fluoride (Scheme 2), which was azeotropically dried using the kryptofix/potassium carbonate system and acetonitrile (ACN). Based on our former experiments with compound [ 18 F]1 [14], we selected dimethyl sulfoxide (DMSO) as the most promising solvent. All radiolabeling experiments were performed with a very low amount of precursor (about 0.5 mg).

Automated Radiosynthesis and Characterization of [ 18 F]9
The radiolabeling of [ 18 F]9 was performed via a nucleophilic aromatic substitution reaction of the nitro group of precursor 10 and [ 18 F]fluoride (Scheme 2), which was azeotropically dried using the kryptofix/potassium carbonate system and acetonitrile (ACN). Based on our former experiments with compound [ 18 F]1 [14], we selected dimethyl sulfoxide (DMSO) as the most promising solvent. All radiolabeling experiments were performed with a very low amount of precursor (about 0.5 mg). The complete radiosynthesis of [ 18 F]9 was accomplished in a remotely controlled synthesis module (TRACERlab FX2 N from GE Healthcare). The synthesizer setup is described in detail in the Materials and Methods. Briefly, [ 18 F]fluoride was trapped on an anion exchange cartridge, eluted with aqueous potassium carbonate, and then azeotropically dried to obtain the [ 18 F]F − /K2.2.2./K2CO3 complex. Radiofluorination of the precursor 10 was performed in DMSO for 15 min at 120 °C. In order to isolate [ 18 F]9, the crude reaction mixture was diluted with aqueous methanol and applied to a semi-preparative HPLC system (chromatogram in Figure S1 in Supplementary Material). The radiotracer fraction was collected at retention times in the range of 22-25 min and purified by solidphase extraction using a C18 cartridge. Afterwards, [ 18 F]9 was eluted from the cartridge and transferred out of the hot cell for concentration and formulation in sterile isotonic saline containing 10% ethanol. The entire process lasted about 80 min. With this procedure, [ 18 F]9 could be reproducibly generated with a high radiochemical purity of ≥99%, reasonable radiochemical yields of 48% ± 4% (n = 4) and molar activities of 10-32 GBq/µmol (n = 4) at starting activities of 2-7 GBq. The identity of [ 18 F]9 was confirmed by analytical radio-HPLC ( Figure S2 in Supplementary Material) with coinjection of the corresponding non-radioactive reference compound 9.
For estimation of the lipophilicity of [ 18 F]9, the shake-flask method with n-octanol and phosphate-buffered saline as a partition system was used, resulting in a logD value of 1.4 ± 0.1 (n = 4).

In Vivo Studies of [ 18 F]9 in Mice
The metabolism of [ 18 F]9 in vivo was investigated in CD-1 mice by taking blood and urine samples at 30 min post i.v. injection of the radiotracer (n = 2). Analysis of the respective samples was performed by micellar (MLC) and reversed-phase chromatography (RP-HPLC) as two The complete radiosynthesis of [ 18 F]9 was accomplished in a remotely controlled synthesis module (TRACERlab FX2 N from GE Healthcare). The synthesizer setup is described in detail in the Materials and Methods. Briefly, [ 18 F]fluoride was trapped on an anion exchange cartridge, eluted with aqueous potassium carbonate, and then azeotropically dried to obtain the [ 18 Radiofluorination of the precursor 10 was performed in DMSO for 15 min at 120 • C. In order to isolate [ 18 F]9, the crude reaction mixture was diluted with aqueous methanol and applied to a semi-preparative HPLC system (chromatogram in Figure S1 in Supplementary Material). The radiotracer fraction was collected at retention times in the range of 22-25 min and purified by solid-phase extraction using a C18 cartridge. Afterwards, [ 18 F]9 was eluted from the cartridge and transferred out of the hot cell for concentration and formulation in sterile isotonic saline containing 10% ethanol. The entire process lasted about 80 min. With this procedure, [ 18 F]9 could be reproducibly generated with a high radiochemical purity of ≥99%, reasonable radiochemical yields of 48% ± 4% (n = 4) and molar activities of 10-32 GBq/µmol (n = 4) at starting activities of 2-7 GBq. The identity of [ 18 F]9 was confirmed by analytical radio-HPLC ( Figure S2 in Supplementary Material) with co-injection of the corresponding non-radioactive reference compound 9.
For estimation of the lipophilicity of [ 18 F]9, the shake-flask method with n-octanol and phosphate-buffered saline as a partition system was used, resulting in a logD value of 1.4 ± 0.1 (n = 4).

In Vivo Studies of [ 18 F]9 in Mice
The metabolism of [ 18 F]9 in vivo was investigated in CD-1 mice by taking blood and urine samples at 30 min post i.v. injection of the radiotracer (n = 2). Analysis of the respective samples was performed by micellar (MLC) and reversed-phase chromatography (RP-HPLC) as two complementary HPLC methods. MLC is also a reversed-phase HPLC mode, but it uses a mobile phase consisting of an aqueous solution with a surfactant above its critical micellar concentration. The formed micelles are able to dissolve the proteins and other components of a biological sample and therefore enable a direct injection of the plasma probe into the HPLC system without eliminating the tissue matrix [24]. In contrast, for RP-HPLC analysis, aliquots of the plasma samples were treated with a mixture of ACN/water to precipitate the proteins, resulting in recoveries of activity of about 93%.
In general, the results obtained with both methods were comparable. Figure 2 shows exemplary RP-HPLC and MLC chromatograms of a plasma and urine sample at 30 min p.i. Accordingly, intact radiotracer accounted for~60% of the total activity in plasma, which was a considerably higher fraction of intact radiotracer in comparison to our former radiotracer [ 18 F]1 (35% at 30 min p.i.) [14]. The main radiometabolite was slightly more polar than [ 18 F] 9 and was also found in the urine, indicating, along with a very high activity concentration (standardized uptake value (SUV) = 75), a renal excretion route for [ 18 F]9. Besides this main radiometabolite, a small fraction of more polar radioactive metabolites was observed in urine.
When comparing the chromatograms of the two HPLC methods, slight differences in the retention profile became evident, a phenomenon that is caused by the differences in the separation mechanisms of the two chromatographic systems. Moreover, the RP-HPLC chromatograms showed a lower fraction of a very polar radiometabolite (within the first 5 min retention time) compared to the MLC ones. Considering the plasma sample, this influenced the peak quantification and therefore resulted in a supposed slightly higher percentage of intact radiotracer using the RP-HPLC method. However, it is likely that this polar component could not be quantitatively extracted during the plasma work-up procedure used to generate the injection sample for RP-HPLC, as reflected by the non-quantitative recovery of 93%. Considering this small loss resulted in a calculated value of 60% of intact tracer for RP-HPLC, which was comparable to the MLC result. In a separate in vitro experiment, the quantitative recovery of the radiotracer could be shown using the same work-up procedure.
RP-HPLC and MLC chromatograms of a plasma and urine sample at 30 min p.i. Accordingly, intact radiotracer accounted for ~60% of the total activity in plasma, which was a considerably higher fraction of intact radiotracer in comparison to our former radiotracer [ 18 F]1 (35% at 30 min p.i.) [14]. The main radiometabolite was slightly more polar than [ 18 F]9 and was also found in the urine, indicating, along with a very high activity concentration (standardized uptake value (SUV) = 75), a renal excretion route for [ 18 F]9. Besides this main radiometabolite, a small fraction of more polar radioactive metabolites was observed in urine.
When comparing the chromatograms of the two HPLC methods, slight differences in the retention profile became evident, a phenomenon that is caused by the differences in the separation mechanisms of the two chromatographic systems. Moreover, the RP-HPLC chromatograms showed a lower fraction of a very polar radiometabolite (within the first 5 min retention time) compared to the MLC ones. Considering the plasma sample, this influenced the peak quantification and therefore resulted in a supposed slightly higher percentage of intact radiotracer using the RP-HPLC method. However, it is likely that this polar component could not be quantitatively extracted during the plasma work-up procedure used to generate the injection sample for RP-HPLC, as reflected by the non-quantitative recovery of 93%. Considering this small loss resulted in a calculated value of 60% of intact tracer for RP-HPLC, which was comparable to the MLC result. In a separate in vitro experiment, the quantitative recovery of the radiotracer could be shown using the same work-up procedure. We then performed a pilot study to estimate very basic characteristics of the pharmacokinetics of the new radiotracer via dynamic small-animal PET imaging with [ 18 F]9 in female CD-1 mice. The We then performed a pilot study to estimate very basic characteristics of the pharmacokinetics of the new radiotracer via dynamic small-animal PET imaging with [ 18 F]9 in female CD-1 mice. The time-activity curves (TACs) obtained in different tissues under both baseline (administration of vehicle, n = 2) and blocking (administration of compound 2, n = 2) conditions are presented in Figure 3; the whole-body maximum intensity projections are shown in Figure S3 in the Supplementary Material. The overall biodistribution of [ 18 F]9 under baseline conditions corresponded to the known quite ubiquitous but physiologically not very dense expression of A 2B receptors [25]. Although higher levels of the A 2B receptor protein have been reported in the literature for certain cell types, such as alveolar epithelial cells and endothelial cells [3,25], the resemblance of the TACs in peripheral organs such as lungs under baseline and blocking conditions indicated no target-specific retention of [ 18 F]9 (TAC peak SUV lung : 3.09 and 2.88 vs. 2.30 and 1.57, resp.; SUV 60 min p.i., lung : 0.44 and 0.25 vs. 0.32 and 0.20, resp.). This was slightly different to the results of Petroni et al. obtained for the carbon-11-labeled compound [ 11 C]-4 in rats [13], because those authors described a significant decrease of uptake of [ 11 C]-4 in the lung induced by the pretreatment of the A 2B receptor agonist BAY60-6583. This discrepancy between our and Petroni's results might indicate differences in the biological effects exerted by agonistic (BAY60-6583) or antagonistic (compound 2) A 2B receptor ligands, as well as model-related differences such as the expression of A 2B receptors. The latter option could also explain the about 10-fold lower uptake of [ 18 F]9 in the brown adipose tissue (BAT) of female CD-1 mice in comparison to the uptake of [ 11 C]-4 in the BAT of male Wistar rats [13].
The time-activity distribution in the kidney, along with the concentration of activity determined in the urine during the metabolism study, indicated a renal excretion route of the radiotracer. In addition, the remarkably high and constantly increasing uptake determined in the gall bladder under blocking conditions indicated accumulation of [ 18 F]9 in the bile, promoted by compound 2. A low tracer accumulation was detected in the brain, with an initial TAC peak SUV of 0.19 (0.20 and 0.17) and subsequent slow washout to an SUV of 0.11 (0.14 and 0.08) after 60 min p.i. Co-administration of compound 2 resulted in a slight reduction of the brain uptake, with a TAC peak SUV of 0.12 (0.11 and 0.13). Although this study was performed in healthy animals, thus in an animal model not characterized by disease-related high expressions of A2B receptors, these initial results are promising regarding a low non-specific binding of [ 18 F]9 under physiological conditions. Accordingly, [ 18 F]9 seems to be applicable for imaging of A2B receptors in future studies in disease models, e.g., of inflammation or cancer. However, further studies are required in order to answer open questions such as the unexpected increase of activity in the gall bladder after pretreatment with antagonist 2. Although this study was performed in healthy animals, thus in an animal model not characterized by disease-related high expressions of A 2B receptors, these initial results are promising regarding a low non-specific binding of [ 18 F]9 under physiological conditions. Accordingly, [ 18 F]9 seems to be applicable for imaging of A 2B receptors in future studies in disease models, e.g., of inflammation or cancer. However, further studies are required in order to answer open questions such as the unexpected increase of activity in the gall bladder after pretreatment with antagonist 2.

General
Analysis of all compounds was performed via mass spectroscopy (MS), thin-layer chromatography (TLC), and NMR spectroscopy. The final compounds 2, 9, and 10 were also subjected to HPLC analysis.
For purification of final products, flash-column chromatography was used with silica gel ZEOsorb 60/40-63 µm from Apollo Scientific Ltd., Cheshire, UK and silica gel 40-63 µm from VWR International GmbH, Darmstadt, Germany.
The chemical purity of the final compounds was ≥ 96% and was controlled by HPLC using a 150 × 3 mm Reprosil-Pur Basic HD 3 µm column (Dr. Maisch HPLC GmbH, Ammerbuch-Entingen, Germany). These analytical chromatographic separations were performed on a Dionex Ultimate 3000 system, incorporating a LPG-3400SD pump, an autosampler WPS-3000 TSL, a column compartment TCC-3000SD, a diode array detector DAD3000 (monitoring from 254-720 nm), and a low-resolution mass spectrometer MSQ 3000 (Thermo Fisher Scientific, Darmstadt, Germany). A mixture of acetonitrile (ACN) and aqueous 20 mM NH 4 OAc was used as eluent in a linear gradient system (0-2.5 min at 25% ACN, 2.

Compounds 3-6
These compounds were synthesized as described in the Supplementary Material.

2-Amino-5-bromo-6-(furan-2-yl)nicotinamide (7)
A suspension of compound 6 (0.74 g, 2.8 mmol) in 22 mL ethanol, 15 mL water, 9.3 mL 6M aqueous NaOH and 0.8 mL H 2 O 2 (50% in H 2 O, 14.0 mmol) was stirred for 5 h at 50 • C. Afterwards, 200 mL water was added and the mixture was extracted with ethyl acetate (4 × 25 mL). The combined organic phases were dried over sodium sulfate, filtered, and the solvent was removed under reduced pressure. The obtained residue was purified by column chromatography (SiO 2 , ethyl acetate/n-hexane, 2/1, v/v) to obtain 7 with a yield of 45% (0.36 g, 1.26 mmol).  (8) To a solution of 7 (100 mg, 0.35 mmol) in 2 mL dimethyl sulfoxide (DMSO), sodium hydride (35.4 mg, 0.88 mmol, 60% in oil) was added and the mixture was stirred for 30 min at room temperature. Afterwards, N,N-carbonyldiimidazole (CDI, 60 mg, 0.39 mmol) dissolved in 1 mL N,N-dimethylformamide (DMF) was added and the reaction mixture heated at 90 • C for two hours. After cooling to room temperature, the solution was treated with 100 mL aqueous 1M HCl and extracted with ethyl acetate (4 × 25 mL). The combined organic phases were dried over sodium sulfate, filtered, and the solvent was removed under reduced pressure. The obtained residue was purified by column chromatography (SiO 2 , CH 2 Cl 2 /CH 3 OH, 39/1, v/v) to obtain 8 with a yield of 50% (55 mg, 0.18 mmol). The mixture was evacuated three times under reduced pressure and refilled with argon to remove air. Subsequently, [Pd(dppf)Cl 2 ] (12 mg, 0.016 mmol) was added under argon and the resulting suspension was stirred at 90 • C for two hours. After cooling to room temperature, the solution was treated with 100 mL aqueous 1 M HCl and extracted with ethyl acetate (4 × 25 mL). The combined organic phases were dried over sodium sulfate, filtered, and the solvent was removed under reduced pressure. The obtained residue was pre-purified by column chromatography (SiO 2 , CH 2 Cl 2 /CH 3 OH/ethyl acetate, 99/1/1, v/v/v) to obtain 31 mg (0.06 mmol, 59%) of crude 9, which was then purified by semi-preparative HPLC (Reprosil-Pur Basic C18-HD, 250 × 20 mm, 35% ACN/aq. 20 mM NH 4 OAc, flow 7.5 mL/min, t R~1  were suspended in 2 mL 1,4-dioxane under an inert atmosphere and the reaction mixture was heated to 90 • C. After an hour, the reaction mixture was cooled to room temperature and compound 8 (307 mg, 1.0 mmol), [Pd(dppf)Cl 2 ] (37 mg, 0.05 mmol), and Cs 2 CO 3 (3.25 g, 10.0 mmol) were added under inert atmosphere and heated for an hour at 90 • C. After cooling to room temperature, the suspension was filtered over celite, the solvent was removed under reducer pressure, and the residue was purified by column chromatography (SiO 2 , CH 2 Cl 2 /CH 3 OH, 99.5/0.5 to 98/2, v/v) to give 10 with a yield of 9% (31 mg, 0.088 mmol).

In Vitro Radioligand Binding Experiments
Membrane preparations of recombinant CHO or HEK (human embryonic kidney) cells expressing the respective human adenosine receptor subtype were obtained according to Borrman et al. [26] or purchased from Perkin Elmer, Germany. Radioligand binding assays using human A 1 , A 2A , A 2B and A 3 receptors were performed according to Alnouri et al. [27].
The following radioligands were employed:  Figure 4, entry 1, Waters GmbH, Eschborn, Germany) and eluted into the reactor with 400 µL of an aqueous solution of potassium carbonate (K 2 CO 3 , 1.8 mg, 13 µmol, entry 2). After addition of Kryptofix 2.2.2. in 1.5 mL ACN (11 mg, 29 µmol, entry 3), the mixture was azeotropically dried for approximately 10 min. Thereafter, 0.5-0.6 mg of the nitro precursor (10) dissolved in 800 µL DMSO (entry 4) was added, and the reaction mixture was stirred at 120 • C for 15 min. After cooling, the reaction mixture was diluted with 1.5 mL H 2 O and 2.0 mL MeOH/H 2 O 1/1 (v/v, entry 5) and transferred into the injection vial (entry 6). Semi-preparative HPLC was performed using a Reprosil-Pur 120 AQ column (250 × 20 mm; 10 µm; Dr. Maisch HPLC GmbH; Ammerbuch-Entingen, Germany) with a solvent composition of 60% MeOH/aq. 20 mM NH 4 OAc at a flow rate of 8.0 mL/min (entry 7). [ 18 F]9 was collected in the dilution vessel (entry 8) previously loaded with 40 mL H 2 O. Final purification was performed by passing the solution through a Sep-Pak ® C18 light cartridge (entry 9), followed by washing with 2 mL water (entry 10) and elution of [ 18 F]9 with 1.3 mL EtOH (entry 11) into the product vial (entry 12). The ethanolic solution was transferred out of the hot cell and the solvent was reduced under a gentle argon stream at 70 • C to a final volume of 10-50 µL. Afterwards, the radiotracer was diluted in isotonic saline to obtain a final product containing 10% of EtOH (v/v).
The molar activity was determined on the basis of a calibration curve created under isocratic HPLC conditions (22% ACN/aq. 20 mM NH 4 OAc; Reprosil-Pur 120 AQ 250 × 4.6 mm, flow 1.0 mL/min), using chromatograms obtained at 254 nM as an appropriate maximum of UV absorbance.  The partition coefficient of [ 18 F]9 was experimentally determined for the n-octanol/PBS system by the shake-flask method as described previously [14].

In Vivo Studies in Mice
For metabolism studies (n = 2), [ 18 F]9 (30-40 MBq) was administered as a bolus in awake female CD-1 mice (10-12 weeks, 25-32 g) via the tail vein. At 30 min p.i., the animals were slightly anesthetized with isoflurane, and blood was sampled from retro-orbital plexus. Blood plasma was obtained as a supernatant after centrifugation of the whole blood samples (14,000 rpm, 1 min; Centrifuge 5418, Eppendorf Vertrieb Deutschland GmbH, Wesseling-Berzdorf, Germany). After cervical luxation of the animals, urine samples were obtained. All samples were weighed and the respective activity measured in a dose calibrator (ISOMED 2010; MED Nuklear-Medizintechnik Dresden GmbH, Dresden, Germany). The partition coefficient of [ 18 F]9 was experimentally determined for the n-octanol/PBS system by the shake-flask method as described previously [14].

In Vivo Studies in Mice
For metabolism studies (n = 2), [ 18 F]9 (30-40 MBq) was administered as a bolus in awake female CD-1 mice (10-12 weeks, 25-32 g) via the tail vein. At 30 min p.i., the animals were slightly anesthetized with isoflurane, and blood was sampled from retro-orbital plexus. Blood plasma was obtained as a supernatant after centrifugation of the whole blood samples (14,000 rpm, 1 min; Centrifuge 5418,